I'm trying to improve the thermal management in the Power Amplifier Section of my integrated amp. There is a particular hot spot I'm worried about. I am measuring 80C+ on the case of a SOT-223 transistor (Q12/112 - ZVP2120G). Is this too hot? This puts the junction near 100C, according to the SOT-223 junction-to-case thermal resistance of around 12C/ Watt. I've calculated about 3/4 Watt (Vds * I) of heat being dissipated. The copper padding area under the drain and PCB thickness are not dissipating the heat quick enough and heating up nearby components, like q19/119 and VR1/101. The rest of the PCB is typically 50-60c after warming up. Another motivator is noticeable sound degradation, when ambient temperatures are above 25C hotter, the amp sounds strained. The main output device heatsinks usually stay below 45C. As a temporary aid I’ve added a small surface mount heatsink on top of the Q12/112 case via thermal tape. This helps a bit (~5C?) but is not a permanent solution. The tape will wear out. I've considered adding a heatsink under the PCB which again means glue that will break down over time. No fans, since they will add dust and I don't want to turn them on and off.
I’m thinking of moving the Q12/112 to a DPAK breakout board and securing that to the main heatsink above the PCB. The heatsinked breakout board will be mounted to the main heatsink via a breadboard - see the attached photo.
In efforts to improve thermal management at the PCB level this setup seems to be adding some noise.
1. An inch of wiring from the DPAK breakout board to the PCB pads.
Will this create considerable noise and or other issues.
2. I plan to connect the heatsink to the Drain which as I understand introduces EMI via the heatsink and can cause other issues. I don't know if these would be noticeable. I figure at best this setup will emulate the same MOSFET , just in a THT package. Or do I electrically insulate with a SIL pad heatsink and connect it to ground? It turns out the heatsink for Q14/114 is connected to its Drain.
In the end will I trade less heat (in the PCB) for noticeable noise and or other issues? This is just one component...
Notes:
I've converted several surface mount power resistors to THT above board to cool down the PCB, positioned other THT power resistors higher above the PCB. My goals are 70-80C junction temperature for Q12/112 at ambient temperature 25C, and PCB temperatures around 50-60C.
ZVP2120G datasheet:
https://www.diodes.com/assets/Datasheets/ZVP2120G.pdf
Breakout board:
https://oshpark.com/projects/SmLNbAES/view_design
Breadboard:
https://www.digikey.com/en/products/detail/sparkfun-electronics/PRT-08808/7387401
Heatsink:
https://www.mouser.com/ProductDetail/532-574502B33G
Stock photo of the amp:
https://www.hifi-inside.com/images/...fier/thumbs/phoca_thumb_l_destiny-1-creek.jpg
Photos of Q112 covered by the current heatsink, silk screen, and schematics attached.
I'm trying to get a perspective and relevant details of the scale of noise or any other issues I might be introducing. The chassis is 3mm thick aluminum which serves as a good heatsink.
I’m thinking of moving the Q12/112 to a DPAK breakout board and securing that to the main heatsink above the PCB. The heatsinked breakout board will be mounted to the main heatsink via a breadboard - see the attached photo.
In efforts to improve thermal management at the PCB level this setup seems to be adding some noise.
1. An inch of wiring from the DPAK breakout board to the PCB pads.
Will this create considerable noise and or other issues.
2. I plan to connect the heatsink to the Drain which as I understand introduces EMI via the heatsink and can cause other issues. I don't know if these would be noticeable. I figure at best this setup will emulate the same MOSFET , just in a THT package. Or do I electrically insulate with a SIL pad heatsink and connect it to ground? It turns out the heatsink for Q14/114 is connected to its Drain.
In the end will I trade less heat (in the PCB) for noticeable noise and or other issues? This is just one component...
Notes:
I've converted several surface mount power resistors to THT above board to cool down the PCB, positioned other THT power resistors higher above the PCB. My goals are 70-80C junction temperature for Q12/112 at ambient temperature 25C, and PCB temperatures around 50-60C.
ZVP2120G datasheet:
https://www.diodes.com/assets/Datasheets/ZVP2120G.pdf
Breakout board:
https://oshpark.com/projects/SmLNbAES/view_design
Breadboard:
https://www.digikey.com/en/products/detail/sparkfun-electronics/PRT-08808/7387401
Heatsink:
https://www.mouser.com/ProductDetail/532-574502B33G
Stock photo of the amp:
https://www.hifi-inside.com/images/...fier/thumbs/phoca_thumb_l_destiny-1-creek.jpg
Photos of Q112 covered by the current heatsink, silk screen, and schematics attached.
I'm trying to get a perspective and relevant details of the scale of noise or any other issues I might be introducing. The chassis is 3mm thick aluminum which serves as a good heatsink.
Attachments
Last edited:
Is there any way to make more room around the transistor so you can glue on a bigger heatsink? I'd do some research to find the best adhesive for this job.
Otherwise you're going to have to make a small board for it and mount it on some metalwork. If the wires are short and tightly twisted, this may work OK.
Also check your amplifier isn't oscillating!
Otherwise you're going to have to make a small board for it and mount it on some metalwork. If the wires are short and tightly twisted, this may work OK.
Also check your amplifier isn't oscillating!
Here’s what I’m thinking. I can drill a small hole in the DPAK breakout board or use a clip to secure it to the heatsink. Then solder the three leads of the breakout board to the breadboard. The breadboard would be screwed to the main heatsink in the second photo. I tried to draw in the new heatsink. It would rest about 10mm from the heatsink for Q14/114.
What is oscillation? How would I check for that? Thanks!
What is oscillation? How would I check for that? Thanks!
Attachments
Last edited:
Regarding oscillations, I can verify dc offset is stable around 1mV, idle voltage is also stable around 18mV. I don’t have an oscilloscope to look for oscillations. The amp does use 40 watts on idle.
It would be a good idea to move the gate resistor, R120, to the breakout board, so it is close to the gate terminal of the transistor.
Is the breakout board an aluminum PCB? That would be best for heat transfer to the heatsink.
Is the breakout board an aluminum PCB? That would be best for heat transfer to the heatsink.
It looks like a copper PCB. R120 is also SMT. Should the gate resistors should be thermally coupled to the transistor? If so, it looks like I would need a THT package and this might not be worth the trouble. Looking again at the Q12 without the surface mount heatsink, and the case temperature is maxing out at 80C. Here’s a photo of the breakout board.
Attachments
Also re-calculated voltage and current for Q12
Vds=48 volts
V across R26 (62 ohms) = 1 volt
P ~ (48*1/62) = 0.77 W
This puts the junction temperature at
80C + (12C/W) * 0.77W = 90C
Ambient temperature is 25C today.
I’m having trouble finding solid info on if 90C is too hot. It seems anything above 100C is considered too hot.
Vds=48 volts
V across R26 (62 ohms) = 1 volt
P ~ (48*1/62) = 0.77 W
This puts the junction temperature at
80C + (12C/W) * 0.77W = 90C
Ambient temperature is 25C today.
I’m having trouble finding solid info on if 90C is too hot. It seems anything above 100C is considered too hot.
Last edited:
An example of a breakout board using aluminum PCB substrate:
https://www.diyaudio.com/community/threads/assembling-the-lu1014d-ims-to-247-adapter.373273/
https://www.diyaudio.com/community/threads/assembling-the-lu1014d-ims-to-247-adapter.373273/
Thanks for the information! I’ll work on designing a breakout board. I had some experience with designing PCBs at my university.
Would you recommend using the aluminum heatsink from the previous photo?
Would you recommend using the aluminum heatsink from the previous photo?
Last edited:
Mounting the MOSFET on an Aluminum substrate PCB, and then mounting the PCB on a heatsink is a very effective way of dissipating heat. The thermal resistance of the aluminum PCB to the heatsink is much lower than that of a FR-4 PCB mounted onto a heatsink.
If by the "previous photo", you mean the one in my link, I have used that particular PCB in an amp that I built. Because the top copper layer is insulated from the aluminum substrate, the PCB can be mounted directly to a heatsink without an insulating pad, so the heat transfer is very good. And the price for a batch of 5 or 10 breakout boards is not very expensive at all at places like JLCPCB.
If by the "previous photo", you mean the one in my link, I have used that particular PCB in an amp that I built. Because the top copper layer is insulated from the aluminum substrate, the PCB can be mounted directly to a heatsink without an insulating pad, so the heat transfer is very good. And the price for a batch of 5 or 10 breakout boards is not very expensive at all at places like JLCPCB.
