Hi All,
I'm wanting someone to "clear" up the rather differing opinions on the net about heatsinking, etc....
I was looking at
http://www.conradheatsinks.com/products/double_f.html
the last one on that page, it's PERFECT - what i want!
However, @ 0.25c/w @ 80deg above ambient, it doesn't seem so good to me...
What i've got is MJ21193's/94's that have a a junction of 0.7c/w.... PLUS mica washer and heatsink compound (people quote anywhere from 0.4c/w up to 1.2c/w) as the "rating"... So 'best' case scenario is 0.7+0.4+0.25 = 1.35c/w... Surely that can't be right? I'd assume that it would be better than that....
I WAS looking to run class A up to about 150w... - say 140 and have the last little bit as a/b - this means i have to get rid of a fair bit of heat, if i have a heatsink that has 0.05c/w+0.7c/w+0.4c/w still equals 1.05c/w...... which still puts my heatsink's above maximum temp... It's NEVER negative temp here, often 30+!! So that just doesn't work! I'm using 4 pairs of outputs per channel (i will jump to 5 pairs if it means better cooling!)... So that heatsink is excellent for the job, i could use 2 of them, but then it's the same price as another heatsink that i was looking at (which has an unknown c/w till we test it!). But even with 0c/w it's still not possible to run that much power through the heatsink! Maybe i'm misreading it, or miscalculating the values? The 'other' heatsink works out to about 333mmx500mm per channel, so it's BIG, but would require mounting brackets, etc for to-3's, however this has them already on there so looks good to me (and would work out at half the price) someone advise please!
Thanks heaps all
Aaron
I'm wanting someone to "clear" up the rather differing opinions on the net about heatsinking, etc....
I was looking at
http://www.conradheatsinks.com/products/double_f.html
the last one on that page, it's PERFECT - what i want!
However, @ 0.25c/w @ 80deg above ambient, it doesn't seem so good to me...
What i've got is MJ21193's/94's that have a a junction of 0.7c/w.... PLUS mica washer and heatsink compound (people quote anywhere from 0.4c/w up to 1.2c/w) as the "rating"... So 'best' case scenario is 0.7+0.4+0.25 = 1.35c/w... Surely that can't be right? I'd assume that it would be better than that....
I WAS looking to run class A up to about 150w... - say 140 and have the last little bit as a/b - this means i have to get rid of a fair bit of heat, if i have a heatsink that has 0.05c/w+0.7c/w+0.4c/w still equals 1.05c/w...... which still puts my heatsink's above maximum temp... It's NEVER negative temp here, often 30+!! So that just doesn't work! I'm using 4 pairs of outputs per channel (i will jump to 5 pairs if it means better cooling!)... So that heatsink is excellent for the job, i could use 2 of them, but then it's the same price as another heatsink that i was looking at (which has an unknown c/w till we test it!). But even with 0c/w it's still not possible to run that much power through the heatsink! Maybe i'm misreading it, or miscalculating the values? The 'other' heatsink works out to about 333mmx500mm per channel, so it's BIG, but would require mounting brackets, etc for to-3's, however this has them already on there so looks good to me (and would work out at half the price) someone advise please!
Thanks heaps all
Aaron
Hi,
your heatsink formula is correct, FOR ONE transistor disspating all the power.
Build yourself a calculation table to insert values into
total power, Heatsink Rsa,Td=Pt*Rsa
device power, device Rcs,Td=Pd*Rcs
device power, device Rjc,Td=Pd*Rjc
choose a Ta to suit your location, season(summer?) & adjacent heat sources 25C to 40C.
Now start adding up temperature differences(Td) to ambient
to give heatsink(Ts), case(Tc), junction(Tj).
hint Ts max 50C to 70C depending on risk of burning?
Tc determines your derating factor from datasheets
Tj max 100C to 125C depending on reliability.
BTW you may need to get two or three of those 0.25C/W sinks per channel to meet your requirements.
regards Andrew T.
your heatsink formula is correct, FOR ONE transistor disspating all the power.
Build yourself a calculation table to insert values into
total power, Heatsink Rsa,Td=Pt*Rsa
device power, device Rcs,Td=Pd*Rcs
device power, device Rjc,Td=Pd*Rjc
choose a Ta to suit your location, season(summer?) & adjacent heat sources 25C to 40C.
Now start adding up temperature differences(Td) to ambient
to give heatsink(Ts), case(Tc), junction(Tj).
hint Ts max 50C to 70C depending on risk of burning?
Tc determines your derating factor from datasheets
Tj max 100C to 125C depending on reliability.
BTW you may need to get two or three of those 0.25C/W sinks per channel to meet your requirements.
regards Andrew T.
