Ultra stable TEC controller as Arduino Shield

May 31, 2016

I didn’t have time for my hobby for a long time, but recently I got around restarting some lingering laser related projects. One thing I felt was needed is an update of my stable laser diode and TEC drivers for holography applications. See here for background. In short, one needs to avoid mode hops by having an exceedingly tight diode current and temperature control. We are talking about stability of the order of 1/1000 degree for extended periods of time. This is in particular critical for ECDL lasers.

This new project was partly motivated by requiring bi-directional drive (so cooling AND heating), so one does not need to run the laser below room temperature. This necessitates a H-bridge PWM drive mode, and this is already close to full digital control. The other motivation was the design of solarfire and dnar and others described in this forum. However that design is not stable enough and also I am not familiar with the microprocessor. So I decided to come up with a stability-improved design, based on the Arduino. It is dirt cheap (whole ready-to-use boards for less than 3 Euro incl shipping..) and there exists lots of useful software for it.

However in order to meet design goals, one has to make some efforts in the hardware and software design. Hardware means precision opamps, voltage references, and <=5ppm/C stable resistors. Fortunately I have a large stockpile of those.
Software means to bring inputs and PWM outputs to like 16bit resolution, which is far more than the usual 10bit ADC input and 8bit PWM output resolutions. Strictly speaking 16bit PWM output is easily possible but then the PWM frequency is a few kHz and this requires an enormous amount of filtering increasing size and also noise pollution. In fact I managed to cook up a software solution for the PWM driver that gives 16bit resolution while still running at 62.5khz! And input wise with suitable oversampling and "dithering" the resolution can increased to 16bit as well.
This avoids expensive 16bit ADC’s.

So all in all, here are the main features of the design that are working as of now:

-full H-bridge for bi-directional TEC drive
-up to +/-3A current
-temperature stability < 1/1000 degree for hours
-noise level about 2/10.000 C
-control via USB serial terminal, optionally LabVIEW
-LCD display
-automatic PID control loop configuration per pushbutton
-saving basic operating data in EEPROM
-ready interface for companion diode driver

So far only a development prototype exists, but I am about designing
PCBs which could be used as add-on shields for the Arduino Uno.

My plan later is to make an update for the LD driver too, giving another Arduino shield which can be stacked on top. This setup is supposed to be suitable for desktop laser diode controllers.

Finally a miniature version is planned with less power (1.5A), where both TEC and laser drivers are on one board and a stacked-on Arduino Nano is used for control. This setup is supposed to go into compact laser heads.

Here a picture of the current hardware setup (no comments please),
and further below the LabVIEW interface showing stability at a few 1/10.000 degrees.

Arduino PWM Proto

Website being moved

April 27, 2015

I was changing provider and so my web site of many years will have a new address:


Some links might not yet work but I will slowly correct it.

Some more diode tests

January 5, 2014

Finally I completed over the holidays a bunch of further diode tests with regard to single mode operation, mostly in ECDL setups. I had collected them over the years but didn’t have enough time to actually do it. Hopefully I’ll be more active again in the future.

In brief, the results are as follows:

  • Mitsubishi ML520G71 300mW/635nm, Opnext HL6388MG 250mW/640nm, Mitsubishi LPC-836 300mW/655nm diodes:
  • Opnext HL63603TG 120mW/638nm/3.8mm laser diode:
    very good ECDL to 60-70mW
  • Mitsubishi ML520G54 110mW/638nm laser diode:
    very good ECDL to 60-70mW
  • Osram PL520 50mW/520nm laser diode:
    good ECDL to 40mW
  • Further details can be found here. As a spoiler, here a brief movie (2.7MB) of a test of the green Osram, which shows stable single mode operation at around 40mW:

    Happy New Year!

It’s been some while….

June 13, 2013

since my last postings. Well one of the reasons was my professional work, and another was that my workhorse for measurements, my LeCroy 9314M scope, was broken. And my backup scope, a LeCroy 9450, was broken already from the day I rescued it from the scrapyard: the screen went dark after warmup.

So what to do – shelling out close to 1K for a decent 4-channel scope or trying a repair? This question kept me blocked for many months, till I kicked myself and started to see whether I can do something. Two of the 4 channels of the 9314M went bad already a few years before, and now the trigger stopped working, but sometimes it came back, apparently depending on the temperature. This is the worst kind of errors to have! I was afraid that one of the numerous custom ICs from LeCroy went bad, or that there would be some error in the logic board. Nevertheless I gave it a try, partly because the schematics are available in the service manual.

In fact, as silly as it may sound, looking for errors in a fantastically complicated circuitry has something to it. It is a bit like reading a criminal thriller…who is the culprit? How can we use logical reasoning combined with intuition and experience to corner it? And then, can we do something against it? Actually in the past I had successfully repaired quite a number of measuring equipment, from Tex to HP, and given up only on one thing, an intractable Schlumberger Stabilock 4040.