This is the heatsink I have been planning to use:
https://www.mouser.com/ProductDetail/532-574502B33G
Can you recommend an small batch Aluminum substrate PCB fabricator?
https://www.mouser.com/ProductDetail/532-574502B33G
Can you recommend an small batch Aluminum substrate PCB fabricator?
I have an extra THT 100ohm resistor that will fit on the breadboard I showed a photo of in post #3 with the breakout board and heatsink. I can drill a hole in the breakout board to mount to the heatsink, and solder the breadboard to the breakout board. Will that work?
I have ordered aluminum PCBs from JLCPCB. You can upload Gerber files to their website and it will calculate a price.
What you propose should work. Just make sure the terminals are not shorting to an unintended point.
What you propose should work. Just make sure the terminals are not shorting to an unintended point.
would you recommend connecting the heatsink to the drain? The drain for Q114 is connected to its heatsink. Or keeping the heatsink electrically insulated? If the later , then will it need to be grounded? I have SIL pads. Thanks.
The drain tab needs to be attached to the copper of the PCB for heat transfer. The details of the PCB, breadboard, and heatsink will determine what insulation is needed to make the correct connections.
Would you recommend electrically connecting the drain tab to the heatsink, or connecting the heatsink to ground? I am finding both scenarios can introduce parasitic oscillations depending on the circuit. But I don’t know enough about this circuit to decide. It seems electrically connecting the drain tab to the heatsink is the best option. The Q114 drain is electrically connected to its heatsink.
Last edited:
Here’s version 1. Unfortunately it won’t fit without a hex headed M3 screw, which I can’t find locally. Leaning more towards making my own PCB. There’s also more concern about noise via the wires as I read about oscillations and design… thanks again for the information.
Attachments
Hi Folks,
I'm revisiting moving Q112 above board. I couldn't find a free PCB design software to create a breakout board smaller than 25.4mm square, and it turns out the prototype photo shown in post #18 is too tall. I may have found a better prototype for mounting the transistor above board. I'm planning to use the following PCB and heatsink.
http://www.proto-advantage.com/store/product_info.php?products_id=2400044
https://www.digikey.com/en/products/detail/wakefield-vette/219-263A/6155915
The PCB will mount to the same screw shown in Post #1, except it will be oriented horizontally. I'll have just enough clearance to adjust the idle voltage VR1/101 after warm up.
I've been looking closer at circuit for Q100 and Q112 and the proximity of R129, R120 and C145. If you look at the silkscreen: R112, R120, and C145 are grouped just right of Q119. There's currently about 10mm-20mm lead distance from R120 to the gate of Q112. I understand its ideal to minimize this distance to reduce parasitic oscillation and ringing. Has the circuit already been designed to minimize these two, and will adding lead distance cause more parasitic ringing and oscillation, via induction in the lead wire for one? Or will moving R112, R120, and C145 closer to the new location of Q112 be advised? The complete schematics can be found in Post #1.
I'm revisiting moving Q112 above board. I couldn't find a free PCB design software to create a breakout board smaller than 25.4mm square, and it turns out the prototype photo shown in post #18 is too tall. I may have found a better prototype for mounting the transistor above board. I'm planning to use the following PCB and heatsink.
http://www.proto-advantage.com/store/product_info.php?products_id=2400044
https://www.digikey.com/en/products/detail/wakefield-vette/219-263A/6155915
The PCB will mount to the same screw shown in Post #1, except it will be oriented horizontally. I'll have just enough clearance to adjust the idle voltage VR1/101 after warm up.
I've been looking closer at circuit for Q100 and Q112 and the proximity of R129, R120 and C145. If you look at the silkscreen: R112, R120, and C145 are grouped just right of Q119. There's currently about 10mm-20mm lead distance from R120 to the gate of Q112. I understand its ideal to minimize this distance to reduce parasitic oscillation and ringing. Has the circuit already been designed to minimize these two, and will adding lead distance cause more parasitic ringing and oscillation, via induction in the lead wire for one? Or will moving R112, R120, and C145 closer to the new location of Q112 be advised? The complete schematics can be found in Post #1.
Attachments
Last edited:
