Katie and dad--I forgeo you were the original posters on this thread! Sorry.....
I'd recommend you look at the Pass First Watt amps, and those built by competent DIY'ers. All good designs mount the output semi's directly on the heatsink. Use mica + grease, or kerafoil, for insulators (they insulate the MOSFETs electrically, but are good thermal conductors). Use the largest heatsinks practical (I used Conrads, from Australia). If you MUST use a steel case, cut holes large enough to still mount the semi directly to the heatsink (i.e., don't try to sandwich the steel chassis/case between the semi and heatsink).
If you still have doubts, ask the same question on the Pass forum, indicating you are building an F5. I'm certain you'll get get a flood of consistent feedback--and that it will underscore what I'm saying.
I'd recommend you look at the Pass First Watt amps, and those built by competent DIY'ers. All good designs mount the output semi's directly on the heatsink. Use mica + grease, or kerafoil, for insulators (they insulate the MOSFETs electrically, but are good thermal conductors). Use the largest heatsinks practical (I used Conrads, from Australia). If you MUST use a steel case, cut holes large enough to still mount the semi directly to the heatsink (i.e., don't try to sandwich the steel chassis/case between the semi and heatsink).
If you still have doubts, ask the same question on the Pass forum, indicating you are building an F5. I'm certain you'll get get a flood of consistent feedback--and that it will underscore what I'm saying.
Sofaspud..... Fear not. You're right on the mark. And for any continued "doubters", the thermal conductivity coefficients for varous materials are readily available on the internet. In a quick nutshell: diamond is about the best (but I don't think you'll find many audio chassis made of it....!); copper is very good, aluminum is good, and steel.....is steel.
sreten must have been nipping at the "rat cheese" perhaps?
1) going through more layers *must* decrease thermal conductivity
2) steel sheet metal chassis are *non-flat* further decreasing the thermal conductivity
3) steel as has been noted several times before is a *less good* thermal conductor
4) the tabs on these plastic pak Mosfets, NOT TO-3 style transistors, are small enough and as such require the best possible thermal path
5) Most of the tabs (back surface) on the plastic paks are plated copper not steel afaik, or worse case steel plated with copper then cadnium (probably cadnium)
6) connected to a steel plate (or sheet metal) the ability of the aluminum heatsink's flat back surface to shed heat is substantially reduced (the case where there are holes punched for the devices, for example) which somewhat reduces the total ability of the heatsink to shed heat, upping the temperature rise.
6a) The lower the conductivity of the thermal path, the higher the internal temperature of the device and bonding wires can reach, for example on peak current draw or in the case of Class A devices, on turn-on. Which is why heatsinks are needed in the first place...
7) Physics: hotter devices have lower SOA.
PS, for some RF type amps a copper "spreader plate" is required because even a direct connection to the aluminum isn't sufficient.
1) going through more layers *must* decrease thermal conductivity
2) steel sheet metal chassis are *non-flat* further decreasing the thermal conductivity
3) steel as has been noted several times before is a *less good* thermal conductor
4) the tabs on these plastic pak Mosfets, NOT TO-3 style transistors, are small enough and as such require the best possible thermal path
5) Most of the tabs (back surface) on the plastic paks are plated copper not steel afaik, or worse case steel plated with copper then cadnium (probably cadnium)
6) connected to a steel plate (or sheet metal) the ability of the aluminum heatsink's flat back surface to shed heat is substantially reduced (the case where there are holes punched for the devices, for example) which somewhat reduces the total ability of the heatsink to shed heat, upping the temperature rise.
6a) The lower the conductivity of the thermal path, the higher the internal temperature of the device and bonding wires can reach, for example on peak current draw or in the case of Class A devices, on turn-on. Which is why heatsinks are needed in the first place...
7) Physics: hotter devices have lower SOA.
PS, for some RF type amps a copper "spreader plate" is required because even a direct connection to the aluminum isn't sufficient.
Last edited:
It would be nice to see some numbers that substantiate this. Can the heatsink shed heat into the air easier than into the steel?
edit: IOW... Without looking, I think the thermal conductivity of air is worse than steel, but ΔT may begin to dominate the equation giving air the advantage.
Last edited:
The surface to air resistance is practically unchanged for the part of the heatsink covered by the steel chassis even with high air flow, so long as the junction between them is greased. Don't worry about that. One of the problems though, is it does have to be greased all over for that, and very flat, and clamped without causing distortion. In the case of sinking a power device Through the chassis and into the aluminum extrusion, you need Much lower resistance.. to the point of necessarily removing the finish (because it's a good thermal resistor) quite far around the device. If it's epoxy or powder coat, that'll suck. It's not that the steel gives a useful "spreading" effect, it's resistance is much too high for that, but the thickness of the finish will cause distortion and voids around the edges, so it has to go away and far. Getting an even squash over the double junction with no distortion would take more and better distributed pressure than you can get by a single screw through the tab and proper torque. In the end, the extra junction, even if everthing is attented to with extreme care, makes it a loser by comparison. What's to debate?
