The big TEC driver thread!

[dave]

The reason its done the other way is to make dust sealing the optical section easier

I'm no where remotely close to being an engineer, but that sounds completely logical to me unless there is a reason it has normally done the other way.

[laser_freak]

General thoughts to TEC-cooling/heating of laser diodes..


When TEC cooling laser diodes a quick reacting chain of control (heat emission> sensing> counteracting) is essential for high efficiency temperature regulation. As I’ve read in most cases here on the forum people tend to place the TEC between the diode mount and the base plate. From a physics point of view, a very ineffective setup.


Bearing in mind that heat rises and cold falls: With the mentioned setup, TEC between the diode mount and the base plate, the TEC Is physically working against the laws of nature by trying to cool upward and dissipate heat downward...
...
3.) Use the natural flow of heat/cold to increase efficiency and minimize unnecessary heat dissipation. Place the TEC on top of the diode mount (cooling side down, heating side up) and use the laser case/lid as a heat sink. I’ve constructed my lasers so that the case/lid is a heat sink (doesn’t even need ventilation!).

4.) If possible turn the diode so that the die carrier is towards the top of the mount (heat dissipates away from the die instead of around it).

The effect of heat rising and cold "falling" only applies to fluids. Since the atoms can not flow within the metal, this effect is negligible. It is unlikely positioning the TEC above the diode will have any effect in terms of TEC efficiency. However as you mentioned, dumping heat into your baseplate will screw up your optics if it is not dissipated well.

http://www.photonlexicon.com/forums/...7&d=1297269947
Furthermore, isn't it always best to insulate the diode mounts when using a TEC? With your setup, it doesn't appear to be insulated. In fact, I would think the TEC would be less efficient because you are trying to cool more mass (the diode's mount + baseplate + the scanner's baseplate).

[MechEng3]

The effect of heat rising and cold "falling" only applies to fluids. Since the atoms can not flow within the metal, this effect is negligible. It is unlikely positioning the TEC above the diode will have any effect in terms of TEC efficiency. However as you mentioned, dumping heat into your baseplate will screw up your optics if it is not dissipated well.

Agreed as well...one caveat....if all the optics are aligned and kept at a set temp{ operating temps}...things don't change enough to notice for most builder/users.There are a few US laser manu's who chose to install the optics onto .230" glass...stays in alignment for ages and the diode is cooled on another block protruding through the glass. So it may be relevant to the design......the big thing I have found is combating the latency time...back to thermal/mass transit calcs.

[dnar]

I just pulled the datasheet on that one and I'm always open for discussion, but did you notice the conversion time @ 12 bits of max 750ms even the 9 bits @ 93.75ms compared to an NTC is just way to slow.

Agreed.

That is way to slow compared to the ADC sample period. I need a fast and constant sample rate for PID.

As far as sensor position is concerned, I find positioning the sensor directly opposite the TEC best with proportional control, however with PID you can place the sensor at the part you wish to thermally regulate ie. LD.

[RedlumX]

..
http://www.photonlexicon.com/forums/...7&d=1297269947
Furthermore, isn't it always best to insulate the diode mounts when using a TEC? With your setup, it doesn't appear to be insulated. In fact, I would think the TEC would be less efficient because you are trying to cool more mass (the diode's mount + baseplate + the scanner's baseplate).

Probably it is meant to be isolated, otherwise it would make little sense. I usually clamp the diode mount with plastic screws against the TEC.

I wonder whether all this effort is really necessary if one uses the laser just as a lamp, essentially....

[Solarfire]

Furthermore, isn't it always best to insulate the diode mounts when using a TEC? With your setup, it doesn't appear to be insulated. In fact, I would think the TEC would be less efficient because you are trying to cool more mass (the diode's mount + baseplate + the scanner's baseplate).

That would require a lot of thermal energy for the cooling/heating effect of the TEC to reach that far into the system. This is the problem though with the TEC mounted under the mount. Extensiv thermal measurements were made on the obove shown setup with the conclusion that a thermal balance sets in at some point within the mount dependend of the ambient temperature. The key being that the mass between diode (heat source) and TEC (cooling/heat source) is keept small. In the above design the size of the mount actually detemind the size of the TEC (15mm x 15mm 3.9W) if the mount were smaller an even smaller TEC could be used.

