FET drivers for BJT outputs?

Dave Zan

Member
2010-11-21 7:12 am
The OnSemi ThermalTrak outputs work well for my application but comparably excellent drivers are hard to find. For 6 output pairs the 2SA1930+complement is not quite adequate and the best seems to be the MJE15034/35.
This is barely faster than the outputs and it occurs that this could be a suitable spot for a fast FET, even if it is the reverse of a common layout.
I have little experience with FETs but expect there are smaller (TO220?) FETs with lower capacitance than a typical output FET.
Does anyone have recommendations?

David
 
even if it is the reverse of a common layout
You're in good company. IIRC, John Curl does it that way in his power amp designs. Why not go over to his Blowtorch thread and ask for recommendations there?

Don't worry about OT - that thread's permanently OT. Actually, now's a good time to barge in with your question, IMO. They're just between arguments at the moment and this might start some useful discussion.
 

Dave Zan

Member
2010-11-21 7:12 am
The FET don't have truly complementary pairs. Ciss and few specs different a lot. Try IRF610 and IRF9610.

Thanks, there are no real complements even in BJT and these look the best so far.

You're in good company. IIRC, John Curl does it that way in his power amp designs.

There are some prominent amplifier designers that claim that BJT output is the best way to go and then add that driving these with a FET stage is better yet.

It makes sense as a circuit so I am pleased learn I am not the alone.

My bad, they're at it again.:(

Go for it anyway, that thread needs a good technical question to get it unstuck.
:2c:

LOL! thanks for that, I really did!

Best wishes
David
 
The FET don't have truly complementary pairs. Ciss and few specs different a lot.
Try IRF610 and IRF9610.

Thanks, there are no real complements even in BJT and these look the best so far.

Use IRF710/IRF9610, they are better complements than IRF610/9610.

Or (4x)BD512/(3x)BD522 from ITT if you can find them...mind low Vds!
 
Last edited:

Dave Zan

Member
2010-11-21 7:12 am
A substancial drawback by using MOSFET drivers
is of course the voltage drop across G-S.
It can be 3.5-4.5 Volts per halfwave.
This means you either need higher power supply or you will get less power out.

It makes sense to me to have driver FETs on elevated supply rails.
But elevated rails are usually reserved for the VAS, so I am a bit out of step here too;)
Bart Locanthi's seminal T-circuit had two extra pairs of separate elevated rails for the driver and the pre-driver respectively so I have some company. Obviously a bit more expensive so not much done any more.
I plan a CFP pre-drive+driver to the EF output so only one pair of elevated rails will be needed.

Best wishes
David
 

Dave Zan

Member
2010-11-21 7:12 am
Mind the clipping behavior, you don't want to saturate the output devices.

If I don't elevate the driver rails too much then all should be well.
Anyway, I have 115 dB sensitive compression drivers. Probably bleed from the ears before they ever clip and saturated outputs will be the least of my problems;)

Indeed, laterals have much lower Vds ...
See post number 9 for a very Nice overview.
http://www.diyaudio.com/forums/solid-state/217528-low-power-mosfets-complementary-pairs.html

Since I plan elevated rails the lower Vds is of little benefit but the lower transconductance remains a drawback. But thanks for the useful link, it also confirms the IRF9610+610 or 710 as the winner so far.

Best wishes
David
 
What you might want to look into are separate rails for the input stages up to the FET drivers, these would be rectified from the same secondary but with separate filters. This makes the ripple lower and also lowers the rail droop at full load for these extra rails. What you want to do is get the driver to have just the right voltage drop WRT rail to get clipping of the output stage a bit below the bottom of the ripple sawtooth of the output stage rails. In this manner clipping becomes cleaner - and the rail loss of the drivers is actually used to an advantage.
Laterals have lower treshold but also lower transconductance, which means at higher currents more Vgs is required than with HEXFETs. Although 3-4V will be needed for the treshold of a HEXFET like the IRF710/9610, 1.5V higher will give you more current than it can actually handle. With laterals, 1-2V (they can vary a lot and it's different for P and N ch) is enough to reach the treshold, but you need a good 7-8V for the maximum current.
 
A substancial drawback by using MOSFET drivers
is of course the voltage drop across G-S.
It can be 3.5-4.5 Volts per halfwave.
This means you either need higher power supply or you will get less power out.

OR you can bootstrap the driver stage and forget all the complex need for two power supply's.;)

Driving the fets to Vds saturation isn't a bad thing as long as the gate drive has a low enough Z to control the increase in Ciss.
 

Dave Zan

Member
2010-11-21 7:12 am
What you might want to look into are separate rails for the input stages up to the FET drivers, these would be rectified from the same secondary but with separate filters...
Laterals have lower treshold but also lower transconductance...

Yes, that is how I planned to do the rails.
I meant to imply that laterals are not attractive in this application because the lower transconductance is a drawback, while their lower threshold provides no benefit that I can't obtain with elevated rails.
So thanks and sorry if it was not clear we are in accord!

Best wishes
David