Last edited:
K&D......stay away from paint on heatsinks if conductivity is your concern. Same goes for powder coating. Anodizing is not too bad, in that is an extremely thin layer. In many cases, even your untreated aluminum heatsinks are coated--with a clear anodize. Don't sweat all of this too much--you're not building the first-ever DIY class A amp. Use the Pass forum as a source, and the prior Pass DIY amp threads and pics as a guide. You can see how I did my heatsinking in my F5, using Conrad (Australian) heatsinks; with these 'sinks and plenum air flow, I probably can build a higher-power Pass F5 Turbo V2 using the very same chassis and sinks. Note, too, that I'm not actually USING a chassis. I've cut a baseplate, rear apron and front panel from 3/16" thick aluminum plate, and the two heatsinks are the two sides (left and right) of the encolosure. No need for a steel chassis. No need for a "box" enclosure. My "chassis" is extremely rigid...
My F5 build is here: http://www.diyaudio.com/forums/pass...-build-beautiful-music-different-drummer.html
Note that I did "lap" the MOSFETS and the heatsink surface where they mate, for best possible heat transfer. Note, also that I did paint the base plate of the heatsinks for esthetics, but NOT the fins--they were anodized by the manufacturer. I also left an unpainted area where I mounted each MOSFET.
Last edited:
Oh, it has been discussed before.
The difference between having the heatsink "free air" where convection can take place on the back side and having it clamped to a steel surface, or worse still a painted steel chassis is that unless there is some provision for the air to move across the steel *inside* the chassis - and there is a very very good connection between the steel chassis and the heatsink - the back side of the heatsink is going to try very hard to get *hot*.
Which is the opposite of what is wanted.
The amount of work required to "fit" the backside of the heatsink properly to a steel chassis has to be 10x the work required to just mount it 1/4" or better off the surface and permit airflow. Or cut a big hole
Makes no sense to me to even have to discuss this in any depth... fact is that a simple (not machined, flat, clean and dosed with thermal compound) sandwich of heatsink and chassis is just never going to be good.
My advice, avoid these issues, don't mount flush to the surface of the chassis - the F5 runs hot
The difference between having the heatsink "free air" where convection can take place on the back side and having it clamped to a steel surface, or worse still a painted steel chassis is that unless there is some provision for the air to move across the steel *inside* the chassis - and there is a very very good connection between the steel chassis and the heatsink - the back side of the heatsink is going to try very hard to get *hot*.
Which is the opposite of what is wanted.
The amount of work required to "fit" the backside of the heatsink properly to a steel chassis has to be 10x the work required to just mount it 1/4" or better off the surface and permit airflow. Or cut a big hole
Makes no sense to me to even have to discuss this in any depth... fact is that a simple (not machined, flat, clean and dosed with thermal compound) sandwich of heatsink and chassis is just never going to be good.
My advice, avoid these issues, don't mount flush to the surface of the chassis - the F5 runs hot
Last edited:
Thermal Conductivity of some common Materials and Gases
gold is the way to go, or copper for the poor folks
gold is the way to go, or copper for the poor folks
Last edited:
Not to repeatedly strike a deceased equine, but for discussion and understanding I have to wonder by what mechanism "the back side of the heatsink is going to try very hard to get *hot*." It seems to go against Fourier's heat flow formula I posted previously, and the heat/electric current analogy.
The backside of the heatsink is the entry of the heat path, so it's expected to be at a higher temperature than the furthest area of the heatsink, but the statement appears to be saying something else.
Diamond is even better than gold--but it's so darned hard to drill and tap a diamond for a 4-40 machine screw.....(!)
Agreed, that the poor horse is now in post-mortem, but to help with this lingering comment, consider this:
1. Yes. An aluminum heatsink can shed heat faster into steel, than into air.....but think of this in "electrical terms".
2. Consider the semi and heatsink to be a power supply.
3. Consider both the steel and air to be capacitors..... Steel, as a small rated capacitance, and air as a HUGE capacitor.
4. The thermal "resistance" between the power supply and steel is lower, so therefore heat (voltage, charge...) will transfer more quickly. However, the steel (unless you have a HUGE block of it....) will ultimately warm up. And as the temp of the steel increases, the transfer of heat (voltage, charge) to the steel will become less, as the temperature (voltage) of the power supply and steel (small capacitor) approach equilibrium.
5. Now consider the case of a heatsink and air. The "resistance" of the heatsink to air is higher (less thermal conductivity), but as the air is heated by the heatsink it will rise (thermal convection) TO BE REPLACED BY COOLER AIR. And so the cycle continues (.....until the temp of the room matches the temp of the semi and heatsink--fortunately, since the air mass of a typical room is large (large capacitor), although a few degree rise is possible, it's normally not an unbearable condition for the inhabitants).
I hope that helps, on this point. Yes--a heatsink to steel has good conductivity, but unless you have several auto engine blocks (i.e., a lot of steel and/or iron), you are dealing with a condition that will ultimately give you thermal equilibrium. The less steel you have, the quicker this will occur. In contrast, with heatsink and air, the air is replenished via the movement of the air, so the heatsink-to-air conductivity is not as good, but the volume of the airflow (convective flow) makes this an effective way to keep an amp cool.
The same analogy goes for water cooled amps (even more efficient than air cooling), as long as you have an endless supply of water, or a radiator to ultimately move the heat to the air, after all.
Amen. Le cheval est morte....!!!
Ok, place a thermal insulator sheet over the backside.
Compare the device temp in that case to the one where the backside has free convective air flow.
Are you saying that the temps will be the same?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.