Some new design ideas w/TEC plates and High Speed fans…

[Amak87]

I’ve got a lot of experience in a lot of different fields.. if I had a lathe I would’ve already dropped the NUBB24/28 laser…

 

[Amak87]

Another discussion side topic, with using my li-po batteries to run drivers for the NUB diode array series of the 445nm blue lasers, I really haven’t seen a Vf increase when pushing. 4-5+ Amps? I have in my possession, NUBM31, NUBM37, NUBM35, NUBM38, NUBB23 and NUBB24 arrays and so far, none of them have had the same increase as we see with the NUBM47, NUBMOF, or Osram 8W diodes? With pushing 5+Amps, you would think you would see 4.4-4.6Vf? For the NUBB24, you would think you when then see the diode array pulling closer to 33-34Vf? But that’s not the case at all, with my experience, with the rise in current, and driving them for 30-40seconds, I see a drop in Vf and then they stabilize around 0.12V lower, they should pull 4V at 3A, but they actually pull less.. I’m not sure if this is happening because I’m running TEC plates and fans for cooling, and I don’t let the temps ever exceed 50-60*C at the heatsink.. I’m going to hook up my NUBB24 setup tonight and post my results pushing 3A, 4A, 5A, and maybe 6A.. with the data sheet info, they appear to be pretty close to the NUBMOF in terms of output.. that would be awesome if they are because pushing them to 5+ A should yield 8W+.. putting the NUBB24 around 60W, and the NUBB28 in the 80W range, and that’s modest, they could easily handle 5.5-6A and yield 8.5-9W per diode.. I wish I had a bigger LPM..

Ok. So the reason I didn’t see a voltage increase is because I had connected the negative line on the voltage meter portion of the voltage/amperage meter. When running these meters from a common power supply, that negative line needs to be left disconnected. Since I have now tested the NUBB24 with this negative line left offf the circuit, I am seeing proper forward voltage increase.

Here’s a picture of the 9-in-1 meter running the NUBB24 with a 4S/16.8v LiPo battery, I only get about 2-3 minutes of run time with a 1500mah capacity..

Voltage: 32.57
Current: 3.00
Watts: 97.89

 

[Amak87]

Here’s my results of driving the NUBB24 diode array at different power levels.

I’m using a 6S, 22.2v (25.2v), 1300mAh Li-Po battery as my power source. The driver is a non-isolated DC-DC booster, 600w, 20A input. I’ll post pics of the entire setup later. I’m running 20AWG coated copper wiring. The NUBB24 is mounted on an aluminum heatsink with a 60mm 12v cooling fan being ran at the same voltage as the diode, they are wired in parallel. So there’s some current being measured off the fan as well. I’m going to separate the fans power source later on and I’ll post the results. But it’s good to know the difference because when building a handheld laser with fans, it’s nice to be able to simply power the fans from the same driver/circuit.

I also noticed that the diode doesn’t like more than 5A.. the heatsink runs effectively and keeps the diode decently cool, allowing for a minimum of 1 minute run times so far.

You can see the last test at 4Amps was ran for 42 seconds and the heatsink wasn’t even warm to the touch. But higher speed fans would allow for much better heat removal.

3Amps:

3.5Amps:
4Amps:

 

[farbe2]

I am not sure what exactly you are doing with the tec there. It frankly looks like it wont help the thermals at all.

You cant just mill a slot in a solid block of aluminum and put a tec in there, it will just thermally short circuit the tec adding a shit ton of heat load to the heatsink.

If mounted correctly, a 1/2 of a NUBM28 needs a tec with roughly QcMax = 43W just to pump the heat.
if you want any significant temp difference from hot to cold side of the tec - to have the heatsink hotter than the diode - you need much more than that.

Say you want to keep the diode at 20°C while your heatsink is at 50°C. This means 30°C temp difference across the tec while it still pumps 43W.
A !good! TEC1-12706 will just barely do that: pump 43W at 30°C difference while using a whoping 85W to do so. (resulting in an abismal ~0,5 COP)
Thats 14V@6A!

Doing that will result in 43W Pumped + 85W Tec power usage being dissipated into your heatsink.

