Laser Kids
RUS

. : LASER 3.0 (Saga of Carborundum) : .

 

First time I knew about the inconvinient properties of the carborundum from Jon Singer web reports, where he tried to use this substance to make a preionizer for a nitrogen laser. However the success was rather poor,
and it is understandable that no will for attempts to reproduce this scheme was arised. Moreover i already had a negative experience with attempts to use a distributed resistance (in form of Bealieau schematics or capacitive discharge). While the energy is low - the beautifull glow, and when You try to reach laser level of energy deposition, the discharge contracts and nothing helps.

Then a canadian article [1] appeared before my eyes. They praised silicon carbide and a preionizer made of it. Their preionizer worked not worse than a multi-spark one, but was much simplier and easy to build. However they used a helium rich mixture (He:N2:CO2=70:15:15) and transition to air substituted ones looked at least doubtfull.

At the third time silicon carbide hit my nose, when i tried to find in literature some light on the process of huge energy dissipation in nitrogen lasers. As usual: when You are looking for something, You find the very other. The French article [2] reports that they succeeded to make working laser not only using a heliumless mixture, but using a CO2:N2:O2=48:42:10, and that is pretty close to using the air. (Actually the air containing mixture differes from that one only by uncontrollable contaminations.)

Such a kick is hard to ignore so I had to give it a try. The silicon carbide was found in form of grinding discs for hand engraver tool (dremel tool). All that was needed, was to connect them to a little test circuit:

           SG
      +--->   <---+-----+---------+
      !           !     !         !
      !           !     !         !
      +----||-----+     \         !
      !    Cp           / Rl    XXXXX
      !                 \ 
     ---                /       XXXXX
     --- Cs             \         !
      !                 !         !
      !                 !         !
      +-----------------+---------+

SG - air filled spark gap. Cp - peaker capacitor Murata 590 pf x 40 kv,
Cs - storage capacitor Murata 2 nf x 40 kv. Rl - leakage resistor (connects the opposite to the storage capacitor plate of the peaker to the ground wire) 5.6 kiloohms,
The 'XXX' means silicon carbide electrodes.

The result impressed me (see photo). Such an even and bright rain of sparks I havent ever seen from any Bealieau circuit. And this sparky rain is accompanied with a violet fog of a diffusion discharge.

sic_test

For the comparison here are the photoes of a discharge with analogous grinding discs made of corund (Al2O3).

al2o3_test

Despite the fact that the material is porous, it does not want to disperse the spark. I.e. the reason of such a nice discharge with carborundum does not hide in the fact that the discharge goes through thin pores between abradant grains. The matter of fact is the conductivity of the carborundum itself.

The test had clearly shown that at least as a preionizer this thing is more than acceptable. And in better case it is good as a laser electrode. However the hand engraver tool grinding discs arent best fit to make laser electrodes. The search for the carborundum bricks in all nearby hardware stores brought nothing. I had to make some web serfing. Finally those bricks vere purchased:

brusok1 brusok2 brusok3

Despite that all of them (if trust the seller's price list) were made of carborundum, in the test circuit they behaved differently. The bricks of the first type (dark greenish grey, with large grains and pores size) give dispersed rain of sparks as it was with Dremel's discs. The bricks of the second and third types (greenish white, finely grit and nonporous looking) behave like dielectric. It means the spark goes way around and
there's nothing between the bricks.

The bricks of the first type are 100x10x3.5 mm sized. The first test laser was assembled with their use.

 

TEST 1

A laser tube was assembled. Its electrodes were formed by 100x10x3.5 mm carborundum bricks. Three bricks per each electrode. The bricks have rather foolish shape, so I slightly cut them to better fit the electrode keeper. It appears that the carborundum bricks can be very easily cut with any diamond tool. After the assemblage the active length of electrodes appeared to be 285 mm.

The gap between electrode keepers was set by a props 10 mm thick. As the result, the spacing between the laser electrodes became slightly less than 3 mm (were the bricks 3 mm thick, as it was said in the catalogue, it would be 4 mm, but in reality the bricks appeared to be thicker).

No preionizer was introduced at this step. It was interesting whether the silicon carbide could make all by itself.

The gas-discharge unit was put into a cut of plastic sewe pipe having diameter of 50 mm and length of 300 mm. The ends of the pipe were shut by perspex alignment mirror mounts. The back mirror was (as usual) the aluminium coated concave authomobile rear view mirror with its protective painting having been removed. The front mirror was dielectric coated ZnSe with 75% reflection. The mirror was noticeably contaminated by a gas torch flame in the vain attemts to make vacuum tight seal with a hot glue (glue-gun). The distance between mirrors is 350 mm.

The laser tube was installed onto a frame with capacitors and spark gap (that was left from a previous laser project). The resulting scheme is:

           SG
      +--->   <---+-----+---------+
      !           !     !         !
      !           !     !         !
      +----||-----+     )         !
      !    Cp           ) Ll    XXXXX
      !                 ) 
     ---                )       XXXXX
     --- Cs             )         !
      !                 !         !
      !                 !         !
      +-----------------+---------+

SG - air filled spark gap (the gap value is usually set to doubled value of the spacing between the laser electrodes - matched load mode)
Cp - peaker capacitor 4 pcs Murata 590pf x 40 kV in parallel,
Cs - storage capacitor, from 3 to 7 pcs Murata 2 nf x 40 kV in parallel.
L1 - leakage inductor (shortens the circuit of the peaker capacitor charging and limits the discharge time - 4 turns of 1 mm copper wire in PVC insulation wound on a 45 mm corpse).
"XXX" - means carborundum (laser electordes)

   //You should better get used to the ascii graphics. Since it is a report,
   not a guide, I'm not going to slave Your habits - all the schematics will
   be in the form that I usually do: no "gif" or "jpg" for simple schemes.
   "txt" only. Try to adjust Your browser for not to eat whitespaces, or 
   just open this in a text viewer/editor.//

The laser tube run at gap values (in the SG) from 6 to 11 mm, with the storage capacity from 6 nf to 14 nf on mixtures of CO2:air 2:1, 1:1 and 1:2. At pressures fom 500 mbar to 100 mbar. No lasing was observed.
Possible reasons:

  • high diffraction losses (CO2 laser radiation with its long wavelength poorly goes throuhg the narrow spacing between laser electrodes)
  • scattering on the discharge inhomogenities (The volumetic discharge is accompanied by more than 10 000 sparkies. Each of them is weak, but the total scattering may appear to be considerable. And the question of the possibility of lasing using multitude of sparks only is still dark. Bealieu himself had reported that his laser operated "on sparks" but nobody later reported the same. So it was possibly misunderstanding or bad translation or bas phrase.)
  • the resistance of the electrodes gives high discharge time constant, so the current prefers to go through L1. This is enstressed by the fact that L1 was specifically choosen for fast lasers (sqrt(CsL1)~1mcs).

L1 was replaced by a high voltage wirewound resistor 5.6 kiloOhm rated. The sparkies in the discharge became less weak more white. And at larger storage capacities (above 10 nF) the strength of sparks was enough to burn through th brick of carborundum. There appeared a burnout hole with black border. Happily was only in single place, not all over the surface. It was decided to add a preionizer. But before telling that results a pair of words on the other interesting observations.

 

It is curious that the discharge in this laser tends to contract not at high pressures, but on the contrary at low ones. Below 500 mbar I was not able to obtain the volumetic discharge between the electrodes. Below 300 mbar I was not able to get ANY kind of discharge between the electrodes - any discharge goes around them as if they were dielectric.

Another thing seeming unusual to me is a black color around the burnhole. One may suppose that it is carbon, but... the brick easily endures gas torch heating to about white glow. And when it is cold again it regains its colour and properties. So there is no organic glue in the brick's content. One may suupose that SiC decomposes itself, but see above... I doubt that the spark can heat it to comparable themperatures.

 

TEST 2

The laser tube was disassembled and a preionizer was added. The desighn of the preionizer resembles Lamberton-Pearson scheme - two wires pulled alongside the electrodes.

           SG
      +--->   <---+-----+---------+-----+-----+
      !           !     !         !     !     !
      !           !     !     Cpi---    !    ---Cpi
      +----||-----+     \        ---    !    ---
      !    Cp           / Rl      !   XXXXX   !
      !                 \         +---     ---+
     ---                /             XXXXX
     --- Cs             \               !
      !                 !               !
      !                 !               !
      +-----------------+---------------+

SG - air spark gap of the enclosed type, Cp - peaker capacitor 4 pcs of Murata 590 pf x 40 kv in parallel,
Cs - storage capacitor from 3 to 7 pcs of Murata 2 nf x 40 kv in parallel, Rl - 5.6 kOhm - leakage resistor (shortens the charge contour of the peaker capacitor but unlike the previous inductance almost does not affect the discharge time)
"XXX" as usual means carborundum, Cpi - preionizer capacitors, 50 pf each
KVI-2 type.

The preionizer wire has diameter of 0.5 mm. It is stretched along the electrodes and placed in height coordinate exactly in the middle between the electrode keepers (i.e. in 5 mm from the keeper plate surface). The distance in the horizontal direction from the axis of wire to the edge of the electrode is 2 mm. The wires are connected to the anode throug the capacitors goes not at both sides of laser (as one could suppose) but at the end corresponding to the output mirror (I was low on capacitors that moment).

Here I will again take an effort to remind You that this kind of preionizer coincides the Lamberton-Pearson's one only by exterrior and design. The logic and principles of operation are completely different. Here the preionizer uses thick wire and large capacity. And utilizes multi-spark discharge type, that becomes possible due to the electrodes are resistive. Considering the Lamberton's preionizer usually they say that utilizes corona discharge. And I could simply say that it is completely different thing. However it would be somewhat reduced thinking. The real corona is spatially limited discharge. The ionization stops where the field is low enough. Unlike this in the Lamberton-Pearson laser's working area there's no place where the field is low enough for the ionization to stop. Actually the discharge does cover all the path between the wire and the adjacent electrode. And since there are no means to force the discharge to take distributed form, it tends to become
a "single bright arc" whenever it possible. In a very special cases when the preionizer feedeing capacitors have low value and the wire is thin the discharge takes form of diffusion-streamer one. This one was studied in [3]
and they showed that the number of streamers increases with increasement of voltage rising rate, with lowering of the wire diameter and with decreasement of electron affinity of the gases. The single bright arc can not preionize all the volume properly, so it's crucial for the discharge to have diffusion
streamer form. In practice it leads to very thin wires (below 100 mcm), low feeding capacities, helium rich mixtures and low inductive power supply circuits.

