Funny that you used the same HS a OPC did
I always wondered why OPC did not use 2 pairs as you did. I guess it is because they are dual die devices.
Looks good BTW, where in Ontario do you live?
OPC chose a good heat sink for the device.
There is a worksheet on the first page of the thread showing 1 pair is good for 400WRMS, or at least that's how I read it. Now I'm no expert but that seems like it's pushing it for 1 pair. I just wanted a little wiggle room. Has anybody tested a single pair up to 400WRMS? I just ordered a 1.5kva transformer that will give me +/-62vdc for the fets. Should get me close to 400WRMS into 4. BTW I'm in Ontario, California
the maximum output power is roughly predicted as:
Poutput = total Pmax of output mosFETs divided by 4.
i.e. two 250W mosfets allows ~ 125W of maximum output power into a severe reactance speaker when driven to hot heatsink temperatures.
Some will treat their amplifier with a lot more care by running at lower temperatures and can achieve reliable maximum power of nearly double that prediction if one never uses severe reactances speakers AND never stresses the amplifier with hot heatsinks.
Using this last "get out clause" then I'd suggest the absolute maximum power for domestic listening, into low to medium reactance speaker and with adequate cooling one could design for maximum output power of 250W.
Poutput = total Pmax of output mosFETs divided by 4.
i.e. two 250W mosfets allows ~ 125W of maximum output power into a severe reactance speaker when driven to hot heatsink temperatures.
Some will treat their amplifier with a lot more care by running at lower temperatures and can achieve reliable maximum power of nearly double that prediction if one never uses severe reactances speakers AND never stresses the amplifier with hot heatsinks.
Using this last "get out clause" then I'd suggest the absolute maximum power for domestic listening, into low to medium reactance speaker and with adequate cooling one could design for maximum output power of 250W.
Interesting. 170W into 8 ohms requires 52volts peak 6.52 Amp peak. 13 volts short of your 65v rail. I was testing with an 8 ohm resistor with +/-45. Clipping occurred at 41.5v with the rail sagging to 43v. But I had 2 pair(exicon). When my new transformer comes in I can test with the higher rail into 4ohms.
Last edited:
Heatsink Ts = 50°C is not hot.
70°C is hot.
90°C is too hot. For a 150°C device. A 200°C device could and should survive running quite a bit hotter.
Then add on the deltaT between different locations on the heatsink.
Then add on the deltaT from Tc to Ts.
Then add on the deltaT from Tj to Tc.
Then add on tolerances.
Tc=90°C gives a power de-rating factor of 0.48
Tc=100°C, DF=0.4
Tc=110°C, DF=0.32
Each increment of 10Cdegrees for Tc massively reduces the remaining power available to take the stresses of the audio signal.
70°C is hot.
90°C is too hot. For a 150°C device. A 200°C device could and should survive running quite a bit hotter.
Then add on the deltaT between different locations on the heatsink.
Then add on the deltaT from Tc to Ts.
Then add on the deltaT from Tj to Tc.
Then add on tolerances.
Tc=90°C gives a power de-rating factor of 0.48
Tc=100°C, DF=0.4
Tc=110°C, DF=0.32
Each increment of 10Cdegrees for Tc massively reduces the remaining power available to take the stresses of the audio signal.
Last edited:
Hi Guys,
This discussion is mostly academic since you're all discussing the limits of operation of a device, and there are too many conditions to say with absolute certainty exactly what the upper limit is for a pair of devices.
What about liquid cooling? I'd imagine with both devices mounted to a copper HX and a few large actively cooled rads you could likely push these packages well beyond what would possible with a big hunk of aluminum. Also, are we talking about real world loads, or purely resistive load? Is it worst case, best case, or somewhere in between? Are we talking about short term transient power, or sustained output?
The point is, it doesn't really matter. If you need 400W of power, then this really isn't the amplifier or even topology you should be looking at. Anything over 200W puts you squarely into class D territory, where the benefits of efficiency and decreased need for fidelity start to become impossible to argue with.
The only case I can see that much power being needed is if you're driving a subwoofer with relatively low efficiency to relatively high output levels. In that case, the amplifier is no longer the limiting factor in sound quality, so you might as well make the move to class D at that point. Even a super cheap Behringer multi-kilowatt will do the trick. At these power levels driver distortion will be approaching 50%, so the amplifier hardly matters.
The Wire lateral fet amp was designed to drive loudspeakers of reasonable efficiency to more than adequate power levels with utmost fidelity being the priority. This inherently limits the power requirements to between 100-200W. Going beyond this with normal hifi drivers puts you quickly into power compression, and that's not exactly a desirable way to listen to music.
Regards,
Owen
This discussion is mostly academic since you're all discussing the limits of operation of a device, and there are too many conditions to say with absolute certainty exactly what the upper limit is for a pair of devices.
What about liquid cooling? I'd imagine with both devices mounted to a copper HX and a few large actively cooled rads you could likely push these packages well beyond what would possible with a big hunk of aluminum. Also, are we talking about real world loads, or purely resistive load? Is it worst case, best case, or somewhere in between? Are we talking about short term transient power, or sustained output?
The point is, it doesn't really matter. If you need 400W of power, then this really isn't the amplifier or even topology you should be looking at. Anything over 200W puts you squarely into class D territory, where the benefits of efficiency and decreased need for fidelity start to become impossible to argue with.
The only case I can see that much power being needed is if you're driving a subwoofer with relatively low efficiency to relatively high output levels. In that case, the amplifier is no longer the limiting factor in sound quality, so you might as well make the move to class D at that point. Even a super cheap Behringer multi-kilowatt will do the trick. At these power levels driver distortion will be approaching 50%, so the amplifier hardly matters.
The Wire lateral fet amp was designed to drive loudspeakers of reasonable efficiency to more than adequate power levels with utmost fidelity being the priority. This inherently limits the power requirements to between 100-200W. Going beyond this with normal hifi drivers puts you quickly into power compression, and that's not exactly a desirable way to listen to music.
Regards,
Owen
Hi Owen,
Totally agree this is pretty much academic.
To give an example with more numbers rather than generalities here is my use case for The Wire lateral fet amp.
DPS600 with 65V on LME and 55V on the lateral fets. The fets are mounted a little below half way up a Conrad MF30-151.5.
Because summer months can get warm here in Australia and given the other heat sources in the AV equipment rack ambient temps I expect might get to a worst case of say 40degC.
Based on Andrew's figures I figure an average heatsink temp of say 60degC is 'safe'. So we've got 20 deg of temp rise.
Which using Conrad's temp correction factors gives R(20) of 0.375degC/W.
This allows max of say 53W =20/0.375.
This is average heatsink temp assuming even heat input over the surface area, so temp at heat source will be maybe 5degC higher again. To compensate for that I'll set bias at 400mA which produces 44W at idle.
The load is parallel Peerless 8in nomex woofers in an active speaker. These are being used 70Hz to 1.7kHz so I would judge the impedance to be fairly benign (fs is 30Hz).
My output target from each speaker would be 105dB SPL peak (85dB average with 20dB headroom) at the listening position. Compensating for listening distance I need 112dB from the speaker at 1m. These are 93dB/2.83V sensitivity and some more math tells me I need to swing about 24V RMS to hit that output level giving approx 140W into 4ohm.
So in this case I have a benign impedance, am safely within power limits and temp limits.
I could go higher bias with lower ambient temps. I could also go for higher output if I wanted.
Happy days!
Chris
Last edited:
I guess we will agree to disagree about that.
Or you could take a stab at justifying your viewpoint from a technical perspective. Perhaps I've missed something.
What are you driving? What listening levels do you need?
One of the fundamental problems with having "more power" is the almost certain need for more gain. If you don't need the extra power, but live with the extra gain, then you're doing a disservice to your entire system.
I'm not trying to be donkey, I'm honestly curious if your setup falls into the 1 in 1000 that need that kind of power. I can already think of an example that would, but for the most part, people just don't understand how much power they actually need, and fall into the trap of thinking "well I can never have too much" which is unfortunately not true.
Regards,
Owen
I didn't quote anything about power. Let me reiterate. I disagree that anything over 200watts is only for class D because of the reduced requirement for fidelity. Oh gee, I forgot to quote the fidelity part. Sorry. I agree with the power part of your argument. I think I haven't made my self clear on the goal for my amp. I am building for 400w into 4 as a worst case scenario. I want it to be able to handle such a load reliably should that occur. Let's not forget that 400w into 4 is 200 into 8 which is what the amp will see 99% of the time...well within your Specs for 1 pair. My board has 2 pair just for added safety. And yes my board(s) will be used for a subwoofer even though you feel that LatFet fidelity is not needed at that power level for a subwoofer. To each his own.
What is the sensitivity of your speaker system? What is your target peak SPL output level? Owen's right, overspecifying power leads to too much gain and that's a much bigger sacrifice of fidelity than difference of amplifier topology.
Chris