It looks like the best we can do is look at temperature differentials that are common for our class-A amplifiers: somewhere around 25C ambient and not more than about 55C heatsink temperatures. Importantly we do not want to exceed the case temperatures of the devices mounted to the heatsinks.
Is there a derating factor for the manufacturer C/W data that would apply to this range of temperatures?
Ben Mah, I agree. it's not apples/apples. There is a certain amount of work occurring that might be described as a critical mass for a particular 'system'. ie the 'hotter' is more energetic, the difference isn't large however, I fall back upon the 1 pound is good for 10 watts mantra.
Last edited:
I agree that we should limit maximum temperature to 55C, so that would be a differential of 30C maximum.
The interface material also has a resistance rating so for best protection of the devices, we should choose the material with the lowest resistance.
However, only the heat sink manufacturer/provider can provide information on their rating. It is unfortunate that that information is not commonly provided.
The interface material also has a resistance rating so for best protection of the devices, we should choose the material with the lowest resistance.
However, only the heat sink manufacturer/provider can provide information on their rating. It is unfortunate that that information is not commonly provided.
Last edited:
Correction to post #401.
Today I remeasured power dissipation and ΔC of my ZD25 amplifier which uses natural convection with the HeatsinkUSA sinks. The power dissipation is 81.1W per heatsink and the temperature rise is 28C, for ΔC/W = .345C/W. The HeatsinkUSA figure of .34C/W/3" is for a sink of length 3". If all were linear it would imply .34C/W*3/7 = .146C/W for a 7" long sink.
Today I remeasured power dissipation and ΔC of my ZD25 amplifier which uses natural convection with the HeatsinkUSA sinks. The power dissipation is 81.1W per heatsink and the temperature rise is 28C, for ΔC/W = .345C/W. The HeatsinkUSA figure of .34C/W/3" is for a sink of length 3". If all were linear it would imply .34C/W*3/7 = .146C/W for a 7" long sink.
Here is something I found on the web:
https://www.infineon.com/dgdl/an-1057.pdf?fileId=5546d462533600a401535591d3170fbd
It states that C/W/3" is based on a temperature differential of 75C.
It also gives correction factors for various temperature differentials and heat sink lengths.
https://www.infineon.com/dgdl/an-1057.pdf?fileId=5546d462533600a401535591d3170fbd
It states that C/W/3" is based on a temperature differential of 75C.
It also gives correction factors for various temperature differentials and heat sink lengths.
.....
ZM - Apologies if I implied there was a problem. There is not. Amp runs flawlessly. I can see how you'd think I may have had an issue given my general dodoness.........
relax, Bro, that was strict reply on what you asked in #414
frankly , in your long posts above, I even didn't see which amp is in case , looking just on hard data, on which I didn't have anything to object
importance of not making silly mistakes with thermal transfer thingies
few months ago, I had a mishap with two consecutive BigBadaBooms - two of my Babelfish M25 amps gone Dodo
after thorough rethinking of everything and again of everything, I couldn't find a flaw neither in damn schematic which some Wakoo drawn, nor in physical execution of amps (which same some Wakoo did)
in the end - culprit is mistake made by one well known EU parts vendor - I ordered proper Keratherm, I paid proper Keratherm, they took money for proper Keratherm, while I finally got some other -lousier for our needs- Keratherm - being so close in color that I didn't even check what's written on bag sticker
only when I went in desparation mode, looking for clues, I read what's written on sticker
result- checked in vivo, using decent DMM with Temp. function, equipped with proper contact sensor (no more IR guns for me - not precise enough) - difference between Keratherm 86/82 and that one I got was - exactly 20C , measuring on very Drain mosfet pin
not big difference, if lazily looking , but world big difference , when mosfet is deciding to sing forever happy , or go fast in flames
no biasing mech capable to follow that thermal runaway and hold it
go figure - I was supersmart - going to spend some more greens, instead of using simple but crude Mica+Goop ............ while being also supersmarta$$, not checking usual things needing check ....... now see work involved, and I'm superhappy that blown woofers were of relatively small and less expensive variety
