-
Looking for help with improving chip resistor thermal management
Posted by gjpmho on 22 Jun, 2022 23:13 -
Hi,
The heat from these chip resistors (put out ~0.8Watts) is dissipating through the PCB and heating nearby components to 80+C, as measured by an IR temp gun. It appears the circuit design is too dense thus heating up nearby components to 80C, which reduces their longevity. I want to explore ways to improve the heatsinking without transplanting the resistors. Ideally reducing by 10C. Since the resistors are not perfectly flat (leads are slightly raised above the body) I don't know which interface material to use in between the resistors and the heatsinks. Should the material cover the whole resistor including the leads, or just the body? I then plan to place the heatsink on top of the interface material. Would you recommend the following thermal gap filler and heatsinks? I also don't want to run the risk of shorting the leads. Would a heatsink this size (8x8x8 and 32C/Watt) actually improve thermal management by several degrees. Any help appreciated. Thank you.
https://www.digikey.com/en/products/detail/würth-elektronik/407150045015/15197563
https://www.digikey.com/en/products/detail/assmann-wsw-components/V2016B/8826901
-
I don't remember what it's called, but they make thick rubbery heat transfer material for exactly this sort of application. It's usually gray and feels somewhat clammy to the touch.
-
Is this the material you're talking about?
https://www.digikey.com/en/products/detail/cui-devices/SF100-202005/9805509 -
Is this a one off modification or for manufacture of many of these units?
If it's for manufacture: you should really redesign the PCB to have more space, copper and resistors in parallel. This will be the cheapest and most reliable method.
If it's for one off modification: can you solder on some tiny heatsinks (eg U-shape bent metal) on each of the pads of the resistors? What's on the other side of the PCB, would that be a better spot? Perhaps even drill some small holes and install some long M2.5 fasteners as cheap and easy heatsinks. -
It's for a one off modification. I'm trying to add some life to my amplifier. Accessing the other side of the PCB is a last resort. Can you tell me more about drilling holes and using fasteners as the heatsinks? I haven't drilled through a PCB before. Are you implying the thermal pad and heatsink combo are ineffective? The heat sink volume is 8x8x8mm^3. I'm guessing worst case scenario is the heatsinks and thermal pad are easy enough to remove.
-
I'm not sure how you would secure the heatsink and thermal pad without using screws anyway. Some thermal pads are sticky, but your surface area is going to be irregular so I don't think it will hold on its own.
Drilling holes: you would have to first confirm that it's only a two-layer board. Hold it up to the light and look through it, see if its opaque (multilayer with ground planes) or semi-transparent (2 layer). After confirming there are no hidden secrets inside the layers: choose a convenient spot on the pads near the end of each resistor and very gently drill (you don't want to risk removing the pad or damaging the resistors).
It might be easier/better to solder heatsinks on.
EDIT: After some thought: your issue is with the caps nearby, not the resistors themselves? In that case adding heatsinks probably won't help, you likely need airflow.
I previously modified a old warm audio amplifier that I wanted to last longer by adding a 120mm fan and running it on a lower voltage (so it was silent). Any passively cooled amp will run dramatically cooler if you do this, even a tiny amount of forced airflow is much bigger than passive convection in a small case. -
Since your problem isn't resistors failing, but other components getting too much heat, any change on the resistor brand/rating/type is not going to work.
The only way is to heatsink enough heat away from the board. I would look at heatsinking the board from below, through a thermal pad material (silpad), to an aluminium heatsink or chassis. This of course requires having thermal vias to the bottom side, and having copper fills there, and enough area with no components. From the picture I think there might be like half a square inch free on the bottom side, yes? A simple block of metal could be used to conduct heat from this area, to chassis. -
I'm not sure how you would secure the heatsink and thermal pad without using screws anyway. Some thermal pads are sticky, but your surface area is going to be irregular so I don't think it will hold on its own.
Drilling holes: you would have to first confirm that it's only a two-layer board. Hold it up to the light and look through it, see if its opaque (multilayer with ground planes) or semi-transparent (2 layer). After confirming there are no hidden secrets inside the layers: choose a convenient spot on the pads near the end of each resistor and very gently drill (you don't want to risk removing the pad or damaging the resistors).
It might be easier/better to solder heatsinks on.
EDIT: After some thought: your issue is with the caps nearby, not the resistors themselves? In that case adding heatsinks probably won't help, you likely need airflow.
I previously modified a old warm audio amplifier that I wanted to last longer by adding a 120mm fan and running it on a lower voltage (so it was silent). Any passively cooled amp will run dramatically cooler if you do this, even a tiny amount of forced airflow is much bigger than passive convection in a small case.
Thanks, The amp was manufactured around 2006. Initially I thought the caps were the heat source and after replaced those with rated 105c caps with no change in temperature. Some members on another forum pointed out the chip resistors as the source of heat. I measured the voltage across each resistor (R38/39 and R138/139) ~38 volts (38*38/2.2K = ~0.7Watts). Removing the PCB is not an easy task, it entails unsoldering some wires, unfastening the output devices from the heatsinks, and removing several nuts and bolts. Yes it looks like there are thermal vias. I've attached the schematics which include the top and bottom layers. Can you tell if there is a ground plane? The chip resistors are in the Left and Right main amp boards (pg. 2 and 3 of 7) and the power supply for the DC Servo and Current Sense (pg. 6 of 7).
I'll look into the 120mm fan option. Perhaps the 5 volts digital power supply can also drive the fans?
What about soldering the exposed end of insulated wire to resistor pads and using sheathed end to tie down the heatsinks to the resistors?
-
Soldering relatively thick copper wires to the pads can work to add some heat sinking, though just a piece of wire is not such a good heat sink. Instead one could as well replace the resistor with THT parts and solder them to the SMD pads. The pads are quite large and part of the heat would go directly to the air. THT resistors in 2.2 K 2 W rated are not that rare. Depending on the use in the circuit would may be able to use slightly higher resistance (e.g. 2.4 K or even 2.7 K) and reduce the power this way. It looks like this is an auxilliary supply that is stablized with zener diodes.
-
Is this the material you're talking about?
https://www.digikey.com/en/products/detail/cui-devices/SF100-202005/9805509
No, that's just a thin insulator pad for mounting transistors and such. The stuff I'm talking about is much thicker, 5mm or more in some cases. -
Is this the material you're talking about?
https://www.digikey.com/en/products/detail/cui-devices/SF100-202005/9805509
No, that's just a thin insulator pad for mounting transistors and such. The stuff I'm talking about is much thicker, 5mm or more in some cases.
Thanks, I ended up purchasing these instead:
https://www.digikey.com/en/products/detail/t-global-technology/TG-A6200-30-30-1-5/11617042
The resistors are 3x7mm, with the leads less than 0.5mm above the resistor body. Isn't 5mm at this scale too thick? -
Thanks, I ended up purchasing these instead:
https://www.digikey.com/en/products/detail/t-global-technology/TG-A6200-30-30-1-5/11617042
The resistors are 3x7mm, with the leads less than 0.5mm above the resistor body. Isn't 5mm at this scale too thick?
These "thermal pads", "silpads", "gap pads" are manufactured in wide range of thicknesses, and while they are decent thermal conductors compared to air, they still suck compared to metal. So you want to keep them as thin as possible. But sometimes you need something like 5mm. These thick products are usually softer, too, so they can conform to the components.
Now if you can place this to underside of the board in area where there are no components or TH legs at all, and you can couple it to a metal surface like chassis, I recommend approx. 0.5 mm thickness. This is thick enough to smooth out the tiny gaps and bumps in PCB / enclosure surfaces or compensate for slightly angled mounting by squeezing out and filling the gap; yet not too thick to hinder heat transfer.
Sometimes you actually use this stuff on component side (with thin as possible components of course), and then you might want to use those thicker variants, up to 5mm.
But if you have two smooth surfaces like PCB + enclosure separated by 5.0mm, do not use 5-6mm thick thermal pad, instead use a 4.5mm thick aluminum block + 0.7mm thick thermal pad or so. Best to have mounting holes near to apply decent pressure.
The thicker it is, the more it pays of to select higher possible thermal conductivity. Best products are around 15 W/(mK).
Also note voltage ratings. If you are working with mains voltages and need isolation, you may need glassfiber reinforced and certified silpads. The reinforcement compromises thermal conductivity, you may have to use one rated for just 3-4 W/(mK) and then add some thickness.
For comparison, aluminum is around 200W/(mK). -
It's for a one off modification. I'm trying to add some life to my amplifier. Accessing the other side of the PCB is a last resort. Can you tell me more about drilling holes and using fasteners as the heatsinks? I haven't drilled through a PCB before. Are you implying the thermal pad and heatsink combo are ineffective? The heat sink volume is 8x8x8mm^3. I'm guessing worst case scenario is the heatsinks and thermal pad are easy enough to remove.
Just take the SMD resistors off and replace with leaded (THT) resistors. Much safer than attempting to drill the PCB and you could do all the work from the top side of the PCB.
You can space the THT resistors well above the PCB and shape the leads to fit the existing pads. With THT resistors on sufficiently long leads a lot of the heat will be carried away by convection before it gets down to the PCB. -
If there is no airflow: I don't think replacing the resistors with through-hole axials or adding heatsinks will make the nearby capacitors any cooler. The heat has to go somewhere, and without airflow it will just diffuse into nearby objects just the same regardless of what heatsink you use.
You need to take the heat away from the local area where the capacitors are. Forced airflow works, as would a very long heatsink/heatpipe that takes it somewhere else (eg all the way to the metal chassis). -
If there is no airflow: I don't think replacing the resistors with through-hole axials or adding heatsinks will make the nearby capacitors any cooler. The heat has to go somewhere, and without airflow it will just diffuse into nearby objects just the same regardless of what heatsink you use.
This: chances are high the TH resistor would heat up the capacitors even more, unless you can wire the resistors elsewhere, physically far away from the capacitors.
There is no other way to either reduce dissipation (look at this first: why are you dissipating so much power? Does it absolutely need to be that way?), or actually remove the heat, either by fan cooling, or passive heatsinking. -
If there is no airflow: I don't think replacing the resistors with through-hole axials or adding heatsinks will make the nearby capacitors any cooler. The heat has to go somewhere, and without airflow it will just diffuse into nearby objects just the same regardless of what heatsink you use.
You need to take the heat away from the local area where the capacitors are. Forced airflow works, as would a very long heatsink/heatpipe that takes it somewhere else (eg all the way to the metal chassis).
Thanks, there is some convective airflow through top and bottom case vents. I've already purchased the heatsinks and thermal pads at minimal cost, and will check for a temperature drop in the surrounding components. Then I will look into adding fans. I appreciate the help. -
If there is no airflow: I don't think replacing the resistors with through-hole axials or adding heatsinks will make the nearby capacitors any cooler. The heat has to go somewhere, and without airflow it will just diffuse into nearby objects just the same regardless of what heatsink you use.
There is always airflow even if the case is totally enclosed there will still be internal convective currents carrying the heat away from the hotter components to distribute it throughout the case before that heat is lost through the sides of the case. The overall average temperature within the case will be the same as there is still the same amount of heat generated but the hot-spots will not be quite as severe adjacent to the sensitive components.
You need to take the heat away from the local area where the capacitors are. Forced airflow works, as would a very long heatsink/heatpipe that takes it somewhere else (eg all the way to the metal chassis).
The advantage of using THT components in this case is that more of the heat is carried away by convection and less through the leads and into the PCB. Having less heat transferred through the leads means that there is less heat conducted through the tracks and laminate to adjacent heat sensitive components such as the electrolytics. -
If there is no airflow: I don't think replacing the resistors with through-hole axials or adding heatsinks will make the nearby capacitors any cooler. The heat has to go somewhere, and without airflow it will just diffuse into nearby objects just the same regardless of what heatsink you use.
There is always airflow even if the case is totally enclosed there will still be internal convective currents carrying the heat away from the hotter components to distribute it throughout the case before that heat is lost through the sides of the case. The overall average temperature within the case will be the same as there is still the same amount of heat generated but the hot-spots will not be quite as severe adjacent to the sensitive components.
You need to take the heat away from the local area where the capacitors are. Forced airflow works, as would a very long heatsink/heatpipe that takes it somewhere else (eg all the way to the metal chassis).
The advantage of using THT components in this case is that more of the heat is carried away by convection and less through the leads and into the PCB. Having less heat transferred through the leads means that there is less heat conducted through the tracks and laminate to adjacent heat sensitive components such as the electrolytics.
Are there certain types of THT resistors you recommend for this scenario? I've finding wirewound and
metal film. How about this one?
https://www.mouser.com/ProductDetail/Ohmite/45F2K2E?qs=sGAEpiMZZMsPqMdJzcrNwvQIkRbkS0x3a1H%2FTQHaOQs%3D
Soldering THT leads to the pads seems unstable and like a hack. Maybe that's just superficial. Do you bend the leads into little feet and cover them in solder on the pad? Or is there a technique you would recommend?
-
Yes, bend the leads into little feet to increase the contact area with the pad. Needle nose pliers are useful.
The final solution will be more frail, you will have to use your own judgement on whether you think it's enough. I'd happily do it to my own gear, but I wouldn't do it to gear that I then intend to ship somewhere else (bumps & shocks). -
There is always airflow even if the case is totally enclosed there will still be internal convective currents carrying the heat away from the hotter components to distribute it throughout the case before that heat is lost through the sides of the case. The overall average temperature within the case will be the same as there is still the same amount of heat generated but the hot-spots will not be quite as severe adjacent to the sensitive components.
The advantage of using THT components in this case is that more of the heat is carried away by convection and less through the leads and into the PCB. Having less heat transferred through the leads means that there is less heat conducted through the tracks and laminate to adjacent heat sensitive components such as the electrolytics.
This is true, but it's hard to predict without complex thermal simulations or testing. SMD design uses board heatsinking more, TH design uses component body heatsinking more. Now it's very hard to say whichever results in higher capacitor temperature, because using board as heatsink also spreads part of the heat away from the capacitors, but as you say, part of the heat goes into the capacitor through the legs. If the TH resistor bodies are next to the capacitors, they will directly radiate heat into the caps. OTOH, if you make the resistors large enough, surface temperature then drops and dominating heat transfer method changes from radiation to convection and if you now can bend the resistors a bit further from the caps, the cap might run cooler than with SMD.
I do prefer full-SMD designs with single-sided load not only because it saves in manufacturing cost, but allows the whole underside of the board used as a heatsink - either just the board alone, or with thermal interface material + aluminum heatsink. TH components have the advantage that you can actually mount them to a heatsink directly, which is relevant when dissipating many watts within a single component, where the thermal impedance of about 10K/W of SMD FR4 w/ thermal vias design is still too much. With separately mounted TH parts, you can get below 1K/W. -
Also if the caps are in low-ripple bus i.e. they don't self-heat much and all the heat is from resistors, you can try shielding the caps. You could try a small piece of reflective aluminum tape just on the resistor side of the caps, but not on the opposite side. Remember that the aluminum tape will prevent both absorbing (incoming) thermal radiation but also prevent radiative cooling, this is why you use it only on the hot side, and it's no good if the capacitor itself runs hot because then it will reduce the cooling.
Aluminum tape will almost completely drop the radiation component to zero, of course doing nothing to thermal conduction by air. But if the resistors are running very hot, say like 100degC surface temperature, radiative part can be pretty large. -
Hi All,
Thanks again for the help. For now, I decided to stick (pun intended
)with the thermal pads and heatsinks. I cleaned all surfaces with IPA first and cut the pads to size using a razor. The thermal pads (1mm) are tacky on both side and it's surprising how much heat they conduct without added pressure. Maybe the weight of the heatsink and adhesive is sufficient. The heatsinks are too hot to the touch within a few minutes. This works, at least for now. The heatsinks are removing some heat from the board! I have to borrow an IR thermometer to see if the electrolytic caps are cooler. Will wait and see how this pans out after a few days of listening. So far so good...
I found this webpage which put things into perspective regarding heat sink volume versus air flow:
https://celsiainc.com/resources/calculators/heat-sink-size-calculator/
According to this site (and I haven't verified yet) the heat sink volume needed for each chip resistor with convection cooling is 3cm^3. The largest heatsinks I could fit on the chip resistors are 1/5 -1/3 of that volume.
http://www.tglobaltechnology.com/uploads/files/tds/TG-A6200.pdf
https://www.digikey.com/en/products/detail/assmann-wsw-components/V2016B/8826901
-
Also if the caps are in low-ripple bus i.e. they don't self-heat much and all the heat is from resistors, you can try shielding the caps. You could try a small piece of reflective aluminum tape just on the resistor side of the caps, but not on the opposite side. Remember that the aluminum tape will prevent both absorbing (incoming) thermal radiation but also prevent radiative cooling, this is why you use it only on the hot side, and it's no good if the capacitor itself runs hot because then it will reduce the cooling.
Aluminum tape will almost completely drop the radiation component to zero, of course doing nothing to thermal conduction by air. But if the resistors are running very hot, say like 100degC surface temperature, radiative part can be pretty large.
Here are the caps.
https://www.mouser.com/ProductDetail/Nichicon/UCD1V100MCQ1GS?qs=j%252B1pi9TdxUa53cZR3ibT1A%3D%3D -
Now if you could place those heatsinks on the opposite side of the board, that would definitely remove the heat from the vicinity of the caps.
But what you did likely helps, too. The problem just is, those heatsinks are still right next to the capacitors, so it's hard to say what exactly happens. One thing is sure, the resistors will run cooler.
You do need some physical mounting of the heatsinks, thermal pad material is no adhesive, even if it seems to kind-of stick. For prototype testing, gravity is enough for short term. -
Thanks, so far this is the least expensive and least invasive modification. Accessing the other side of the board is a real PITA, and I don't plan to sell or move the amp. Otherwise, I don't see how the heatsinks will fall off the thermal pads. I overestimated clearance on the resistor pads to solder wires that would anchor the heatsinks. Also this runs the risk of heat breaking down the wire insulation eventually shorting the heatsink to the resistor. I don't think drilling is a safe option, and still would have to access the opposite side of the board. The electrolytic capacitors are at least 1 mm from the heatsinks. I've attached a photo of another modification to improve heat sinking: Aluminum can folded 4X and keyring, which works quite well. I realize both mods are metastable, so will check on them periodically. Thanks again for the insights and recommendations. I hope to share before and after temperature readings soon. The proof is in the pudding!