Hi,
Is it OK to use the amp chassis as heatsink for discrete semiconductors like B+ voltage regulating MOSFETs (insulated package, like TO-220F), or should I always put a dedicated heatsink? I'm not looking at extreme cases like 6C33-based amps, which get extremely hot, but something based on 6L6GC/EL34 in class AB. Any advice is much appreciated!
Thanks,
Jose
Is it OK to use the amp chassis as heatsink for discrete semiconductors like B+ voltage regulating MOSFETs (insulated package, like TO-220F), or should I always put a dedicated heatsink? I'm not looking at extreme cases like 6C33-based amps, which get extremely hot, but something based on 6L6GC/EL34 in class AB. Any advice is much appreciated!
Thanks,
Jose
Depends on the power that needs cooling and the internal temps. If in doubt
mount a dedicated heatsink at the amp outside or other cool place.
What will be the dissipation of the MOSFET? Measure voltage in, voltage out and output current and you will know: (vin - vout) * current.
Sometimes I use the chassis, sometimes I add an aluminium strip of 3 mm thick and 3 or 4 cm. wide for extra cooling surface. Sometimes I have to add a “real” heatsink.
Regards, Gerrit
Sometimes I use the chassis, sometimes I add an aluminium strip of 3 mm thick and 3 or 4 cm. wide for extra cooling surface. Sometimes I have to add a “real” heatsink.
Regards, Gerrit
I use П-shape heatsinks bolted to aluminum chassis
If some thingy is called "heatsink" it does not mean that it works better. It has own area, the smaller, the higher is the thermal resistance between it and an air. Also, if it is installed on a board, the board closes convection path. Then, most of them that I see on DIY pictures, were designed for forced air cooling, too small distances between fins to be optimal. And finally, the heat dissipated by them must some how go from under the chassis.
If some thingy is called "heatsink" it does not mean that it works better. It has own area, the smaller, the higher is the thermal resistance between it and an air. Also, if it is installed on a board, the board closes convection path. Then, most of them that I see on DIY pictures, were designed for forced air cooling, too small distances between fins to be optimal. And finally, the heat dissipated by them must some how go from under the chassis.
Attachments
I'm using an aluminium chassis, 17x10x2 inches, 1.5mm thick sheet. I expect the dissipation to be 8W or less based on screen current datasheets, let's say 100V across, 80mA max for an AB1 6L6GC stereo amp. I will now better when I build the thing 🙂
Attachments
I agree with you, I will definitely not put heatsinks inside the chassis, the thermals would be horrible. I was planning to use an old CPU heatsink, bolted on top of the chassis, and the transistors attached from the bottom side, through a rectangular opening. But I'm sick of drilling and cutting 🙁
Attachments
Is a relatively big one, 1.5mm aluminium sheet, 17x10x2 inches. I hope is enough.
Attachments
Last edited:
I have a MOSFET dissipating about 45W under the 833C in my amps. It is attached to the inside of the chassis side extrusion with clips, and finned heatsinks are mounted on the outside of the chassis side extrusion in the same location. So both the chassis and the heatsinks dissipate the heat. See the finned heatsinks on the left side of the chassis in this picture.
Attachments
I'm sure that chassis would be suitable as a heatsink, no worries.
I've ran regulators supplying 1 amp or so in the same way, the chassis got barely warm in that area.
When oriented such a way it becomes much less efficient that when the convection path is open all the way along the fins, when hot air goes up through the bottom to the top of them. And the optimal distance between fins for convection cooling is 8 mm. For forced air they are denser. Look how they are installed in professional equipment. Either at the rear panel, or inside between bottom and top covers with vent holes.
The area is good for 8W, but 1.5mm thickness can make a single TO-220 hotter than optimum.
If you have two packages, spaced out across the chassis it's OK. If you have one TO-220, a heat 'spreader-bar' is helpful to keep the junction temperature down. You can get pieces of 5mm thick alu on ebay. Even 50mm x 100mm will help.
TO-220FP packages can have high Theta-JC (thermal resistance from junction to case). 6 °C per Watt is quite common.
Example: If the heatsink surface is at 45 °C, with 8W to dissipate: the junction will rise maybe 48 °C for the package, and another 10-12 °C for the heatsink interface (using thermal grease). That's 105 °C for this example - already a bit uncomfortable for high reliability. High voltage FETs in the standard TO-220 package can easily achieve theta-JC of 0.5 to 0.9 °C/W. The high voltage is no problem if clips are used instead of screws, to mount the transistor.
Agreed, for me the ideal would be mounted at the rear, or keeping a fan on top like . I thought about vent holes, but not keen on them, also do not like the possible vibrations introduced by a fan.
For the moment I have decided to take the easy way and go for chassis as heatsink.
I have to admit I did not take into account the thermal resistance of the TO-220FP, and for the STP3NK60ZFP I got is 6.25, vs 2.78 for the normal one. Rookie mistake, thank you for pointing that out.
I'm actually adding a 3mm angled piece to the chassis to prevent bending (I was advised that 1.5mm aluminium sheet will probably bend on the long run), hopefully this will help with the dissipation.
Thanks again for your advice!
You can a heat pipe - mount the component and then use the heat pipe to move the heat away to a separated heatsink.
Also with direct chassis mounting - heat is lost both sides of the chassis - outside and inside, depending on the ambient temperature of outside and inside. Cooling inside needs to be able to cope with the rise internally.
I like the idea that the sides of an amp are the heatsink with full length fins. The result is that all the side walls act to cool the inside of the amp if nothing is mounted directly, but I prefer mounting to the sides so the gradient allows faster cooling than inside the case.
Also with direct chassis mounting - heat is lost both sides of the chassis - outside and inside, depending on the ambient temperature of outside and inside. Cooling inside needs to be able to cope with the rise internally.
I like the idea that the sides of an amp are the heatsink with full length fins. The result is that all the side walls act to cool the inside of the amp if nothing is mounted directly, but I prefer mounting to the sides so the gradient allows faster cooling than inside the case.
> a 3mm angled piece to the chassis
3mm should be OK, allow a good contact area, at least 50 x 75mm, with thermal grease.
> STP3NK60ZFP I got is 6.25, vs 2.78 for the normal one.
2.78 K/W is better for sure, or try the STP10NK60Z for 1.1K/W, if the bigger-area FET does not degrade phase too much.
3mm should be OK, allow a good contact area, at least 50 x 75mm, with thermal grease.
> STP3NK60ZFP I got is 6.25, vs 2.78 for the normal one.
2.78 K/W is better for sure, or try the STP10NK60Z for 1.1K/W, if the bigger-area FET does not degrade phase too much.
That would be better indeed. Unfortunately I got 19 STP3NK60ZFP, £3.80 for the lot, I'm married to them I'm afraid 🙂