from now on, every amp I'm going to make for someone else is going out of my workshop with loudspeaker protection
excuse my longish rant
few months ago, I had a mishap with two consecutive BigBadaBooms - two of my Babelfish M25 amps gone Dodo
after thorough rethinking of everything and again of everything, I couldn't find a flaw neither in damn schematic which some Wakoo drawn, nor in physical execution of amps (which same some Wakoo did)
in the end - culprit is mistake made by one well known EU parts vendor - I ordered proper Keratherm, I paid proper Keratherm, they took money for proper Keratherm, while I finally got some other -lousier for our needs- Keratherm - being so close in color that I didn't even check what's written on bag sticker
only when I went in desparation mode, looking for clues, I read what's written on sticker
result- checked in vivo, using decent DMM with Temp. function, equipped with proper contact sensor (no more IR guns for me - not precise enough) - difference between Keratherm 86/82 and that one I got was - exactly 20C , measuring on very Drain mosfet pin
not big difference, if lazily looking , but world big difference , when mosfet is deciding to sing forever happy , or go fast in flames
no biasing mech capable to follow that thermal runaway and hold it
go figure - I was supersmart - going to spend some more greens, instead of using simple but crude Mica+Goop ............ while being also supersmarta$$, not checking usual things needing check ....... now see work involved, and I'm superhappy that blown woofers were of relatively small and less expensive variety
from now on, every amp I'm going to make for someone else is going out of my workshop with loudspeaker protection
excuse my longish rant
Ben - Awesome paper! Thank you. Saved to my education file.
That seems to jive reasonably well with the article Ben posted.
0.34 x 1.25 (derating for ΔT from 30C) = 0.43
0.43 x 0.67 (correction factor from 3 to 7 inches) = 0.29
So, in your real world example, you're far closer to ideal / expected performance if we assume the HUSA sinks behave close enough to the paper.
ZM - Very sorry to hear about the kaboom.
I'm totally relaxed, I took no offense at all. I truly appreciate all posts. Thus the emoji. I'll work on being a bit more concise. Walls of text aren't the easiest to digest.
Lynn - thanks for including me in the discussion / experiments. I've learned a ton!
Off for another coffee. My brain hurts. On a happy note, the BA-3 at 0A75 sounds just lovely with early morning tunes played softly to not wake wifey. Been listening for a few hours while reading the article Ben posted and fiddling.
That seems to jive reasonably well with the article Ben posted.
0.34 x 1.25 (derating for ΔT from 30C) = 0.43
0.43 x 0.67 (correction factor from 3 to 7 inches) = 0.29
So, in your real world example, you're far closer to ideal / expected performance if we assume the HUSA sinks behave close enough to the paper.
ZM - Very sorry to hear about the kaboom.
I'm totally relaxed, I took no offense at all. I truly appreciate all posts. Thus the emoji. I'll work on being a bit more concise. Walls of text aren't the easiest to digest.Lynn - thanks for including me in the discussion / experiments. I've learned a ton!
Off for another coffee. My brain hurts. On a happy note, the BA-3 at 0A75 sounds just lovely with early morning tunes played softly to not wake wifey. Been listening for a few hours while reading the article Ben posted and fiddling.
Lynn - this is the current summary. I don't think I'm going to press too much further in terms of temps. It's totally fine as personal preference for me re: the ability to touch the sinks for as long as I'd like. I can see ~50W of dissipation per sink => 100W per side being sort of a "sweet spot" for me, but I could push it further. I just might for my own curiosity. 
Summary - If I use the 25% de-rating along with the measured ambient temps close to the amp after it's been sitting all night with the temp on the fins vs. on the flat body of the sinks; the measurements agree reasonably well with the prediction.
Hope that's helped a bit. Hopefully some others can share some measurements from their amps. Again, thanks!
Summary - If I use the 25% de-rating along with the measured ambient temps close to the amp after it's been sitting all night with the temp on the fins vs. on the flat body of the sinks; the measurements agree reasonably well with the prediction.
Hope that's helped a bit. Hopefully some others can share some measurements from their amps. Again, thanks!
Last edited:
It's impossible to measure core temp of FET, so I recently tested 3 different thermal paste on my GPU (GTX1070Ti). Stock paste, MX-4 and Kryonaut. It did not make any meaningful difference at low load (45-55C), but when overclock is set to max, 75C with stock paste and 65C with Kryonaut. I see about 10C difference. So I do believe it's a good idea to use lowest thermal resistant material if core temp of FET is expected to be very high. Even pressure is also important, so I'm making simple DIY FET clamps for Tokin.
https://www.nxp.com/docs/en/application-note/AN3789.pdf
I know I'm a bit paranoid about this, but there is no long term report for high biased Tokin.
few months ago, I had a mishap with two consecutive BigBadaBooms - two of my Babelfish M25 amps gone Dodo
after thorough rethinking of everything and again of everything, I couldn't find a flaw neither in damn schematic which some Wakoo drawn, nor in physical execution of amps (which same some Wakoo did)
in the end - culprit is mistake made by one well known EU parts vendor - I ordered proper Keratherm, I paid proper Keratherm, they took money for proper Keratherm, while I finally got some other -lousier for our needs- Keratherm - being so close in color that I didn't even check what's written on bag sticker
only when I went in desparation mode, looking for clues, I read what's written on sticker
result- checked in vivo, using decent DMM with Temp. function, equipped with proper contact sensor (no more IR guns for me - not precise enough) - difference between Keratherm 86/82 and that one I got was - exactly 20C , measuring on very Drain mosfet pin
not big difference, if lazily looking , but world big difference , when mosfet is deciding to sing forever happy , or go fast in flames
no biasing mech capable to follow that thermal runaway and hold it
go figure - I was supersmart - going to spend some more greens, instead of using simple but crude Mica+Goop ............ while being also supersmarta$$, not checking usual things needing check ....... now see work involved, and I'm superhappy that blown woofers were of relatively small and less expensive variety
from now on, every amp I'm going to make for someone else is going out of my workshop with loudspeaker protection
excuse my longish rant
What circuit do you use for this speaker protection? I've always suspected (no real data to support this, just ocd likely) they color sound. If you selected one, I feel certain there are no or minimal as possible sound effects?
Thanks for any help on this, I am familiar with store PCB but wondering what Mighty Zenmod uses in his creations! You can PM if off topic
Thanks in advance,
Russellc
Tokin is so damn big , so no worries with 50-60-70W of dissipation per case
you can use cardboard as isolator, it'll still survive
Plain vanilla uPC1237 circuit, but my own pcb - everything on board, including small 3VA pcb xformer
relays in speaker hot
I don't care for that difference someone can hear when including decent relays in speaker line
life is too short ..... or, I'm choosing other areas to fully develop my OCD
It is still possible to find mica sheets on Ebay.com. I use it for the Tokin 2SK182ES SITs. It is easy to cut with scissors, and with a razor blade or Exacto knife the sheet can be separated into layers less than 2mils thick. Care must be taken to make holes for the screws; Dremel tool works here. I found that the Mouser 561-SW6 nylon shoulder washers would work for insulating the mounting screws from the Tokin case.
you can use cardboard as isolator, it'll still survive
LOL. Tokin is stress-free FET, indeed. I sometimes find 10A is going though it by my typical ADHD mistake, but I don't have to panic.
I know I am late to the heatsink party, but wanted to provide some more info for triangulation:
I have built my SissySIT into a 5U/400 case. In my build, it runs at up to about 110W per side. Even then, I can hold my hands on the heatsink just still indefinitely. So that should be close to 50 deg C at the top of the sinks, or a little over 25 deg C temperature rise.
Regards, Claas
I have built my SissySIT into a 5U/400 case. In my build, it runs at up to about 110W per side. Even then, I can hold my hands on the heatsink just still indefinitely. So that should be close to 50 deg C at the top of the sinks, or a little over 25 deg C temperature rise.
Regards, Claas