Electronics > Projects, Designs, and Technical Stuff
Unexpected amount of heat
(1/5) > >>
OM222O:
I built an evaluation board for a current sink with an op amp driving an NPN darlington pair transistor, via a 1k\$\Omega\$ base resistor.

The Collector is connected to +12V and the emitter is kept at 1V with the shunt resistor being 1\$\Omega\$. I calculated the temperature rise to be 11 degrees which I knew is extremely optimistic, but it was based on the data sheet values.
Thermal resistance is given to be 1C/W and the power dissipation is (12-1)*(1/1) = 11 watt. 11 volts at 1A.

here is the data sheet for the darlington pair:
https://www.st.com/content/ccc/resource/technical/document/datasheet/34/7a/29/f6/d1/3e/4c/58/CD00001247.pdf/files/CD00001247.pdf/jcr:content/translations/en.CD00001247.pdf

I also can see that 1A at 11V is well within the DC SOA of the D2PAK package (which is the one I'm using).
but in practice after about 5 seconds the package gets very warm and after about 10 seconds I can't touch the part!

I have upgraded to this part since the previous pass element I used was the MJD6039T4G which ended up failing shorted and killing an expensive op amp with it (pulled the base to +12 which was connected to the output of the op amp which had a max VCC spec of 6V and was powered from 5)  :palm: I'm worried that the same thing will happen with this part too. Please let me know where I'm making mistakes in my calculations and how can I prevent the parts from going nuclear?

Edit: in case a schematic helps:

Vset = 1.024 (just round it to 1)
Rset = 1\$\Omega\$
Rload = 0
Vcc = 12v
Rbase = 1k\$\Omega\$
Siwastaja:
Thermal resistance is 1C/W from junction to case. You need to add the thermal resistance of your heatsink in series to the calculation. If your PCB is acting as a heatsink, I recommend you google some similar D2PAK datasheets to find example PCB layouts with their evaluated thermal resistances junction to ambient. Some device datasheets or appnotes list full junction-to-ambient resistances for two or even three different PCB layouts.

Or, you can just measure the temperature you are reaching, and depending on that, improve the PCB heatsinking layout and iteratively test if it's getting good enough:

- Make the collector pad bigger. After you have stretched it about an inch in each direction, the effect starts to be fairly limited, since copper is thin and thermal resistance builds up going sideways.
- Add another collector copper region on the bottom, and add thermal vias to connect the regions together,

If this still isn't enough,
- Mount the PCB to a metal case, add thermally conductive silpad material as the interface material. You need screws close to the transistor so that there is no board flex causing gap. Heat flows from the transistor to the top copper fill, then through the vias to the bottom copper fill, from there, through the silpad, to the case.

Guesstimating from experience, dissipating 11W is possible with these means, but not trivial; a small standard D2PAK footprint definitely won't do it.
OM222O:
I can't add thermal vias as the board is single sided (other side must be used as the front panel) and the box is a plastic one.
In actual operation the device won't be on more than 2seconds on the 1A range and the maximum voltage will be 9 instead of 12, so a maximum power dissipation of 8watts. I'm still worried about the device being damaged if stressed repeatedly. I try to add more copper area next to it for sure. Top and bottom have massive ground planes which are in very close proximity of the device as well, I'm just not sure how much the thermal resistance of the FR-4 material is. the data sheet also mentions nothing about junction to ambient thermal resistance ...
can you please provide some links to layouts with similar packages which provide temperature calculations / thermal resistance for a D2PAK package with different layouts? that'll be extremely helpful.

Also can I use some of those cheap SMD heat sinks (basic aluminum fins) to further protect the device from thermal runaway?
for example this one: https://www.mouser.co.uk/ProductDetail/Aavid/7109DG?qs=sGAEpiMZZMttgyDkZ5WiujTlBOILjl7xi4aoVUO77j4%3D
I'm not sure how large is the thermal resistance of the plastic casing ... I doubt this will help at all?

