Your boxed Perreaux has +/- 82Vdc rails, and 160Vds power devices.
Under full load, the rail voltages sag to 75-76Vdc, Perreaux numbers.
A reason why the output stages of the amp survive your favorite transients.
The power supply with the monster EI and unboosted capacity has to sag, in order to save the power TO3's.
Mains power regulation in Australia is +10%, and mains still is 240Vac in most parts of Australia, even after 2001.
Perreaux models for Europe have 230Vac transformers, and mains regulation is also +10%.
Which means that the rail voltages can go up to as high as +90Vdc.
It doesn't help that the output stages of the Perreaux power amps have no source resistors.
Lateral MOSFETs are awesome devices, they're as good as indestructible.
But there's one thing they choke to death on, and that's too much voltage between drain and source.
So please go ahead and make it a real stiffy.
(personally, I'd swap the J50/K135 out for J55/K175, J56/K176 or 200Vds equivalent Exicon numbers before commencing any Ishiwata wiz of Oz games)
AAH, from president/owner of PS audio and the Threshold corporation to canning beer.
Good ol' Randy, where would we be without marketing moguls.
(or was that mongols ?)
Mogul mongols ......
You reckon his retirement package is a bit thinish ?
( Randall Patton - Address, Phone, Public Records - Radaris )
( Randall Patton - Address, Phone, Public Records - Radaris )
What practical use would that kind of power be?
For 30,000 euros you could certainly find some efficient speakers.
Things like that seem silly, not impressive.
Then again I am naive and ignorant. It is staggering to see the extremes of what is considered "good power" in audio.
For now I am heading in the direction of single digit watts and milliwatts, while others go extreme the other way.
What gives?
Thanks for the push, jacco, that's got me to grab the service manual to check things out, and there are some surprising aspects to the 2150B. Nominally, those output devices can only withstand 160V, yet Perreaux quite happily quotes voltage swings of 164V, they were perfectly OK about shipping them with those specs. Since it is very easy to generate a test signal which doesn't "pre-sag" the rails, which also includes full level transients, it would have been very easy to kill brand new units, on day 1 of use, if it were truly a major weakness.
I can only imagine that Perreaux must have done their own evaluations, to sort out the more robust units - to make them comfortable with the working conditions of the transistors. Certainly, with my unit the working rails were constantly in the high 80's, and occasionally low 90's - no complaints.
As jacco suggests, if anyone is thinking of improving the integrity of the power supply, and is concerned about max. voltages, change the output transistors, or, what would be my preferred way, drop the voltage rail level to that which you're more comfortable with, by some means. The few extra volts means a tiny amount in the big picture, so nothing really lost there from the POV of SQ.
It's the ability of the amp to deliver transient high power cleanly, not constant, dirty high power that matters. It would be easy to get an amp with big mutha bits to roast a chicken by driving a heater element, but what would the quality of the waveform be while doing this? So, the amp is made to handle the sharp burst of power needed to get the crescendo happening correctly.
Of course, you'll always have the numbskulls, wanting to show off to their friends, or "rock out", keeping the accelerator flat to the floor, "just because they can". So, you give the amp the smarts to shut itself down, to protect itself from idiot use. Active studio monitors have this as a matter of course, otherwise they wouldn't last a day, from the rounds of "enthusiastic users" playing with them ...
Three output pairs, would never, ever be able to do 1R sensibly - the other major problem was the big screw terminal smoothing caps, I tried my best to work with them initially, but binned them to get major improvements ...
My curiosity aroused, I looked further afield: amazingly, pro amps were using those particular Perreaux output transistors with rails up to +-100V - like the C-audio SR707, used in very harsh conditions, and driven hard. Obviously, the Hitachi parts were very tough, and conservatively rated ...
Yes, the devices used have a very good beta figure, and more importantly, their beta curve looks more like a straight line.
In terms of power handling and especially heat handling, I would prefer to use MJ 2115/21196, but unfortunately, their beta curve is not nearly as good.
Lastly, I remind you that while 7 pairs were shown, I'd like to see 8 pairs. More can't hurt, I'd feel safer, and the price differnce is miniscule overall.
Actually, in truth, your questions are irrelevant, because we are trying to figure out a way to solve Wayne's problems with what he already has.
To be perfectly clear, I have no idea what business the designer had by letting it drop down that low, into the short circuit territory.
Nevertheless, I wonder what happens with such tightly packed devices in terms of the Early effect?
The ones you mention seem to be a very close shave ...
The ones you mention seem to be a very close shave ...
My epistemological ears can't help but prick up at such statements. How do you know more can't hurt?
I've heard a fairly consistent refrain around these parts that a single output pair sounds best. I speculate that's down to poorer PSRR with lots of paralleled pairs, but I have no data to go on. More pairs has to hurt the PSRR - if my hypothesis is correct then this'll also hurt the sound.
Until somebody makes a complementary pair of output devices with a rating of say 1,000 Watts, Irms = 25 A, Ipeak = 40A, and so forth, paralleling output devices is a given for high powers and load tolerance.
As for the PSSR, that's something a designer has to deal with once he is aware of what he must use to obtain the desired specs. Tight, massive PSUs are an obvious must.
The REAL question here for me is why are our PSUs' specs continuously sliding downwards? We see ever smaller transformers, ever smaller capacitors, it's as if they are treying to find the giving up now point.
Just compare the PSUs used in say 1995, with those made in 1975, with those made last year and you see a sad picture. For a start, hardly anybody bothers to separate and regulate the power lines for the input stage(s) and VAS - now, when the parts for it are dirt cheap and abundant. At best, this is offered as an optional extra, if at all. Add it all up and you arrive at truly unholy figures. But even so, there are VERY few who offer it even as an option.
In short, Rick, I think you are barking at the wrong tree. However extreme and rare Wayne's case may be, tell me how I can get the kind of power he needs without paralleling transistor pairs.
You missed the point totally dvv. I wasn't saying 'don't have N pairs of output devices', rather I was questioning how you had arrived at your 'another pair can't hurt' statement.
Why not just admit you said it off the cuff without experimental support and move on? Or that "yeah it does hurt a little but the upsides in reliability and load tolerance are more than worth the slight drag on SQ"?
Why not just admit you said it off the cuff without experimental support and move on? Or that "yeah it does hurt a little but the upsides in reliability and load tolerance are more than worth the slight drag on SQ"?
Would you like to share a datasheet ?
I've been thinking about an output stage with just lots of SOT-89 or D2PAK on a PCB squished on a heatsink with silpads (or a metal core PCB). These devices are cheap, you can get very high hFe. Total power dissipation wouldn't be so great though. That could be nice for a low power amp...
- Status
- Not open for further replies.