DIY Thermal LPM for under $50

lasersbee · Aug 14, 2011

KreAture said:
Do you really believe that?
Wasting time laying out a circuit from scratch is silly.

Oh really.... Designing a PCB Layout from scratch is a silly
waste of time....:thinking:

We design PCB layouts all the time for ourselves and our
clients...
Perhaps it's a silly waste of time for you because you don't
have the skills to pull it off...

KreAture said:
Everyone isn't in the us...
Anyways, The known wattage may vary just as much as the thermal interface to a heater. I like to use thermal silicone pads at work for such situations. They are very stable and do not leave any residue. They are also verry pliable and one simply cuts them to size.
http://www.stockwell.com/data_sheets/thermal/thermal_mgmt_products.pdf

You also have to remember that you are applying a specific ammount of heat to the hot side, not a specific temperature. Tis makes the C/w rating less important and the dT in the junction doesn't matter. What does matter is that you insulate the top and sides so as not to leak energy that way. You want all your energy to build up in the resistor and transfer through the TEC. The temperature will stabelize based on the c/w of the boundry and the base-conductance of the TEC together with it's load. Higher load on the TEC will mean a lower c/w for it and thus also for the entire stack. (Good to know if stuff gets too hot; Just increase the load on the TEC to cool... We've used it at work but somehow I can't find the numbers for how much of a difference it had on c/w. I'll see if I can find it later.)

Oh, and with my line of work I could just order up a 50W laser from one of our suppliers but the prices I wouldn't dream of it.

I'd like to see that massive insulsation when you are done....

Don't forget also through the Sil-Pad....

Now you are talking about applying power to the TEC for cooling..
Take it from someone that knows just a bit about LPMs... that's
not the way to use it for a Sensor....

Very confusing on this end....:cryyy:

Jerry

http://laserpointerforums.com/attac...2594t-diy-thermal-lpm-under-50-mmeasylpm1.jpg

random person · Aug 14, 2011

lasersbee said:
Oh really.... Designing a PCB Layout from scratch is a silly
waste of time....:thinking:

We design PCB layouts all the time for ourselves and our
clients...
Perhaps it's a silly waste of time for you because you don't
have the skills to pull it off...

I'd agree with laserbee.... we have people on here that have a very limited knowledge of electronics and are able to pull off successful and productive PCB builds. Its very valuable for the experience as well as the end product. Designing a PCB from scratch is cake and cheap(with respect to other things that this hobby involves). Only disadvantage is that it takes 3 weeks which is too long for normal businesses that do rapid pace designs. For us hobbyists, is no problem.

As for using a silpad... your till relying on contact between the silpad and the LPM for heat conduction. In addition, heat will be lossed in that transfer process. I'd assume that would work worse then a grease or a adhesive. However, it would most certainly be something to try.

KreAture · Aug 15, 2011

Laserbee don't take my sentences out of context.
I am not a native english speaker and if you read on you see I state that it is not always fun to "redo something already done.". That was what I ment. When you just want to tweak a bit, you don't want to start over.

Also, I am in the same line of work as you. We develop for regular customers and military/healthcare on a daily basis.

As for the cooling, I do NOT suggest applying power. In fact I suggest drawing power.
The reason you can use the TEC as a sensor is it is acting like a generator. Albeit only 4% efficient but still. There is a small effect when you load it, where the c/w of the element changes. This allows you to apply more power on hot side without it overheating. The effect is small though, atleast at the powerlevels we use it for at work.

Now that you mention it, I like your idea though. By cooling hot side to ambient you can measure how much power you need on the TEC to do that. Since there is no loss or influx from ambient to ambient the input must come from the laser. One might need very accurate datasheets or measurements to make such a system work though, but it would allow higher powers.

Random Person:
There won't be any actual loss of power through the silpad. In fact it doesn't matter what C/w you have as long as you insulate the stack. Some thin sheets of foam or neoprene may work. You apply a certain power and this has to flow somewhere. Temp will increase untill the flow is stabelized, any thermal simulator will show you that.

