Hi Gaborbela,
You can still use a big heatsink for the single MOSFET can't you (100W per MOSFET)? Besides, can you really match three pairs of MOSFETs to the levels needed? The problem with releasing a board that takes 3 pairs of $2 MOSFETS, people will just stuff unmatched units in there and then one of them will melt because it is taking all the heat and then we have unhappy builders.
You can still use a big heatsink for the single MOSFET can't you (100W per MOSFET)? Besides, can you really match three pairs of MOSFETs to the levels needed? The problem with releasing a board that takes 3 pairs of $2 MOSFETS, people will just stuff unmatched units in there and then one of them will melt because it is taking all the heat and then we have unhappy builders.
Yes, 1 cooler per MOSFET as each is putting out 100W! Coolers are rated 125W.
So 4 CPU coolers for a stereo amp. At $52 for the cooling, it is much more cost effective than any passive aluminum fin thingy that can take 400W.
Yes, that is much more economic, both in expenses and in space vs passive aluminum thingy... and as for weight, opting for passive sinks capable of handling 400W might not be the best option for the back!
Folks,
Let me give the explanation for the change of course with the Big Boy output stage.
The output stage will be dealing with a quiescent current of 3A and a rail voltage of 36.
This corresponds to 108W of low grade, continuous heat for EACH DEVICE. The total module will create 216W, and as a stereo, 432W. This is 0.58 horsepower, 0.43Kw. This is a behemoth and still air cooling would require a heatsink with 50mm fins of dimension 350mm x 150mm - EACH DEVICE.
Let us examine the options. If we accept that each of four heatsinks will each cost about $USD25, we look at $100 and a huge amount of metal, 350mm long and at least 150mm tall and maybe around 400mm width. It would be somewhat imposing and require careful locating for ventilation. And a lot of drilling and fitting.
Water cooling is an option. Modern computing affords us this option; but again, the costs are high, and in any case the radiator, the heat exchanger, needs fans, and usually they are 80x80 for moderate cost. These fans are noisy because they spin at around 2000rpm, and besides, the is a gurgling noise which might please scuba divers but not audiophiles......
The next option is heat pipe, and direct heat exchanging using larger, 120x120 fans which spin with PWM control down to about 1000rpm and carefully angled fins which minimise turbulence, and hence noise. Examine a nuclear sub prop, and you have an idea of the work behind this technology to reduce noise, which is critical in such vessels. These CPU coolers are a direct beneficial of recent turbo fan design, and they are very, very quiet, down to 10dBA. And they are cheap, X found some beauties which cost only $USD12.50 each, and for $50 and a bit of US shipping you have around 520W of cooling at your fingertips for almost nothing. And it will be very quiet, and one each can be allocated its only device and all is apples.......
This is the cheapest cooling option for a very large, veritably a gargantuan Class A with more than 50Wx2 audio and 430W heat. This amp would not be at home in Costa Rica, but in Alaska it would give clear meaning to 'warm' in long listening sessions.
The cooling issues are identical regardless of using a single pair, using large IXYS mosfets as listed here. If you cannot find these, find one with these parameters:
IXTK88N30P TO264 600W 88A 6.3nF 300V 55S
IXTK90P20P TO264 890W 90A 12nF 200V 50S
I know Gaborbela has done a lot of work on the three pair BB. However, on reflection, and knowing how many IRFs must be tested to get four triples at high levels of match, THE IXYS CHOICE IS BY FAR THE BEST OPTION. There are no issues with the matching, and you only have to buy four devices, two nmos and two pmos, and know your parts will work as designed. I have to say that this entire thread is a development of a large, unorthodox Class A audio, and there are many choices to make during the process, many of which waste time, and money, and energy, and often more. Such is life in a product development.
I commend all of you to these amplifiers. They are proving, in X's energetic and able endeavours, to sound very, very good, perhaps as good as you could ever get in the DIY category. We all owe a debt of gratitude to Nelson Pass's brilliant dynamic constant current source, it makes a good amplifier stratospheric.
Cheers,
Hugh
Let me give the explanation for the change of course with the Big Boy output stage.
The output stage will be dealing with a quiescent current of 3A and a rail voltage of 36.
This corresponds to 108W of low grade, continuous heat for EACH DEVICE. The total module will create 216W, and as a stereo, 432W. This is 0.58 horsepower, 0.43Kw. This is a behemoth and still air cooling would require a heatsink with 50mm fins of dimension 350mm x 150mm - EACH DEVICE.
Let us examine the options. If we accept that each of four heatsinks will each cost about $USD25, we look at $100 and a huge amount of metal, 350mm long and at least 150mm tall and maybe around 400mm width. It would be somewhat imposing and require careful locating for ventilation. And a lot of drilling and fitting.
Water cooling is an option. Modern computing affords us this option; but again, the costs are high, and in any case the radiator, the heat exchanger, needs fans, and usually they are 80x80 for moderate cost. These fans are noisy because they spin at around 2000rpm, and besides, the is a gurgling noise which might please scuba divers but not audiophiles......
The next option is heat pipe, and direct heat exchanging using larger, 120x120 fans which spin with PWM control down to about 1000rpm and carefully angled fins which minimise turbulence, and hence noise. Examine a nuclear sub prop, and you have an idea of the work behind this technology to reduce noise, which is critical in such vessels. These CPU coolers are a direct beneficial of recent turbo fan design, and they are very, very quiet, down to 10dBA. And they are cheap, X found some beauties which cost only $USD12.50 each, and for $50 and a bit of US shipping you have around 520W of cooling at your fingertips for almost nothing. And it will be very quiet, and one each can be allocated its only device and all is apples.......
