Ok...as far as placement on a heatsink goes...where is the best place to put transistors? Say the heatsink is upright, fins perpendicular to the floor ...is the best place in the middle..height wise? lowish? or highish of center? I would think lowish of center would work best but i've heard highish works the best. Highish being on the jxn of the middle and top 1/3 heightwise of the heatsink. Thanks ya'll! Hope I was clear enough 
-Matthew K. Olson
-Matthew K. Olson
Hi Matthew,
transistor placement on a heatsink has been discussed before on this forum. One of the members gave a clear argument that a transistor should be placed at top of the sink, while noting that most designs he had seen to date had placed the devices in the centre or lower part of the sink. Try a search on the Pass or solid state forums. You may be able to find the original and not have to rely on my dodgy recollections...
James
transistor placement on a heatsink has been discussed before on this forum. One of the members gave a clear argument that a transistor should be placed at top of the sink, while noting that most designs he had seen to date had placed the devices in the centre or lower part of the sink. Try a search on the Pass or solid state forums. You may be able to find the original and not have to rely on my dodgy recollections...
James
I think right in the middle is best. If we assume the thermal resistance of the heat sink and the mounting pad is nil, then placement at the center allows for the most even distribution of heat, which will therefore allow for the best cooling via radiation.
I can see why some would think mounting a little low might be better, as heat tends to rise. If the thermal resistance of the heat sink is not close to 0, then this might make sense. But, most of us DIY'ers use massive heat sinks so I think that the ideal location would be very close to the center anyway. I suppose one could experiment with both locations, but I think that for a transistor mounted very much off center, the heat distribution will not be so uniform, and this equates hot areas.
I am just guessing here - so others in the know might have some real data to provide.
I can see why some would think mounting a little low might be better, as heat tends to rise. If the thermal resistance of the heat sink is not close to 0, then this might make sense. But, most of us DIY'ers use massive heat sinks so I think that the ideal location would be very close to the center anyway. I suppose one could experiment with both locations, but I think that for a transistor mounted very much off center, the heat distribution will not be so uniform, and this equates hot areas.
I am just guessing here - so others in the know might have some real data to provide.
Mattyo5 said:I have searched actually...but didn't find anything...but thanks for the reply. A friend of mine said placing them towards the top was a good idea. I will do just that. Thanks
-Matthew K. Olson
Hi Matthew,
found it -
http://www.diyaudio.com/forums/showthread.php?postid=40578#post40578
You may want to ask wuffwaff for clarification.
James
heatsinks use both radiation and convection to get rid of heat. (well they also use conduction of course...). i'm guessing the reason placing a transistor toward the top is favored is because it helps with the chimney effect (convection). but i don't know any of the formulae involved, so maybe there's a tradeoff between convection and radiation there.
/andrew
/andrew
I think mounting to geometric center is best. That way you would get the best conduction of heat from the device to the sink. As for convection, I doubt moving the device around affects the convection heat transfer coefficient very much. And radiation is minor at the temperatures we are talking about and not worth considering.
If you examine quality car audio amplifiers, their
transistor placement is not in the center, if any
thing, it's located on the the edge of the pcb,
but quality car amplifiers are pretty robust and
take major abuse, plus ambient temperatures
due to climate pre-heat the fins, in any case,
these abnormal placement of transistors don't
seem to bother these amplifiers, even kilowatt
ones
If your heatsinks are high in mass, then I wouldn't
trip out on location.
Do the propane torch test.
Take a piece of metal. fire up your propane torch and
heat the edge of the metal. Place finger on metal 1 inch below
where the flame hits, tell me if your finger burns
Do the same by heating up the center of the metal,
does your finger burn ?
But if I had to choose, I choose center just because .... lol ....
transistor placement is not in the center, if any
thing, it's located on the the edge of the pcb,
but quality car amplifiers are pretty robust and
take major abuse, plus ambient temperatures
due to climate pre-heat the fins, in any case,
these abnormal placement of transistors don't
seem to bother these amplifiers, even kilowatt
ones
If your heatsinks are high in mass, then I wouldn't
trip out on location.
Do the propane torch test.
Take a piece of metal. fire up your propane torch and
heat the edge of the metal. Place finger on metal 1 inch below
where the flame hits, tell me if your finger burns
Do the same by heating up the center of the metal,
does your finger burn ?
But if I had to choose, I choose center just because .... lol ....
Also I believe it can`t be very wrong placing transistors into the geometrical middle of a heatsink but after reading and rethinking that idea about placing them rather in the "upper" half of a sink, this makes more and more sense to me.
