I use as predrivers 2SC2240/2SA970 100MHz 3pF Cob, drivers MJE15034/MJE15035, and output are ThermalTrack devices. This amp is in daily use with no stability problem, schematic here: http://www.diyaudio.com/forums/solid-state/182554-thermaltrak-tmc-amp-15.html#post3065943. It is not so hard to stabilize the TRIPLE.
OS wrote "Using a real low Cob fast predriver 1381/3503 resulted in the most unstable triple.
It could be made to run on some IPS's by including it in a highly compensated FB loop."
I had a bit more luck - I used 1381/3503 on my e-Amp triple without problems. In that design I did decouple carefully and include base and collector stoppers on the pre-drivers, drivers and OP devices. I guess the take away point here is triples need to be handled with care.
I have taken the decision not to use these as pre-drivers moving forward simply because they are getting difficult to get and will stick with the MJE340/350 - the pre-dirvers do not need to be as fast as the TIS devices.
(NB - its cascodes that can also be a problem - base stopper absolutely needed - had mine going along at 180 MHz - 2N5550/5551 :0 )
It could be made to run on some IPS's by including it in a highly compensated FB loop."
I had a bit more luck - I used 1381/3503 on my e-Amp triple without problems. In that design I did decouple carefully and include base and collector stoppers on the pre-drivers, drivers and OP devices. I guess the take away point here is triples need to be handled with care.
I have taken the decision not to use these as pre-drivers moving forward simply because they are getting difficult to get and will stick with the MJE340/350 - the pre-dirvers do not need to be as fast as the TIS devices.
(NB - its cascodes that can also be a problem - base stopper absolutely needed - had mine going along at 180 MHz - 2N5550/5551 :0 )
Last edited:
OS wrote "Using a real low Cob fast predriver 1381/3503 resulted in the most unstable triple.
It could be made to run on some IPS's by including it in a highly compensated FB loop."
I had a bit more luck - I used 1381/3503 on my e-Amp triple without problems. In that design I did decouple carefully and include base and collector stoppers on the pre-drivers, drivers and OP devices. I guess the take away point here is triples need to be handled with care.
I have taken the decision not to use these as pre-drivers moving forward simply because they are getting difficult to get and will stick with the MJE340/350 - the pre-dirvers do not need to be as fast as the TIS devices.
(NB - its cascodes that can also be a problem - base stopper absolutely needed - had mine going along at 180 MHz - 2N5550/5551 :0 )
In ALL the 100+ OEM EF3's in my "collection" , the ones that use a fast
pre-driver always use B-C shunts (33-100p).
Dadod may of had good luck with shunting the driver , but I have never
seen this in my collection of schematics.
By far the most common was the 70mhz toshiba drivers sc4793/sa1837.
"fast/ fast/ fast " would be highly unlikely. Very few of the Sanken OP
lineup are above 50mhz Ft.
(3) examples below - all are "fast - Faster -slow " , all use 22p -33p to
shunt the predrivers ..... all I either own/recapped and are 20-30 YO !
For my slewmaster , I have some fast to-3p sankens for drivers , the
good ol' NJW's for OP and the mje's for predriver. Should be ideal.
OS
Attachments
Interesting collection of schematics OS.
On my Ovation 250 amp I used BF471/2, 15032/3 and 21193/4. Base stoppers on the OP devices, and then a heavy snubber of 27 Ohms and 1 nF to GND in the bases of the driver. A bit of an overkill - 220 pF would also have done it ok. But, without the snubber, I could easily get the OPS to oscillate.
The e-Amp uses a network from the base of the driver to its collector, plus a bunch of other stuff - I spent a lot more time optimizing that part of the amp.
With EF2, base stoppers are usually all you need - but for me about 100 W is the upper limit then.
I think the biggest issue here is probably the change in input capacitance with voltage. When I started getting back into audio again about 12 years ago, an early prototype used IRF240/9240 (part numbering?). Mistakenly, MJE for the VAS. Lots or parasitics at the top of the waveforms near pk-pk OP and if it clipped, stick rail and a real mess as it exited. Move to BF471/2.made a big difference as did the OPS stoppers. Then I moved onto bipolars.
I may revisit the FET thing again. The idea of a 4 or 5 MHz ULGF FET output CFA excites me.
I think that it's not really necessary. Check out the Marantz MA9, or PM11, or SM11. No B-C shunt was applied.
Sajti
Confirmed ... this triple can be very stable without the 33p B-C shunts while
simulating a CFA IPS. Hook up a "blameless" , and SW tests will show VERY
HF ringing.
I would suggest leaving the 2 33pF's alone , as one might want to use
many IPS's. Better safe than sorry ...
PS - It could also be a symmetrical/asymmetrical phenomena ... H/K's
990 has no shunts but has ferrite beads and inductors on the
output stage devices.
Edit - Actually , looking more closely ... they shunt with 270p at the VAS.
OS
Attachments
Last edited:
Damir I did have a look and I downloaded your asc file which looked fine. However you didnt make use of a triple so I wouldnt speculate. Ive never seen or tried the compensation that you used but Ill admit Im very curious whether it will work. I compared to me using shunt on the same asc file, your OIC is better in THD terms. I got best of 0.0046 to your 0.0039 if I remember the numbers correctly now. I could get better with shunt but I kept the phase and gain margins about equal for better comparison. I should mention that the way I use shunt you basically get a two pole response, I showed it in a post in the other thread.
OIC now only needs to be tested on practical amp, it may turn out to be a very good scheme.
Thanks Manso for reply, please go to GainWire and tell me if you still want the PCB.