You should be considering using forced air cooling because there are too many heatsinks.
At 150W dissipation per channel, you would need two 300 x 150mm Conrad sinks per channel, each around 0.3C/watt. At 75W dissipated per heatsink, this is quite acceptable for four devices per rail suitably placed across the ledge or backing as it's technically only about a 25C rise above ambient....
The better option is to use two Conrad MF18 series, which are designed for forced air circulation. End on, fins facing each other, you have a 120mm x 120mm window ideally suited to a fan of the same dimensions, operating via a plenum chamber to cut turbulent flow and noise. You can then dissipate up to 175W per sink - 350W for the two - with only a measured 27C rise above ambient using a 48V fan operating at half voltage. You won't hear it beyond about five feet if the fan is insulated and the plenum set up properly (use a Papst for less than 20dbA of noise). I've done this work, checked all the power ratings, and it's very satisfactory, compact, and quiet as a churchmouse. And two MF18-151.5 are much cheaper than four MF30-151.5!!
Cheers,
Hugh
At 150W dissipation per channel, you would need two 300 x 150mm Conrad sinks per channel, each around 0.3C/watt. At 75W dissipated per heatsink, this is quite acceptable for four devices per rail suitably placed across the ledge or backing as it's technically only about a 25C rise above ambient....
The better option is to use two Conrad MF18 series, which are designed for forced air circulation. End on, fins facing each other, you have a 120mm x 120mm window ideally suited to a fan of the same dimensions, operating via a plenum chamber to cut turbulent flow and noise. You can then dissipate up to 175W per sink - 350W for the two - with only a measured 27C rise above ambient using a 48V fan operating at half voltage. You won't hear it beyond about five feet if the fan is insulated and the plenum set up properly (use a Papst for less than 20dbA of noise). I've done this work, checked all the power ratings, and it's very satisfactory, compact, and quiet as a churchmouse. And two MF18-151.5 are much cheaper than four MF30-151.5!!
Cheers,
Hugh
Thanks for the formula Andrew 
i'll have to work it out!
Hugh - thanks, i HAD thought of that, i'm TRYING to avoid using fan's, but it looks like it's the only choice... My amp has 3 channels in it, so there would have to be at least 2 fans i'd guess, the amp is able to be ANYTHING up to 500x500x500, that's why i've got so much heatsink area to work with!!!
I could use 3 MF18's (151) and then have fans on each, in a sealed case (with only entry and exit for air, which was planned to be at the base of the case, so the toroids, etc get cooled as well...)
However, how do i mount to-3 devices to the MF18's? Do i have to get it "made" that way, or do i get something made up that allows me to bolt them to it? That's one of the things that's been getting me about the other ones, they already have everything there ready to go (flange)...?
Thanks for the help so far!
Aaron
i'll have to work it out!Hugh - thanks, i HAD thought of that, i'm TRYING to avoid using fan's, but it looks like it's the only choice... My amp has 3 channels in it, so there would have to be at least 2 fans i'd guess, the amp is able to be ANYTHING up to 500x500x500, that's why i've got so much heatsink area to work with!!!
I could use 3 MF18's (151) and then have fans on each, in a sealed case (with only entry and exit for air, which was planned to be at the base of the case, so the toroids, etc get cooled as well...)
However, how do i mount to-3 devices to the MF18's? Do i have to get it "made" that way, or do i get something made up that allows me to bolt them to it? That's one of the things that's been getting me about the other ones, they already have everything there ready to go (flange)...?
Thanks for the help so far!
Aaron
With a bit of care you might be able to run live heatsinks and dispense with the mica washer, especially if your output stage is an emitter follower meaning the collectors and therefore the heatsinks don't flap up and down with the signal.
A fan will improve the efficacy of a heatsink by 5:1 or even 6:1.
90% of the improvement occurs by the time the fan is running at 75% of rated voltage. Increasing from 100% to 200% rated fan voltage
makes barely 5% further improvement.
A fan will improve the efficacy of a heatsink by 5:1 or even 6:1.
90% of the improvement occurs by the time the fan is running at 75% of rated voltage. Increasing from 100% to 200% rated fan voltage
makes barely 5% further improvement.
here's a link to Thermalloy's "Thermal Resistance Tool"
http://www.aavidthermalloy.com/technical/thermal.shtml
-- for ambient use 32 LFM. In my 2x120 Watt Bridged Amp I use a pair of 11" x4" heat sinks with 10 fins (each fin is 1.3" high) -- the two heat sinks are joined together with aluminum so that they allow a tunnel in which a computer fan directs traffic. The heat sinks stay at about 106 degrees Fahrenheit.
if you run a device too close to its maximum the mean-time-between failure jumps.
http://www.aavidthermalloy.com/technical/thermal.shtml
-- for ambient use 32 LFM. In my 2x120 Watt Bridged Amp I use a pair of 11" x4" heat sinks with 10 fins (each fin is 1.3" high) -- the two heat sinks are joined together with aluminum so that they allow a tunnel in which a computer fan directs traffic. The heat sinks stay at about 106 degrees Fahrenheit.
if you run a device too close to its maximum the mean-time-between failure jumps.
