Pro Audio Design Forum

Thanks for sharing this very interesting work.

I understand the thermal advantages of direct-metal mounting.
Hypothetically what would be the device temperature difference if the same exact design (at 200W idle) had conventional thermal pads?
I suspect that the difference would be significant.
It might be worth measuring.

I agree that the use of separate transformers or separate windings due to the floating rails isn't an issue given that the users of an amp like this would demand dual mono.

mediatechnology wrote:Thanks for sharing this very interesting work.

I understand the thermal advantages of direct-metal mounting.
Hypothetically what would be the device temperature difference if the same exact design (at 200W idle) had conventional thermal pads?
I suspect that the difference would be significant.
It might be worth measuring.

I agree that the use of separate transformers or separate windings due to the floating rails isn't an issue given that the users of an amp like this would demand dual mono.

I had already measured the temperature difference of between various insulators and also compared that with direct mounting, here are the results. It was an exercise to self convince myself about the superior thermal performance achievable with direct mounting. Believe me, with this we have an added advantage of zero risk of insulator failure as well.
Test were done at ambient room temperature of 30C

Heatsink temperature was monitored for 65C

1. With Mica Insulators the case temperature was 76-78C
2. With Silicon Pads the case temperature was 82-86C
3. With Aluminum nitride ceramic pads the case temperature was 80-82C
4. With Direct metal mounting the case temperature was 68-70C

The case temperature was measured at collector/drain terminal pin of the mosfet which itself is a part of metal tab of device.

I hope this helps

It has been a while since i messed with this but I looked into thermal resistance a bunch when the power amp industry transitioned from two-screw TO-3 metal power devices to one-screw plastic packages.

From memory a good thermal insulator should be well under 1'C/W more like ,5' or less.

The one screw plastic devices had a few strikes against them but being cheaper you can use more in parallel to share the heat dissipation.

Probably the most significant benefit of a hot heat sink design is that it is easier to manufacture, while I recall headaches just trying to get the factory to tension the screws consistently. The modern (value) amps use spring clips to hold the devices to the heat sink, even easier for the factory. In my judgement many commercial amp designs allowed worst case temperatures to get dangerously close to package limitations (unlike old school metal, the plastic package degrades mechanically before the silicon melts), but they get away with it due to low duty cycle of most music.

JR

PS: FWIW my last patent while I was working at Peavey was for a heat sink design (6,515,859 Roberts , et al. February 4, 2003). In a forced air cooled heat sink a design concern is equalizing the temperature across the heat sink from the cold air to hot air end. The amp is only as strong as it's hottest device. By angling a pair of finned heat sinks in toward each other at the hot (outlet) end, the heat transfer is increased relative to the cold end helping to normalize the device temperature across the length of the heat sink.

Heatsink temperature was monitored for 65C

1. With Mica Insulators the case temperature was 76-78C
2. With Silicon Pads the case temperature was 82-86C
3. With Aluminum nitride ceramic pads the case temperature was 80-82C
4. With Direct metal mounting the case temperature was 68-70C

The case temperature was measured at collector/drain terminal pin of the mosfet which itself is a part of metal tab of device.

Wow, that's a significant reduction.

mediatechnology wrote:

Heatsink temperature was monitored for 65C

1. With Mica Insulators the case temperature was 76-78C
2. With Silicon Pads the case temperature was 82-86C
3. With Aluminum nitride ceramic pads the case temperature was 80-82C
4. With Direct metal mounting the case temperature was 68-70C

The case temperature was measured at collector/drain terminal pin of the mosfet which itself is a part of metal tab of device.

Wow, that's a significant reduction.

Yes it is and in case of Class-A its even more significant due to continuous heat dissipation.

FWIW my last patent while I was working at Peavey was for a heat sink design (6,515,859 Roberts , et al. February 4, 2003). In a forced air cooled heat sink a design concern is equalizing the temperature across the heat sink from the cold air to hot air end. The amp is only as strong as it's hottest device. By angling a pair of finned heat sinks in toward each other at the hot (outlet) end, the heat transfer is increased relative to the cold end helping to normalize the device temperature across the length of the heat sink.

John - I recall seeing an article about this IIRC in Sound and Video Contractor. Did you write an article?

Kanwar - What are the power supply currents and voltages to obtain 50W?

mediatechnology wrote:

FWIW my last patent while I was working at Peavey was for a heat sink design (6,515,859 Roberts , et al. February 4, 2003). In a forced air cooled heat sink a design concern is equalizing the temperature across the heat sink from the cold air to hot air end. The amp is only as strong as it's hottest device. By angling a pair of finned heat sinks in toward each other at the hot (outlet) end, the heat transfer is increased relative to the cold end helping to normalize the device temperature across the length of the heat sink.

John - I recall seeing an article about this IIRC in Sound and Video Contractor. Did you write an article?

Kanwar - What are the power supply currents and voltages to obtain 50W?

You may have seen a nice full page ad but I never wrote a proper article (I may have done something short for the Peavey dealer magazine but I don't recall.) I did write a more comprehensive amplifier overview where I described different amp classes and power devices. I still see that old article around the WWW while all the amps I mentioned in the article are long since obsolete.

I recall being irritated by Mackie's advertising campaign for their CS800 killer power amp using a unremarkable forced air heat sink design. They came up with a cute name for the heat sink design and advertised the piss out of like the second coming of Lord Kelvin. My heat sink design actually did something measurable. In the margin reducing the temperature spread across the heat sink results in a real increase in output power (thermal duty cycle) and/or better reliability if outlier hot devices are tolerated. Many manufacturers ran plastic devices hotter than conservative design suggested they should have, mainly because they worked. In my estimate you need about 1.5x the plastic devices vs metal. Few amp designers used that many especially for value designs. While this is all academic in the new class D calculus where devices are not dissipation limited but current-voltage product limited.

JR

mediatechnology wrote: Kanwar - What are the power supply currents and voltages to obtain 50W?

Hi Wayne,

Rail voltage is +/-32VDC and the Class-A quiescent current is 3.5A per rail

Rail voltage is +/-32VDC and the Class-A quiescent current is 3.5A per rail

Ok, so it looks like Iq = Ipeak (50W, 8 Ohms).
And a 48VCT transformer looks to be a about right.

Thanks again for sharing this with us. I hope that you continue to pursue this and share your results.