Hello,
I've been working on a project to build the high powered Aleph-X design and I've been concerned about heat sinks.
According to this site:
http://www.facstaff.bucknell.edu/esantane/movies/aleph-x.html
The high powered version dissipates 320w heat/channel.
He later goes on to describe the process for calculating heat sink requirements, for this case I've calculated:
25c/320w = .078 c/w
0.078/1.25 = .0625 c/w then round down to nearest hundredth:
.0625 = .06 c/w requirement
Is this number correct?
I had been interested in the Conrad Heatsink group buy. These looked like nice heat sinks at a good price. Initially I had thought this would require simply 4 heat sinks, two per monobloc, but reviewing the specs for this heat sink I am now uncertain. According to Conrad the MF35-151.5 model can dissipate 0.21c/w at 80c. I am unfamiliar with heat sink specs, but I am assuming this is a linear measure? That two of these heat sinks could dissipate .105c/w, four dissipating .05025c/w, meaning that I would need 4 of these per monobloc to satisfy the heat sink requirement. Is this correct? At $123/pair (x4 for 8 total) I really am thinking I cannot afford this. If someone could please confirm my calculations, or possibly offer suggestions for a cheaper heat sink source I would be very appreciative.
Ryan
I've been working on a project to build the high powered Aleph-X design and I've been concerned about heat sinks.
According to this site:
http://www.facstaff.bucknell.edu/esantane/movies/aleph-x.html
The high powered version dissipates 320w heat/channel.
He later goes on to describe the process for calculating heat sink requirements, for this case I've calculated:
25c/320w = .078 c/w
0.078/1.25 = .0625 c/w then round down to nearest hundredth:
.0625 = .06 c/w requirement
Is this number correct?
I had been interested in the Conrad Heatsink group buy. These looked like nice heat sinks at a good price. Initially I had thought this would require simply 4 heat sinks, two per monobloc, but reviewing the specs for this heat sink I am now uncertain. According to Conrad the MF35-151.5 model can dissipate 0.21c/w at 80c. I am unfamiliar with heat sink specs, but I am assuming this is a linear measure? That two of these heat sinks could dissipate .105c/w, four dissipating .05025c/w, meaning that I would need 4 of these per monobloc to satisfy the heat sink requirement. Is this correct? At $123/pair (x4 for 8 total) I really am thinking I cannot afford this. If someone could please confirm my calculations, or possibly offer suggestions for a cheaper heat sink source I would be very appreciative.
Ryan
Yeah, I meant to ask about the phenomena as well, cause I had briefly glanced at a post mentioning the need for such correction.
So the way I'm interpreting this now is that the ambient temperature is working against me right? 4 heat sinks now dissipate: .295/4 = .07375c/w, which doesn't meet my requirement of .06c/w, so I would be looking to have potentially another (total 5) of these heat sinks per monobloc?
So the way I'm interpreting this now is that the ambient temperature is working against me right? 4 heat sinks now dissipate: .295/4 = .07375c/w, which doesn't meet my requirement of .06c/w, so I would be looking to have potentially another (total 5) of these heat sinks per monobloc?
Even worse, place 2 heatsinks at a close distance and they'll cross-influence eachother's thermal efficiency number.
The "ambient" factor is not a constant temperature reference, for the thermal resistance calculation it's an average across the (3D) surface of the heatsink.
Put 2 heatsinks in-line and the heat coming from the "touching" side of the 1st heatsink elevates the temperature of the surrounding air for the 2nd heatsink.
Comparable with putting 2 heatsinks on top of eachother, though less dramatic, expect a 10% drop in efficiency for choo-choo heatsinks.
Or in numbers : 2 times 0.295=> 0.16C/W
You can save a lot of cash by going forced air or water cooling at these dissipation levels.
The "ambient" factor is not a constant temperature reference, for the thermal resistance calculation it's an average across the (3D) surface of the heatsink.
