Jacco - we are looking at different pictures. Have you looked at the one I posted a link about?
This one indeed looks much better, also because the HS is huge.
Last edited:
syn you are mixing things now. The heatsink of a class AB amp properly designed isn't 82.5 deg for a long time and if it is the heatsink is too small since you can burn yourself. Also, your design is more or less the same according to pictures. The difference is that Pavel has his output devices more under the pcb.
...Which is exactly what I originally claimed as being a good practice. What's then the contention here?
I would also add that, according to my experience, from 100W/8ohm and up, between oversized heatsinks and smaller size heatsinks with temperature triggered forced cooling as a sefety net, I prefer the later. That's for cost reasons, also acknowledging that wors case dissipation is unlikely to occure to often or for extended periods of time. Pick your poison
Your numbers are without questions correct, Ovidiu...but audio amps are not rf transmitter and they cant run all the time at " full throttle ". Average long time output power is by typical music only 10-20% of full output and it only in case, when you have setup volume control at full level. So typical average output power is still much lower ( I am talking about class AB ). By the years I had build several hundred of these amps and I had never exchange any only one PS cap and any power transistor, believe or not...
Last edited:
Your numbers are without questions correct, Ovidiu...but audio amps are not rf transmitter and they cant run all the time at " full throttle ". Average long time output power is by typical music only 10-20% of full output and it only in case, when you have setup volume control at full level. So typical average output power is still much lower ( I am talking about class AB ). By the years I had build several hundred of these amps and I had never exchange any only one PS cap and any power transistor, believe or not...
Class AB yield is perverse. The maximum dissipation occurs at about 45% of the full power, in my example at 45W output power, which is likely to occur pretty often in normal usage, in particular if you listen (as myself) to 96 - 100dB SPL.
Then is a matter of specification. If one claims 100W RMS power in 8ohm, then I am expecting the amp to survive indefinitely under any circustances or conditions, including max dissipation. In Stereophile, JA is testing such as well, placing the amps for an hour at 50% power before any measurements, close enough to the theoretical limit. And yes, some famous amps overheat and even blow at this test.
Ouch...
People, which made norm for consumer electronic, are fortunately more reasonable than you and redaction of Stereophile : by this testing is amp loaded at 8% of declared output power for the period of four hours and temperature of heatsink must not to be higher than forty degrees up to ambient...
But is true, for somebody is hot heatsink someting like postcoital appeasement....
: by this testing is amp loaded at 8% of declared output power for the period of four hours and temperature of heatsink must not to be higher than forty degrees up to ambient...But is true, for somebody is hot heatsink someting like postcoital appeasement....
People, which made norm for consumer electronic, are fortunately more reasonable than you and redaction of Stereophile
But is true, for somebody is hot heatsink someting like postcoital appeasement....
Wow! Is this some sort of EC standard? If so, can you point me at some documentation about?
For the 100W RMS into 8ohm case above, this standard looks at 8W output power and allows the heatsink at 25 + 40 deg C = 65 deg C; simple math shows that at 8W output power the amp yield is 22% and hence the power dissipation is about 36W, that is, 9W/output device. Following the same thermal budget as above, this would require a heatsink of minimum 1.1 C/W.
Here's an example of an 1.16 C/W heatsink http://media.digikey.com/photos/Wakefield Photos/421-K.jpg Four output devices on this HS, for an 100W/8ohm amp! I wouldn't dare to rely on such, but if this is what it takes to sell stuff in Europe, so be it...
Aluminum, anodized black. Here's the DigiKey link: http://search.digikey.com/scripts/D...ge_link=hp_go_button&KeyWords=421-K&x=26&y=10 and the manufacturer's (Wakefield Engineering) documentation (look under 421K): http://www.wakefield.com/PDF/extruded_heat_sink.pdf The graph in the datasheet confirms the calculations: for 36W dissipation, the temperature raises about 40 deg C over ambient.
Happy now?
Last edited:
I'm sorry syn08, I've been hasty. It is a lot larger than the pic shows it, with the specified size aluminium is enough.