Think about this way: the surface of a heatsink around the transistors is always somewhat hotter than the rest. When the transistors are mounted in the lower half, the upper half of the heatsink is cooler so the heated (and hotter) air in proximity of the "lower" heatsink parts want to move upwards whereas the cooler air around the "upper" area is blocking (reducing) that convection as it rather wants to move down (because it`s cooler).
At least theoretically this makes really sense to me and speaks for mounting transistors in the "upper" area actually.
If at all and how far this accounts for in the real world would be very interesting to know.
Think about this way: the surface of a heatsink around the transistors is always somewhat hotter than the rest. When the transistors are mounted in the lower half, the upper half of the heatsink is cooler so the heated (and hotter) air in proximity of the "lower" heatsink parts want to move upwards whereas the cooler air around the "upper" area is blocking (reducing) that convection as it rather wants to move down (because it`s cooler).
At least theoretically this makes really sense to me and speaks for mounting transistors in the "upper" area actually.
If at all and how far this accounts for in the real world would be very interesting to know.
cocolino,
I think from analyzing the convection aspect of the problem you're right.
But in this problem, there's conduction, convection, and radiation. And, at these temperatures, the heat transfer coefficients for each will be in that order: conduction, convection, and radiation. So, the simple analysis (not always right) is that you get the biggest improvement when you help the biggest contributor. In this case, that is the conduction from device to heatsink. If we improve that the most, then that is the best solution.
I think from analyzing the convection aspect of the problem you're right.
But in this problem, there's conduction, convection, and radiation. And, at these temperatures, the heat transfer coefficients for each will be in that order: conduction, convection, and radiation. So, the simple analysis (not always right) is that you get the biggest improvement when you help the biggest contributor. In this case, that is the conduction from device to heatsink. If we improve that the most, then that is the best solution.
Hi leadbelly,
the conduction of heat from the device through the insulation pad into the heatsink and from there into every part of the sink is one aspect. Without convection the conduction could be as good as it wants - it wouldn`t help much to cool the devices (assumed taking aside radiation effect).
Imagine a transistor/insulation/heatsink arrangement with a thermal conductivity of infinity (of course not possible in the real world) but with zero thermal conductivity from heatsink to air the heatsinks temperature would saturate on the transistor temperature after a while - but it would not cool.
Therefore I would assume that the position of mounting the transistors does not negatively effect conductivity (at least not for steady state power dissipation) but maybe could help convection.
So the abilty of a heatsink to actually cool the devices not only depends on conductivity but it is at least as important how good the heat can be transfered to air and in turn this air away from the heatsink, respectively to bring as much amount as possible of "fresh"-air in contact with the heatsink surface - and this is what convection does.
We all know forced convection (fans) can increase the max. power dissipation ability by a very large extend and it would be great to find a way to increase convection (even only a slight increase might bring a big improvement in the max. dissipation) by mechanical arrangement of heatsinks or transistor mounting only, without using fans (I´m not a fan of fans
).
Hi,
As You can see, I live in Hungary, here we have only one company which makes heatsinks. I know some guys from there, and they gave me some documents about the usage of the heatsinks.
There are some interesting facts.
If You use only one transistor, the best placing is the middle of the heatsink.
If You use two of them, the placing is more interesting, from the top of the heastsink:
Top - 15% - first tr. - 70% - second tr. -15% - Bottom
(Ihope it's clear)
For three devices it's more complicated. If anybody interesting i will scan it, and put up to my homepage.
Sajti
As You can see, I live in Hungary, here we have only one company which makes heatsinks. I know some guys from there, and they gave me some documents about the usage of the heatsinks.
There are some interesting facts.
If You use only one transistor, the best placing is the middle of the heatsink.
If You use two of them, the placing is more interesting, from the top of the heastsink:
Top - 15% - first tr. - 70% - second tr. -15% - Bottom
(Ihope it's clear)For three devices it's more complicated. If anybody interesting i will scan it, and put up to my homepage.
Sajti
sajti said:Top - 15% - first tr. - 70% - second tr. -15% - Bottom
This is the way I positioned the transistors for my AlephX.
It's test running for days now, and they don't get too hot.
I used Sil-Pad2000 from Bergquist, which is fairly expensive (€ 1.85 each) but has good thermal ratings.
It will look a bit like the Peter_Daniel_prototype_setup_and_wiring.
Does the company also have a valid explanation why to put them like they explain? Could you check that? In building Aleph's the wiring would become much longer then I prefer. That’s why I made a 'group' on the top part of the sink.
/Hugo
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.