POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

As in the title ..... in the past, i already experimented with TEC's, with the idea to get some decent (and cheap) sensor for laser power metes, without have to use thermopiles ..... but never got too much good results, cause, mainly, the old type TEC's was too big, and the thermal conductivity of a TEC unit is intrinsecally too high ..... then i got "real" ones, and left the idea vanish.

Recently, i got a pair of working TEC plates from Heruur, mounted on TO3 style bases, and my DIY maniac part woked up and got the feeling to restart some experiment.

First trial, just painted AR black the surface of a 10x10mm one, and stuck it on a heatsink (first pic) ..... stabilization time is still too high for a fast response meter, cause it still take approximatively 15 seconds for stabilize (old ones was around 30 seconds), but the output is a decent amount, also if the sensitivity is not extremely high (it start to read only over 14 / 15 mW) ..... anyway, as you can see in the second pic, it give me almost 7 mV for a 50 mW module (emitting 44 mW), so a simple op amp can take care of it.

But doing different tests, i noticed that it's too much sensitive to air temperature, and that, also with all that heatsink mass, it transfer too much to the "cold" face, deriving too much for long or repeated reading (the reading in the pic is after 20 cycles of 30 seconds from the 44mW module, where at the first reading it was 7,6mV) ..... i also tried using a beamsplitter and the LPM, in conjunction with it, and the 150mW module on the heatsink, with half of the beam on the LPM and half on the TEC ..... keeping the beam for 120 seconds on, after thermal stabilization, and taking readings on LPM each 10 seconds, the LPM was deriving just 4%, where the TEC was deriving almost 30% ..... that was not completely unexpected, cause, after all, TECs are extremely efficent heat transfer plates, when not powered ..... but was anyway more of the expected.

Then i had a crazy idea ..... took my last 8x8mm TEC plate, and soldered it over the 10x10mm TEC that is on the TO3 plate (was both gold plated, and i used some indium alloy, for not risk to melt the junction connections, but was a hell of work the same ) ..... and also soldered a smd NTC chip on the plate of the bigger TEC (pic 3).

My idea is to use the lower tec for keep the "cold" side of the upper one stabilized at a fixed temperature (was planning 18 C or similar, for not have to use too much power) using the NTC as sensor and feedback for a PWM circuit, and enclose the assembly in something that repair it from air currents ..... at this point, the upper TEC must be much more precise and stable in reading

Still developing the circuit and finishing the painting of the "reading face", anyway, but an on-the-fly test, gave me some encouraging result ..... after thermal stabilization at 20 C, as example, the upper TEC was starting to read something with a 9mW module, took 11 seconds for stabilize with the 44mW one, and 14 seconds for stabilize with the 144mW one ..... and remained "almost" stable also after a pair of minutes (almost, cause the driving circuit for the lower TEC is still not perfect, so it have a bit of delay in reaction and a little derive, almost 3% in temperature ..... still experimenting, anyway)

Well, i'm working on it only in the free time, so don't expect all the schematics ready and published for tomorrow, nor keep the breath waiting ..... but, seeing that there was some encouraging results, i've thought to share this, at this point ..... when i get better results, i update this one, and who know, maybe in the meantime also others can have some good ideas in the same direction and share them, too

BTW, i've seen that heruur have on ebay some very small TECs, like 6x6mm ..... ofcourse, using small ones, there are less junctions, so there's also less output voltage, but i'm wondering about what the response time of a plate so small can be, compared to my 8x8mm one (after all, 8x8 is 64 square millimeters, where 6x6mm are just 36 square millimeters, and also if the ceramic plates are the same thickness, the thermal mass must be almost the half )

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

You, sir, are an Italian god of science and awesomeness.

One thing, though. If after some work you decide to rise the voltage with an op.amp., don't you think it'll deviate the readings even more?

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

absolutely awesome!

this easily could result in the best hobbyist-lpm, and be better than the average pro lpm for thousands of dollars too!
once you built a working prototype, assembling a handful of those wouldnt be much work, would it?

awesome.. /subscribed

manuel

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

Take a guess at how the laserbee works

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

I believe SenKat was also planning to do something like this, I remember receiving a TEC with 1 surface painted with special black paint.

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

MarioMaster said:

Take a guess at how the laserbee works

Click to expand...

really? i never saw or read about any lpm with thermally stabilized backside of the sensor-peltier.. my die4thing from robin is just a tec on a heatsink. from pics i guessed the average industrial head is the same too. i assume most heads will be like this?

manuel

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

nikokapo said:

One thing, though. If after some work you decide to rise the voltage with an op.amp., don't you think it'll deviate the readings even more?

Click to expand...

Is basically for this reason that i'm still developing a decent circuit ..... i want to build, first, something that can keep the "cold" surface of the small TEC always at the temperature that i've set (and about this, i just realized that i need a double system ..... must also be able to HEAT it if it go too low, not just freeze it).