But 1.1 C/W is the maximum DF you'd use for any reasonable output level, with those specs. And then we're talknig about the usual home usage - where your amps aren't using even a fraction of their power. They wouldn't heat up dangerously, even if you lived in the tropics.
Sorry, "ambient" has a different meaning up north sometimes. 40C above ambient still could get devices up to only an acceptable tempreature range according to their datasheets, you know that With a proper design, even if involving temperature-sensitive devices, one shouldn't be afraid of a few things warming up, just a bit, isn't that so?
But 1.1 C/W is the maximum DF you'd use for any reasonable output level, with those specs. And then we're talknig about the usual home usage - where your amps aren't using even a fraction of their power. They wouldn't heat up dangerously, even if you lived in the tropics.
Sorry, "ambient" has a different meaning up north sometimes. 40C above ambient still could get devices up to only an acceptable tempreature range according to their datasheets, you know that
With a proper design, even if involving temperature-sensitive devices, one shouldn't be afraid of a few things warming up, just a bit, isn't that so?
But 1.1 C/W is the maximum DF you'd use for any reasonable output level, with those specs. And then we're talknig about the usual home usage - where your amps aren't using even a fraction of their power. They wouldn't heat up dangerously, even if you lived in the tropics.
Sorry, "ambient" has a different meaning up north sometimes. 40C above ambient still could get devices up to only an acceptable tempreature range according to their datasheets, you know that
I was only wondering if this is an EC regulation; YMMV, but personally I wouldn't use such thin cooling, not even for my home amp, not to mention selling and standing beyond the design.
And this is of course for Canada, where the hell freezes over every winter.
"Extending the PCB over the heatsink mounted devices as shown in this post http://www.diyaudio.com/forums/showthread.php?p=1940001 is not the best practice. Due to the high dissipation in the power devices, the PCB (in particular if it has a ground plane) will get pretty hot and, sooner or later, all kind of very hard to debug issues may occur (starting with drying out electrolytics, in particular 85 degs types, and ending with soldering cracks because of the high thermal stress). As much as possible, the PCB should stay off from the top of the power devices like here http://www.diyaudio.com/forums/showthread.php?p=1883792 or, even better, here: http://www.diyaudio.com/forums/showthread.php?p=1866841 , this may provide (according to my measurements) 5-10 degs less at the board level , which could make a significant difference in time. "
Sorry Syn08, I cannot agree with this. I mounted my power devices under the PCB (10 off devices) and actually use the power devices as a 'spacer'. The PCB acts as the washer - so I clamp the output power devices between the PCB and the heatsink using this technique. This has been running for about 2 years with no issues. As Peranders says, 60 or 70 degrees is quite normal in Industrial electronics, although my amp only runs this hot when I really drive it hard on continuous out put testing (I have uController overtemp at 70C anyway).
Sorry Syn08, I cannot agree with this. I mounted my power devices under the PCB (10 off devices) and actually use the power devices as a 'spacer'. The PCB acts as the washer - so I clamp the output power devices between the PCB and the heatsink using this technique. This has been running for about 2 years with no issues. As Peranders says, 60 or 70 degrees is quite normal in Industrial electronics, although my amp only runs this hot when I really drive it hard on continuous out put testing (I have uController overtemp at 70C anyway).
Just wondering, since you have built at least one amp, what is your practical experience regarding the cooling issue? Will your theories get confirmed?
I feel there's certainly something I could learn from this topic. How are you people dealing with heat in these projects? How are you testing your projects?
I agree 70C board temp is acceptable, but I can't make it, not even to the levels I usually use to listen at, not to mention worst case.
Bonsai, can you describe shortly your project?
Mr Popa,
my point was that there's little to complain about Upupa's (aka Kazzatel aka Mr DPaudio ltd aka Pavel-D) 222 model.
If heat would be a problem for the extending 223 board, Pavel can/could space the pcb with equal length stand-offs as he uses on the larger model to be done with a radiation problem.
The heatsink on the DPA223 picture looks a large exposed area model, judging by the many points protruding at the sides of the ribs.
Then, i've always been one for forced induction fans (+ intercoolers and horrible sfc number gasturbines)
Looking forward to the uP YAP supervisor.
I agree.
I agree - I was just suggesting another idea, to avoid extending the PCB over the power devices cases. But then Bonsai is suggesting that using the PCB board as a washer for the power devices screws is not a problem, which totally confuses me. There's a significant gap between my experience and such a practice with power amps of 100W and up.
I'll post a little movie of the current working prototype ASAP, maybe tonight. I'm looking for some feedback, I really suck at designing graphic user interfaces/pretty screens.
What is the matter ? My conception, i.e. placing of power devices at massive insulated copper prism, have every advantages and I do not know any better solution. Absorption of pulse power is the best, because devices are placing at massive prism without insulation, i.e. absorption of heat current is ideal. Bias sensing devices are at the same prism near power devices, so delay are very short and bias stability are precise. Interface between prism and heatsink are very well, because area of prism is much larger than area of power device and copper have better heat conductivity than aluminium. Prism also make rail of PS, which is ideal, because its value of conductivity is manifold higher than equivalent area at printed board. And last advantage very easy assembling, adjusting and servicing - I am wailing, when I see power devices, which are soldered directly to PCB and screwed to heatsink, because in this case is servicing really crazy...
All my experiences show me, that this conception is right and reliability is superb, so again, what is the matter ?
All my experiences show me, that this conception is right and reliability is superb, so again, what is the matter ?
All my experiences show me, that this conception is right and reliability is superb, so again, what is the matter ?
One of the matters is that not everybody have access to "copper prisms", goldened, and the associated machining. So, for economical reasons, there must be a very serious reason why these should to be considered.
(a) The thermal conduction of aluminum and copper are very close. In fact, I am not sure if the aluminum - copper interface thermal resistance is actually better (in the overall thermal budged) to the direct aluminum interface.
(b) The power supply bars - you have a good point. How are the prisms connected electrically to the PCB?
(c) "Absorbtion of pulse power", if you mean by that the copper acting as a heat spreader with a small time constant, this is in contradiction with (a) above. The thermal resistence between the copper prism and the aluminum heatsink has to be high enough so the two thermal capacities (prism and heatsink) would be decoupled and have separate impacts on the overal thermal budget.
Pavel, I am only trying to separate what makes sense from a design perspective from the optically striking effect of your constructions. While I agree that your stuff appears absolutely stunning, as an "infidel" that I am I am not convinced that all the bells and whistles (like golden screws, etc...) do necessary have a positive impact on the final product performance. There are things that have way more impact than the golden screws...
Just as an example, placing the output coil on the PCB. I got myself some heat from fellow DIYers for this (in YAP 2.1, but only because I am targeting 1ppm distortions (THD20, IMD 19+20kHz, etc...), and every drop of 2nd harmonic adds up). They are right, at 400W into 4ohm, the magnetic coupling between the output coil and the ground plane and power supply lines is clearly measurable, and it wouldn't help a iota to further use goldened screws or goldened copper prisms. I see you are also placing the coils on the PCB...
Last edited:
"Bonsai, can you describe shortly your project? "
Here is the link: http://www.diyaudio.com/forums/showthread.php?t=96192
I have in th e interim relayed the power amp board, combining the power supply onto the amplifier board (i.e. just like Upopa), but I have just been too busy to order the boards and build them up. The protection circuit info I posted on the other thread BTW is for this amp. I'm driving B&W 703's and use a Pioneer SACD source and this pre-amp (LME4562 based) http://www.diyaudio.com/forums/showthread.php?p=1702062&highlight=Bonsai#post1702062
Here is the link: http://www.diyaudio.com/forums/showthread.php?t=96192
I have in th e interim relayed the power amp board, combining the power supply onto the amplifier board (i.e. just like Upopa), but I have just been too busy to order the boards and build them up. The protection circuit info I posted on the other thread BTW is for this amp. I'm driving B&W 703's and use a Pioneer SACD source and this pre-amp (LME4562 based) http://www.diyaudio.com/forums/showthread.php?p=1702062&highlight=Bonsai#post1702062
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.