Hmmm....
I'm very much liking the idea of running fan forced now, the capability is much much better, nothing like normal air!
I'd like them to run as cool as i can, i can run 2 heatsinks together, but the 3rd one will be on it's own, which makes me wonder if i should make them all the same design, etc, not really sure how a 12x12cm fan will work with a 12x6 heatsink though...
Aaron
I'm very much liking the idea of running fan forced now, the capability is much much better, nothing like normal air!
I'd like them to run as cool as i can, i can run 2 heatsinks together, but the 3rd one will be on it's own, which makes me wonder if i should make them all the same design, etc, not really sure how a 12x12cm fan will work with a 12x6 heatsink though...
Aaron
the other thing is that you don't have to run the fan at 12V -- it will run much more quietly (but push less air) at a lower voltage -- so get a cooking thermometer (those metal ones they sell at Bed n' Bath are pretty acurate and they go to over 220 degrees F) -- measure temperature on the heat sink with the fan at 12Volts, then reduce the voltage, allow the system to settle for a couple minutes anda take the temperature again.
Here is hot, damn hot, everything different from transistor direct over heatsink will
Create some heat transference loss.
I do all my amplifiers using transistors direct to heatsink, using thermical grease, and sometimes flattening the transistor surface using a very thin sand paper.
3 by 3 inches each 10 watts consumption....this is a fin size to 10 watts heat...not power output...consumption.... amperes multiplied by supply voltage...each rail calculation... at maximum power and minimum possible load.
This works guaranteed...have only to calculate:
9 square inches correspond to 10 watts heat transfer to air
X square inches will correspond to some 300 watts consumption amplifier, for instance.
9 multiplied by 300 = 10 multiplied by (X)
inverting sides...to be more agreable :
10X= 2700 .... do.... 270 square inches will be enought to 30 degrees centigrades environment, without fan.
If heatsink is outside.
If you have some feet under the amplifier.
If you have condition to air flow from down to up behind some platform were the amplifier is installed
And minimum of 3 inches of distance up the cabinet to internal heat sinking
Back heat sinks must be free...nothing covering from its natural position , having space up to hot air move up....air must flow without resistance.
This is 45 years experience result...if not perfect, will work guaranteed
Carlos
Create some heat transference loss.
I do all my amplifiers using transistors direct to heatsink, using thermical grease, and sometimes flattening the transistor surface using a very thin sand paper.
3 by 3 inches each 10 watts consumption....this is a fin size to 10 watts heat...not power output...consumption.... amperes multiplied by supply voltage...each rail calculation... at maximum power and minimum possible load.
This works guaranteed...have only to calculate:
9 square inches correspond to 10 watts heat transfer to air
X square inches will correspond to some 300 watts consumption amplifier, for instance.
9 multiplied by 300 = 10 multiplied by (X)
inverting sides...to be more agreable :
10X= 2700 .... do.... 270 square inches will be enought to 30 degrees centigrades environment, without fan.
If heatsink is outside.
If you have some feet under the amplifier.
If you have condition to air flow from down to up behind some platform were the amplifier is installed
And minimum of 3 inches of distance up the cabinet to internal heat sinking
Back heat sinks must be free...nothing covering from its natural position , having space up to hot air move up....air must flow without resistance.
This is 45 years experience result...if not perfect, will work guaranteed
Carlos
Heatsinks are mounted over wood pieces, this way no conduction, no short between them
They are isolated from chassis and one related the other.
The bad part is that heatsinks turn alive...more than 40 volts positive and 40 volts negative can be a little annoying...if you have amplifiers using 69 positive volts and 69 negative volts, as you already notice 138 volts can kill someone.
So, this is not indicated, not good, to hi voltage amplifiers...this is adequated to class A amplifiers, using low voltage and hi current.
Carlos
They are isolated from chassis and one related the other.
The bad part is that heatsinks turn alive...more than 40 volts positive and 40 volts negative can be a little annoying...if you have amplifiers using 69 positive volts and 69 negative volts, as you already notice 138 volts can kill someone.
So, this is not indicated, not good, to hi voltage amplifiers...this is adequated to class A amplifiers, using low voltage and hi current.
Carlos
Nuttr,
All good advice here!
Like the idea of a cooking thermometer!
Graham (aka Circlotron) is dead right about the improvement in cooling, though I'd suggest the factor is closer to three times.
Using flat backed MF18s, you can't use TO3. An angle mounting section suffers huge thermal impedance at the junction of the section and the heatsink; don't do it. Even Al welded sections are poor transmitters of heat. You need to use plastic pack transistors; TO-3P or TO-247. And if you are using Class A with bipolar, I wouldn't be running each device over about 35W apiece, even if they are rated at 125W.
And Jackinj is right on the money about MTBF. As you reduce junction temperature from 125C, the usual rated max, each 10C increment doubles MTBF. So, if MTBF at 125C junction is 2000 hours, let's say, life at 105C is 4000 hours, and at 95C 8000 hours, and so on. It pays to run 'em cooler, and junction is often about to 25C higher than case temperature........
Cheers,
Hugh
All good advice here!
Like the idea of a cooking thermometer!
Graham (aka Circlotron) is dead right about the improvement in cooling, though I'd suggest the factor is closer to three times.
Using flat backed MF18s, you can't use TO3. An angle mounting section suffers huge thermal impedance at the junction of the section and the heatsink; don't do it. Even Al welded sections are poor transmitters of heat. You need to use plastic pack transistors; TO-3P or TO-247. And if you are using Class A with bipolar, I wouldn't be running each device over about 35W apiece, even if they are rated at 125W.
And Jackinj is right on the money about MTBF. As you reduce junction temperature from 125C, the usual rated max, each 10C increment doubles MTBF. So, if MTBF at 125C junction is 2000 hours, let's say, life at 105C is 4000 hours, and at 95C 8000 hours, and so on. It pays to run 'em cooler, and junction is often about to 25C higher than case temperature........
Cheers,
Hugh
AndrewT said:
as a part-time exegete, I don't know if you can always take the gospel as gospel, but it is "gut spiel".
There's a great book by one of the engineers from Tellabs, Tony Kordyban, called "Hot Air Rises, Heat Sinks" -- he coined the term "Therminator" for the computer simulator which they used to predict failures. After reading this very amusing book I conclude that the MTBF figures are serious but depend upon dozens of exogenous factors -- and the one which the DIYr doesn't have to put at the top of the list is "Cost".
The semiconductor conductors often don't agree among themselves!
Hi
about TO3 mounting this is the way I did it:
Not class A amp but cooling seems to be working, this have only been tested on bench though.
more pics: http://web.comhem.se/~u33113170/p3a/
this is another setup I will try for my future DOZ: http://web.comhem.se/~u33113170/doz/sink1.jpg
it is not finnished but I hope you get the idea, which is to have a big surface between mounting section and heatsink.
about TO3 mounting this is the way I did it:
An externally hosted image should be here but it was not working when we last tested it.
Not class A amp but cooling seems to be working, this have only been tested on bench though.
more pics: http://web.comhem.se/~u33113170/p3a/
this is another setup I will try for my future DOZ: http://web.comhem.se/~u33113170/doz/sink1.jpg
it is not finnished but I hope you get the idea, which is to have a big surface between mounting section and heatsink.
Hi Andrew,
Figures vary widely for different devices, but no, the life quoted at 125 junction C is probably about five times less than it should.
However, all the ratios are correct. The 10C is based on Arrhenius calculations from the speed of organic chemical reactions. In this situation they relate to the migration of doping within a hot junction. Eventually the doping issues compromise operation and the device fails.
Cheers,
Hugh
Figures vary widely for different devices, but no, the life quoted at 125 junction C is probably about five times less than it should.
However, all the ratios are correct. The 10C is based on Arrhenius calculations from the speed of organic chemical reactions. In this situation they relate to the migration of doping within a hot junction. Eventually the doping issues compromise operation and the device fails.
Cheers,
Hugh
Keld, in you environment...hehe....do not need so much heat sinking as here.
If you use some heavy gauge bar....a knife large bar.... screw into your heatsink, and under the window the bar touch the surrounded air.... It maybe turn your transistors cold....as your outside temperature is low in winter.
I think, really!, that in you case... environment can give you enormous help.... fan in summer...and some kind of heat transfer to environment can be great..... flexible copper, from the heatsink, and going outside using some small window aperture?... will that work?
Carlos
If you use some heavy gauge bar....a knife large bar.... screw into your heatsink, and under the window the bar touch the surrounded air.... It maybe turn your transistors cold....as your outside temperature is low in winter.
I think, really!, that in you case... environment can give you enormous help.... fan in summer...and some kind of heat transfer to environment can be great..... flexible copper, from the heatsink, and going outside using some small window aperture?... will that work?
Carlos
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.