Best hi-volt BJT's for output stage?

[K-amps]

What do you guys think is the best BJT's for use in output stages where rails approach 90v? I'd prefer responses limited to flatpak types i.e. not TO-3's but instead to TO-3P's , TO-3PL's, MT-200's etc.

Here are some devices I researched and will welcome responses from anyone with experience with these.

• PS Audio AC Generators have rails of 105vdc and use MJL21193/94.
• Toshiba has the 150 watt replacement for the Ubiquitous 2sc3281/A1302 that many in the Industry used, i.e. Mr. Pass in the Forte amps, Bob carver in his Sunfire’s etc. These are higher volt (Vce=230) and are 2sc5200/2sa1943.
• Toshiba’s own 180 watt version of the above (also Vce=230) the 2sc5359/2sa1987.
• Kenwood used custom Sanken Devices in their M2/M2a (and their other DLD class-G amps) and were 15Amp 210vce devices called DAT1521N/P
• Sanken’s well known 2sa1295/2sc3264 pair. 200w 17A and 230vce.
• Any other devices?
• What does Krell use? Is it a variant of the old MJ15024/25 that is now upgraded as the MJ21195/96?


Thanks!!

[Eva]

Needing such a high output power, have you considered using class-G?

I've ripped some waveforms from audio CDs and simulated power dissipation of a standard class AB circuit driving a resistive load and being feed by these waveforms

Results were outstanding, with +-120V supply and at clipping threshold, power efficiency was about 30% while with +120V,+80V,+40V,-40V-80V,-120V it approached 80-90%

In practice Class-G with music waveforms has power efficiencies comparable to class-D and class-B or AB linearity [It could also be implemented in class-A, anybody tried this?]

Also, when driving reactive loads it's useful to let the positive rail drop under 0V and the negative rail to rise over 0V since this reduces dissipation, improves reliability [SOA] and allows the reactive load to actually return some energy back to the power supply instead of being fully transformed into heat in the output devices

As output devices I prefer Sanken or Toshiba since the rugged style of Motorola usually means using very big dies and making slower devices with huge B-E and C-E capacitances, with large turn-on delays and turn-off current tails that need lots of frequency compensation to lower the frequency-bandwith product of the amplifier and get decent phase margin and little oscillation when recovering from clipping or at zero-crossing with high slew rates [testing the circuit fulll power at 20Khz with some clipping reveals all these things]

Capacitances can be easily measured and compared

Tip: false devices appear to have 15% to 50% of the capacitance found in the original devices and opening them you can see the smaller die size used [I bought some false Sanken 2SC3264 and Toshiba 2SC3281 & 2SA1302 some time ago, false devices marked as Toshiba or Sanken appear to be common those days]

[dine1967]

Have a look at the new transistors from Onsemi... MJL4281A and MJL4302A. They are supposed to be for high power applications.

[K-amps]

Nice,

I have never seen 350vce BJT's for Audio use. Looking at the part numbers, they seem eerily similar to 1302/3281

It is also a 35MHz part. One of the few power devices that go that high... I have heard that the Custom parts Dan D'Agostino ordered from Motorola were 35Mhz parts. These may be downstream products out of that product developement... i am just guessing but this is a nice find!

thanks Dine1967!

[K-amps]

Eva,

I will get to your very interesting post soon, it was too much to assimilate one one reading... I do not want you to think I am ungrateful for your effort.

regards

[AudioBob]

We've had good luck with the Toshiba 2SC3281 and 2SA1302 flat packs.

The TO-3 packages allow larger die sizes, but they also pass the leads through the heat sink, which puts a practical limit on the thickness of the aluminum you attach the transistors to.

[Tony]

Hi EVA,

i am very much interested in class G amps, all these years i have only done classAB types, so can you point to links about this type of amps? or maybe post the schematics here.... thanks a lot in advance!!!

Also, when driving reactive loads it's useful to let the positive rail drop under 0V and the negative rail to rise over 0V since this reduces dissipation, improves reliability [SOA] and allows the reactive load to actually return some energy back to the power supply instead of being fully transformed into heat in the output devices

this is interesting, can you elaborate more please!!!

[K-amps]

Joan2,

One of the best selling Class-G amps (in Absolute volume) are the Kenwood DLD amps. This includes the following integrateds:

KA-770D
KA-880
KA-1100
KA-2200
KA-3300

And power amps

M1/M2 series.

and ofcourse one of the most sought after Japanese integrated (Actually its not integrated because the PSU unit is separate) is the Kenwood L-02a (1983 circa) they still go over $1000 on eBay if you see one.

The M2's are 220watts into 8 ohms and are great for experimenting, you can get these as low as $200 on ebay and play around. They feature medium NFB and a high damping factor of 1000. For the money, they make great subwoofer amps on a eer budget.

[Tony]

K-amps,
thanks for the info, i am interested in a class G diy amp, i saw Hitachi Amps from Audio magazines many years back and i wonder if there are any circuits floating around for a diy project...thanks a lot...

[K-amps]

I guess I should have made it more clear.. what I should have said was that I have not seen any class-G DIY designs for guys like us, but I have has good success modding those Kenwood amps.

They are weak in the PSU section and that can be improved up etc. Just by making the M2 PSU more beefy i.e. 1.5kVa will increase the power from 220wpc to over 350 wpc into 8E.

Peace!