Bob Cordell Interview: BJT vs. MOSFET

[john curl]

Personally, I prefer to use mosfets in the VAS stage, so that I can run higher drive currents directly into the output mosfets. Why use bipolars?

[andy_c]

Originally posted by john curl
Personally, I prefer to use mosfets in the VAS stage, so that I can run higher drive currents directly into the output mosfets. Why use bipolars?

Interesting, John. You mentioned you used boosted supplies for the VAS. How low do you allow the gate-drain voltage of the VAS to get as clipping is approached?

[john curl]

It depends, whatever it takes. It is easy to obtain 200Vmosfets and run them at +/-90V in a VAS stage. This should be enough for almost any practical mosfet output stage discussed here. You do have to heatsink the T-220 drivers, but that isn't too difficult or expensive. The point is that the driver mosfets do not need current bias from the input stage in order to operate effectively. Bipolar transistors lose their input impedance as they are increased in operating current and require a constant bias drive because of their finite beta. This can put a significant load on the input stage, without further buffering.

[andy_c]

Thanks John.

[Ultima Thule]

John and Nelson,

so you like to run the VAS stage at high currents, that requires devices like IRF 610/9610 (SOA about 100mA @ 200V @ DC).

The question now is, how do you see on the VAS stages driving requirements as the "problem" with "high" capacitive drive load treated for a while in this thread is pushed "backwards" from the VAS driving "high capacitive output devices" to input stage "driving high capacitive VAS" instead.
Would the answer simply be "use bigger devices in the input stage as well, which then again pushes backwards the driving requirements to the preamp instead...."?

The mentioned devices above has a Cre of around 15 pF, this in comparison to some nice 200 Volt BJT devices from Sanyo with around only 2pF Cre.

So what is the deal?

Cheers Michael

[Nelson Pass]

Originally posted by Bob Cordell
BTW, in your view is a balanced output stage different than a bridged output? A while back, bridging of amplifiers got a bad rap, it seems. Have we just changed the semantics to protect the innocent, or is there a substantive difference in architecture between an amplifier with a balanced output and one with bridged outputs? I'm assuming both have single-ended inputs, so that that part of the issue does not cloud the discussion.

I think it's a matter of semantics, although at PL we make
balanced amplifiers that use symmetric feedback and cross
couple error from one side to the other to get distortion
cancellation. Errors appearing at one output polarity are
largely replicated in phase on the other polarity output. The
patent on that is #5,376,899 although the circuit has continued
to mutate. You will see a very similar topology in the THS4131
chip, and TI subsequently licensed the patent.

In this case the amplifier is designed as a balanced piece (if you
drive the RCA connector we ground the - input, so that we are
just amplifying the difference between + input and ground).

If you simply bridge a more conventional amplifier you don't get
that effect, but of course there's nothing wrong with that per se.
The bad rap mostly likely comes from amplifiers that simply aren't
designed to drive a load that hard, and some poor designs take
the output of 1/2 and recycle it to drive the input of the other
half, compounding the distortion.

To answer another question, the X600.5 (monoblock) has 48
devices arranged in 4 groups, each 12 in parallel. Anticipating
the next question, we rate the slew rate as 50 V/uS, but I took
a look at one on the bench today and saw about 80 V/uS. You
can get a sloppy 160 V/uS if there's no load. This information is
useful for those who want to distinguish between driving Cgs
versus Cgd - the answer being that both seem to be drawing
about the same current.

[Nelson Pass]

Originally posted by Ultima Thule
The question now is, how do you see on the VAS stages driving requirements as the "problem" with "high" capacitive drive load treated for a while in this thread is pushed "backwards" from the VAS driving "high capacitive output devices" to input stage "driving high capacitive VAS" instead.
Would the answer simply be "use bigger devices in the input stage as well, which then again pushes backwards the driving requirements to the preamp instead...."?

If you use a 610 or 9610 for the Vas you still get quite a bit
of bandwidth. My usual amplifier has a diff input of 2sk289
and 2sj109, these are cascoded, and then they drive Mosfets
which form the Vas. Because the input devices are cascoded,
you can run more current through them which gets you more
drive current, and while I don't find the need to do this, you can
relieve the Vas capacitance by cascoding it also.

In an earlier incarnation, the X600, I used 610's for the input
transistors and followed them with a folded cascode so that the
entire amplifier had only 2 stages - the input transistors as the
Vas and the outputs as followers. It was a very popular
amplifier for about 7 years and was finally replaced by the X600.5
which uses the Jfets for inputs.

[john curl]

I use IRF 610's and 9610's for my VAS drive stage. However, the asymmetry of the complementary devices gives me some 2'd harmonic at low levels. I get more than 100V/us slew rate. Bipolar devices would give me more 3'rd and less 2'nd, and probably lower overall distortion performance. I believe in fets, however. I use them whenever I can.

[Bob Cordell]

Originally posted by Nelson Pass


I think it's a matter of semantics, although at PL we make
balanced amplifiers that use symmetric feedback and cross
couple error from one side to the other to get distortion
cancellation. Errors appearing at one output polarity are
largely replicated in phase on the other polarity output. The
patent on that is #5,376,899 although the circuit has continued
to mutate. You will see a very similar topology in the THS4131
chip, and TI subsequently licensed the patent.

In this case the amplifier is designed as a balanced piece (if you
drive the RCA connector we ground the - input, so that we are
just amplifying the difference between + input and ground).

If you simply bridge a more conventional amplifier you don't get
that effect, but of course there's nothing wrong with that per se.
The bad rap mostly likely comes from amplifiers that simply aren't
designed to drive a load that hard, and some poor designs take
the output of 1/2 and recycle it to drive the input of the other
half, compounding the distortion.

To answer another question, the X600.5 (monoblock) has 48
devices arranged in 4 groups, each 12 in parallel. Anticipating
the next question, we rate the slew rate as 50 V/uS, but I took
a look at one on the bench today and saw about 80 V/uS. You
can get a sloppy 160 V/uS if there's no load. This information is
useful for those who want to distinguish between driving Cgs
versus Cgd - the answer being that both seem to be drawing
about the same current.

Nelson, thanks. Good answer. I'll have to look up that patent. I especially like you're characterizing it as, at least partially, an issue of semantics, and noting that the bad rap may have come from designs that were either poorly executed or not executed with the amount of strength to deal with the higher currents/smaller effective load that bridging brings.

I actually have a gut feeling that balanced or bridged brings with it some interesting power supply advantages (not just in terms of reduced rail voltages), partly in respect to the path taken by the speaker current.

Bob

[john curl]

Duh! I did it for this reason in 1969 at Ampex Research for a motor drive, and I used Nelson's symmetrical ckt too!