On the contrary here we have multi-spark discharge from thick wire to the electrode with distributed resistance. The distribution of the sparks depends mainly to the resistance of this electrode. Thick wire has lower resistance and inductance. The preionizer feeding capacitors can have rather large value. All this allows to increase power and energy of the preionization, and moreover it allows to operate with voltage pulses having gradual front. However despite all those advantages this preionizer is way less effective than the one of barrier-discharge type used in laser with holed electrodes. On the other hand it is way simplier. And when the preionizerless discharge has the form very close to the needed one, it seems logical to begin with the most simple solution.

And again the laser tube run at gap values (in the SG) from 6 to 11 mm, with the storage capacity from 6 nf to 14 nf on mixtures of CO2:air 2:1, 1:1 and 1:2. At pressures fom 500 mbar to 100 mbar. No lasing was observed. The observable look of the discharhe became more smooth but it seems not to be enough.

 

TEST 3 (first light)

The laser tube was rearranged again. In order to increase spacing between the laser electrodes one 1mm thick perspex plate was added to each spacing prop. Hereby the spacing between the laser electrodes became 4 mm. The placement of the preionizer wires was kept unchanged. I.e. they are still at a height of 5 mm from the cathode keeper plate. The laser became asymmetric. The preionizer wires are now placed (in vertical direction coordinate) at 5-3.5=1.5 mm from the cathode cutoff, and at 6-3.5=2.5 mm from the anode cutoff.

The electrical circuit was kept unchanged. With Cs=6nf and with CO2:air=2:1 mixture at air pressure the lasing was observed. Almost accidentally. It was noticed that just after the turning on of the laser the calorimeter
"behaves uneasy". Than it calms down and clearly shows zero.

Then the beam was focused onto a carbon paper. A spherical aluminium coated mirror with focal length f=50 mm was used. Here it became visible that the laser gives only the first 1-2 pulses in the sequence and then dims. Due to my spark gap is of the uncontrolled type, I tried to reproduce repetition rate change by repeatedly plugging the power supply to the mains. It appeared that the laser was able to produce pulses at ~0.5-1 Hz (each one or two seconds). And it was uncapable to do more.

It was supposed that preionizer feeding capacitives got charge from the first pulse, and since the circuit is cut, they cannot discharge further on, rather than through the huge and uncontrollable stray resistance. To take things under control the leakage resistors were added:

           SG
      +--->   <---+-----+-------+-------+-----+-----+------+
      !           !     !       !       !     !     !      !
      !           !     !       /   Cpi---    !    ---Cpi  /
      +----||-----+     \       \      ---    !    ---     \ Rpi
      !    Cp           / Rl    /       !   XXXXX   !      /
      !                 \       \Rpi    +---     ---+      \
     ---                /       !       !   XXXXX   !      !
     --- Cs             \       +-------+     !     +------+
      !                 !                     !
      !                 !                     !
      +-----------------+---------------------+

Laser got run again. With Cs=6nf and CO2:air mixture 2:1 at full athmospheric pressure (SG1 set to 8 mm) it gave more than in the first run. Since power is low and unstable it is hard to say anything in figures but subjectively the output rised by about 30%. With 1:1 mixture the output rised once more and @ 1Hz laser yeld ~6 mW. It still fails if attempting to operate faster than once per second.

The preionizer connection was reworked. Leads were added at the back mirror side of the laser. RpiCpi circuits are similar in all places (50pf x 5.6kOhm). Since the number of such circuits is now 4 (each of two wires has two ends and each end is feeded through RpiCpi from the anode) the total preionizing capacity is now 200 pf.

Got a test run again. Still Cs=6nf and the mixture CO2:air=1:1 was used. Pressure was 1 bar. At 1 Hz the laser gave ~12-15 mW. When turning onto a self sustained repetition rate (about 30 Hz) it gives about ten pulses until dims. The traces on the carbon paper became throughout holes (of course with focusing).

There was an attempt to increase Cpi to 100 pf. Laser became unstable with low and unpredictable probability of lasing. There appeared bright spark discharges inside the laser tube. The sparks run not between the electrodes but between electrode keeping plates, and one can see that the preionizing wires play role of ignition electrode as in trigatron.

The Cpi capacitors were returned to 50 pf and the attempt was made to increase storage capacity. One more 2nf Murata was added. The pulse energy increased but the bearable repetition rate dropped. As the result the average power decreased significantly (to ~eight milliwatts).

Interesting notice: when focusing to a carbon paper the plume (visible spot where the beam hits the target) is not "electric-blue" as for powerfull TEA lasers, but "yellowish-white" as for low pressure longtitudinal CO2 lasers or for solid state free running ones. It probably means that the resistanse of the carborundum electrodes makes the pulse to became long, possibly to several microseconds.

The main conclusion one can make is that most possibly the deal was in diffraction losses. In TEA lasers with metal electrodes the last form a waveguide that keeps wave losses at reasonable level and small spacing between the electrodes becomes tolerable. The other situation is here: the carborundum is coarse and highly absorbing. Why "most possibly" not "surely"? Because with higher spacings a higher energy deposition is possible. There exists a possibility that increasement of energy deposition with keeping the capacity at constant level (i.e. with smoother discharge) might have played a role.

Still incomprehensible the nature of lasing stopping at higher repetition rates. With metal electrodes my TEA's never had demonstrated such a problem, at least up to 30 Hz.

 

TEST 4

Further increasement of the spacing between the electrodes. The gas-discharge unit was reassembled. The props between electrode keeping plates were replaced by ones with 12 mm height. It gives interelectrode spacing of 5 mm. The placement of preionizer wires relatively to the cathode was kept intact. Correspondently the distance to the anode cutoff (by vertical coordinate) has grown to 3.5 mm.

Tests of the unit at the open air with storage capacity equal to 4 nf has shown nice uniform discharge. It seems miraclous to observe a uniform glow of such a size in the air. Especially after such efforts directed to the volumetic discharge ignition.

miracle

For the comparison here is a photo of the discharge with preionizer having been disconnected. The idfference is prominent, isnt it? By the way at smaller spacings the difference is less significant.

miracle_not

The gap in the SG was set to 10 mm and with Cs=6nf the laser got a run on CO2:air=2:1 mixture at athmospheric pressure. Yielded about 6 mW at ~1Hz. Still shuts with repetition rate rising. In addition at the ends of the laser cell there appeared white sparks (jumping from plate to plate via preionizer wires).

Cs was increased more: to 10 nf (5 of Muratas). The power at low repetition rate grew to 15 mW. And at last the laser became operational at free running repetition rate (~10 Hz). It yelds ~12 mW. I.e. relatively to the low rate mode the pulse energy dropped more than 10 times. But it still lases. However the white arcs at the ends of the cell become so strong that it may cause troubles with the integrity of the laser.

One more experiment was made: the storage bank was connected to the laser tube not by a wide metal foil strip as usual, but by a strip having 2 cm width and 20 cm length (imitation of "feed-by-wire" mode). It still lased but the power dropped by 4 times (from 12mW to 3 mW).

Conclusion: further increasement of the space betweeen the laser electrodes from 4 mm to 5 mm wasn't helpfull. With increasement of operation voltage the preionizer began to work worse (to initiate sparks). So one needs to change the place for the wires or to reduce the feeding capacity (which is undesirable) or completely change the preionizer type. In addition there appear round path breakdowns around the laser gap (not between the carborundum bricks but between feeding wirings via the preionizer). And after it have sparked several times there appear charcoal coating on the dielectric (on electrode keeping plates or props). Such a coating fixes the birthplace of sparks and furher on laser becomes inoperational even at lowe voltages. I.e. when the voltage is low enogh its hard to ionize the gap between the laser electrodes and (so the output is low) and when the voltage is high enough the gas discharge unit fails and needs to be repaired. Seems that for the given size of the carborundum bricks the voltage needed for normal operation (~30 kV) appears to be too high.

 

TEST 5 (backstep) (25.10.2014)

Since the increasement of the spacing between the laser electrodes to 5 mm has not brought any positive, the laser was reasssembled back to the previous configuration (with the inter-electrode spacing of 4 mm).

Then an attempt was made to increase the storage capacity with gradual decreasment of voltage (only if needed, when too many sparks appear). Step by step all seven Muratas were installed into the laser frame (14 nf total). Strangely but there was no need to decrease voltage strongly. The optimal gap in the main spark gap is near 7.5 mm.

Obviously the discharge is less stable than with 3 muratas for the same conditions, but way more stable than with 3 muratas and 5 mm spacing between the laser electrodes.

Lasing tests gave the next results (for not to type it everywhere i'll say that the pressure was 1 bar/1000mbar/765torr in all tests herebelow. The mixture was also without any additives such as xylene or any.)

  • on CO2:air=1:1 mixture

    The yeld was 10 mW at 6 Hz. The power drop from the first pulses in the sequence to the further ones is evident by the naked eye. The discharge does also behave correspondently: the share of the violet glow reduces and the share of the streamers grows. One may propose that it is due to the electronegative ions accumulation.
  • on CO2:combustor=1:1 mixture

    Laser gave 42 mW at 6 Hz. Visually there's no drop of power from the first pulses in the sequence to the subsequent ones. May be by a factor of two or evn less. (For ones who haven't read my other files, i should remind that the "combustor" is the common air, where the oxygen was burned out by combustion of a carbohydride fuel. In this particular case it was done by burning ethanol - sometimes the type of fuel makes some difference for the discharge form.)
  • on CO2:N2=1:1 mixture

    Laser gave 96 mW at 6 Hz. In comparison with "combustor" here's even less drop of power from the first pulses in the sequence to the subsequent ones. The plume of the laser spot has got a typical "electric" bluish color. The nitrogen was of the technical grade from a high pressure bottle. (Yes, I finally got one too. Could not resist to the temptation to find out hou much a nitrogen laser will yeld when feeded by pure nitrogen. However it is a different story.)

THE CONCLUSION: for particularly that carborundum bricks the spacing of 4 mm between them is fairly optimal. The simpliest build of laser, having wire type preionizer appears to be operational with air:CO2 mixtures at full athmospheric pressure. When feeded with comparatively pure gases it showed efficiency of 1% by energy deposition. The reason of that rather low efficiency lies not in the ohmic losses inside the electrodes. (Moreover now I don't believe that the low efficiency in Bealieau laser was due to the losses in the resistors. Something else there. Maybe too slow energy deposition in comparison with the upper laser level effective lifetime.)

It is clearly seen that it wants higher preionization and maybe higher energy deposition. For going further more carborundum needed. When I get some the report will continue.

 

28.10.2014

The aluminium spherical mirror (the car rearview one, having been washed out of protective paint) was replaced by a real laser one. The new mirror is gold plated plano. The front mirror was kept intact. Other conditions are unchancged too.

On CO2:combustor=1:1 mixture laser gave 9 mW at 6 Hz. Very careful alignment while laser is running allows to grab as much as 18 mW.

The sense is that with a mirror, having greater reflection, but being plano, it yelds lower energy than with aluminium mirror having focusing properties. It means that at such a spacing between laser electrodes the plano-plano resonator already has significant diffraction losses. Significant enough to lower the output power.

 

30.10.2014

A try to emulate the nitrogen by cleaning the combustor gas. One car tyre was filled by the combustor. Then the gas was allowed to flow into another tyre through a gas-mask filter. Then into that new tyre a technical grade carbon dioxide from Crossman's cartridge was poured until the volume ratio became 1:1.

With this mixture the laser gave 36 mW (rep rate 6 Hz - the spark gap was kept untouched).

After that a (fresh) mixture of Co2 with technical grade nitrogen (from a cylinder) with a ratio of 1:1 was made.

Using this nitrogen mixture the laser gave 60 mW.
It means that cleaning of the combustor gas has gave nothing. The factor of two difference between yelds with nitrogen and with combustor hasnt gone away. Seems it is due to undercombusted oxygen. It cannot be captured by the active charcoal.

However the reason of yeld decreasement in comparison with previous tests does bother. The question why anything gets worse is always bothering. Handlights having been bought a year ago light less brightly even with fresh batterries. Year to year the cellular phone keeps connection at less and less distance from the cell. And even when there it seems to be nothing to die... As in case of this particular laser. Six carborundum bricks, two wires, two mirrors and a few pieces of plastic. What can occure to that?

Its back mirror was removed, slightly and carefully polished with a paste for cellular phones restauration. The front mirror was simply cleaned up. Both mirrors were then placed back and aligned. The laser once again was allowed to run on a CO2:N2=1:1 mixture. It gave 78 mW but rapidly degraded to the previous 60 mW.

Front mirror was replaced by more reflective (and less badly used) one. R=95%. Yield was 18 mW. The front mirror was again replaced. But this case by less reflective (and more badly used) one. With R=50% yield was... yield was... Well... At first it was very high. The light spot on the graphite surface of the calorimeter was large and bright enough to make thoghts of about 100 mJ or even more. But then the laser got died. It happened too fast to get enough time to malke measurements, so no figures, sorry for that.

Disassembling and defectation has shown some cleavages on two (of six) carborundum bricks. Possibly the chips of SiC have shortened the preionizer wires. Probably they could be shakened out without disassemblage of the laser.

 

31.10.2014

Laser was assembled back (without any modifications or repairings) and started on nitrogenic mixture: CO2:N2=1:1. The output coupler was still 50%. It lases. So the pieces of the carborundum definitely could be shakened out and it was possible to continue the use of the laser without the ceremony of disassemblage. The output, however, was only 30 mW. And no light on the graphite at all. After those promising preliminary results it was a bit disappointing.

To the tyre containing the nitrogen with carbon dioxide mixture a pair of xylene drops were added. The drops were really small - may be of a 1.5 cubic millimeters. Then the tyre was shakened thouroughly to ensure even distribution of the xylene vapours. The run on this mixture gave 78 mW. The first 5..10 pulses show bright light on the graphite. Then it dims and after ~3-5 seconds (18-30 pulses) there is no signs of light at all. Up to that moment the calorimeter reading reachs about a half of their static value, so I still cannot say anything definite about the energy of the first pulses.

So in previous experiments that were only my thoughts that the mixture was free of xylene. In the reality the rubber tyre smelled with xylene remaining from the previous tests. And with adding the fresh gas the amount of xylene in the mixture decreased gradually. At first it came closer to the optimum and the power of the laser rised. Further on it decreased away from the optimum and the power began to drop and the laser started to spark. The last fact actually had caused the cleavage of the electrodes.

The next test was an attempt to forsage the combustor:CO2 mixture by adding some xylene too. There was no success though. The optimal concentration of the xylene is rather low and in this case the xylene burns out fast in the remains of oxygen. If we add too much xylene it burns out not so fast but it begans to drop the outbut power by itself. As it was written in most articles: "the discharge became more uniform but no increasement of output power was observed". That phrase just means that they didnt hit the optimum concentration.

 

02.11.2014

Laser had worked for some time and died after that. It did about 1..2*10^3 of pulses with gradial dimming to complete darkness. Even for the mixture of CO2 with technical grade nitrogen. Disassemblage and visual inspection have found out nothing. It looks like everything is allright but it does not lase. Moreover it doesnt even spark. At least it sparks not more than it did when lasing was OK.

 

05.11.2014

A new small variant of laser was made. It has electrode length of exactly one brick (10 cm). The preionization was made by Dumanchin-Farcy scheme:

    HV pulse _/\_  O------------+-------------------------------------+
                                !                                     ! Cpi
    |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX|    --- 50
    |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX|    ---
    |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX|     !
     -----------------------------------------------------------      !
  ^                                                                  -+
  ! 4mm                                                     ________/ !
  V  ___     ___     ___     ___     ___     ___     ___   / ___      !
    |XXX| 0 |XXX| 0 |XXX| 0 |XXX| 0 |XXX| 0 |XXX| 0 |XXX| 0 |XXX|     !
    |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX|     / Rpi
    |XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX|     \ 100k
                                !                                     /
                                !                                     \
                                !                                     /
                                !                                     !
                                +-------------------------------------+
                                !
                               ---

Dumanchin's preionizer is a plurality of insulated metal wires, placed in grooves in the cathode. In the original form that were nickel wires inside quartz tubings having diameter of 1 millimeter. Its clear that DIYer is out of such a resources. If theres none nickel in quartz, how'bout copper in teflon?

One of the carborundum bricks was grooved by means of some abrasive disc. It was done by hand engraving tool, so the grooves aren't very even and straight. However with some fortune one can put nearly 30 grooves over the length of the brick. (The work appeared to be dirty and uneasy: dust and spray flew directly into face. If You ever want to do such a thing be sure to put on glasses and respirator.)

Copper wire in teflon insulation (russian MGTF trademark) was put into the grooves and connected to the busbar of the opposite polarity. (Kinda to the anode. Why "kinda" will be clear from the futher reading.)

At first the spacing between the laser electrodes was set to 3 mm and a test run was made. The discharge glow was very bright and uniform and without any signs of streamers. As they say: "it can't be better!" Then the spacing was increased to 4 mm with correspondent increasement of feeding voltage. And here the bad things started. First of all the insulation of the wire had failed, then the gaps and spacings in the schematic wiring began to spark, then again the insulation of the wire... The geometry of the laser auxiliary circuits was revised in order to increase the gaps, and the attempt was made to reduce the electrical stress on the preionizer. The direct connection to the opposite polarity was changed by the connection through a rather small capacitor (50 pf) exactly as it is shown on the figure above. However it didnt help the wire to survive. After that all available insulated wires of suitable outer diameter (~1mm) were grabbed and tested. Some failed earlier than MGTF, others appeared to be stronger. The one most, strong to the electrical stress, that i have found looks like this:

provod

can not surely detemine its trademark by how it looks like. It is elastic stranded wire, its insulation looks like PVC. And it was found inside a very old LPT printer connection cable. Thats all i can say.

This wire whitstanded about three thousands of pulses whan the main spark gap was set to 12 mm (about 31 kV). Since in our laser the voltage should be lower one may expect that this wire will show certain lifetime. After the preionizer had been made of this wire the breakdowns ceased. The volumetic discharge with some streamers was then obtained for the interelectrode spacing of 4 mm. But for some reason the electrode with the preionizer wishes to be anode. For the other polarity no glow dicharge appeared. Sparks only.

For the tests of lasing this little testbed was sealed into a cut of two-inch PVC tubing with the mirror alignment units on the ends. The mirrors were: the back one - aluminium on glass, spheric R=2 meters, the output one - ZnSe, dielectric, ro=95%, plano. The distance between mirrors was 185 mm. The gas mixture: technical grade nitrogen with carbon dioxide from Crossman's cartridge, 1:1 proportion. Also an addition of xylene was used in amount of ~5 cubic millimeters per 20 liters of the gas mixture.

It lased without any difficulties. At the repetition rate of 1 Hz it yields 12..15 mJ per pulse. At higher rep rates it dims rapidly. As it was with the previous type of laser: the first pulse is powerfull, the second one is weak, and further on is darkness.

Here is how the laser looks like (dont look at the size of the frame, it is just universal, and intended for placement of much greater devices. You better go figure the size of the laser tube itself.)

100mm_dumanchin
100mm_dumanchin_in1 100mm_dumanchin_in2
100mm_dumanchin_in3 100mm_dumanchin_in4

The storage capacity is 4 nf (it can be seen on the photo, that only 2 Muratas are used), charging voltage is 21 kV (8mm in the main spark gap), and if the voltage of the glowing discharge is 3 kV/mm, then the charge, having flown before dimming/contracting of the discharge, will be:

q=4e-9F*(21000V-4mm*3000V/mm)=3.6e-5 Cl

And the endergy deposition in gas:

E=qU=3.6e-5 Cl*4mm*3000V/mm=0.43 J

I.e. the efficiency by the energy deposition is 2.8..3.4%.

And the full efficiency (related to the stored energy CU^2/2) is 1.3..1.7%

Even while remembering that we use rather clean gases, the effeciency value obtained for such a short laser does impress.

In principle hereabove is a (not)great handwaving, because I've taken the field tension for the stationary glow discharge to be equal to the breakdown field tension for the air. (Even, not for the nitrogen!) But, say, in [4] they give 30..35 kV/cm for the field tension for the glowing phase in pure CO2.

 

10.11.2014

An ability of the short (10 cm) laser with Dumanchin's preionizer to operate with mixtures with air and with combustor was proven. On the air:CO2=1:1 laser gave ~6 mJ and for the combustor:CO2=1:1 it yielded less than 5 mJ - its hard to measure due to low power and instabilities, but it gives definitely less than with air. Its curious that the combustor, being usually better than air, gives here such a fail. Most possibly it is due to contamination by water vapours. The laser is short, the gain is small, and in above all that the quenching by H2O appears.

With both mixtures the the top rep rate, when stable operation is possible, is even less than with nitrogen: once per 2 sec for air based mixture and something like once per 3 sec for the combustor based one. Pulse-to-pulse stability became worse too. Consequent pulses may easily differ in energy by the power of two. It looks like the laser sits just over the threshold.

In order to check a possibility to build the laser having no vacuum pump, a test was done to run the laser with filling by washing mode. The laser was evacuated (to less than 100 torr), then filled with air. Then the exhaust pipe (at the rear mirror) was opened to the athmosphere and the inlet pipe (the one at the front mirror side) was fed by the premixed N2:CO2 mixture. When abot 5 liters were flown down, it lased similarly to the case of air with co2 mixture. After that 5 liters more of the mixture were allowed to flow through the laser. After that it gave well formed laser spot when observed by the light on a piece of carbon paper. Further on another 5 liters of gas were passed through the tube. This time the lasing power was almost undistinguishable from the one in case of tests with normal filling.

If the volume of the laser tube is VL=L*pi*d^2/4=13*3.14*5^2=255 ml, then it results that the washing ratio should be k=V0/VL=40..60. If one can provide such washing ratio when filling by gas blow, the laser will work well.

 

11.11.2014 GRIDDIE

A new laser was assembled. As it was in case of Dumanchin's one it has electrode length the same as the length of the brick. Anode was made of one carborundum brick 100x10x3 mm. Cathode - a grid from a kitchen sieve. The power supply was made by this scheme:

       +---+------+-------------+---------------+------+
       !   !      !             !               ! Cpi  !
       !   !   5k6\             !               ! 50   / Rpi
       V   !      /             !              ---     \ 100k
          ---     \             !              ---     /
       ^  ---     /          XXXXXXX            !      !
       !   !4x570 !                          ---+------+
       !   !      !          ........_______/   
       +---+      !          XXXXXXX
           !      !             !
           !      +-------------+
          ---  2 x Murata       !
          ---  doorknob         !
           !  2000pf x 40 kV    !
           +--------------------+

The kitchen sieve looks like this:

sito

It was found out that the mesh can be easily removed just by unbeading its frame. No need to cut anything. To flatten the mesh was also easy, despite any infernal expectations. The size of the mesh cell is 2 mm. It better be smaller, but one uses what he has. The spacing between the mesh plane and the anode is 4 mm.

Behind the grid there is a preionizing electrode - a similar carborundum brick. Size of the mesh is greater than the size of the brick with a reasonable surplus. Pay attention to the fact that the brick, that is placed behind the mesh, is not just a preionizer. Naturally all the supply currrent goes through it. I.e. this circuit is the one with feeding through the preionizer. Ideally it should provide preionization not only by the UV, but also by the electrons having been extracted from the auxiliary discharge into the main discharge by the electric field.

At the very first test the scheme sparked happily. The sparks went not between the anode and the mesh, but between two carborundum bricks, through the cells of the mesh. After a bit rework of the RC circuit of the preionizer some glow was finally obtained. But... As it was seen from the end of the laser the glow appeared not between the planar faces of the electrodes, but around their sharp edges.

The edges of the bricks were rounded. This work can be done by a diamond cutting disc and a hand engraving tool. What do You think, had the glow moved to the planar faces? No. It disappeared completely. Only sparks remained.

In the torture process of the spark surpressing many things were done.
Even a guard discharger was added:

       +---+------+-------------+---------------+------+
       !   !      !             !               ! Cpi  !
       !   !   1k1\             !               ! 50   / Rpi
       V   !      /             !              ---     \ 100k
          ---     \             !              ---     /
       ^  ---     /          XXXXXXX            !      !
       !   !4x570 !                          ---+------+
       !   !      !          ........_______/   
       +---+      !          XXXXXXX    V
           !      !             !       ^
           !      +-------------+-------+
          ---  2 x Murata       !
          ---  doorknob         !
           !  2000pf x 40 kV    !
           +--------------------+

The guard spark gap was known to help under the similar circumstances (US.Pat 3940710). BTW it possibly helps by making a preionization of the preionization spacing. (Yep, the box in the box.) But with this particular laser it didnt help. The solution was found in making the electrodes thicker.

       +---+------+-------------+---------------+------+
       !   !      !             !               ! Cpi  !
       !   !   1k1\             !               ! 50   / Rpi
       V   !      /             !              ---     \ 100k
          ---     \             !              ---     /
       ^  ---     /          XXXXXXX            !      !
       !   !4x570 !          \XXXXX/         ---+------+
       !   !      !                          !
 +30kV !   !      !          ........_______/   
    o--+---+      !          /XXXXX\    !
           !      !          XXXXXXX    V
           !      !             !       ^
           !      +-------------+-------+
          ---  2 x Murata       !
          ---  doorknob         !
 GND 30 kV !  2000pf x 40 kV    !
    o------+--------------------+

The slash characters near the crosses (meaning carborundum) are the silly try to designate that the carborundum bricks have rounded edges. In addition for the things to be more clear it was needed to show the power supply connection points. Usually it is clear even to a fish, that the power supply should be connected to the storage capactior, but here the polarity of the connection is important.

Now the electrodes (the main one and the undermesh one) are both glued of two bricks being placed one onto another. The thickness of the carborundum increased thus to 7 mm. The spark is still present, but at least it is accompanied by the glow. The glow is visible by naked eye and finally takes place between planar faces of the electrodes.

However the test of lasing have shown that the device is inoperational. I.e. the intensity of the glowing discharge is too weak.

At the last despair try the sheme was "turned upside down":

  GND 30 kv 
    o------+----------+-------------+---------+------+
           !          !             !         ! Cpi  !
           !       1k1\             !         ! 200  / Rpi
 2 x Murata!          /             !        ---     \ 100k
 doorknob ---         \             !        ---     /
  2000pf  ---         /          XXXXXXX      !      !
   40 kV   !          !          \XXXXX/      +------+
           !          !                       !     
           !          !      ____.......      !
    o------+---+      !     /    /XXXXX\      !
 +30 kV    !   !      !     !    XXXXXXX      !
           !   !      !     !       !         !
           !   V      !     !       +---------+ 
    4x570 ---         !     !       
          ---  ^      !     !       
           !   !      !     !       
           +---+------+---- +

The mesh became anode, the cathode became carborundum brick and the undermesh electrode became banal preionizing electrode. The full supply current does not flow through it after all.

The discharge test have shown a "sparky rain", and barely seen glow at the background. As always in such cases it is not clear beforehand whether it will lase or the sparks will interfere with lasing.

The laser was started on N2:CO2=1:1 mixture. The resonator was the same as for previous one (back - Al on glass, sphere, measured R>96%; front - plano, ZnSe, R=95%). At rep rate of 1 Hz it gave 8..10 mJ. What is more curious it does not dim with rep rate rising. At ~10Hz it gives ~18 mW, that means 1.8 mJ per pulse.

However with the combustor based mixture it behaves differently. Not only it ceases with rep rate rising, it shuts also with time passed - even at the rep rate less than 1 Hz it gives 5..10 pulses untill fading. After that
even the filling by the pure N2:CO2 mixture does not bring it to consciousness. Only after a long pause (more than 1 hour), and after having been filleg by a fresh pure mixture it begins to lase. I.e. it appears to have a property to be capable to be poisoned by bad gases.

CONCLUSIONS:

  • with good gases the grid-based laser is better than the Dumanchin's one. it gives the same power but lives at higher repetition rates. With bad gases the Dumanchin's laser is better.
  • for the normal laser operation it needs the sum thickness of the carborundum electrodes to be twice greater than the spacing between the electrodes. That's particularly why there was the fail in increasement of the spacing to 5 mm and above.
  • once again it was proved that nothing certain could be said by visual observations of the discharge. In some cases it lases with sparks in the laser channel, in other cases it refuses to lase even with a fine glow.

 

12.11.2014

Here are the photoes of the laser internals. One can see the mesh, the carborundum brick behind the mesh, and the fact that the bricka are put one onto another in order to increase the total thickness of the carborundum.

griddie1 griddie2
griddie3 griddie4

The griddie laser was run on a mixture with air. It lased but weakly. The power was unmeasurable by my calorimeter. The light spot on carbon paper, produced by its focused beam was little - literally having the size of one pixel. And the pixel is weak, and it does not appear any time.

The tests were made with mixtures of CO2 with air in proportions from 2:1 to 1:2.

CO2 : AIR   QUALITIVE CHARACTERISTIC

2

: 1 - weak pixel

1

: 1 - bright, but still pixel, not strip or macroscopic spot

2

: 3 - weak pixel again

1

: 2 - no lasing at all

In any of the tests the lasing was unstable. And it does not seem that the laser wants low repetition rate, i.e. certain delay between pulses. It just hard to say whether in this particular pulse will be any lasing or not. At the rep rate of 1 Hz the laser spot appears with a probability ~10%. And at rep rate of 0.1 Hz it does still appear with a probability of ~10%. Moreover the first pulses in the sequences tend to be without lasing. However in the contrast to the combustor based mixture, the air based ones do not poison laser. I.e. after a certain pause and with fresh batch of gas it was possible to see lasing again and again.

Generally i believe (i cannot prove it seriously, so it remains only to believe) that in lasers with the unlimited speed of energy deposition (ones based on metal electrodes, low inductance capacitors, having low inductance wiring, fast spark gap and correctly choosen peaker) the highest gain takes place on the mixture proportions of 1:1 It seems that the energy here is at top too. In lasers with slow energy deposition (Bealieau type and etc.) it can happen that the optimum will go to the higher nitrogen proportions, since the vibrational levels live in it sufficiently longer

Concerning the mixtures with technical grade nitrogen, it need be said, that this design of laser appeared to be very sensitive to the gas quality. With the mixture having been stored in (rubber) tyre overnight the output drops three times in comparison with the fresh mixture.

In order to increase the preionization effectiveness a try was made to introduce a delay between the preionizing and the main pulse. Due to lazyness the most simple way to do this was selected. (See US Pat. Appl. 10/612848). The anode was simply connected with the cathode by a dumb shunt capacitor, and into the feeding circuit a delay element is insertes. The authors of the mentioned application used an inductor, and here i placed a spark gap)

  GND 30 kv 
    o------+----------+---------------+---------+------+
           !          !               !         ! Cpi  !
           !       1k1\   +---+       V SGD     ! 200  / Rpi
 2 x Murata!          /   !   !       ^        ---     \ 100k
 doorknob ---         \   !   +-------+        ---     /
  2000pf  ---         /  ---       XXXXXXX      !      !
   40 kV   !          !  --- CD    \XXXXX/      +------+
           !          !   !  470pf              !     
           !          !   !    ____.......      !
    o------+---+      !   !   /    /XXXXX\      !
 +30 kV    !   !      !   !   !    XXXXXXX      !
           !   !      !   +---+       !         !
           !   V      !       !       +---------+ 
    4x570 ---         !       !       
          ---  ^      !       !       
           !   !      !       !       
           +---+------+-------+

It run with N2:CO2=1:1 with the previous resonator.
It lases but worse than without CD and SGD. Maybe too much is lost on the spark gap.
A try to replace it with an inductor (LD on the scheme):

  GND 30 kv 
    o------+----------+-------+-------+---------+------+
           !          !       V       !         !      !
           !          !       ^  SGD  )         !      !
           !          !       !       )         ! Cpi  !
           !       1k1\   +---+       ) LD      ! 200  / Rpi
 2 x Murata!          /   !   !       !        ---     \ 100k
 doorknob ---         \   !   +-------+        ---     /
  2000pf  ---         /  ---       XXXXXXX      !      !
   40 kV   !          !  --- CD    \XXXXX/      +------+
           !          !   !  470pf              !     
           !          !   !    ____.......      !
    o------+---+      !   !   /    /XXXXX\      !
 +30 kV    !   !      !   !   !    XXXXXXX      !
           !   !      !   +---+       !         !
           !   V      !       !       +---------+ 
    4x570 ---         !       !       
          ---  ^      !       !       
           !   !      !       !       
           +---+------+-------+

When LD is formed by a copper wire, 1 mm in diameter and 5 cm long, and when this wire is bend nearly to half of a ring the things seem to be quite close to the optimum and the laser yelds the same energy as it did without all this stuff (SGD, CD and LD). It gives less with larger and lower values of the LD. It is Curious that when LD is formed by 3-turns coil, wound on a tubing having 50 mm diameter, the SGD begins to spark. Need to note that SGD naturally had form of a spacing between the laser tube lead and the laser frame base, so it could not easily be removed from the scheme completely. With such a coil (3 turns 50 mm diameter) laser gives energy intermediate between the optimal one and one in case of only SGD usage.

Again I should note that I'm not scientist - I'm a DIYer. With my instruments and resources the difference by a factor of two is "insufficient" one, it should better be treated as subjective rather than objective difference. Only the difference by an order of magnitude does really mean something.

Hence if in addition one succeed in proving that the preionizer of this scheme is also tolerant to inductance, it will appear to be that this scheme can really be "fed by wire". However I have no zeal to do it, dince the griddie laser has shown other shortcomings - first of all the evil sensitivity to the gas quality.

 

17.11.2014 STANDIE

With the griddie laser it was found out that the carborundum should have a certain thickness for the good operation (best if the thickness is nearly equal to the twice of the value of the spacing between the electrodes). Since that it seemed that a good way for increasement of the inter electrode spacing is to increase the thickness of the carborundum, and the very way to do it is to set the bricks in "standing pose." So the bricks were made to stand on their narrow face:

  //////////////////  <-perspex platform
     XXXXXXXXXXX  
     XXXXXXXXXXX <------Carborundum brick  
 ^                  
 ! 4 mm                  
 V                  
    O XXX O XXX O
      XXX   XXX  \  wire d=0.5mm
      XXX   XXX   \-------------
      XXX   XXX 
      XXX   XXX <------Carborundum brick
      XXX   XXX 
  //////////////////  <-perspex platform
         <->
          2 mm

Two down bricks were installed in standing pose and the upper brick was
put as lying on its holder plate. Between the standing bricks, along them,
some wires were stretched. If the wires are insulated the schematic becomes
"longtitudinal Dumanchin's scheme". If they are naked it becomes something
intermediate between Dumanchin's and Lamberton's ones.
If there is no insulation, there is nothing to breakdown. So at first
the naked wires were choosen. The wires were connected to the other polarity
by individual capacitors Cpi1...Cpi3

   +-               -+------+----O _/\_ HV pulse
   ! \_____________/ !      !
   !   XXXXXXXXXXX   !      !
   !   XXXXXXXXXXX   !      !
  ---               ---    ---
  --- Cpi1      Cpi2---    --- Cpi3
   !                 !      !
   +--O XXX O XXX O--+      !
        XXX ! XXX           !
        XXX ! XXX           !
        XXX ! XXX           !
        XXX ! XXX           !
        XXX ! XXX           !
   +-------------           !
   !        !               !
  ---       +---------------+
   - 
(Ground)

Or by common capacitor Cpi:

   +-               -+------+----O _/\_ HV pulse
   ! \_____________/ !      !
   !   XXXXXXXXXXX   !      !
   !   XXXXXXXXXXX   !      !
  ---               ---    ---
  --- Cpi1      Cpi2---    --- Cpi3
   !                 !      !
   +--O XXX O XXX O--+      !
        XXX ! XXX           !
        XXX ! XXX           !
        XXX ! XXX           !
        XXX ! XXX           !
        XXX ! XXX           !
   +-------------           !
   !        !               !
  ---       +---------------+
   - 
(Ground)

The leakage resistors (in parallel to the capacitors) arent shown on the scheme for the sake of clarity. However in reality they were installed.

The assembly was tested in open air. The storage capacitor was 2x2 nF, the main spark gap was set to 8 mm. Paradox, but the second schematics (all wires are soldered together and connected by the common capacitor) shows visually better discharge than the first schematics, where the prionizing wires are fed
by individual capacitors. Even if the condition of Cpi1+Cpi2+Cpi3=Cpi is observed.

The assembly was then installed into the plastic tube with resonator (the same as the above and the above of the above) and tested with the N2:CO2=1:1 mixture. NO LASING. Even with fresh gas. Even at low rep rate. Even at the spacings up to 12 mm in the main spark gap (up to 31 kV). Neither with individual capacitors nor with common one.

OK then. The next is the longtidudinal Dumanchin's scheme. The naked wires were replaced by the insulated ones (see note of 05.11.2014)

          -               --------+----O _/\_ HV pulse         
           \_____________/        !
             XXXXXXXXXXX          !
             XXXXXXXXXXX          !
                                  !
                                  !
                                  !    
  +--------(O)XXX(O)XXX(O)--------+
  !           XXX ! XXX           !    
  !           XXX ! XXX           !
  !           XXX ! XXX           !
  !           XXX ! XXX           !
  !           XXX ! XXX           !
  !      +-------------           !
  !      !        !               !
  !     ---       +---------------+
  !      -                        !
  !   (Ground)                    !
  +-------------------------------+

(o) - means here the insulated wire.

The discharge became more even. The spaks became more weak. But still no lasing at all.

A proposition: the resistance of vertically standing bricks is too high. Or, to be more correct, the field is too weack to cause the avalanche breakdown in the carborundum. (Since the SiC is semiconductor, it seems to be capable to enter the state of high conductivity, when the applied electric field reaches a certain threshold.)

For the checkout only two bricks were left:

          -               --------+----O _/\_ HV pulse         
           \_____________/        !
             XXXXXXXXXXX          !
             XXXXXXXXXXX          !
                                  !
                                  !
                                  !    
  +----------(O) (O) (O)----------+
  !          XXXXX!XXXXX          !    
  !          XXXXX!XXXXX          !
  !         /-----!-----          !
  !      +--      !               !
  !      !        !               !
  !      !        !               !
  !     ---       +---------------+
  !      -                        !
  !   (Ground)                    !
  +-------------------------------+

And along one of them three insulated wires were put. The spacing between wire and the other side brick was set to 4 mm. The assembly in the essence is longtitudibal variant of Dumanchin's scheme without any grooves. It is much simplier to be made than the one described above.

Lasing test has shown the prominent absence of lasing.

Then the spacing between bricks was reduced. It became 4 mm between the bricks themselves (not between wire and brick as above). The central wire was removed for not to make obstacles on the long wave light pass.

          -               --------+----O _/\_ HV pulse         
           \_____________/        !
             XXXXXXXXXXX          !
             XXXXXXXXXXX          !
                                  !
                                  !
                                  !    
  +----------(O)     (O)----------+
  !          XXXXXXXXXXX          !    
  !          XXXXXXXXXXX          !
  !         /-----------          !
  !      +--                      !
  !      !                        !
  !      !                        !
  !     ---                       !
  !      -                        !
  !   (Ground)                    !
  +-------------------------------+

Such a scheme hadn't refused to lase. With the negative polarity on the electrode with the preionizer the laser yielded 3.6 mW @ 1 Hz and 24 mW @ 16 Hz (1.5 mJ for those lazy to use calc.exe). With the positive polarity on the electrode with the preionizer, it gave 4.2 mW @ 1 Hz and 30 mW @ 16 Hz (~2 mJ).

Maybe it is not much, but in comparison with all previous carborundum lasers it shows amazingly low dropdown with the increasment of the rep rate.

By the way, I hope that You are already got used to the ASCII graphics. Up to this moment the file already contains 17 schemes. In the pixel form I'd got mad to draw them, and You - to download them.

 

20.11.2014 SPARKY

OK, we've reached the point to test a "classic" scheme with spark type preionizer. Possibly most of You were quite impatient to wait for these results.

The most simple scheme is going to be assembled:

  +---+------+----+----+----+----+------+
  !   !      !    !    !    !    !      !
  V  ---    ---  ---   !   ---  --- Cpi !
     ---    ---  ---   !   ---  ---     !
  ^   !Cp    !    !  XXXXX  !    !      !
  !SG !      V    V         V    V      /
  +---+      ^    ^         ^    ^ SGpi \ Rb
      !      !    !  XXXXX  !    !      /
      !      !    !    !    !    !      \
     ---     +----+----+----+----+      /
     --- Cs            !                !
      !                !                !
      +----------------+----------------+

For the sake of the simplicity it is not shown that in parallel to each preionizing capacitor a 5.6 kOhm resistor is connected. There are also shown only 4 of the capacitors. In reality there are 8 of them.

As a few previous variants this laser has short electrodes. Their length is equal to the length of one carborundum brick. The inter-electrode spacing will be set to 4 mm.

Additionally in the process of the assemblage it was found out that my caliper was broken. So the sizes in the previous variants of lasers (and especially the interelectrode spacing) may have been measured incorrectly. I hope the error was small, but if You want to reproduce one of the lasers, You should be prepared to +-1 mm variations. Just to stay on the safe side. The preionizing spark gaps are made of M3 screws and placed in 1 cm from the electrode(s) and in 2 cm fom each other. Totally 8 spark gaps (see the figure):

                   20 mm
               !<-------->!

          -    O          O          O          O
          ^
     10mm !
          V
          XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
          XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
          XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
          ^
     10mm !
          V
          -         O          O          O          O
                               !<-------->!
                                   20 mm

The gap in the spark gaps is to be set to 2 mm. It should be mentioned that for correct operation the preionizing spark gaps MUST fire BEFORE the main laser channel ignites. Therefore the gap in the preionizing spark gaps should be choosen (much) shorter than the main interelectrode spacing.

The typical mistake of amateur laser-builders is the hope in some "non-self sustained discharge controlled by the external UV radiation". Drop it. When a necessary voltage is already applied to the electrodes the affect of any external radiation will cause nothing but spark. The preionization is called so exactly due to the fact that it must be applied before. Up to the moment of sufficient voltage application to the main laser electrodes the gas between them MUST ALREADY BE CONDUCTIVE. Only in that case You have a chance to obtain glowing phase of the discharge.

 

24.11.2014

The laser was finalized and tested at the next conditions: Cs=6nf, Cp=4x510pf, Cpi=100pf, Rb=1 kOhm, and in parallel to each Cpi a leakage resistor rated to 5.6 kOhm was installed. The spacing in the main spark gap was 8 mm.
Without artificial limiting of the repetition rate (it did ~16 Hz) the laser gave:

With mixture of technical grade nitrogen:CO2=1:1 78 mW (4.8 mJ)
With mixture of air:CO2=1:1 96 mW (6 mJ)
With mixture of combustor:CO2=1:1 84 mW (5.2 mJ)

Dont think that the things are bad with the technical grade nitrogen. Just the mixture was too old (was mixed abt ~3 days ago). The last two mixtures were fresh. It is more correct to consider that with all mixtures the laser gave the same energy in the limits of measurement errors.

With any mixture the laser spot on carbon paper is yellowish white without any signs of blue "electric" color.

An attempt was made to add an inductor into the feeding circuit:

  +---+----+     +-----+----+----+----+----+------+
  !   !    !     !     !    !    !    !    !      !
  V  ---   )     !    ---  ---   !   ---  --- Cpi !
     ---   ) Lt  !    ---  ---   !   ---  ---     !
  ^   !Cp  )     !     !    !  XXXXX  !    !      !
  !SG !    !     !     V    V         V    V      /
  +---+    +-----+     ^    ^         ^    ^ SGpi \ Rb
      !                !    !  XXXXX  !    !      /
      !                !    !    !    !    !      \
     ---               +----+----+----+----+      /
     --- Cs                      !                !
      !                          !                !
      +--------------------------+----------------+

Lt - was a ~30 cm long cut of flexible copper wire in vynil insulation. The section of the wire was about 1 square mm. The laser was then tested with the air:CO2 mixture. It lased. However the output power was 4 times lower than in the previous tests. After the usual type of connection had been restored the power did not return to the previous level.

Disassemblage and defectation have shown that one of the carborundum bricks was broken (possibly by a spark). I.e. one cannot say that the stray inductance has directly diminished the output. It just made the sparks to appear more easily. And the sparks in their turn damaged the electrode. After the scheme will be repaired I should check the operationability in the "fed-by-wire" mode again, and also I need measure the output at the low repetition rate.

inside_classic01 inside_classic02
inside_classic03 inside_classic04

PRELIMINARY CONCLUSION:

  • THE CLASSIC IS THE CLASSIC. It shows good results. But it is due to some reasons.
  • Very high energy of the preionization (more than 14% of the total in the storage capacitor) helps. Despite the large distance between the laser chanel and the spark gaps it still works.
  • the leakage resistors in parallel to the preionizing capacitors were intentially choosen as low as 5.6 k. The idea is to provide the UV lighting of the laser channel even after the capacitors Cpi have already charged. Even after that with such a low resitance there exists a continious arc discharge. It is known that all negative ions have very low ionization potential and can be decomposed even by the long wave UV radiation. The continious lighting should provide negative ions decomposition and releasement of the free electrons at any stage of the discharge. Judging from the operation of the scheme with the air based and combustor based mixtures the principle really works. I.e. the continious UV lighting does compensate the bad action of the electronegative gases, such as oxygen.

 

25.11.2014

The broken brick in the laser was replaced with another one. The laser got a test run. The yield was miniscule. The cause was the replacement brick. It has a trace of a spark breakdown from previous experiments. Maybe due to this fact or maybe due to its own properties (such as e.g. low conductivity) but when this brick is installed the glow is weak and the spark is strong.

Another replacement of bricks was performed.
This time the laser seemed to resurrect: 84 mW at 16 Hz rep rate.
At 1 Hz it gave 9 mJ. I.e. the energy drop with the rep rate increasement was rather low - less than twice.

In order to test the ability to operate with a "high inductance" power supply the laser was connected as follows:

      +------->  <-----+----+----+----+----+------+
      !        SG      !    !    !    !    !      !
      !               ---  ---   !   ---  --- Cpi !
     ---              ---  ---   !   ---  ---     !
     --- Cs3           !    !  XXXXX  !    !      !
      !                V    V         V    V      /
     ---               ^    ^         ^    ^ SGpi \ Rb
     --- Cs2           !    !  XXXXX  !    !      /
      !                !    !    !    !    !      \
     ---               +----+----+----+----+      /
     --- Cs1                     !                !
      !                          !                !
      +--------------------------+----------------+

The storage bank was formed by three mylar film rolled capacitors of K73-14 type (0.015mcF x 10 kV) having been connected in series. In parallel to the main spark gap SG there REALLY was NO PEAKING CAPACITOR. It was not an error, it was a survival test. Rb=100kOhm. All connections were made by a flexible wire having ~1square mm section. The spacing in the spark gap was set to 8 mm.

high_inductance_power_supply

It's amazing, but the laser was able to operate. It gave 40 mW at 11 Hz repetition rate. It worked for a few minutes (just enough to measure its power) and died.

It was found out that in the bricks there appeared a new hole made by a spark. And instead of the nice glow there are only sparks when trying to turn on the laser.

THE CONCLUSION:
Generally speaking we've got almost perfect scheme of laser.

  • It works at athmospheric pressure with gas mixtures of almost any purity grade, regardless whether they contain oxygen or not.
  • It keeps the energy output with increasing the repetition rate.
  • It provides high gain - even the laser with its electrodes as short as 10 cm was capable to operate with decent efficiency and output.
  • It gives long pulse - the dream of many CO2-laser designers. I probably won't make much error if I say that it has duration of several microseconds.

The scheme has only single one shortcoming:

  • IT IS NOT RELIABLE!
    While there is an even glow discharge - all things are OK, but when a spark appears it means the carborundum is dead. And there's almost no way to completely avoid the appearance of sparks.

The operation of the scheme was so impressive that a proposition appeared: what if this scheme is better than the one with holed electrode and the barrier discharge preionizer. May be all these years we've played fools and the first fails with the classical scheme were nothing more than experimental mistake.

For the test puposes two aluminium electrodes were made. They have shape of parallelepiped with all edges having been rounded. 10 mm wide, 100 mm long and 4 mm in height. The working side of the electrodes was polished.

