AndrewT said:
The devices are very similar, die size should be almost if not exactly the same. Onsemi said the difference is that the newer devices have a little bit improved beta linearity.
60 deg 4 ohms 63V pk on +-70V avg seems to not exceed the 50mS curve derated to 80 degrees tc by much. They are single-shot of course, but the dissipation peaks are shorter than 10mS even at 20Hz IIRC.
I wouldn't worry much as there are reliable PA amps pushing the same power with similar rails with just 4 pairs of 2SC5200/2SA1943 - and these have very bad SOA when compared to the Onsemis.
That's sounds a bit scary. :O But if it has a small enough transformer it might be safe. I wouldn't use it either though. I've seen even worse cheap "PA" amps - 120V idle rails with 4 pairs of those transistors!!! They blow up within a week or so 
What I'm thinking of is the QSC RMX 1450 that has +-78V idle rails - hardly unreliable cheap stuff. 400W max at low freqencies so there is a bit of rail drop.
http://www.qscaudio.com/support/library/schems/Current/RMX Series/rmx1450.pdf
http://www.qscaudio.com/products/amps/rmx/rmx.htm
No insulators between heatsink and transistors so they gain a little there - but the MJLs have 0.625K/W instead of 0.833K/W and much better high voltage SOA easily offsetting the use of insulators for the MJLs.
What I'm thinking of is the QSC RMX 1450 that has +-78V idle rails - hardly unreliable cheap stuff. 400W max at low freqencies so there is a bit of rail drop.
http://www.qscaudio.com/support/library/schems/Current/RMX Series/rmx1450.pdf
http://www.qscaudio.com/products/amps/rmx/rmx.htm
No insulators between heatsink and transistors so they gain a little there - but the MJLs have 0.625K/W instead of 0.833K/W and much better high voltage SOA easily offsetting the use of insulators for the MJLs.
megajocke said:
"Bad" SOA compared to the On devices - based on published curves. Based on what these devices actually do in the field reliably, I'd say the real world capabilities are pretty close. The Toshiba dies are huge also. The old D424's were only rated full power to 70 volts, and they would give even the 2N6259's a run for their money in DC SOA tests. That was back in the 80's.
And if four devices in parallel will last more than a split second on +/-120V they ought to hold up a good long time on +/-80.
Nice to hear! Do you believe replacing 2SC3281/2SA1302 with the 2SC5200/2SA1943 could be safe then SOA-wise? The 2SC3281 curves show better SOA at high voltages, but maybe the 2SC5200 curves are conservative enough?
Do you know anything about the Fairchild FJL4215/FJL4315 which are very similar to the Toshibas? SOA looks similar to 2sc5200/2sa1943 but maybe they aren't as conservative...
Do you know anything about the Fairchild FJL4215/FJL4315 which are very similar to the Toshibas? SOA looks similar to 2sc5200/2sa1943 but maybe they aren't as conservative...
When I first saw the SOA curves, I wondered why Toshiba bothered to replace the 3281 with the 5200. But amp makers have switched over, with no real problems in reliability. Other than the occasional idiot Chinese outfit running +/-120V straight class B which wouldn't survive no matter what they used.
Rumor has it that Crown is buying the FJL4215/4315 by the ton. They're probably a licensed copy. I have a handful of samples, and eventaully I'll get around to doing some comparisons. FWIW, I've been using their HEXFET equivalents, TIP devices, and T0-126 devices for a while now because they're so inexpensive and readily available. No problems as of yet. Some of their IGBTs look pretty interesting too......
Rumor has it that Crown is buying the FJL4215/4315 by the ton. They're probably a licensed copy. I have a handful of samples, and eventaully I'll get around to doing some comparisons. FWIW, I've been using their HEXFET equivalents, TIP devices, and T0-126 devices for a while now because they're so inexpensive and readily available. No problems as of yet. Some of their IGBTs look pretty interesting too......
- OT warning -
That amp I mentioned with +-120V - the power supply actually has taps halfway. Transformer has two 80V CT windings. Capacitors (63V IIRC) are connected in series... The reason is probably so they can use the lower voltage caps. So everything needed for class H except the rail switches are there. If they had used the taps for class H the amps should have worked very well!
The price for those FJLs is very nice I bought 25 of each polarity from digikey to try out, there are no transistors suitable for audio available locally over here except TIP35/36 and MJ15024/5. And both types cost about $7.50 a piece over here - and that does not include the 25% of VAT...
Pretty nice at $1.6 per piece:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FJL4215OTU-ND
I also got some FJA4214/FJA4313 in the smaller TO3P case. A little bit worse heat conduction, but they take less space and are cheaper. Should be perfect for replacing transistors in equipment where the originals are obsolete.
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FJA4213OTU-ND
Have you tried the FQA32N20C mosfets?
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FQA32N20C-ND
I'm wondering if they are suitable for linear operation as I'm thinking of trying a class G (tracking) amp with mosfets for the outer rails and BJTs for the inner devices. One small/medium size BJT per negative mosfet should allow one to use N-channels even there. If the amp needs multiple output pairs each output and rail booster will have it's own commutation diode so that the BJT output devices force current sharing for the mosfets despite gate voltage/transconductance mismatches.
The HGTP/HGTGxxN60A4D IGBTs look pretty nice. I'm thinking of trying a ZCS unregulated half-bridge where the capacitor is chosen so that current flows in half-sine pulses shorter than the on time with these. Maybe match this with the class G? Hmm...
That amp I mentioned with +-120V - the power supply actually has taps halfway. Transformer has two 80V CT windings. Capacitors (63V IIRC) are connected in series... The reason is probably so they can use the lower voltage caps. So everything needed for class H except the rail switches are there. If they had used the taps for class H the amps should have worked very well!
The price for those FJLs is very nice
I bought 25 of each polarity from digikey to try out, there are no transistors suitable for audio available locally over here except TIP35/36 and MJ15024/5. And both types cost about $7.50 a piece over here - and that does not include the 25% of VAT...Pretty nice at $1.6 per piece:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FJL4215OTU-ND
I also got some FJA4214/FJA4313 in the smaller TO3P case. A little bit worse heat conduction, but they take less space and are cheaper. Should be perfect for replacing transistors in equipment where the originals are obsolete.
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FJA4213OTU-ND
Have you tried the FQA32N20C mosfets?
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FQA32N20C-ND
I'm wondering if they are suitable for linear operation as I'm thinking of trying a class G (tracking) amp with mosfets for the outer rails and BJTs for the inner devices. One small/medium size BJT per negative mosfet should allow one to use N-channels even there. If the amp needs multiple output pairs each output and rail booster will have it's own commutation diode so that the BJT output devices force current sharing for the mosfets despite gate voltage/transconductance mismatches.
The HGTP/HGTGxxN60A4D IGBTs look pretty nice. I'm thinking of trying a ZCS unregulated half-bridge where the capacitor is chosen so that current flows in half-sine pulses shorter than the on time with these. Maybe match this with the class G? Hmm...
megajocke said:
I tried them in my A+C+C for last C, but for an opposite rail I had to parallel a pair of FQA12P20 devices, but it still is not enough...
Later I've replaced that FQA32N20 by IRFP4231 that is much better for my application, but a P type device I need seems like still don't exist...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.