Testing of materials for laser diode housings

[Bbe]

An interesting discussion broke out in this thread. I have little more time now, and I modeled the heating process on a Dave's housing model and Nichia 9mm diode.

Legend:
Baseplate - 15mm aluminium with stable room temp (25°С) down side
Laser diode - silver package with a maximum thermal conductivity
Other components, such as wires or lens retainers are not involved in the model due to their weak thermal effect.
We use excellent thermal grease

1-Diode in nominal mode.

Current - 1400mA,
Forward voltage - 4.6V
Total power - 6400mW
Laser emission power - 2100mW
Heat emission power - 4300mW

1.a - Diode temp results for brass housing


Chip temp - 35.3°С
Diode package temp - 29.1°С

1.b - Diode temp results for aluminium housing


Chip temp - 34.3°С
Diode package temp - 28.1°С

1.c - Diode temp results for copper housing


Chip temp - 32.4°С
Diode package temp - 27°С


2-Diode in overload mode.

Current - 2100mA,
Forward voltage - 4.8V
Total power - 10000mW
Laser emission power - 2650mW
Heat emission power - 7350mW

2.a - Diode temp results for brass housing


Chip temp - 42.6°С
Diode package temp - 32°С

2.b - Diode temp results for aluminium housing


Chip temp - 40.9°С
Diode package temp - 30.4°С

2.c - Diode temp results for copper housing


Chip temp - 39°С
Diode package temp - 28.4°С

To sum up, we have a 3.6°C difference on chip between using brass and copper housings

[planters]

Bbe,

I do not understand the results. If you were substitute a "perfect" thermal conductor into the model there would obviously be no thermal gradient. Now, if you were to introduce two thermal conductors into the program; one with exactly twice the thermal conductivity of the other the temperature differential between any two points with the lower conductivity material would have to be twice the differential between the same two points in the higher conductivity material. This has to be true because thermal conductivity is defined as energy flow/temperature gradient. If the models are otherwise identical then the conducting cross section is also identical and to carry the same energy flow (power) to the base plate the temperature gradients must be inversely and linearly proportional.

Copper has approximately 3.5 times the thermal conductivity of brass. If the diode package temperature is 32C and the base plate is 25C, this is a differential of 7C and with the copper model the differential is 3.4C. Do you think the interface with the base plate is causing a 2C barrier? A 2C barrier would be consistent with that differential.

What is a silver package and what is the cross section of your mount. Do you allow for the missing material from the large threaded bore for the lens?

[Bbe]

Thanks planters,
We have a thermal gradient, but it's difficult to see in this bandwidth of colors. For example, I measured several points for brass housing with "overloaded" diode

I do not understand the results. If you were substitute a "perfect" thermal conductor into the model there would obviously be no thermal gradient.

That is what we see in this model

This has to be true because thermal conductivity is defined as energy flow/temperature gradient.

I see that the differential for the aluminum plate around 0.6C (in proximity close to diode housing), these results are consistent with practical measurements.

Copper has approximately 3.5 times the thermal conductivity of brass. If the diode package temperature is 32C and the base plate is 25C, this is a differential of 7C and with the copper model the differential is 3.4C. Do you think the interface with the base plate is causing a 2C barrier? A 2C barrier would be consistent with that differential.

I have no idea about Nichia 9mm package, I took the silver to idealize my model The cross-section is a gap between the pressure plate and housing

What is a silver package and what is the cross section of your mount.

Do you allow for the missing material from the large threaded bore for the lens?

Of course, the program took into account all the holes in the test

[dave]

Interesting stuff.

Will be interesting to test the accuracy in a real world environment

[-bart-]

@bbe

I would like to see if heatpipes could have a significant effect on the transfer of heat.
Is that something that could be easily simulated ?

[danielbriggs]

Superb work. We need more like this.

I assume this is steady state?

If an non-infinite sink is assumed, (lets say the projector base/enclosure) heats to 40C, could you run the numbers again?

[planters]

Bbe,

Sorry, but I still don't understand a few things. What is a silver package? Is this some kind of idealized diode module? When I asked what the cross section was I was unclear. I meant, what are the X,Y an Z dimensions of the mount that you modeled?

Nevertheless, this is very good work and it will be interesting to compare real world measurements to this model.

[Bbe]

In the real world we can not measure the chip temp

Will be interesting to test the accuracy in a real world environment

You need to simulate flows inside pipe? Usually heat pipes have known characteristics and specified in the program as the temperature difference at certain heating power

I would like to see if heatpipes could have a significant effect on the transfer of heat.
Is that something that could be easily simulated ?

If neglect the heat transfer in the infrared and air convection, just add 15C to my result. It's a state when baseplate has a fixed temperature, for example with TEC cooling.

I assume this is steady state?
If an non-infinite sink is assumed, (lets say the projector base/enclosure) heats to 40C, could you run the numbers again?

I have no idea about the diode package material, it is not copper, not ceramics ... Most likely it is some alloy. If you have any info, let me know and I refine the model.

What is a silver package? Is this some kind of idealized diode module?

This is Dave's housing model, 14x16x17mm

I meant, what are the X,Y an Z dimensions of the mount that you modeled?

[-bart-]

No, I mean a diodemount with heatpipes.
Can you simulate that to investigate if this would be a feasible idea ?

Maybe an aluminium mount with one or two heatpipes could perform better than a pure copper model.

[planters]

Actually, you can make a pretty good estimate of chip temperature from the wavelength shift. At a fixed current, the GaN semiconductor red shifts approximately 0.05nm/C and the AlGaAs red shifts 0.25nm/C

Yet, I was suggesting that the simulation could be compared to real world measurements of the mount with a thermistor attached to the surface. When I operated one of Dave's mounts with approximately the same heat load as your over driven model the top of the brass mount was hot. I had not bothered to measure it at the time, but it was well above body temperature (37.0 C). One of the confounding elements was this laser was using an O-like lens and this has since been replaced by a lens that is provided with the harvested diode and preforms better with a higher power throughput. It is possible that some of the heating should be attributed to the O-like lens.

I have no idea about the diode package material, it is not copper, not ceramics

I am almost certain that the diode package is copper, but that is not my question. You state "laser diode-silver package" What do you mean?