Once this is done, the derive of the TEC itself must be low enough for not influence the output, so a decent op amp stage must be enough for have a stable reading ..... the alternative, is to make an op amp stage with a variable gain, with a feedback from a second NTC, so, it can self-compensate the thermal derive from the 2 faces of the TEC, but i left this possibility for last, cause is much more complex to achieve the exact compensation in this way ..... after all, my idea is to make something that anyone can reply as DIY, if wanted, so i think that i have to try to keep it the more easy possible.

Krutz said:

once you built a working prototype, assembling a handful of those wouldnt be much work, would it?

Click to expand...

Uh, never thought about a commercial line, if this is that what you mean, LOL ..... plus, i'm in Italy ..... shipping around internationally something from here is a damn pain you-know-where, in fact of shipping fees.

MarioMaster said:

Take a guess at how the laserbee works

Click to expand...

Laserbee use a sorta of ceramic "thermopile" (and, unfortunately, the last time i've seen these types of thermopiles in a shop, was 20 years ago ..... guess they are not more in production) ..... thermopiles are more efficent and more precise and sensitive than TEC, at least if you don't find a company that build for you a custom design TEC, but look below for the costs .

BTW, i don't have one for examine it, so, i may be in error, but if that is a TEC, from the images, is for sure the more thin TEC that i've ever seen :wtf:

Anyway, personally i think that laserbee is a great alternative to professional units, especially for the data logging capability (i have a pair of LPMs, and none of them have this function, also if i've built an external data logging unit that go connected to the output of a normal LPM, for this ..... and if i had not those already at home, probably it was my choice, too)

Things said:

I believe SenKat was also planning to do something like this, I remember receiving a TEC with 1 surface painted with special black paint.

Click to expand...

Senkat ? (i need to use search button, perhaps ) ..... do you know if they (or other sellers) have small TECs, like, 5 or 6 mm of size, at decent prices ?

I searched for thin and small TECs, but nothing to do in the "normal" market, and the only company that replied me positively, said me "Yes, we can made custom TECs for you, minimum surface is 30 square mm, shape at custom design, minimum thickness is 1 mm (hell this btw mean 1/20 of the thermal mass of a common TEC), minimum quantity is 5000 pieces (AAAAAAARGH !!) and sampling price for 5000 pieces will be 15.000 $, plus taxes" ..... YIKE !

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

There really is no need for 2 TECs. What you are measuring is heatflow through one TEC, and this does not depend on the temperature of its cold side all that much in this application. The laser will just cause a temperature -difference- between back and front, and as long as you dont loose a significant of heat by radiation or convection, this difference is the same regardless of the temperature of the cold side.

One thing to investigate upfront though: Is the voltage/current produced linear with laser power at all?

I suspect it will turn out not to be, and you will need corrections for it. A meter like the laserbee has a microprocessor in it that can easily do this correction as long as the curve is known.

Going fully DIY poses a problem of calibration here: you will need several lasers of known power to get the calibration curve right, and its exact values will depend on the precise construction of the sensor. This will make the whole unit difficult to reproduce for others.

I'm interested in any progress you make though, please keep the info coming

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

@ Benm: you're right about the linearity problem, it's the first thing that i want look after the thermal circuit problem ..... but, sorry, you're partially wrong about the thermal difference from the 2 sides (let me explain what i mean)

You probably know how a TEC work in the normal use (heat pump) ..... but, for being able to do this, junction cubes are made with material that is not just a semiconductor, but also have a very good thermal transfer index ..... and there's a lot of them

When the TEC is not powered, like in this case, it act like a single body ..... is true, it don't have the same thermal transfer index of metal or ceramic, but is almost half of it, and, trust me, it's more than enough ..... when you apply heat on a face, it transfer it through the junction cubes to the other at a too fast rate, for keep the difference of the temperature enough constant (this is the derive that i was mentioning before) ..... and cause the output is caused from the difference of temperature from the 2 faces (the seebeck effect), more heat is transferred to the "cold" face, less voltage you have at the output with the same power applied ..... is for this reason that i want to thermally stabilize the cold face.

One consideration, in fact IS enough to glue the cold face to a BIG heatsink (must be big, for keep its temperature stable) for quick and occasional reading, but there are some factors that you have to considerate also in this way ..... doing this, for have a precise and stable reading, you need to have a temperature sensor that detect the temperature of the whole assembly, and change the response curve of the amplifier, in accord with the ambient temperature .

Basically, it's true that if i do a measure from the same laser at 20 and at 30 degrees of ambient temperature, as example, the amount of energy is the same, and cause the plates are both at the same temperature, at the start of the reading, in an ideal assembly you must have the same reading ..... unfortunately, the real environment is all except ideal, and the efficence in heat dissipation of a heatsink change based to the ambient temperature, too, and not in linear way, so in a practical ambient, you never get the same results, at different ambient temperatures.