So I undertook this journey for a couple of evenings. First thing was to take everything apart and create extension cables for the front panel, otherwise one never would be able to come close to the live main board. And then finding the way through the SMD circuit with another scope, voltmeters, and most importantly, freeze spray and heat gun. Unfortunately the latter produced most of the time misleading and contradicting results, partly because there were two independent thermal errors. Finally the culprits were found, both very close to hot IC’s which again confirms the expectation that most likely a component fails due to thermal stress. The one responsible for dropping 2 of 4 channels was IC(*) A4/74HCT138 and the one for loosing trigger was diode(*) CR402/SM4004 which ran extremely hot, so that it had partly unsoldered itself, creating in intermittent contact (almost the worst thing to find…). Anyway, to make a long story short, after exchanging these parts the scope worked like new!

And since I was in the mood, I also took on the other scope, LC 9450. There again a thermal error, this time it was the mosfet(*) Q76/IRF830 responsible for emergency blanking the screen. It turned out to leak current at higher temperatures. After exchange all was well!

So now were things back to normal and I restarted activities by investigating a few new diodes for ECDL operation; among them the green 520nm Osram PL520. Stay tuned for results!

(*) refers to schematics in maintenance manuals


February 12, 2012

Holidays are a good time for upgrades, etc, and as always that involves much more work than anticipated. In short, upgrading to the Lion version of Mac OSX for all my computers had some severe fallout. Namely, LabVIEW 8.2 is not supposed to work under Lion, and after upgrading to LabVIEW 11 it turned out that the CIN (code interface node) construct is not any more supported… but that is needed as interface to the IOWarrior chip I am using for controlling my electronic measurement setup (via I2C bus and USB).

So, in effect I decided to scrap the IOWarrior hardware interface alltogether and switch to the Arduino Uno, which is very common, cheap, easy to program and has a lot of support, and in particular, LabVIEW and I2C support. Rewriting the hardware drivers for all my ADC/DAC chips kept me busy for a couple of days, thereby completely overthrowing other projects that I intended to do during the Christmas break… here a pic of my universal laser diode/TEC controller now working with the Arduino and LabVIEW11:

Of course, after all was done, it turned out that not only does LabVIEW 8.2 still work under Lion, but also that CINs still work under LabVEW 11.. well so it goes, but at least I have now a streamlined setup with some hope for future compatibility and extensibility.

Since I now got to appreciate the Arduino, I toy with the idea of building a new, completely Arduino-controlled laser diode controller and analyzer, with all sorts of fancy LCD displays etc… seems great fun ahead😉

As for other news on the electronics/instrumentation front, the friendly greek friends, whom I sold a few lasers, donated me a nice spectrometer (many thanks to them!) These small, fiber coupled and grating-based spectrometers are sold by Science-Surplus, but they come unaligned. Their site has notes for doing the alignment which are pretty clear to follow, but it still took me a day to get the spectrometer to work. I also extended their LabVIEW interface and it now looks like this:

For some more comments see here.

More diode tests

October 10, 2011

After a long break I found time to check a few diodes for single longitudinal mode operation.

1) Alleged new, cheap 635/638nm diodes on ebay. These were actually 658nm diodes and also not new. I returned them and so far am still waiting for a refund…..

2) Mitsubishi ML520G71 300mW/635nm laser diode. Unsuitable both in free-running and ECDL mode.

3) HL63133DG 170mW/638nm laser diode. Excellent in ECDL mode, so far the best diode! Clean stable output beyond 100mW can be achieved. See the pic at 635nm and around 100mW:

4) Update: The diode currently sold as Opnext HL45023TG on ebay turned out to be the PL450 from Osram. It has quite similar data, so most wouldn’t complain, but for purposes of single longitudinal mode operation there may be a major difference, potentially. Let’s hope for it, because the diode seems pretty useless for holography. I found it completely unsuitable in free running mode, and (after a lot of tinkering) quite OK in ECDL mode. Initial attempts produced less than than 20mW in single mode operation, but with some optimizations more than 40mW was achieved. And the beam quality is good, especially compared to the common 445nm/1W diodes which do not provide much more power in stable single mode operation.

Here a brief movie (2MB) showing single mode operation from the CCD spectrum analyzer at around 40mW (popping sounds indicate mode jumps):

Further Update: Now after all worked well I built everything “ready to go” into a nice case (see below), and for unknown reasons I can’t get any more stable SLM operation to more than 25-30mW. The only thing that was changed was the collimator, and it seems that the adjustment is extremely critical for stability. Indeed it turns out that the stability depends crucially on how far the focus point, or beam waist, is from the laser; changing it between 1m, 2m, and 5m makes a huge difference, and this amounts to very minor rotations of the collimator.

In order to more systematically investigate the effects of feedback, I made it variable by using a waveplate and a polarizing beam splitter cube; preliminary results show that indeed the feedback must be in a relatively narrow window, but this alone does not guarantee SLM operation…. more work is necessary! Here a pic of the first test setup:

From left to right: grating, beamsplitter, waveplate in rotation mount, diode mount.