This is the cheapest cooling option for a very large, veritably a gargantuan Class A with more than 50Wx2 audio and 430W heat. This amp would not be at home in Costa Rica, but in Alaska it would give clear meaning to 'warm' in long listening sessions.
The cooling issues are identical regardless of using a single pair, using large IXYS mosfets as listed here. If you cannot find these, find one with these parameters:
IXTK88N30P TO264 600W 88A 6.3nF 300V 55S
IXTK90P20P TO264 890W 90A 12nF 200V 50S
I know Gaborbela has done a lot of work on the three pair BB. However, on reflection, and knowing how many IRFs must be tested to get four triples at high levels of match, THE IXYS CHOICE IS BY FAR THE BEST OPTION. There are no issues with the matching, and you only have to buy four devices, two nmos and two pmos, and know your parts will work as designed. I have to say that this entire thread is a development of a large, unorthodox Class A audio, and there are many choices to make during the process, many of which waste time, and money, and energy, and often more. Such is life in a product development.
I commend all of you to these amplifiers. They are proving, in X's energetic and able endeavours, to sound very, very good, perhaps as good as you could ever get in the DIY category. We all owe a debt of gratitude to Nelson Pass's brilliant dynamic constant current source, it makes a good amplifier stratospheric.
Cheers,
Hugh
Hi,
Is there an option for 4ohm high sensitive speakers (with a dip to 3 ohm) ?
I once tested my speakers and with 1.5V I get 91-92dbc at listening position.
I don't need the high voltage just a lot of current.
I thought the 3 pair output with lower voltage would be an option.
Regards,
Danny
Is there an option for 4ohm high sensitive speakers (with a dip to 3 ohm) ?
I once tested my speakers and with 1.5V I get 91-92dbc at listening position.
I don't need the high voltage just a lot of current.
I thought the 3 pair output with lower voltage would be an option.
Regards,
Danny
Hi Danny_66,
You have some really nice (world class) speakers there. Raal ribbons and Accuton mids, 94dB sensitive. You probably don’t need the 52W B.B., the 20W Alpha is probably more than enough. I think we could change outputs on Alpha 20 to the bigger IXIS devices and instead of 1.35amps and +/-24v, maybe drop to +/-18v and run 2.7amps. I think that would get about 24wrms into 4ohms. Each output MOSFET would be at about 50w but 100wpc dissipation is a lot more manageable. This all needs to be confirmed with a simulation...
You have some really nice (world class) speakers there. Raal ribbons and Accuton mids, 94dB sensitive. You probably don’t need the 52W B.B., the 20W Alpha is probably more than enough. I think we could change outputs on Alpha 20 to the bigger IXIS devices and instead of 1.35amps and +/-24v, maybe drop to +/-18v and run 2.7amps. I think that would get about 24wrms into 4ohms. Each output MOSFET would be at about 50w but 100wpc dissipation is a lot more manageable. This all needs to be confirmed with a simulation...
Thanks!
Yes, for my speakers 52W BB is overkill, especially the heat.
I just need something like the Alpha 20, but one that can deliver the current into a 3-4 ohm load with the same stability and distortion, not much asked
Good idea to try the IXYS devices at higher current and lower voltage.
I'll startup ltspice and simulate +/-18v and 2.7amps.
Regards,
Danny
Yes, for my speakers 52W BB is overkill, especially the heat.
I just need something like the Alpha 20, but one that can deliver the current into a 3-4 ohm load with the same stability and distortion, not much asked
Good idea to try the IXYS devices at higher current and lower voltage.
I'll startup ltspice and simulate +/-18v and 2.7amps.
Regards,
Danny
Last edited:
Thanks!
Yes, for my speakers 52W BB is overkill, especially the heat.
I just need something like the Alpha 20, but one that can deliver the current into a 3-4 ohm load with the same stability and distortion, not much asked
Why not two amps? ALPHA with mids+tweeter and class D or class AB with woofers.
My RefSpeaker
Alcone-Lautsprecher AC 10 HE
[PDF] http://www.lautsprechershop.de/pdf/alcone/ac10he.pdf
You do need very good amplifiers to make sound with < 220 Hz
Alcone-Lautsprecher AC 10 HE
[PDF] http://www.lautsprechershop.de/pdf/alcone/ac10he.pdf
Last edited:
I'm going for one huge heatsink per channel with BB, and I still like single device solution might try out that copper heat spreader and was thinking that I add fan, just in case, that blows directly to devices rather than fins itself when and if it reaches very high temperatures. Those sinks were handling 60w Aleph 5's before without overheat problems.
Can anyone confirm what output devices BB will use, so I can order them already?
Oh, got mail from Toroidy, it seems my Alpha20 trafo is ready
might try out that copper heat spreader and was thinking that I add fan, just in case, that blows directly to devices rather than fins itself when and if it reaches very high temperatures. Those sinks were handling 60w Aleph 5's before without overheat problems.Can anyone confirm what output devices BB will use, so I can order them already?
Oh, got mail from Toroidy, it seems my Alpha20 trafo is ready
Last edited:
Yes, biamping like this will be best, with an Alpha for mids and tweets.
But the Accuton mid in combination with 2 Alcone woofers in the crossover region pulls it a little lower.
Last edited: