Hello to all. On the smaller transistors like the 220s why do we use insulator pads when were trying to do is get rid of the heat into a heatsink? Why not just apply heat compound to the transistor and bolt it right to the heatsink? Most of the time the heatsink justs sit on top of the circut board anyway.Thanks, tom
if your heatsink cannot accidentally be shorted to supply or output or ground, then direct mounting with thermal paste is the best way to keep the device cool.
However, due to vibration, the extra weight of the sink will cause fatigue of the device connections to the PCB. If this is serious then it will eventually break the device leg/s.
However, due to vibration, the extra weight of the sink will cause fatigue of the device connections to the PCB. If this is serious then it will eventually break the device leg/s.
another consideration, one i've always wondered about. if you bolt a transistor directly to the heatsink, the heatsink now becomes an extension of the transistor collector. it's also a huge stray capacitance. as long as the collector is at rail voltage, this isn't much of a problem, but if the collector is the output element (in a common emitter amp for instance) it's a huge problem. with an insulator, you now have a very small capacitance in series with the larger capacitance, and the total capacitance is smaller than either capacitance. also, live heatsinks are a safety hazard. we recently got a couple of large Pioneer receivers in for repair. the output devices were mounted on two heat sinks on insulated plastic brackets. there were no warning stickers or indication otherwise that these heatsinks were live. sure enough one was at +60V and the other at -60V. those are large heatsinks and it's difficult to work around them. one tech got bit by brushing up against one. i found out they were live by clipping the -lead of my meter to one of them and starting to take voltage measurements in the amp section. the voltages i was seeing didn't make any sense the speaker wires coming off the amp board appeared to be at +60V as well as the ground wire to the amp board. it was then i figured out that the heatsink wasn't grounded. good thing i didn't connect a scope ground to it.... SMPSs are another place to beware of live heatsinks, i.ve been bit by live heatsinks on those a couple of times.
if you connect a heatsink directly to a rail voltage. put a sticker or something on it..... there have been some pro-audio manufacturers that used those little box heatsinks for TO-3's and they mark them with stickers
if you connect a heatsink directly to a rail voltage. put a sticker or something on it..... there have been some pro-audio manufacturers that used those little box heatsinks for TO-3's and they mark them with stickers
there are a lot of manufacturers making TO-220-iso devices where the whole package is plastic and can be mounted directly to a heat sink. there are a lot of TO-3P packaged devices with the heat sink tab insulated as well. you still need heat sink grease or sil pads to fill gaps between the device and heat sink. the sil pads aren't just there as electrical insulators, but to fill gaps and actually transfer more heat than you would if you just bolted the device directly to the heat sink.
dognut said:
Bolting the component straight to the heatsink will cause it to get the same potential (voltage) as the metal part of that component. If the heatsink is positioned in a way that there is no electrical connection to other parts with different potentials, then you can leave out the insulation pad. You may want to consider the remark unclejed613 made about parasitic capacitance: in some cases you might be better off with the pad.
Often more components with different potentials are bolted to one big heatsink. Then insulating them is mandatory, otherwise you will have one big shortcut, obviously.
Yes, the pad has some thermal resistance, but that's small enough not to be of importance, UNLESS the chosen size of the heatsink is barely able to keep things cool. Undersizing is never a good idea here (just ask Microsoft about the Red Ring of Death with their XBOX's), overheating is one of the most common causes of shortened component life.
Thermal paste will only be beneficial if used sparingly, otherwise it will actually make things worse, not better. If you take a look at guys assembling a PC (the ones that know what they're doing), you will notice they use no more thermal paste than the size of a grain of rice between CPU and heatsink. The pressure of the heatsink will spread this tiny bit out over the whole surface.
IMHO the thermal resistance of the small heatsink inside a non-active-cooled case is too high for the insulating pad's thermal resistance to be of any significance. We're not talking about 130W CPUs that need to be kept below 60 degrees by a huge active cooler.
And personally I hate anything live. Imagine poking around while the thing is running, and... zap. I've also been bitten by live heatsinks of SMPS b4.
And personally I hate anything live. Imagine poking around while the thing is running, and... zap. I've also been bitten by live heatsinks of SMPS b4.
IMO depends on your application. Many times, you want the HS to be isolated, but other times, heat transfer is priority.
On a large class AB CFP amp I built, the collectors go to the leads of the output resistors, so there's only a few mV between the sinks, and they are not grounded. Being there's no thermal insulator at all, the transistors never feel hotter than the heatsinks, and the heatsinks are also fan cooled as well. I've had no problems with stability, and I use a good zobel network also.
However, most of the time, it's best to use a thermal insulator.
On a large class AB CFP amp I built, the collectors go to the leads of the output resistors, so there's only a few mV between the sinks, and they are not grounded. Being there's no thermal insulator at all, the transistors never feel hotter than the heatsinks, and the heatsinks are also fan cooled as well. I've had no problems with stability, and I use a good zobel network also.
However, most of the time, it's best to use a thermal insulator.
some manufacturers glop on the heatsink grease, too, like Pioneer and their output chips in their "mid-fi" receivers. i usually don't even need to add any grease when i repair those amps, just spread out what's already there. if you wonder how bio labs let viruses leak out, just consider how fast just a tiny bit of heatsink grease gets all over everything when you're working on a repair. a company i used to work for actually used clear heatsink grease. it still got on your clothing, etc..., but it didn't leave white splotches when it did. we wear grey work uniforms where i work, and one of my team members was replacing one of those Pioneer amp chips, and left it "buttered" side up on his work bench. he turned to answer the phone, and caught the leads on his uniform and it left a large rectangular white blotch. at break time i said "i can tell you were working on a Pioneer......". he told me later his wife was a bit upset, because she washed that shirt twice to get the blotch out.
Sony tends to use it sparingly, and it looks like some machine stripes it on their heat sinks. Yamaha seems to use about the right amount, and a lot of manufacturers have gone to using strips of sil-pad. i'm starting to see these very thick, but VERY heat conductive squishy rubber pads on a lot of HDMI and DVD processor chips, as well as on class D amp chips. according to what i have seen in trade magazines, they're not silicone, but some other type of rubber that is so heat conductive that it even draws away the heat from the chip's leads as it gets squished over the edge of the chip. as i NERTS (scrap) DVD players that aren't cost effective to repair, i've been gathering a few of these pads to test for their thermal conductivity. right away, i've noticed that when they're on a heat sink, they are "colder" to the touch than sil-pads are, but i want to do actual temperature gradient testing to be sure. i'll probably use a couple of diodes as thermal sensors, and a transistor as the heat source (or even better, monitor the transistor Vbe directly as a built-in thermal sensor). the pads are squishy enough that i could use 1N4148s as sensors and not create any large gaps between the pads and metal surfaces.
Sony tends to use it sparingly, and it looks like some machine stripes it on their heat sinks. Yamaha seems to use about the right amount, and a lot of manufacturers have gone to using strips of sil-pad. i'm starting to see these very thick, but VERY heat conductive squishy rubber pads on a lot of HDMI and DVD processor chips, as well as on class D amp chips. according to what i have seen in trade magazines, they're not silicone, but some other type of rubber that is so heat conductive that it even draws away the heat from the chip's leads as it gets squished over the edge of the chip. as i NERTS (scrap) DVD players that aren't cost effective to repair, i've been gathering a few of these pads to test for their thermal conductivity. right away, i've noticed that when they're on a heat sink, they are "colder" to the touch than sil-pads are, but i want to do actual temperature gradient testing to be sure. i'll probably use a couple of diodes as thermal sensors, and a transistor as the heat source (or even better, monitor the transistor Vbe directly as a built-in thermal sensor). the pads are squishy enough that i could use 1N4148s as sensors and not create any large gaps between the pads and metal surfaces.
wwenze said:
Actually, the application is not so relevant, sizing of heatsink and sufficient airflow are.
I once, foolishly, used an insulation pad instead of thermal paste between a hothead CPU ("Prescott" Pentium D) and the actively cooled heatsink. It died within days. Here the pad's thermal reisistance was the determining factor. The stock cooler was borderline acceptable and barely able to keep things cool. It's no wonder a huge (after)market emerged aimed at CPU-cooling. We're talking about huge power dissipation through a small surface area here, so there's very little margin for error.
When you're talking about a class AB amp, there's only appreciable power dissipation at high power levels during a long time (and if you have neighbours, that's unlikely to happen).
A class A amp is a different story. The heatsink needs to be properly sized and the transistors mounted with electrical insulation and thermal paste. My Marantz PM80SE is a good example, a large part of space inside this amp is used for the heatsinks. In class A mode (it can also be switched into AB mode) it needs a lot of free space around it and especially above it to ventilate.
Attachments
dave_gerecke said:
found it, they're called Gap-Pads.
http://www.bergquistcompany.com/thermal_materials/gap-pad.htm
same company that makes Sil-Pads
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.