Revitalizing the Coherent Sapphire 488-20

September 11, 2011

There was a moderately good ebay deal to be had and so I decided to bring a Coherent Sapphire 488-20 laser head plus driver board back from my trip to the US – along with 20kg of other laser goodies😉. This is an interesting, though not too powerful optically pumped semiconductor laser (OPSL). The board looked new but the head was used and sold as “probably not up to specs”. Indeed so, I measured 9.6mW instead of 20mW, but I figured out, by some reverse engineering, how to bring it up back to specs, ie, 20mW with clean stable single longitudinal-mode operation. Adding a housing for the driver board and a few analog and digital controls, et voila… system ready for holography! For details see here.

… been a busy fall

December 5, 2010

Well the title says it all, I have spent the last couple of months in developing/building/testing a new series of driver boards, and complete ECDLs. While the boards and red ECDLs run very well, the 445nm ECDLs still make headache for stability reasons; 40-60mW can be done pretty well but everything beyond that is hard to keep stable over extended periods of time.

As soon as I have found a good way to have the die cast aluminum cases painted, I will sell off a few lasers, to make room for new projects. Here some eye candy:

PS: Drivers are sold put, most lasers too. I am not sure
whether I will build more, as the effort is barely worthwhile ;-( Just in case, keep an eye on my blog!

General Technical Discussions

October 24, 2010

The original post was about the closure of the holography forum, which has now been replaced by the holoforum created by Ahmet.

I changed the title and topic of the thread, as various general and interesting comments were made that had nothing to do with the closure of the forum. I think in this way the information will be easier accessible.

Blue frenzy hits the holographer too…

June 12, 2010

There has been for the last 10 days or so an incredible frenzy over the 1W 445nm Nichia diodes that can be harvested from certain projectors; see the turmoil at the various light show and ballon popper forums. Well the holographer, being not particularly interested in using his lasers as lamps in discotheques, or in wielding 1W laser pointers in public, is more interested in the mode stucture of these little beasties. So I had to get a few of those, and today, within the hour of receiving it, I set it up and checked the mode structure with my CCD spectrum analyzer (due to lack of a 1 Amp capable driver I had first to mod my ol’ proven SDL800 such as to allow it to run blue diodes with higher forward voltage). And voila – without any systematic search, I found that the diode runs single longitudinal mode up to 55mW (with Lens-27 at 221mA and 15.4C) ! This is truly exciting, not the least because the diodes are transverse multimode and apparently have many emitters. Here is a quick sample overview:


From top to bottom the current was 221mA, 225mA, 300mA, and 400mA. I didn’t turn the diode up to more than 500mA where it was yielding 430mW (with non-optimal Lens-27), because single mode cannot be expected at high powers anyway. Note the beautiful single mode at 221mA!

PS1: I now was completing the first systematic landscape scan and the result is shown here. A sweet spot at 233mA and 14.9C yielded 64mW! I tested also an ECDL setup but so far the power wasn’t significantly higher, but this can probably be improved as well.

PS2: A first first test of an ECDL didn’t pan out quite well, there are many parameters into this game indeed and perhaps things can be improved. Indeed a test with better collimator, different grating and careful adjustment, more than 160200mW single longitudinal mode was achieved, but this was quite unstable. Stay tuned for more results.

In fact beam quality appears the main downside of these diodes and may be the deciding factor of their suitability for holography. Firsts tests of shooting through a spatial filter were the opposite of exciting…. probably, for a holography application, best would be to knife-edge split up the beam and use the pieces as object and reference beams.

PS3: Right now I was spending time in designing a higher power version of my super stable driver boards, which would be optimized for ECDLs using the 445nm diodes. Since I was running out of the HY5600 TEC controllers I was forced to cook up a completely new circuit, this time however PWM based and with readily available parts. It runs at least as well as the old circuit.

PS4: I now verified reports here and here that even a simple glass plate can give sufficient feedback in order to provoke stable single mode operation; indeed the results look promising, see here for details.

Below a pic of a test setup with glass plate feedback:

PS5: After a couple of weeks where I have spend quite some effort to try various different configurations for glass plate feedback, the following scheme seems to emerge. Within limits the amount of feedback does not matter much, what changes is how much the threshold is current is lowered. But with a reduced threshold also the maximal current for stable or single mode operation is reduced. It turns out that no matter what I tried, the maximal power for stable single mode operation stays always at about 60-80mW. This is not much different from the free running diode (except that the proportion of single mode zone modes is higher and the general stability is much better).

Also with grating feedback the generic situation is not much different, but there are lucky expections. But even there higher power operation is quite a gamble.

So here a brief summary: single mode operation with the blue Nichia diode is generically possible up to 60-80mW, quite independently of the setup. With sacrifices on stability and mode purity, more than 200mW are possible, also relatively independent of the feedback. However, there is then a tendency for several supermodes to lase simultaneusly. For an extended summary see here.