Furthermore tests also showed that in order to get aproximatly equal results with TEC mounted below the mount a 6-10 fold on energy increase was necessary, not to mention the high thermal oscillation measured at the diode.

The part about heat rising and cold falling only applying to liquids or air is not entirely true (not as much in solids). Heat is a radiating energy which is passed as excitation from molecule to molecule until a thermal energy balance sets in.

It's the summ of the little things which can make a big diffrence and there is actually no extra effort involved in the build.

Did I miss anything??

Ahh yes... Cheers

[dnar]

The part about heat rising and cold falling only applying to liquids or air is not entirely true (not as much in solids).

You are correct if and when you consider heated air on the TEC hot side.

I have in the past used insulator sheets to minimize thermal feedback from hot to cold sides.

[RedlumX]

Furthermore tests also showed that in order to get aproximatly equal results with TEC mounted below the mount a 6-10 fold on energy increase was necessary, not to mention the high thermal oscillation measured at the diode.

Well that is not what my measurements show; I just need a little bit more TEC power than the diode dissipates. However, I think there may be a misunderstanding and the point is as follows. It does not matter whether the diode mount is cooled from above or below. What matters is where the heat sink is: whether it is in free air where convection can transport the heat away without obstruction. If the heat sink is below the laser, the cooling is not so efficient and the case temperature will be higher at a given ambient temperature. This applies to small heat sinks; for large sinks these issues play little role.

However this becomes much more serious if the driver sits in the same case. Despite the powers involved are not high, this can lead to thermal runaway if the ambient temp is high. One needs a really large heat sink in this situation, more than one may think at first; I mean the order of 1K/W.

As for thermal oscillations: this is a sign of a ill-designed feedback loop and it can be always avoided by a proper design. For this one needs to dermine the open-loop response and then adapt the PID feedback parameters appropriately.

[dnar]

As for thermal oscillations: this is a sign of a ill-designed feedback loop and it can be always avoided by a proper design. For this one needs to dermine the open-loop response and then adapt the PID feedback parameters appropriately.

Exactly. .

I just need a little bit more TEC power than the diode dissipates.

This is true. You only require more TEC power to get the diode below ambient.

[Solarfire]

Well that is not what my measurements show; I just need a little bit more TEC power than the diode dissipates. However, I think there may be a misunderstanding and the point is as follows. It does not matter whether the diode mount is cooled from above or below. What matters is where the heat sink is: whether it is in free air where convection can transport the heat away without obstruction. If the heat sink is below the laser, the cooling is not so efficient and the case temperature will be higher at a given ambient temperature. This applies to small heat sinks; for large sinks these issues play little role.

However this becomes much more serious if the driver sits in the same case. Despite the powers involved are not high, this can lead to thermal runaway if the ambient temp is high. One needs a really large heat sink in this situation, more than one may think at first; I mean the order of 1K/W.

As for thermal oscillations: this is a sign of a ill-designed feedback loop and it can be always avoided by a proper design. For this one needs to dermine the open-loop response and then adapt the PID feedback parameters appropriately.

I’m already regretting posting my findings.


To keep it simple:


Fact is: Pumping heat into the laser base plate has absolutely no pros, because the dissipated heat is just going to end up heating up the entire laser housing unnecessarily screwing with the ambient temperature, Optics and whatever else. How much is irrelevant, it’s simply unnecessary. A Quick analogy: Putting the heat sinking between CPU and Motherboard.


Fact is: Dissipating heat away from the laser/diode via the shortest path is indisputably the best way to go (basic physics 101). A Quick analogy: I prefer my heat sink on top of the CPU carrying heat away from the motherboard.


Of course thermal oscillation is a sign of an ill-designed feedback loop, and of course one can level this out, the point is at cost of a much slower reaction time. With a TEC under the mount it is farthest from the heat source to be cooled thus predetermined to be an inert control loop.


Fact is: This is a thermal feedback loop meaning it is already pretty inert as is. Optimal design of a thermal feedback loop requires that heat source, thermal sensor and TEC be as close to one another as possible otherwise the control loop will become more and more inert with increasing distance of these 3 elements. Quick analogy: Heating and air-conditioning.


Take a closer look at high end DPSS lasers, in a lot of cases the diode is mounted directly on the TEC and the NTC is in the TEC, you just can’t get much closer than that, also basic physics 101.


Cheers!