Based on all that I've read the 1 common thing needed in class-A amp's is mega heatsink's There have been a few group buy's ect... But these HS are are a major cost problem! I am no engineer as most probabily know from my other post's! But let's say for example you have 6 mosfet's running at 35v high bias ect.. bolted to a 6" x 12" x 2" HS with a 3/8 thick plate & 1/8 fins 1/4 apart. Each mosfet is 2" apart, now do these create a hotspot on the aluminum, or does the aluminum pull heat fast enough to even thing's out? I have seen people use angle for mounting devices ect.. I have this idea or question if you will* If copper is a better conductor could we then take a 1/2" x 2" x 8" plate ect.. bolt it to the 3/8 back of Hs and better distribute and/or dissipate heat ?
As far as calculations on heat sink size, I can't help you. Passdiy.com has a gallery that would give you an idea. 35V, 6 devices, biased @ 3 amps total, will give you about 35W dissipation per device, and about 210W dissipation for one channel. This would be considered a "high bias" Aleph 5 running 64W music power; which at first doesn't sound like a lot, but trust me, its a high quality 60 watts, and usually more than enough. There are plenty of heat sink simulators on the net that will give you an idea of the size you need. The best way is to find the biggest heat sinks you can afford, maybe on ebay, or surplus. That way you can experiment, and upscale the design if necessary. Your idea of adding another Cu plate to an existing heat sink is not recommended. With that, you’re increasing thermal resistance and actually lowering efficiency.
Hi,
a heatsink works fairly well if the thickness to radius is ~>=1/10.
your 3/8inch plate will dissipate well out to a radius from the device of nearly 4inches.
The same rule applies to the fins. If they are thinner than 1/10 their height, then the far end runs significantly cooler than the hot end and this reduces the heatsink efficiency significantly. if the fins are tapered one does not lose much dissipation capacity but saves a lot of aluminium and cost. Similarly some backplates are tapered to save weight & cost.
For best performance from a sink the whole sink should be at the same temperature. That's the way the manufacturer tests them. Your devices would need to be spread over the majority of the backplate surface to match the spec. The closer you can approach this the cooler your devices will run. But then your wires start to get too long. Compromise.
If you had a thin backplate then a copper spreader would be of some use. But your multiple devices spread along the 12inch width would see little if any benefit. (it's what the semi manufacturers do inside their transistors/FETs).
The devices will run cooler if the single row is located just below halfway. About 40% up from the bottom is about right for a row.
Look up papers by Wakefield et all for details of heatsinks theory and practice.
a heatsink works fairly well if the thickness to radius is ~>=1/10.
your 3/8inch plate will dissipate well out to a radius from the device of nearly 4inches.
The same rule applies to the fins. If they are thinner than 1/10 their height, then the far end runs significantly cooler than the hot end and this reduces the heatsink efficiency significantly. if the fins are tapered one does not lose much dissipation capacity but saves a lot of aluminium and cost. Similarly some backplates are tapered to save weight & cost.
For best performance from a sink the whole sink should be at the same temperature. That's the way the manufacturer tests them. Your devices would need to be spread over the majority of the backplate surface to match the spec. The closer you can approach this the cooler your devices will run. But then your wires start to get too long. Compromise.
If you had a thin backplate then a copper spreader would be of some use. But your multiple devices spread along the 12inch width would see little if any benefit. (it's what the semi manufacturers do inside their transistors/FETs).
The devices will run cooler if the single row is located just below halfway. About 40% up from the bottom is about right for a row.
Look up papers by Wakefield et all for details of heatsinks theory and practice.
The short answer is yes, devices will develop hot spots. The heat does conduct away from the devices, but not as quickly as it is produced. Thicker is better, but still not a cure all.
I am continually amazed when I see photos here showing devices mounted side by side, presumably in an attempt to have a "short signal path" or to avoid paying for a longer/larger PC board. Those who build like this invariably say something like, "Well, it hasn't blown up yet!" as though that excuses abusing the devices.
There was a thread not too long ago regarding a buy on heat sinks. Numerous posts in that thread showed that many involved had no idea about how to handle heat. They would even go so far as to include images in their posts from one manufacturer's online heat sink simulation program, depicting clear hot spots and (relatively) cold spots where little heat was being dissipated. The disappointing thing was that the posters were showing these images as 'proof' that they were managing heat well. Oh dear, oh dearie, dearie me...quite the contrary.
The punchline being that when I pointed out that all was not right in Denmark I was reviled and told to mind my own business.
Oh, well...
Grey
I am continually amazed when I see photos here showing devices mounted side by side, presumably in an attempt to have a "short signal path" or to avoid paying for a longer/larger PC board. Those who build like this invariably say something like, "Well, it hasn't blown up yet!" as though that excuses abusing the devices.
There was a thread not too long ago regarding a buy on heat sinks. Numerous posts in that thread showed that many involved had no idea about how to handle heat. They would even go so far as to include images in their posts from one manufacturer's online heat sink simulation program, depicting clear hot spots and (relatively) cold spots where little heat was being dissipated. The disappointing thing was that the posters were showing these images as 'proof' that they were managing heat well. Oh dear, oh dearie, dearie me...quite the contrary.
The punchline being that when I pointed out that all was not right in Denmark I was reviled and told to mind my own business.
Oh, well...
Grey
A number of our Forum members cannot accept constructive criticism.GRollins said:when I pointed out that all was not right in Denmark I was reviled and told to mind my own business.
But, that is their problem not your's.
It's just a pity many newbies follow their misinformed advice.
AndrewT said:A number of our Forum members cannot accept constructive criticism.
But, that is their problem not your's.
It's just a pity many newbies follow their misinformed advice.
Well, that's part of the explanation. The other half of the reason, that some strive to prove that their POV on thermal management is right.....is that it's usually pretty expensive to buy heatsinks and so forth, that actually are adequate
Magura
cloud said:Based on all that I've read the 1 common thing needed in class-A amp's is mega heatsink's There have been a few group buy's ect... But these HS are are a major cost problem! I am no engineer as most probabily know from my other post's! But let's say for example you have 6 mosfet's running at 35v high bias ect.. bolted to a 6" x 12" x 2" HS with a 3/8 thick plate & 1/8 fins 1/4 apart. Each mosfet is 2" apart, now do these create a hotspot on the aluminum, or does the aluminum pull heat fast enough to even thing's out? I have seen people use angle for mounting devices ect.. I have this idea or question if you will* If copper is a better conductor could we then take a 1/2" x 2" x 8" plate ect.. bolt it to the 3/8 back of Hs and better distribute and/or dissipate heat ?
Hello Cloud,
Consider looking at commercial products rather than other DIY projects for guidance. There are benefits of using a copper "spreader" bar for heat, but IMHO the benefits out weight the added cost of a heavy copper bar. There are also issues with making absolutely sure that both surfaces are machined perfectly. Also the spreader bar would need the mounting bolts torqued the same across. Unevenly torqued mounting bolts and any imperfections in the machining will create hot or cool spots.
A pair of Conrad Heatsink's 350x151.1 are nearly ideal for a stereo A30, F5, or mono A60. Manufacturer's C/W data is typically uncorrected for ambient room temperature. As an example, using the Conrad 350x151.1 has a factory C/W rating of .21. Corrected for typical ambient room temp with a factor of 1.4 gives a more real world C/W of .295.
I use Silicone Elastomer pads from Bergquist to mount Mosfets to the vertical mid-point of the heatsink. The results have been excellent trouble free listening for long periods of time.
-David
heat sink
http://www.aavidthermalloy.com/technical/thermal.shtml
This is a great tool. There is also a calculator to convert cubic feet to linear feet. Some quick experimentation will show the advantages of active cooling (fan).
http://www.aavidthermalloy.com/technical/thermal.shtml
This is a great tool. There is also a calculator to convert cubic feet to linear feet. Some quick experimentation will show the advantages of active cooling (fan).
Re: Conrad Heatsinks
Here's a picture of my A30 with the Conrad heat sinks: http://www.diyaudio.com/forums/showthread.php?s=&threadid=70622&highlight=1st+Aleph
Conrad products are well made, low cost, and have a powder coated finish rather than anodized. The finish will stay black forever rather than aging to a blue color over time with high heat.
The heat sink is actually 350mmx151.5 rather than 151.1 as quoted earlier. The part number is MF35-151.5, and located at the bottom of this page: http://www.conradheatsinks.com/products/flat100_350.html
I'd call Conrad rather than emailing them for faster service. Shipment by Aussie Post took a week to get to California.
The Bergquist Silicone Elastomer pads are available from Digikey, and far better than mica and does not require any goop.
Good luck on your build,
-David
SCD said:Hello David:
Do you have a reasonable source for those lovely heatsinks. I was hoping to get some with a recent group buy, but things appear to have fallen off the rails with group buy.
Any help would be appreciated.
Here's a picture of my A30 with the Conrad heat sinks: http://www.diyaudio.com/forums/showthread.php?s=&threadid=70622&highlight=1st+Aleph
Conrad products are well made, low cost, and have a powder coated finish rather than anodized. The finish will stay black forever rather than aging to a blue color over time with high heat.
The heat sink is actually 350mmx151.5 rather than 151.1 as quoted earlier. The part number is MF35-151.5, and located at the bottom of this page: http://www.conradheatsinks.com/products/flat100_350.html
I'd call Conrad rather than emailing them for faster service. Shipment by Aussie Post took a week to get to California.
The Bergquist Silicone Elastomer pads are available from Digikey, and far better than mica and does not require any goop.
Good luck on your build,
-David
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Amplifiers
- Pass Labs
- Heatsink Ideas 101