Bob Cordell Interview: BJT vs. MOSFET

Jump to Latest
N

nikwal

Member
Joined 2005
dsl.mine.nu
Hav'nt seen this kind of "mosfet" (hex i my case irfp somethng) stage anywhere..
http://dsl.mine.nu/out.gif
Is it because it's cräp or did I not search enough or did I do a useful thing for a change?..
I'm using it and I'm pleased with the sound hmm there's a bunch of theories that made me connect them in that way a while ago.. something with current and linearity in the component.. hmm or I might have misunderstood the datasheet hmm
 
A

AndrewT

R.I.P.
Joined 2004
nikwal said:
Hav'nt seen this kind of "mosfet" (hex i my case irfp somethng) stage anywhere..
http://dsl.mine.nu/out.gif
Is it because it's cräp or did I not search enough or did I do a useful thing for a change?..
I'm using it and I'm pleased with the sound hmm there's a bunch of theories that made me connect them in that way a while ago.. something with current and linearity in the component.. hmm or I might have misunderstood the datasheet hmm
Click to expand...

Hi,
that looks like a cascode that swings with output voltage.
It reduces the voltage across any device by about 50% compared to the non-cascoded version.
See Leach super amp or Nelson's own circuit for BJT versions.
 
x-pro

x-pro

Member
Joined 2003
ant-audio.co.uk
AndrewT said:
Hi,
that looks like a cascode that swings with output voltage.
It reduces the voltage across any device by about 50% compared to the non-cascoded version.
Click to expand...

It is called usually "totem pole" and was used before, especially when higher voltage transistors were not available. My Creek 5350 design (and A52 before that) also used a similar configuration in the output stage, thought it was N-channel only. It is an alternative to paralleling devices to dissipate more power and I found it is better than a parallel connection for MOSFETs as it keeps only one pair of devices in control of the output.

Cheers

Alex
 
B

Bob Cordell

Member
Joined 2006
cordellaudio.com
nikwal said:
Hav'nt seen this kind of "mosfet" (hex i my case irfp somethng) stage anywhere..
http://dsl.mine.nu/out.gif
Is it because it's cräp or did I not search enough or did I do a useful thing for a change?..
I'm using it and I'm pleased with the sound hmm there's a bunch of theories that made me connect them in that way a while ago.. something with current and linearity in the component.. hmm or I might have misunderstood the datasheet hmm
Click to expand...


I haven't seen this with MOSFETs either, but it is a useful configuration, as others here have mentioned. I believe that Bryston still uses it to this day in their hi-powered amps. It originated in earlier times when SOA of output transistors was not as good as it is today. By halving the voltage across the output transistors, it helps SOA a lot. It also has a partial cascode benefit AND it also has a partial benefit of shielding the inner output transistors (that really do the audio work) from garbage on the power supply rails by a factor of about two.

Cheers,
Bob
 
Nelson Pass

Nelson Pass

The one and only
Joined 2001
Paid Member

Attachments

  • 800a schem.jpg
    800a schem.jpg
    91.8 KB · Views: 806
Terry Demol

Terry Demol

Member
Joined 2002
Bob Cordell said:



I haven't seen this with MOSFETs either, but it is a useful configuration, as others here have mentioned. I believe that Bryston still uses it to this day in their hi-powered amps. It originated in earlier times when SOA of output transistors was not as good as it is today. By halving the voltage across the output transistors, it helps SOA a lot. It also has a partial cascode benefit AND it also has a partial benefit of shielding the inner output transistors (that really do the audio work) from garbage on the power supply rails by a factor of about two.

Cheers,
Bob
Click to expand...

Bob

Bryston use something different which is a combination of emitter
follower and CFP in the same OP stage. Their config is quite unique.

I'm thinking the cascoded OP stage isn't very popular probably due
to the fact that it really eats into headroom which is precious in the
OP stage.

cheers

Terry
 
A

AndrewT

R.I.P.
Joined 2004
Terry Demol said:

I'm thinking the cascoded OP stage isn't very popular probably due to the fact that it really eats into headroom which is precious in the OP stage.
Click to expand...
If the supply voltage is that high then the loss of an extra couple of volts is hardly noticed.
Compare that to the same two transistors in parallel operating from half the supply voltage. The losses here will be higher and the result will be less overhead.
 
B

Bob Cordell

Member
Joined 2006
cordellaudio.com
Terry Demol said:


Bob

Bryston use something different which is a combination of emitter
follower and CFP in the same OP stage. Their config is quite unique.

I'm thinking the cascoded OP stage isn't very popular probably due
to the fact that it really eats into headroom which is precious in the
OP stage.

cheers

Terry
Click to expand...


Hi Terry,

Yes and no. In their lower power amplifiers they use a single level of output transistors, but they are arranged in a rather unique composite emitter-follower/CFP configuration. However, when they go to their higher power amplifiers, they insert another layer of output transistors in series. That's what I was referring to. You are right in that what is in those layers is not just a simple emitter follower structure.

Cheers,
Bob
 
Terry Demol

Terry Demol

Member
Joined 2002
Bob Cordell said:



Hi Terry,

Yes and no. In their lower power amplifiers they use a single level of output transistors, but they are arranged in a rather unique composite emitter-follower/CFP configuration. However, when they go to their higher power amplifiers, they insert another layer of output transistors in series. That's what I was referring to. You are right in that what is in those layers is not just a simple emitter follower structure.

Cheers,
Bob
Click to expand...

Thanks Bob,

Very interesting, I haven't seen these other designs.

So they still use the CFP+EF but insert a cascoded ollower 'above'
this whole OP structure.

To me it does seem like added complexity for relatively small gain.
I would think that the current modulation due to OP load causes
way more non linearity than voltage modulation across OP
devices.

One thing it would do is give much higher PSRR for OP stage
(sans FB), maybe that is a design motivation.

cheers

Terry
 
anatech

anatech

Administrator
Joined 2004
Paid Member
Hi Jim,
That document is about as much advertising as it is factual.

One thing I disagree with strongly is the "matched output transistor" nonsense. Having serviced a few of these, I can tell you that the outputs are matched only in part number. :devilr: The devil seems to be in the details. Details they are not giving the reader.

-Chris
 
PMA

PMA

Member
Joined 2002
pmacura.cz/audiopage.html
Paid Member
Terry Demol said:



Kanwar

WRT the amp mentioned, which OP devices x conduct, the followers,
the common emitter devices, or both?

BTW, I haven't had time to analyse this OP stage closely.

cheers

Terry
Click to expand...

Normally CFP is more prone to cross-conduction. The reason is that they are only the resistors to drain charge from the B-E junction of 2nd device.
 
kASD

kASD

Member
Joined 2004
Nelson Pass said:
Is that a feature?
:cool:
Click to expand...
No, its a flawed feature

Terry Demol said:

Kanwar
WRT the amp mentioned, which OP devices x conduct, the followers,the common emitter devices, or both?
cheers
Terry
Click to expand...

CE devices!!


PMA said:
Kanwar is quite right, though we found problems slightly higher - like 20kHz - 30kHz and more.
Click to expand...
I once tested 7B SST which was showing cross-conduction at open circuit current in excess of 5 amperes at 10KHZ
 
You must log in or register to reply here.
Top Bottom