I want to try to obtain a system that is independent from the external temperature, or at least more independent possible, but keeping it at hobbyst level (those systems already exists, in precision instruments, but a hobbyst cannot throw away hundreds of $ just for get one of them )

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

Amazing project! I will be following this one closely for updates
+1 for awesomeness

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

HIMNL9 said:

Basically, it's true that if i do a measure from the same laser at 20 and at 30 degrees of ambient temperature, as example, the amount of energy is the same, and cause the plates are both at the same temperature, at the start of the reading, in an ideal assembly you must have the same reading ..... unfortunately, the real environment is all except ideal, and the efficence in heat dissipation of a heatsink change based to the ambient temperature, too, and not in linear way, so in a practical ambient, you never get the same results, at different ambient temperatures.

Click to expand...

I wonder how big the difference in reading will be - with the cold side just mounted to a very large heatsink, and measurements taken at 20 and 30 centigrade. I believe that it will not make much difference in the current produced at all. Perhaps there are practial factors that somehow do cause a difference, but that could be sorted later.

As far as i can see going with 'just a big heatsink' first would be the best way to explore viability of the setup. Once you get it to display a close value to the actual laser power over a decent range, you can then see how it reacts to different ambient temperatures.

I suspect it won't be a very linear relation ship, and if thats the case you will need a uC to do the conversion, at which point you could also just add a temperature sensor to the heatsink to determine ambient temperature and correct for it... still sounds easier than a setup with one TEC keeping the other ones cold side at a constant temperature

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

Laserbee does use TECs for his meters, I've seen links to the ones he used to use. So this certainly is a doable project. He just likes to call them "thermopiles" real thermopiles don't look like that. They have a large number of circular junctions mounted on the rear of the absorbtion plate.

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

^ I remember, like, 20 years ago, a company was selling sensors made like the ones that laserbee use, and they calls them "thermopile" ..... basically, was photodetectors, but made specifically for the "thermal" band of the IR, so they was used for measure radiant temperatures, more than light (as far as i remember, the indication from the maker was to paint the "reactive" surface with black matte paint, for take away the disturb caused from the visible light) ..... they was looking as little ceramic plates, half millimeter in thickness and 5 x 6 millimeters in sizes, with a side in white ceramic, and the other dark blue, like a photodiode, with 2 soldering pads on one of the borders, and the dark part covered with somethng like hard transparent paint.

I was using them for some hobby projects, but approximatively 20 years ago, they disappeared from any catalog and list, and from that time i was simply unable to find them anymore ..... anyway, i'm sure at least that these ones was not TECs at all, not in half millimeter of thickness, and not produced more than 20 years ago.

Anyway, i don't have one of the ones used in laserbe, so i cannot say if they are or not TECs ..... from the pictures of the LPM that i've seen around, they looks just too thin for be TECs, and i don't see the junctions on the borders, but there's no macro of the sensor in any place, so i cannot say any definitive yes or no, sorry.



@Benm, i'm not sure, is the reason for which i made also some test with a single TEC plate, cause i want to check both the alternatives and then decide what one is the better ..... as i said, probably for occasional readings, also just a tec glued on a good heatsink is enough ..... once you power up the meter and set the zero, just a pair of cycles of 30 seconds or similar, don't transfer enough heat to the heatsink for false the reading ..... is the change of the temperature in long times, or for long reading cycles (like, as example, if you want to test the stability of a diode in a heatsink for minutes or hours), that gave me that idea about thermal stabilization.

Anyway, your idea is good too, so i use the single TEC for made some experiments in the next days, about the stability and deviation, and post also these data (real life permitting ..... as i've sid, please don't hold the breath in waiting, ok ? )

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

Yes, that's true - we should have someone get some good macro shots of the sensor to be sure. If you can find some of these IR reading devices I'd appreciate it.

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

You also do realize that in order to get accurate results with a thermopile, you need the proper broadband coatings. While flat black paint may work accurately for some wavelengths, it may not for others.

Re: POSSIBLE (just possible) "cheap LPM" using small TEC's (first experiments)

@ MarioMaster: LOL, i've searched them like a desperate, when they took them out from the market ..... they was acting basically like thermopiles, but had a cost, like, less than half dollar ..... i searched them for years, everywhere, before give up.


@ digital_blue: I know, but we can't have professional paints at hobby level, and lamp black coating is so delicate that you can ruin it just looking it bad , so i had to experiment wih different paints, til i found a decent alternative ..... actually, the better solution that i've found is a black matte paint used for "exhaust tubes and silencers" in bikes motors ..... it's not the better solution, but i discovered a decent way for "take away" the most part of the reflection (it require 3 passes, and have success only 80% of the times, anyway it require some pics and explanations, so i try to do this with some images, if i reach, in the future)

BTW, anyone that have suggestions for a decent broadband adsorbent and antireflection black cover, is welcome to post it ..... as example, i discovered that some (but not all) black matte oxydation processes for aluminium, are less reflective of any black paint i've found, but have doubts that a similar process can be done on a ceramic surface