For a heatsink thats say 0,23K/W you would end up with a heatsink temperature of just around 50°C (at 20°C ambient and 20°C diode temperature).

That same heatsink used without a tec would keep the diode at 30°C while using 85W less power!
So you get 10°C lower diode temperature for huge cost, complexity and and additional 85W of power input and just 1/3 of the runtime using the same batterys.

That imho just makes no sense.

 

[Amak87]

I am not sure what exactly you are doing with the tec there. It frankly looks like it wont help the thermals at all.

You cant just mill a slot in a solid block of aluminum and put a tec in there, it will just thermally short circuit the tec adding a shit ton of heat load to the heatsink.

If mounted correctly, a 1/2 of a NUBM28 needs a tec with roughly QcMax = 43W just to pump the heat.
if you want any significant temp difference from hot to cold side of the tec - to have the heatsink hotter than the diode - you need much more than that.

Say you want to keep the diode at 20°C while your heatsink is at 50°C. This means 30°C temp difference across the tec while it still pumps 43W.
A !good! TEC1-12706 will just barely do that: pump 43W at 30°C difference while using a whoping 85W to do so. (resulting in an abismal ~0,5 COP)
Thats 14V@6A!

Doing that will result in 43W Pumped + 85W Tec power usage being dissipated into your heatsink.

For a heatsink thats say 0,23K/W you would end up with a heatsink temperature of just around 50°C (at 20°C ambient and 20°C diode temperature).

That same heatsink used without a tec would keep the diode at 30°C while using 85W less power!
So you get 10°C lower diode temperature for huge cost, complexity and and additional 85W of power input and just 1/3 of the runtime using the same batterys.

That imho just makes no sense.

If you read all the way thru, it was the start of milling the slot, the design is supposed to be a fully milled slot about 42mm deep…

And no, you’re 100% wrong about short circuiting the TeC plate lmfao, where is the contact point ?? There is none, not thru ceramic plates bud… I think you’re greatly confused..


And fyi, I’ve done a crap load of testing with TEC plates, the specs on the manufacturer sheets are hardly anything to go by, unless you’re not applying any heat load.. but applying heat load and proper heat dissipation from the hot side, they work completely differently, most of the time, they only need a fraction of their max power supply ratings..

And I don’t care about saving power, we are using such a small amount to begin with, I have lasers that operate for 10-12minutes without falling below 7W.. I have a lot of R&D behind what im stating and you clearly don’t…

iMho, you clearly hav no idea what you’re talking about.. TEC cooling is used and has been used to cool laser diodes for over 20 years bud.. lol it’s a proven method and soon I’ll be posting the final results of my TEC/fan cooled NUBM0F and Osram 8w diodes that will clear the 10W mark and stay there, not just peak ratings where it barley hits and then falls off dramatically..

The 20x40 TEC plate performs wonderfully with a heat load with just 5v/2A, just 10W of power..its minimal, and using a time delay device will further the performance, having the TEC plates turn on for 20-30seconds before hand…

You stating that the TEC plate will short out being sandwiched between aluminum, astounds me… like.. what the f… lol you really think that? Lmfao

I mean, that’s literally what they’re designed to do……

Also, did you just look at the pics and read a few sentences here and there ? Did you skip past all the other vastly experienced laser builders that have commented in this thread ? Strange how they’re not saying anything that you’re saying..

I’ve got over 60-70 tec plates in my possession, different sizes and manufacturers. I’ve tested all of them, not just like the YouTube guys who don’t apply a heat load, I’ve ran them sandwiched Between aluminum heatsinks, copper heatsinks, and some steel as well. I’ve tested 5 different types of thermal grease, found the best combinations and found the sweet spots between heat transfer and cooling power, of which, is really not something you can do using the spec charts. They’re usually not even close unless you buy soMe high end TEC plates, and even then, the performance changes dramatically with a heat load applied while supplying different levels of current.

And TEC plates , MUST BE SANDWICHED BETWEEN A HIGH THERMALLY CONDUCTIVE METAL.. you saying otherwise really tells me you’re clueless.. I mean… I’m lost on that one.. please enlighten me

 

[farbe2]

Yeah, sure…

I was talking about a thermal short, not a electrical one but you seem to completely miss that.

Maybe you can show the heat load calculations to proof your point?

Take for example the:

The World Leader in Thermal Management Solutions

lairdthermal.com

A high performance tec from a very well known company.

It doesn’t really make sense to me to discus the application if you aren’t willing to learn how to really use tecs and how to calculate all thermal aspects of a system.
Just to let you know, it’s very much possible to just mathematically derive a solution that will show you exactly how a tec will perform, which temperatures you will get at which parts of a assembly and how much power it draws. If you really want, you could even simulate heat flows on your PC without spending time on the mill just to find out it doesn’t work.

Not sure why you are hurt.

BTW, I do this (every day) for a living, so I am fairly sure I know what I am talking about. Simulation of a multi diode laser module:

 

[Alien Laser]

Years ago many other members try something similar only the used different method since the are more experienced in this field.I also was curious and used similar heatsink like yours and added a cheep Tec and tiny fans i believe the are 5X5 and the problem was one side will heat the heatsink and one side was cooling it so it was a fail, So what i did was i add a Tec right behind the diode and added a heatsink on the Hot end and it worked ok the only problem was the diode failed and was a tiny bit wet or moister build no idea.
farbe2 is not pulling your leg he is correct.
Anyway good luck i keep an eye to see if you’re projects will work

 

[Amak87]

again, the TEC plate current needs to be dialed in with a specific setup..

The point is to have the inner diameter of the heatsink cooled, while having the heat created from the diode pulled from the center And pushed to the outer portion of the heatsink, where the fans will cool it.. and no, the specs on the TEC plates are almost always incorrect due to varying heat sources.. it’s crazy to me that you think this won’t work when it’s the exact design of the TEC plates… applying just enough current so that the temp isn’t so low and the temperature difference is smaller, which will keep any moisture from developing.. I’ve tested it for hours already.. I’ve never been able to get anywhere near the spec claims, and I’ve watched hours and hours of others doing good tests on them and they don’t either.

I actually have 2 types of the Laird TEC plates.. funny cuz they’re a laser company that uses them to cool their laser heatsinks.. lol

I’ll make my tec/fan cooled heatsinks , and you’ll see..

And if you were talking about a thermal short, then why not just state that? I’ve never seen a thermal short occur during my tests.. again, there’s a sweet spot for TEC plates, just the right amount of current and they perform amazing.

It’s crazy to me that you don’t understand that having TEC plates, surrounding a diode, wouldn’t transfer the heat away from it? My first pic only showed 1 partial slot.. my final design has 3, in a triangluar layout, and fans on the outer portion that keep the hot side of the heatsink cooled.

In my tests, i am able to keep around 55*F inner wall temp, and barely 85*F outer wall temp with 2 high speed fans, that’s measuring the heatsink temp.. the diode is absolutely operating at a cooler temperature and remains stable.

I don’t need a computer to tell me that my actual tests aren’t working lol when they are..

 

[Amak87]

Years ago many other members try something similar only the used different method since the are more experienced in this field.I also was curious and used similar heatsink like yours and added a cheep Tec and tiny fans i believe the are 5X5 and the problem was one side will heat the heatsink and one side was cooling it so it was a fail, So what i did was i add a Tec right behind the diode and added a heatsink on the Hot end and it worked ok the only problem was the diode failed and was a tiny bit wet or moister build no idea.
farbe2 is not pulling your leg he is correct.
Anyway good luck i keep an eye to see if you’re projects will work

If you don’t use enough TeC plates at the correct current, then yeah, the diode heat will overcome the whole attempt to cool it.. I’ve tested this using NUBM37 diodes, NUBB23/24, and the smaller 7-8W diodes..

The heatsink has a much lower temperature and remains stable..

The 20x40 TEC plates I use, are run in series, 12.6v at 4.8A, that’s the sweet spot for them, but even with that knowledge, I’ll still be dialing in each setup I build, as all the TEC plates will perform differently..

just stay seated and watch..

 

[Amak87]

Yeah, sure…

I was talking about a thermal short, not a electrical one but you seem to completely miss that.

Maybe you can show the heat load calculations to proof your point?

Take for example the:

The World Leader in Thermal Management Solutions

lairdthermal.com

A high performance tec from a very well known company.

It doesn’t really make sense to me to discus the application if you aren’t willing to learn how to really use tecs and how to calculate all thermal aspects of a system.
Just to let you know, it’s very much possible to just mathematically derive a solution that will show you exactly how a tec will perform, which temperatures you will get at which parts of a assembly and how much power it draws. If you really want, you could even simulate heat flows on your PC without spending time on the mill just to find out it doesn’t work.

Not sure why you are hurt.

BTW, I do this (every day) for a living, so I am fairly sure I know what I am talking about. Simulation of a multi diode laser module:
View attachment 78399

Go ahead and explain a thermal short circuit and how it would occur in my heatsink/TEC setup.. cuz I don’t see how it could even happen, and I believe you’re using incorrect terminology for this concept..

But I’m willing to learn, so go ahead and explain……

 

[farbe2]

And if you were talking about a thermal short, then why not just state that?

Didnt i do that in my first post? Just for it to be not read?

You cant just mill a slot in a solid block of aluminum and put a tec in there, it will just thermally short circuit the tec adding a shit ton of heat load to the heatsink.

How it works:


The 20x40mm Tec you are using will move heat, it doesn't produce cold. It just moves heat from its cold side to its hot side (reversible by polarization change but that does not matter).
If you power the tec and make the cold side face the bore where the diode sits, it tries to move the heat to the outside but:
The outside and the inside are one piece. The aluminum will just conduct the heat from the outside back into the middle part essentially thermally shorting out the tec.



Just like if you try to fill a swimming pool with a pump but the pump is sucking water from the same swimming pool you are trying to fill. The pool will not get fuller!
It also doesn't matter if you use 2 or even three tecs, same concept: trying to fill the pool with multiple pumps all sucking from the pool they are trying to fill.

The right thing to to would be to divide the pool into two sections by using a barrier, that barrier is the insulator between hot and cold side.
With this barrier, and the pumps sucking on one side and pumping to the other, they would be able to produce a water hight (or temperature) difference between both sides: thats what you tec is doing but with no barrier.

Normally, tecs are clamped between two thermally conductive surfaces with screws and thermally insulating washers that prevent heat from flowing back into the cold surface.
Like this:


One image specifies "SS screws" thats not something German, that stands for stainless steel.. Why? Because it is a very poor thermal conductor, which keeps the "short" to a minimum, a minimum amount of heat flowing back into the cold side. The other drawing specifies Nylon spacers or fiber washers, another type of thermal insulator. (Think about it like the pool: the screws are holes in the pool barrier, having big holes will make the pumps struggle to create a water hight difference because the water will just flow back to the lower side)

Also: your tec is just "slid" right into the slot, there is no mounting pressure at all on the surfaces. Tecs need around 75-200PSI of pressure to work right and dont break over time. The pressure also keeps their thermal resistance between the tec and the mounting surface in check, too low pressure means too high resistance to efficiently remove the heat from the hotside.
The Belleville spring washer sets this pressure and keeps it under different thermal expansion situations.

Ever used a CPU cooler on a CPU without its spring clamps? -> similar concept -> not good

Now to your setup.
I used this tec: https://lairdthermal.com/de/product...peltier-cp-series/CP10-131-04-L1-TOW-W4.5/pdf
and just run it at a 20 watts of power you will get this: (redder is hotter, bluer is colder, see scale on the right side)


You can see the tec in the slot tries its best to create a temp different but only reaches 30,113°C-29,852°C = 0,26°C!
So the core is a whopping 0,26°C colder than the outside while wasting 20W of electrical power.

Its even clearer if we look at the heat flow: (redder = more heat flow, blue = less heat flow)

You can clearly see the red/orange/yellow areas around the slot ends where the aluminum heatsink connects the two tec sides. The heat flow is the highest there because it's the shortest route between hot and cold side of the tec.


Disclaimer: I just used guessed at the dimensions for the CAD model, the tec is 40x20mm and the bore is 9mm, outside is 35mm, this looked like his pics, but it could be slightly different.

Sure having more power into the tec will result in higher temp differences, but come on, its like using a fire department type pump in a kids sized inflatable pool and saying: "See, I can get the water to splash around a little more, so it must be filling..."



I suggest the following: Try replacing the Tec with an aluminum plate, keep the fan and measure the diode temperature and power again. Temperature will be lower, power higher and battery will last longer!
Just removing the tec might be slightly worse as the tec acts as an essentially perfect thermal conductor in this scenario, so leaving the slot empty would introduce thermal resistance.

 

[Ricker]

Didnt i do that in my first post? Just for it to be not read?


How it works:

View attachment 78408
The 20x40mm Tec you are using will move heat, it doesn't produce cold. It just moves heat from its cold side to its hot side (reversible by polarization change but that does not matter).
If you power the tec and make the cold side face the bore where the diode sits, it tries to move the heat to the outside but:
The outside and the inside are one piece. The aluminum will just conduct the heat from the outside back into the middle part essentially thermally shorting out the tec.

View attachment 78409

Just like if you try to fill a swimming pool with a pump but the pump is sucking water from the same swimming pool you are trying to fill. The pool will not get fuller!
It also doesn't matter if you use 2 or even three tecs, same concept: trying to fill the pool with multiple pumps all sucking from the pool they are trying to fill.

The right thing to to would be to divide the pool into two sections by using a barrier, that barrier is the insulator between hot and cold side.
With this barrier, and the pumps sucking on one side and pumping to the other, they would be able to produce a water hight (or temperature) difference between both sides: thats what you tec is doing but with no barrier.

Normally, tecs are clamped between two thermally conductive surfaces with screws and thermally insulating washers that prevent heat from flowing back into the cold surface.
Like this:
View attachment 78410View attachment 78411

One image specifies "SS screws" thats not something German, that stands for stainless steel.. Why? Because it is a very poor thermal conductor, which keeps the "short" to a minimum, a minimum amount of heat flowing back into the cold side. The other drawing specifies Nylon spacers or fiber washers, another type of thermal insulator. (Think about it like the pool: the screws are holes in the pool barrier, having big holes will make the pumps struggle to create a water hight difference because the water will just flow back to the lower side)

Also: your tec is just "slid" right into the slot, there is no mounting pressure at all on the surfaces. Tecs need around 75-200PSI of pressure to work right and dont break over time. The pressure also keeps their thermal resistance between the tec and the mounting surface in check, too low pressure means too high resistance to efficiently remove the heat from the hotside.
The Belleville spring washer sets this pressure and keeps it under different thermal expansion situations.

Ever used a CPU cooler on a CPU without its spring clamps? -> similar concept -> not good

Now to your setup.
I used this tec: https://lairdthermal.com/de/product...peltier-cp-series/CP10-131-04-L1-TOW-W4.5/pdf
and just run it at a 20 watts of power you will get this: (redder is hotter, bluer is colder, see scale on the right side)

View attachment 78412
You can see the tec in the slot tries its best to create a temp different but only reaches 30,113°C-29,852°C = 0,26°C!
So the core is a whopping 0,26°C colder than the outside while wasting 20W of electrical power.

Its even clearer if we look at the heat flow: (redder = more heat flow, blue = less heat flow)
View attachment 78413
You can clearly see the red/orange/yellow areas around the slot ends where the aluminum heatsink connects the two tec sides. The heat flow is the highest there because it's the shortest route between hot and cold side of the tec.


Disclaimer: I just used guessed at the dimensions for the CAD model, the tec is 40x20mm and the bore is 9mm, outside is 35mm, this looked like his pics, but it could be slightly different.

Sure having more power into the tec will result in higher temp differences, but come on, its like using a fire department type pump in a kids sized inflatable pool and saying: "See, I can get the water to splash around a little more, so it must be filling..."



I suggest the following: Try replacing the Tec with an aluminum plate, keep the fan and measure the diode temperature and power again. Temperature will be lower, power higher and battery will last longer!
Just removing the tec might be slightly worse as the tec acts as an essentially perfect thermal conductor in this scenario, so leaving the slot empty would introduce thermal resistance.

Very informative, thank you!