Also, I just love putting microcontrollers in stuff. The dvm/tec solution seems to work very well, but it doesn't contain any avr's making me try and find an excuse to put one in. I can't help it. It's what I do...

qumefox · Aug 15, 2011

Where are you getting 'powering the tec' from? None of the schematics here do that. The tec is merely the signal input to the opamp amplifier circuit.

And here's proof the circuit works. Though granted, this is a bit more advanced than the simple non-inverting amplifier mentioned.. Mainly because i'm going about compensating for the voltage offset in the op-amp the proper way.. which involves using negative voltage (what the 555 circuit on the other end of the breadboard is doing) applied properly to the op-amp inputs and not screwing with adding voltage to the ground. The fact i'm also having to impedance match the circuit so the ophir head will work properly also necessitated using both sides of the 358. But it's still at heart a simple non-inverting op-amp amplifier circuit.

KreAture · Aug 15, 2011

qumefox who are you referring to?
Nice breadboard btw. I also like your videogen. If you don't need it I can provide you with my address...

I found my TECs today. They are 15x15mm 3.4A 3.8V elements capable of pumping 8.1w and has a dTmax of 74 and a Rac of 1.03 Ohm.
Supplied with 3.67w heat (well most of it) and experiencing a cold side of 29.6c and hot side of 41.4c giving a dT of 11.8 they produce about 262mV with approx 3.2 c/w. To get some numbers on the change in c/w with heat applied I shorted/loaded it with a 1.1 Ohm resistor and let it settle. It then measured Tcold=28.4c and Thot=36.8c making dt=8.4c and giving a 2.3 c/w. Power in remains approx the same give or take a 2'nd decimal.
It now generated only 13.5 mV though.
A 2.2 Ohm resistor gave me Tcold=28.3c Thot=37.5c dT=9.2c P=3.55w c/w=2.59 and Vout=143mV.

Just thought it might be interesting with some numbers on what I was gabbing about.
The wuestion is though; Will the element still behave linearly if one loads it a bit to gain some coolness on hot side?

lasersbee · Aug 15, 2011

KreAture said:
qumefox who are you referring to?
Nice breadboard btw. I also like your videogen. If you don't need it I can provide you with my address...

I found my TECs today. They are 15x15mm 3.4A 3.8V elements capable of pumping 8.1w and has a dTmax of 74 and a Rac of 1.03 Ohm.
Supplied with 3.67w heat (well most of it) and experiencing a cold side of 29.6c and hot side of 41.4c giving a dT of 11.8 they produce about 262mV with approx 3.2 c/w. To get some numbers on the change in c/w with heat applied I shorted/loaded it with a 1.1 Ohm resistor and let it settle. It then measured Tcold=28.4c and Thot=36.8c making dt=8.4c and giving a 2.3 c/w. Power in remains approx the same give or take a 2'nd decimal.
It now generated only 13.5 mV though.
A 2.2 Ohm resistor gave me Tcold=28.3c Thot=37.5c dT=9.2c P=3.55w c/w=2.59 and Vout=143mV.

Just thought it might be interesting with some numbers on what I was gabbing about.
The wuestion is though; Will the element still behave linearly if one loads it a bit to gain some coolness on hot side?

So how much current does your whole circuit draw
from what voltage power supply...:thinking:

Jerry

You can contact us at any time on our Website: J.BAUER Electronics

KreAture · Aug 15, 2011

lasersbee said:
So how much current does your whole circuit draw
from what voltage power supply...:thinking:
Jerry

What circuit? These measurements were done with a lab setup and using a Fluke thermal-cam to get average hotside and coldside temps. (Quicker than finding the hot probed blocks, or making some.)

I am only supplying power to the heater to simulate heavy power input.
(And supplying 12v to my cold-side fan for consistent readings.)

What I wanted to show, was the load on the peltier by "shorting" it with a resistor, although dropping the sensitivity will also cool the hot side.
I think the effect is related to the seebeck effect but I can't really explain the physics behind it. It is real though.

On my comment of actually cooling the hot side (making it in essence the cold side but only too room temp): That would use power. In fact, it would require about 30% more power than the input laser power due to less than 100% efficiency in the TEC. I guess it wouldn't work well either as it appears it's not the actual laser-power that is the problem, but the fact that it has to be absorbed by a thin layer on the surface of the TEC and lasers are good at burning stuff... Fun thought experiment though.

On a different (slightly) note: I did test the TEC with blackened surface as a detector today. Defocusing the beam to be about 90% of the width of the TEC and placed the TEC under it with it's HSF cooling the back-side.
I used a diode from a SH-S223Q Super Writemaster 22x DVD-RW drive (silly diode that needed lots of work to fit a normal 5.6mm host) and ran it at a few fixed resistor-values. Here's the raw data:

Code:

Id(mA)  Vtec(mV)
139     2 (lasing error/flicker)
192     12
227     16.3
288     17.2
312     16.7
340     17.7 (flickering again)

I used a standard linear reg setup as constant-current.
Looks like I should have used a few more values between 139 and 227, but based on the curve I think the diode peaks there? From what I read here on the forums theoutput goes down again after you reach max out and from there it's not far to ruin. I guess I better stay below 300 mA?

The high current needed for stable lasing was odd to me too, but I firt noticed it when testing the laser the first time. I couldn't get it to light stabely at 50 mA. In fact it didn't light at all. Luckally it just needed more...

I must say that the TEC as sensor principle works quite well.
Cudos for that! I guess I'll settle for using an avr to deliver serial data as I want to log it.

Using my power inputs earlier I estimated a "ideal world max power" for the laser. If my 3.6w today produced 262 mV and the losses there were about equal % wise as what I have with the laser, then 17 mV should equal 233mW.
Most likely that is optimistic and I may not even be beyond 200.
All I know is it's barely enough to cut my stencils, but it does actually work.

qumefox · Aug 15, 2011

KreAture said:
qumefox who are you referring to?
Nice breadboard btw. I also like your videogen. If you don't need it I can provide you with my address...

Since i'm out of the TV repair business, it's for sale. $200+S&H & pp fees.

lasersbee · Aug 16, 2011

KreAture said:
What circuit? These measurements were done with a lab setup and using a Fluke thermal-cam to get average hotside and coldside temps. (Quicker than finding the hot probed blocks, or making some.)

I am only supplying power to the heater to simulate heavy power input.
(And supplying 12v to my cold-side fan for consistent readings.)

What I wanted to show, was the load on the peltier by "shorting" it with a resistor, although dropping the sensitivity will also cool the hot side.
I think the effect is related to the seebeck effect but I can't really explain the physics behind it. It is real though.

On my comment of actually cooling the hot side (making it in essence the cold side but only too room temp): That would use power. In fact, it would require about 30% more power than the input laser power due to less than 100% efficiency in the TEC. I guess it wouldn't work well either as it appears it's not the actual laser-power that is the problem, but the fact that it has to be absorbed by a thin layer on the surface of the TEC and lasers are good at burning stuff... Fun thought experiment though.

On a different (slightly) note: I did test the TEC with blackened surface as a detector today. Defocusing the beam to be about 90% of the width of the TEC and placed the TEC under it with it's HSF cooling the back-side.
I used a diode from a SH-S223Q Super Writemaster 22x DVD-RW drive (silly diode that needed lots of work to fit a normal 5.6mm host) and ran it at a few fixed resistor-values. Here's the raw data:

Code:

Id(mA) Vtec(mV) 139 2 (lasing error/flicker) 192 12 227 16.3 288 17.2 312 16.7 340 17.7 (flickering again)

I used a standard linear reg setup as constant-current.
Looks like I should have used a few more values between 139 and 227, but based on the curve I think the diode peaks there? From what I read here on the forums theoutput goes down again after you reach max out and from there it's not far to ruin. I guess I better stay below 300 mA?

