Complementary Pre-Driver Circuit

[Samuel Groner]

Hi

Some time back I came across the output stage shown on page 1 here: MP108U_C.pdf

What is the advantage of driving the power transistors (Q3, Q22) with a complementary pre-driver (Q6, Q9 and Q18, Q21) in contrast to a more conventional darlington arrangement? Faster switch-off? Any advantage to use that for BJT power transistors as well?

Samuel

 

[burdij]

To provide a path to rapidly discharge the output transistor gates on either polarity (enhanced switching time, as you said)?

 

[JohnRoberts]

+1

That was a pretty common shortcoming of early lateral mosfet power amp designs. They didn't turn off quickly and could get into a mutual conduction mode at high level, high frequency (both power devices on).

Bipolar devices when used as switches benefit from driving the base to emitter potential but in many audio applications a more gradual class AB transition intentionally forces a slower transition, to mitigate effects of turn-on and turn-off delays beyond simple gate charge.

JR

 

[PRR]

> Any advantage to use that for BJT power transistors as well?

Inkjet head driver?

BJTs suck their own capacitance. Base current "quickly" discharges stored charge in the base. In addition you have a 100 ohm resistor B-E to absorb C-B leakage at high temperature. That's normally all you need for audio.

"Quickly" is relative. But modest pull-downs at the base don't improve things much due to parasitic resistance between the actual base charge zone and the base pin. In hard-fast duty like TV sweep, you throw a big negative spike at the base to help suck charge out. For audio, this is too much like "helping" the speaker cone with a hammer. The older Germaniums would pass audio OK. Full-power 20KHz would fry the amp, but who wants to listen to that? (OR to a kick-off spike?) Since the early (late?) 1970s we have had transistors fast enuff for jazz.

 

[bcarso]

[quote author="Samuel Groner"]What is the advantage of driving the power transistors (Q3, Q22) with a complementary pre-driver (Q6, Q9 and Q18, Q21) in contrast to a more conventional darlington arrangement? Faster switch-off? Any advantage to use that for BJT power transistors as well?

Samuel[/quote]

See Stochino's power amp work as presented in Electronics World a good while back. It even got an approving nod from Self (and those don't happen often). Search for "non-slewing" IIRC. I think Self lists some references too.

Agree with PRR that this is probably unnecessary for real audio stuff, until and unless ultrasonic response becomes the Next Big Thing...

 

[Bo Hansen]

Samuel,

I remember I have a magazine article somewhare that explain just this complementary drive topology.
Maybe it was from Wirless World or AES magazine sometime under the 1990´s.

I let you know if I find it in my archive.

--Bo

 

[bcarso]

Stochino, EW+WW, March 1996, cited in Ben Duncan's book High Performance Audio Power Amplifiers.

 

[Gus]

One of the moto brown data books "POWER MOSFET TRANSISTOR DATA" 1984 has a few pages on circuits to charge and discharged the gate cap of power mosfets (can be a few 1000pf).

One of the circuits is somewhat like Q6,Q9 Q18,Q21.

 

[Guest]

Using Okkacm's Razor, why don't use complementary emitter followers directly instead of using them to drive source followers?
What is advantage of FETs instead of BJTs? If anyway complementary BJT pairs drive FETs, do we get any advantage such a way?

 

[JohnRoberts]

[quote author="Wavebourn"]Using Okkacm's Razor, why don't use complementary emitter followers directly instead of using them to drive source followers?
What is advantage of FETs instead of BJTs? If anyway complementary BJT pairs drive FETs, do we get any advantage such a way?[/quote]

For one reason, consumers wanted to buy mosfet amps.... and the customer is always right, even when they're wrong.

Power mosfets do have one real advantage for the power application in that the temperature coefficient at higher current, leads to local current sharing rather than small hot spots inside the power devices. So literally the entire device must overheat to fail, unlike bipolar devices that can have pinhole points of failure.

Some early mosfet designs may have benefited from relaxed protection circuitry due to this perceive extra ruggedness. I suspect some also suffered from inadequate interstage drive capability. The mosfets, being voltage rather than current devices could be driven very simply at lower slew rates, until the gate capacitance became significant. Then you had to deal with turn on time, and the perhaps worse "mutual conduction" mode when you don't turn them off at high frequency.

I think lateral Mosfets (Hitachi) have mostly gone away, as these have fallen out of favor.

JR

 

[bcarso]

[quote author="JohnRoberts"]For one reason, consumers wanted to buy mosfet amps.... and the customer is always right, even when they're wrong.[/quote]

Rich May once told me that the decision to use MOSFET output devices at Sumo was primarily a marketing one :roll:

[quote author="JohnRoberts"]
Power mosfets do have one real advantage for the power application in that the temperature coefficient at higher current, leads to local current sharing rather than small hot spots inside the power devices. So literally the entire device must overheat to fail, unlike bipolar devices that can have pinhole points of failure. [/quote]

At least one other advantage is the absence of storage time per se, which one gets in bipolars that are allowed to go into saturation. And it's a bit tricky to come close to but just avoid saturation---one of the potential downfalls of having your output devices common-emitter instead of the more common emitter-followers. Of course you can still saturate emitter followers if you really work at it (higher secondary rails for the drivers).

[quote author="JohnRoberts"]
I think lateral Mosfets (Hitachi) have mostly gone away, as these have fallen out of favor.[/quote]
They are popular now for RF use: http://www.nxp.com/products/rf/rf_power/basestation/ldmos/

 

[Bo Hansen]

Samuel,

I find it, but I can not scan the article for you just now, so ask Electronics World / Wirless World for a copy.

"Building better buffers"
November 1992
page 931 to 938

also a other article in same magazine, about same topology.
"Ultra fast amplifers"
October 1995
page 835 to 841

--Bo

 

[JohnRoberts]

[quote author="bcarso"]

At least one other advantage is the absence of storage time per se, which one gets in bipolars that are allowed to go into saturation. And it's a bit tricky to come close to but just avoid saturation---one of the potential downfalls of having your output devices common-emitter instead of the more common emitter-followers. Of course you can still saturate emitter followers if you really work at it (higher secondary rails for the drivers). [/quote]

It's pretty common to use anti-sat diodes on intermediate common emitter stages. I once had a reliability issue related to the reverse breakdown voltage in one application. These diodes want to be low capacitance because of where they are in the path, so you don't just throw in a 4003.

I was dismissing the carrier storage for class AB audio operation but indeed if an output stage is allowed to saturate it can stick to the rail. Amp designers can be a little reluctant to clamp to much below full output swing since customers read more into output watts than dBW and a volt of output swing could be several watts while insignificant dB..

If I remember correctly Hitachi may have used these parts for commutation of their class H power amp rails where switching time could be an issue.

JR

 

[Larrchild]

Most modern tv transmitters use LDMOS devices, albeit with DSP precorrection loops.

 

[Samuel Groner]

Thanks for the discussion and information. I'll check the suggested literature, the Stochino work was on my to-read list anyway.

Samuel