Those electrodes were installed in place of the carborundum ones and the laser tube was connected to the low inductance pulse supply (the one with Murata capacitors). Cs=4nf.

classic_with_metal_electrodes1 classic_with_metal_electrodes2

The tests with gas mixtres based on air and on combustor have shown no miracle. With the metal electrodes the scheme produces sparks only. No glow, no lasing. No light inside the tunnel. It does not produce the glow even when having been pumped out to 25 torr. A bit disappointing. But on the other hand it means that the laser with drilled electrodes and with barrier discharge preionization was dicovered not in vain. In its most primitive modifications it gives volumetic discharge with air based gas mixtures at 200..300 torr, and the last variants of this laser are operational with air:CO2=1:1 at slightly higher pressures than a full athmosphere.

 

03.12.2014 LYRICAL RECESSION - CAPACITIVE LASER

When speaking regardless to the laser paraneters it is quite easy to obtain a volumetic discharge in the air at athmospheric pressure. Take two metal pieces without any sharp edges, wrap them into dielectric foil. (The dielectric should be thin and electrically strong. Mylar is a good choice.) Connect these electrodes to a source of high voltage alternating or pulsed current. Enjoy the result. In some papers they call this kind of the discharge as "corona on (at) the dielectric surface". As for me, I like the term "barrier discharge". In sounds more reasonably.

However any attempts to fill the volume of the discharge with laser gas mixture and to put all this into the resonator will only lead to fail. The energy deposition is too low here.

To ensure this we can make some simple estimations. Let Egas - is the field tension corresponding to the discharge glow in the gas (let me remind that it is equal to the breakdown voltage in the uniform field if we pretend the accuracy range not better than 10..20%. For the air it will be 30 kV/cm.)
Let Ed - is the electric strength of the dielectric and e - its dielectric permeability. Apparently the top value of the electric charge q, that can be passed throug a unit of dielectric surface (without breaking it) is equal
to the specific (to the unit of area) capacity multiplied by the charging voltage (for the obvious reasons the latter cannot exceed Ed*delta):

q = Ed*delta * e0*e/delta = 8.85e-12*e*Ed

Curiously that the thickness of the dielectric delta has gone. And the laser feeding voltage hasnt been left in the expression too. The limiting charge per unit of area appears to be related only to the properties of the
dielectric.

From the experience of nitrogen lasers we know, that the mylar foil can hold as much as 10 kv per each 100 mcm of thickness. I.e. 100 kv/mm or 1e8 V/m. The reference value of the dielectric permeability of the polyethyleneterephtalate (this is the true name of the mylar) is e=4. However my own measurements of the capacity approve only e=2. So:

q = 8.85e-12*2*1e8=0.0018 Кл/кв.м.=0.18 mcCl/sq.cm.

The electric power is definitely the electric current having been multiplied by the voltage drop (on the part of the circuit). The voltage drop on the gas discharge is a very conservative value. So conservative that some time ago some gas discharge stabilitrons were produced. I.e. the circuit with the gas discharge can be approximated with a circuit having a stabilitron with a reference voltage of Egas*Delta:

                  o                  o
                  !                  !
                  !                 ---
               -------              ---
               =======               !  Ust=Egas*Delta
                                     +---L<|---+
               =======                         !
               -------               +---------+
                  !                  !
                  !                 --- 
                  o                 ---
                                     !
                                     o

The electric energy deposited in the gas (on the stabilitron) is:

W=S*q*Ust=S*q*Egas*Delta

Where S - is the area of the electrodes (area of the discharge cross section) and Delta is the value of spacing between the laser electrodes.
The energy specific to the unit of volume is:

w=W/(S*Delta)=q*Egas ; w=8.85e-12*e*Ed*Egas

As we can see all the technical parameters (area and height of the discharge) have gone again and we left with the electric parameters of the dielectric and the gas only.

For q = 0.18 mcCl/sq.cm, and Egas=30 kV/cm (the discharge in tha air between mylar insulated electrodes): w=5.4 mJ/cub.cm=5.4 J/liter.

Most TEA lasers have thethreshold of lasing at 50..70 J/liter. And from the formulas above it can be clearly seen that neither voltage variations nor dielectric thickness changes can help to improve the situation. Only other dielectric type can help. Such as barium titanate, that is used in the ceramic capacitors. But it is the very other story.

Thinking on the obtained formulas gives another curious conclusion. One should remember that the Ed decreases with the dielectric thickness increasement. If we have something about 100 kV/mm for the thin mylar foil, it will drop to somewhere near 30..50 kV/mm when we use the mylar with the thickness of about 1 cm. It means that the capacitive laser should be kinda low voltage with thin dielectric rather than high voltage with the thick one.

The said above is true for the case of single charge. And if we can recharge the capacities N times during the necessary time, the energy deposition will grow N times. The value of N is understandable: if we want the energy deposition to be at the level of "good" lasers then N =200J/l / 5J/l =40.

And how long can be is that "necessary time"? The energy must be put in the gas before the carbon dioxide will loose it. In other words the energy deposition time must be less than the lifetime of CO2 molecula at the upper laser level. The lifetime of CO2 having been mixed with air and at athmospheric pressure is kinda hard to find in literature. The very rough estimations give something like 1 mcs. I.e. one should be able to recharge the laser electrodes 40 times during 1 mcs of time. Thus the feeding frequency will be 10 MHz (4 recharges per full period of oscillations). If we orient to the feeding energy of 1 J it will mean the power of the HV generator of 1 J/1mcs=1 MW. Its a dead end. The results of attempts to create 1 MW rated HV HF power supply can be seen here

Apparently its unreal to make the capacitive discharge laser based on the mylar dielectric. However the previous test laser (the sparky) has shown that things can be not so bad. It was operational when fed by the capacitors that just cannot discharge faster than in 4 mcs. So the estimation of the necessary energy deposition time can be corrected to the better direction.

One way or another, but the capacitive laser was created.

capas1 capas_in1

Here is the scheme of its design.

capacitive1.GIF
capacitive2.GIF

Its body is the old good plastic sewer pipe, 2 inch in diameter. The laser electrodes are made of the pieces of another platic tubing. This one has 20 mm outer diameter. The conductive part of the electrodes (the plates) is formed by sticky aluminium tape. And the barrier layer is formed by 120-mcm mylar foil. The woring length of the electrodes is 100 mm. The full length of the laser (mirror-to-mirror) is 205 mm. The mirrors are just the same as were in the previous and the pre-previous lasers.
Here is the connection schematics:

                  +--/\/\/---+
                  ! Cpi      ! +-/\/\/\--+
   +---||---+     +----||----+ !         !
   !        !     !          ! !         !
   !        !     !        +---+-||-+----+
   !        !     !        ! !      !
   +-->  <--+-----+------o)   (o----+
   !   SG   !     !        ! !      !
   !        !     !        +---+-||-+----+
   !        !     !          ! !         !
   !        !     +----||----+ !         !
  --- Cs    )     ! Rpi      ! +-/\/\/\--+
  ---       )     +--/\/\/---+           !
   !        ) Ll                         !
   !        )                            !
   !        !                            !
   !        !                            !
   !        !                            !
   +--------+----------------------------+

The preionizers here are in 2 mm from the mylar surface. Two wires per electrode. The connection is "cross" type, i.e. the wire, that is near the electrode of one polarity, is connected to the opposite polarity terminal (through the RpiCpi circuit though).

Cpi=23 pf, Rpi=100 kOhm
L1 is wound on a cut of 50 mm plastic tubing by an enameled copper wire having 2 mm in diameter. The best results were obtained when L1=1..2 turns.
Cs was always 4 nf.
The most amazing thing is that it lased. Despite all doubts and fears.

The lasing is very weak and with low probability (one lasing per 10..30 pulses) so i can not measure its energy. But visually (by the brightness of the plasma spot when focused on a piece of carbon paper) i can estimate it as 300 or maybe 500 microjoules. Also need to say that it works only with N2:CO2 mixture. Still no lasing was with air:CO2. But at least it lases.

 

4.12.2014

Here are a few results of attempts to optimize the preionizing circuit
(the main spark gap SG was set to 10 mm)

Rpi Cpi probability of lasing
100k 23pf <10%
5k6 23pf >10%
5k6 34pf >80%
5k6 68pf ~50%

Note that the right column contains neither power nor the energy of lasing. Its just a probability, i.e. the number of pulses with lasing divided by the total number of pulses. For the smallest values the probability itself tends to vary.

The main result is that with Rpi=5.6kOhm Cpi=34pf it lases almost everytime. The output power however can change seriously from pulse to pulse. With Rpi=5.6kOhm Cpi=34pf the laser got a run on a CO2:air=1:1 mixture. The lasing probability was about 10%. I.e. similar to the behaviour with N2:Co2 mixture with unoptimized preionizer subcircuit.

Yes this is neither my first attempt to make a capacitive laser based on mylar dielectric, nor the first successfull. However prevous ones vere operational only at the reduced pressure, when the gain is higher at the same level of the energy deposition. Somewhat unexpected was the fact that such a short laser did lase after numerous previous ones fefused to operate even when being 1.5 meter long.

The key to the success was the usage of the preionization. And it is especially strange when taking into account that the capacitive type of laser never had any problems with the ignition of diffuse type of the discharge. Apparently the preionization affects the laser operation by much more complex way than only by helping to make the discharge to be diffuse.

Take for example the reduced pressure nitrogen laser. The pressure is low. The times are in nanosecond range. No problems with volumetic discharge ignition. However the preionizations is more than welcome here. I can refer to the excellent work of Jon Singer [5], or can share my own experience: a reduced pressure nitrogen laser having the air as the lasing medium and equipped with a preionizer of medium strength yields MORE energy than the similar laser with pure nitrogen but without any preionizer.

Another example is carborundum (see the beginning of this web-report). It does not lase without preionizer, though the volumetic discharge can be obtained here without the preionizer. On the other hand even with the preionizer there still are some sparks and streamers.

The third thing is the capasitive laser being discussed here. In order for it to lase it appeared to be enough to add some preionizer. Moreover, when being visually observed, the discharge with preionizer looks worse than
without it: there are more streamers and ununiformities.

One time is contingacy, two times - coincidence, three times is the regularity. In three different types of lasers the usage of the preionization did either enlarge the output or even helped to make working laser from inoperational one.

What is it related to? The first thing that comes to the mind is the increasement of the discharge formation speed and lowering the energy deposition time. It might be sufficient for nitrogen lasers or to some extent for the capacitive ones (just imagine if the first half period of the oscillations comes without the discharge - half of energy goes to trash if we remember that the oscillations are decaying.)

However the last two lasers, having been described here, show that the speed is not so important. The gain and the output could be high even with the energy depodition time of several microseconds.

The next is increasement of the discharge uniformity. As they say, any kind of volumetic discharge, even the most visually uniform one, consists of microstreamers, nanostreamers, picostreamers and so on to the level of separate moleculas. And the preionization will increase the uniformity just until the discharge will fill all the volume with the precision to separate moleculas. Can it work or not? In principle if the energy density in the streamers is above the saturation level, then the average gain will grow with the uniformity increasement. (Such a situation can take place e.g. when all the CO2 moleculas in the streamer are excited and there's nothing to excite further.) If the energy density is well below the saturation, all the values can be added linearly and the gain will depend only to the average energy deposition and wont depend to the details of its distribution. Most of my lasers have the energy density by one or two orders of magnitude lower than the level of saturation. Can the current in the streamers be one or two orders of magnitude higher than the average one? Who knows... Maybe it can, however it looks doubtfull since in that case the streamers would tend to become sparks.

Another reason could be the change of electron temperature when changing the discharge mode. However the preionization affects only the beginning stage of the discharge. So the influence should probably be small.

One can device many other mechanisms and without deep research and precise measurements it always will be hard to say which of them works and why. So I put this inference here only to illustrate how complex things can be rather than to say what exactly goes on.

8.12.2014

A new sample of the capacitive laser was assembled. Th working length of the electrodes was increased to 22 cm. The total length of the laser 30 cm. The preionization wires became fed from both ends. Since there are four wires and each of them has two ends, there are eight preionization feeding circuits Rpi Cpi now. Cpi=34pf, Rpi=100 kOhm. The mirrors are taken from the previous laser. Cs=6nf. L1 was also taken from the previous design.

capas_v2

The laser works much better than its shorter variant. The probability of lasing rised up to 80% at 1 Hz with air based gas mixture. It was even possible to measure the output. It was 2.5 mJ with air:CO2=1:1 and 4.5 mJ with nitrogen:CO2=1:1.

Unlikely the shorter one, this sample of the laser behaved positively when the storage capacity Cs was increased from 4 nf to 6 nf. Both the stability and the energy output were increased.

Here are the laser spot traces on a thermal sensitive paper for faxes.
(Was focused by a ZnSe lense with f=50 mm):

termpaper termpaper_dots

One can see that the spot consists of two independent elliptical parts. The parts are placed at the sides of preionizers. It makes me to think that the preionization is too weak to ionize the center of the tube properly. I.e. there appeared two independent lasers inside. The attempt to incrase the distance between the preionizer wires (with keeping constant the distance between wire and electrode) has realy lead to the drop of output and probability of lasing. So one should probably better decrease that distance. However theres no limit for perfection. I already have the idea how tho make this laser powerfull. Im too lazy to do the detaile draw so try to perceive from the text. The essense is to increase the area of the electrodes without increasing their radii (the latter would increase the width of the discharge column and it is completely unnecessary here). Along all the increased area one should place the similar preionization wires. Their action is not only to preionize, but also to ignite the discharge in places where the electrodes are far from each other. The electric charge from all this area will flow through the plasma and feed the main discharge column:

     +-------------------------------+
     !                               !
     !              +----------------)------+
     !              !                !      !
     !  +--+--+--+--+    +--+--+--+--+      !
     !  !  !  !  !  !    !  !  !  !  !      !
     !  -  -  -  -  -    -  -  -  -  -      !
     !  -  -  -  -  -    -  -  -  -  -      !
     !  !  !  !  !  !    !  !  !  !  !      !
     !  o  o  o  o  o    o  o  o  o  o      !
     ---------------      -------------------
 +                  )    (                     -
  ------------------      ----------------------
     !  o  o  o  o  o    o  o  o  o  o      !
     !  !  !  !  !  !    !  !  !  !  !      !
     !  -  -  -  -  -    -  -  -  -  -      !
     !  -  -  -  -  -    -  -  -  -  -      !
     !  !  !  !  !  !    !  !  !  !  !      !
     !  +--+--+--+--+    +--+--+--+--+      !
     !              !                !      !
     !              +----------------)------+
     !                               !
     +-------------------------------+

Kinda that. Mylar on the electrodes is not shown. The resistors in parallel to the capacitors arent shown too.

Sorry, but I have no resources to make a sample of this to check the principle. This one will definitely not suit a plastic pipe of a reasonable diameter. And in the limits of visibility there is no plastic box tight and strong enough to endure the pressure. I should note here that I had some attempts to make a CO2 laser inside a body being not vacuum-tight but kinda "gas-limited". None of that were successfull. Its understandable after the
success with filling a laser by a vacuumless method (by gas washing). The required washing ratio of 30..60 can easily be provided only for very small to small lasers. For the medium to large ones the available volume of the gas mixture is just not enough.

Thats all for this type of laser for now.

THE MAIN CONCLUSIONS:

  • the operational laser based on the barrier discharge between mylar insulated electrodes was successfully designed and tested.
  • the laser is operational with CO2 gas mixtures based on the air and on technical grade nitrogen. The output power has measurable level.
  • the ways how to make the laser to be powerfull are clear.

One more notice. If You want to reproduce such a laser, it is very desirable to get a dielectric coated mirror having 95% reflectance. The laser will definitely lase inside a resonator consisting of aluminium mirrors only (my measurements say that my, not very good, mirrors have reflectance from 92% to 96% at the wave of CO2 laser), but the part of radiation coming through the coupling hole will be so little that it will be hard to comprehend that the laser does work at all.

 

19.01.2015

A capacitor laser was built, based upon barium titanate as the dielectric. Some capacitors 2000 pf x 30 kv were taken (those were China-made so "30KV" was just an advertisment. Even their design does not allow more than 15 kv when being used at open air <15 mm of the distance between wire leads > and even when put in oil those 30 kv look doubtfull).

First of all it was needed to get rid of the plastic housings of the capacitors. Three weeks bathing in acetone helped it out.
ceramic_01 ceramic_02 ceramic_03 ceramic_04
The ceramic pill inside the capacitors were 6 mm thickand had 22 mm diameter.

It seemed uneasy to erect a laser with the plasma column 20 mm wide, so I desided to cut the ceramic pills into pairs of halves. It was done by hand engraver tool with a diamond cutoff disc. Then the obtained semicircles were put along a straight line to form an elongated shape:

      --
  |)   ^
  (|   !
  |) ~150 mm
  (|   !
  |)   V
      ---
 >--< ~10 mm

A pair of such electrodes was installed into a laser with a Lamberton-Pearson type of preionizer (i.e. with wires strectched along the electrodes). The interelectrode spacing was set to 4 mm/ The laser was installed onto a usual (low inductive) frame having storage capacity of 4 nf (2 of muratas). Here was no success in obtaining a volumetic discharge. Neither with air:CO2 mixture nor with N2:CO2 one. There was multitude of sparks but no violet glow.

The electrodes were then moved into a laser with spark type preionizer (being referred here like "classic scheme"). The design of the laser was a bit different from the described above, but the main parameters (like distance to preionizers or their capacity) were kept untouched. There appeared difficulties with a sliding spark from electrode keeper, along the ceramic surface and to preionizer pins. (One should keep in mind that the sparks along the surface of something with high dielectric constant can easily become very very long.)

So the electrodes were put into epoxy compound (its obvious, that the facial surface was kept clean and open).
ceramic_electrode
This helped to surpress the shortening breakdowns, but the volumetic dischagre still was absent at athmospheric pressure. Neither with air nor with nitrogen. With the epoxy protected electrodes the electric strength of the laser rised to the extent that allowed tests with reduced pressure.

At 900 mbar (90% of the normal air pressure) a strong voled glow appeared between the electrodes. No the rain of sparks didnt disappear. But at the background of it there was a strong volumetic discharge.

After installing into a resonator and proper alignment the lasing was obtained. However despite all expectations it was weak and unstable. The energy was ~ 1 mJ, and the probability of lasing was ~10%. Judging from the comparatively high output energy the instability was due to discharge instability rather than due to the proximity of the lasing threshold.

CONCLUSION: the distributed impedance of the electrodes (their effective resistance) created by the specific capacity of 526 pf/sq.cm. has appeared to be too small to stabilize the discharge. The surface capacity was increased by 30 times in comparison with mylar, but it brought not only the rise of the energy deposition, but also the rise of plasma instability.

If one wants to advanse in this direction I would advise to increase the ceramic thickness by three times (to 15..20 mm). However it will possibly require profiling of the electrodes.

This laser appears to be kinda expensive and the result is "kinda disappointing".

 


  1. P. Pace, P. Mathieu, J. Cruickshank. Miniature, sealed TEA-CO2 lasers with integral semiconductive preionization. Rev. Sci. Instrum. 53 (1982) No. 12, 1861-1863
  2. D. Beaupere, G. Helias, A. Bettinger. TEA CO2 laser with a SiC anode. Rev. Sci. Instrum. 56 (1985), No. 12, 2251-2258
  3. A.I. Pavlovsky, M.A.Voinov V.V.Gorohov et.al. Transversal spatial structure of high voltage diffusive discharges. Journal of Technical Physics, Vol.60,N 1, 1990, pp 64-71
  4. V.Y. Baranov, V.M. Bporisov, Y.B. Krokhin et. al. Free Running Stimulated Emission from an Electric Discharge CO2 Laser in the Nanosecond Range of the Light Pulse Duration. Sov. J. of Quantum Electronics, v5, N5, 1978, pp 1141-1143
  5. Jon Singer. TJIIRRS: Number 5C [New] of an Ongoing Series; “Theorie und Praxis IIA”: Revamping the “DKDIY” Laser Part 1.

 

<< HOME PAGE