The high current needed for stable lasing was odd to me too, but I firt noticed it when testing the laser the first time. I couldn't get it to light stabely at 50 mA. In fact it didn't light at all. Luckally it just needed more...

I must say that the TEC as sensor principle works quite well.
Cudos for that! I guess I'll settle for using an avr to deliver serial data as I want to log it.

Using my power inputs earlier I estimated a "ideal world max power" for the laser. If my 3.6w today produced 262 mV and the losses there were about equal % wise as what I have with the laser, then 17 mV should equal 233mW.
Most likely that is optimistic and I may not even be beyond 200.
All I know is it's barely enough to cut my stencils, but it does actually work.

Well....:thinking: You seem to be convinced to be on the right track...

Jerry

You can contact us at any time on our Website: J.BAUER Electronics

qumefox · Aug 16, 2011

I got lost while he was wanting to power the tec while using it as a sensor.

lasersbee · Aug 16, 2011

I refuse to argue with someone that "thinks" he's
got it all figured out...

Jerry

You can contact us at any time on our Website: J.BAUER Electronics

KreAture · Aug 16, 2011

Well. I am not telling you what to do, nor that you are wrong in what you do.
I am asking questions, and coming out with ideas. Mostly to hear honest informed opinions as they are worth more than hours of thinking by oneself.

I am also airing ideas as that's what I do.
I am used to my ideas being shot down, as most of them are and will be either far fetched, ludicrous, impractical, dangerous and sometimes all at once or a subset of said. However, once in a while there is something that can be good and that makes it worth while. It's what I do, and how I am.

Don't think for a second the feedback is ignored. I may defend my ideas, but I don't necessarily stop thinking about what you guys say. Some stuff takes longer to sink in though.

Nowe for some actual questions again...

This calibrated and offset method seems to work well, but it's not how I am used to working. I tend to get datasheets and have to stick to em or face reviews with endless changes untill i have. In the TEC case I would probably have used the C/W of the element unloaded coupled with the V/dT factor for running as a generator. That would have let me calculate any voltage out into a thermal delta, and again turn that into the power flowing through the element. After some consideration I come to the conclusion that this would work as the C/W rating is the elements leakage power and does not cover the current out when loaded. Any additional power out has to be met by "consuming" more power from the hot side thus changing the elements C/W. What I am saying is, both unloaded and loaded methods would work, but the unloaded is simpler (you are right) as one part of the equation can be ignored. In this case I would hit the wall with the power in, as I can only assume how good the transfer of energy is into the element from the laser.

So the questions:
Since the relationship of voltage out to power for a specific element can be known, is the absorption-factor of the laserbeam the only unknown in the setup?
If so, I assume it has to be defined by testing as every user with the same element may use a different absorption-layer?

Long-winded I know. I just want you to understand where the questios are coming from.

qumefox · Aug 16, 2011

The answer to your question is yes and no. The absorption of the laser is the main unknown. However the absorption of different wavelengths will vary a bit from material to material as well. However, for all intents and purposes for hobbyists, about all we can do about the latter is just paint the TEC surface flat black.

KreAture · Aug 16, 2011

Ok. Since I don't have a red lasrer with a known output I am a bit in the dark then. (hehe) Atleast I can compare my diodes and settings to find the highest power for my cnc. I just swapped the SH-S223Q silly laserpackage for a proper 5.6mm open can from a AD-7170S and it immediately gave me 18% more output according to a simple mV reading of the matted TEC.

qumefox · Aug 16, 2011

Yes. There aren't really any easy methods of calibrating a LPM without having another calibrated LPM to compare and calibrate the DIY one to.

freeloader700 · Oct 3, 2011

Thanks! I think I am going to give this a try

DIY Thermal LPM for under $50

lasersbee

0

random person

0

KreAture

0

qumefox

0

KreAture

0

lasersbee

0

KreAture

0

qumefox

0

lasersbee

0

qumefox

0

lasersbee

0

KreAture

0

qumefox

0

KreAture

0

qumefox

0

freeloader700

0

Online statistics

Forum statistics

Share this page