Put 2 heatsinks in-line and the heat coming from the "touching" side of the 1st heatsink elevates the temperature of the surrounding air for the 2nd heatsink.
Comparable with putting 2 heatsinks on top of eachother, though less dramatic, expect a 10% drop in efficiency for choo-choo heatsinks.
Or in numbers : 2 times 0.295=> 0.16C/W
You can save a lot of cash by going forced air or water cooling at these dissipation levels.
Two of these heat sinks forming one side each of your amp box works great for most current stereo F series FirstWatt designs, Aleph3, Aleph30, stereo AlephX (HifiZen's boards), stereo Aleph J 30 watt, and mono Aleph60.
If you are planing on pushing the envelope of heat dissipation, make sure there is very good cooling inside the cabinet, lots of clearance around the chassis, and with the bottom of the heat sinks a couple of inches above the shelf/resting surface. For example, ventilation holes along the bottom of the chassis and a rather open top. A low speed box fan is also a good idea. There are a number of very good quite box fans that if run a a lower voltage are extremely quite.
A thermal cut off switch on each heat sink is always a good idea.
-David
If you are planing on pushing the envelope of heat dissipation, make sure there is very good cooling inside the cabinet, lots of clearance around the chassis, and with the bottom of the heat sinks a couple of inches above the shelf/resting surface. For example, ventilation holes along the bottom of the chassis and a rather open top. A low speed box fan is also a good idea. There are a number of very good quite box fans that if run a a lower voltage are extremely quite.
A thermal cut off switch on each heat sink is always a good idea.
-David
I have found that the Aleph-X is very sensitive to air moving over the front end circuitry and particularly the input diff pair. This causes significant drifting and shifting of the absolute DC offsets. I would be cautious too, of uneven heating of the output heatsinks.
So any use of a fan requires special attention to fan location and air flow patterns. The only fan cooling that I could see as having minimal issues would be the use of a tunnel type heatsink arrangement for the output section. I would also pay attention as to how the output transistors are arranged on the sink so that cooling is even for all transistors of all four quadrants.
Regards,
Graeme
So any use of a fan requires special attention to fan location and air flow patterns. The only fan cooling that I could see as having minimal issues would be the use of a tunnel type heatsink arrangement for the output section. I would also pay attention as to how the output transistors are arranged on the sink so that cooling is even for all transistors of all four quadrants.
Regards,
Graeme
Graeme is correct in his advice regarding air circulation. If your AlephX does not exceed 100w dissipation per heatsink, then you're in fat city and no additional cooling fan is needed.
My prior thought on adding a fan was to keep the other components like the capacitors from cooking, and not so much to provide additional cooling for the heat sinks.
Also the orientation of the heat sinks is critical. Heat sinks designed for horizontal mounting that are mounted vertical can reduce thermal efficiency by 10% or more.
-David
My prior thought on adding a fan was to keep the other components like the capacitors from cooking, and not so much to provide additional cooling for the heat sinks.
Also the orientation of the heat sinks is critical. Heat sinks designed for horizontal mounting that are mounted vertical can reduce thermal efficiency by 10% or more.
-David
For you reference only, I have built monoblocks of Aleph-X with 100W Class A output, with +/-24V rails and 7.7A bias, per channel. Power suplly is CRCRC, and the R's were also mounted on the same heatsink as the power devices.
I used 2 large heatsinks which were rated at below 0.2 C/W, and the heatsink temperature at steady state (after 2 hours) is 40 degC above ambient. They are vertically mounted, black anodised, as they are designed to be operated. Using a low noise 120x120mm PC fan on each sink will reduce the temperature to 25 degC above ambient, with ease. But it is personal decision to use fans or not. I don't, as I have very efficient heat transfer from heatsink to transistor.
If you want to have the temperature rise at 25 degC without fans, you need to improve the heatsink rating (which is normally defined at 60 degC above ambient) by about a factor of 2; factor of 1.6 to go down from +40 degC to +25 degC, and another 30%+ or so for the reduced temperature difference.
A good heatsink supplier (like the one I used) would tell you that the only way to predict temperature rise properly would be to used full 3D simulation, or a published rating curve plotting dissipation vs temperature rise, based on a predefined heatsink geometry.
Or go to Pass Labs site and have a look at how big the sinks are for an AX100. That will give you a very good idea as to whether your 2x Conrad is sufficient. (I am not using Conrad, as you might have guessed.)
Patrick
PS I have been able to predict my heatsink and transistor case temperature to within 3 degC between calculation and measurement, in case you might be interested to know.
I used 2 large heatsinks which were rated at below 0.2 C/W, and the heatsink temperature at steady state (after 2 hours) is 40 degC above ambient. They are vertically mounted, black anodised, as they are designed to be operated. Using a low noise 120x120mm PC fan on each sink will reduce the temperature to 25 degC above ambient, with ease. But it is personal decision to use fans or not. I don't, as I have very efficient heat transfer from heatsink to transistor.
If you want to have the temperature rise at 25 degC without fans, you need to improve the heatsink rating (which is normally defined at 60 degC above ambient) by about a factor of 2; factor of 1.6 to go down from +40 degC to +25 degC, and another 30%+ or so for the reduced temperature difference.
A good heatsink supplier (like the one I used) would tell you that the only way to predict temperature rise properly would be to used full 3D simulation, or a published rating curve plotting dissipation vs temperature rise, based on a predefined heatsink geometry.
Or go to Pass Labs site and have a look at how big the sinks are for an AX100. That will give you a very good idea as to whether your 2x Conrad is sufficient. (I am not using Conrad, as you might have guessed.)
Patrick
PS I have been able to predict my heatsink and transistor case temperature to within 3 degC between calculation and measurement, in case you might be interested to know.
For those interested in doing their own thermal simulations, you may wish to try the "natsink" software at :
http://www.frigprim.com/
Patrick
http://www.frigprim.com/
Patrick
I buy .187 aluminum sheet stock from onlinemetals and cut it up on the table saw to make my own DIY sinks. This has always worked well for me. The biggest advantage is the lowered stress of knowing that I have unlimited supplies of sink material available off-the-shelf and at the lowest possible price.
Graeme
Graeme
I'm looking into the sheet metal idea. One thing that I'm confused about is anodization, is this good or bad? Since anodization creates a thin layer of oxide over the surface (oxide being a thermal insulator) is this helpful or harmful in a heatsink? Obviously natural oxidation is not a good thing, so I'm assuming if I don't get it anodized, any aluminum should be painted.
Aluminum anodizes almost instantly when cut and exposed to air. The layer of oxide is extremely thin buts it's there and protects the metal from oxidizing further. Anodizing uses electric current to build the oxide layer up. The oxide is harder than the bare metal so it's a bit less prone to damage. Anodizing also presents the opportunity to color the metal. It doesn't make that much difference to the heat radiating properties of the sink.
There is no need to paint the bare aluminum to protect it. Bare aluminum will, however, pick up fingerprints quite easily and these are a pain to remove. Painting, however, is a good cosmetic thing to do when WAF is important.
As for black paint being a better radiator of heat than bare aluminum - well I'm not sure. I build big heatsinks and for class A operation I haven't noticed a major difference one way or another. So I take the lazy way out and apply paint only in the places where it keeps things looking nice.
Graeme
There is no need to paint the bare aluminum to protect it. Bare aluminum will, however, pick up fingerprints quite easily and these are a pain to remove. Painting, however, is a good cosmetic thing to do when WAF is important.
As for black paint being a better radiator of heat than bare aluminum - well I'm not sure. I build big heatsinks and for class A operation I haven't noticed a major difference one way or another. So I take the lazy way out and apply paint only in the places where it keeps things looking nice.
Graeme
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.