I really don't want the device to fail, but it's ok if it operates at a high temperature (less than 100c just to be on the safe side).
Siwastaja:
A solderable heatsink is a good idea. The datasheet for that one provides two contradicting unitless numbers for thermal resistance, magical "9" and "11", but thermal product manufacturers are notorious for messing up the basics of documentation, including not understanding the basic units of their trade, so we just need to assume it's 11 degC/W, and finally, test to verify this assumption.

The datasheet says this is valid at 75 degC heatsink dT, in freely moving air. This is a bit demanding condition: your "ambient" would be inside the box I guess, your air probably wouldn't be completely free to move, and you probably couldn't accept dt=75, especially if you are aiming for 100degC transistor temperature - you would need to guarantee your box insides staying at 25degC max! So in practice, the Rth for that particular heatsink would be higher than 11 degC/W in your conditions.

Still assuming 11 degC/W, the total thermal resistance junction-to-ambient would be 11 + 1 = 12 (degC/W) then. 11W dissipated would cause a 12degC/W * 11W = 132 degC temperature rise, clearly beyond unacceptable. So, you'd need a bigger heatsink.

I guess the thermal resistance number they give already includes what the case of the D2PAK, and their recommended footprint through FR4 already dissipates, so you can't expect any extra from it, unless you add a lot of more copper. With thermal vias out of question, you are starting to lose your options.

Pulse duty with short pulses is helping - D2PAK has quite a lot of thermal mass. But then, you'd need to failsafe your design somehow to prevent duty cycles too high. Add a thermistor closely coupled to the transistor, preventing operation while too hot? This would likely be the cheapest solution.

Finally, I'd look into reducing the dissipation by changing what you are actually doing, and how. 11W is a lot of dissipation in a small integrated device, and very difficult to achieve in SMD design, especially when you only have a single side available. You can look at larger SMD heatsinks, of course, but then you'd need to ventilate the case as well, if you expect the possibility of large operation duty cycles, that is.


--- Quote ---can you please provide some links to layouts with similar packages which provide temperature calculations / thermal resistance for a D2PAK package with different layouts?
--- End quote ---

First result on Google with "D2PAK MOSFET":
https://www.vishay.com/docs/91055/sihf740s.pdf
Juction-to-case 1 degC/W (no surprise, it's the same case)
Junction-to-ambient, bare device, 62 degC/W
Junction-to-ambient with example PCB layout of 25.4 x 25.4 mm copper pad on FR-4 - I'd assume 2 oz copper as they are too lazy to define it, 40 degC/W

Another good one, see Figure 7:
https://www.microsemi.com/document-portal/doc_download/124720-an-208-pd70101-pd70201layout-guidelines-application-note
Basically, on a single layer without thermal vias, you are limited to a bit over 30 degC/W, even reaching infinite copper area, and this is on 2oz copper. You likely have 1oz, I'd guess, making things worse.
OM222O:
thanks for the detailed answer! I can place a hard lock of 5s cool down after each measurement (which again, only lasts for 2 seconds and dissipates about 8 watts). I can set up a test and measure the temperature over a time period to confirm that is enough cool down time or increase it if necessary.

however a hardware solution such as a PTC would be amazing just in case. I'm just not sure if I can find one with low enough value. as I said I have to pass about 1A through it and I only have about 8 volt headroom. can you recommend a circuit which allows for limiting the temperature rise without causing too much burden voltage?

also those figures seems pretty awful! There is a display on the board and I can use it's under side for vias, but it's at a really awkward position and I have to re route the PCB for that, but I'll give it a shot.

edit: on a second look, this part seems fairly decent for the task at hand:
https://www.mouser.co.uk/ProductDetail/Murata-Electronics/PRG21BC0R2MM1RA?qs=makukexe9nzHJ%252Bvxp0S09w%3D%3D
0.2 ohm isn't bad at all and it has a "hold current" of 750mA and a max of 10A. I'm assuming combined with the pulse method this should guarantee that the D2PAK device is protected against temperature damage?
even the 0.6 or 1 ohm units don't seem too bad and the curve seem to fit a lot better for the 100c max target?
Navigation
Message Index
Next page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod