Hi,
Whatever bias tracking difficulties BJT's have, which can be
ameliorated with arranging a higher optimum bias, they still
wandering all over the place have far lower low level x/o
distortion than MOSFETs can ever have.
And its not just THD, 0.3% with a nice spectrum is
far better than 0.1% with a much worse spectrum.
MOSFETs at low levels have high THD and a bad spectrum.
Consequently three BJT output devices per channel with an emitter
follower output, and 0.1R emitter resistors and optimum bias for
large signal THD, will give you the lowest low power THD and
FWIW the nicest low power spectrum short of running the
output stage in some form of enhanced class A class AB,
which results in gm doubling and higher high level THD.
Since you can't optimally bias MOSFETs your best bet
is to run them as warm as is sensible with a class A
region for low levels, though this will compromise
the very spurious THD figures at high output.
rgds, sreten.
Whatever bias tracking difficulties BJT's have, which can be
ameliorated with arranging a higher optimum bias, they still
wandering all over the place have far lower low level x/o
distortion than MOSFETs can ever have.
And its not just THD, 0.3% with a nice spectrum is
far better than 0.1% with a much worse spectrum.
MOSFETs at low levels have high THD and a bad spectrum.
Consequently three BJT output devices per channel with an emitter
follower output, and 0.1R emitter resistors and optimum bias for
large signal THD, will give you the lowest low power THD and
FWIW the nicest low power spectrum short of running the
output stage in some form of enhanced class A class AB,
which results in gm doubling and higher high level THD.
Since you can't optimally bias MOSFETs your best bet
is to run them as warm as is sensible with a class A
region for low levels, though this will compromise
the very spurious THD figures at high output.
rgds, sreten.
Last edited:
Often wonder, whether V FETs setup to be very fast could allow an amp to be designed for a higher ULGF. This could give more feedback and reduce high frequency distortion. Would BJTs still have the edge?
If we bring error correction like Hawksford to the table are BJTs, in your opinion, still better?
So far in my limited experience, BJTs and V-FETs can sound good but L-FETs have been a little disappointing.
I find this to be a myth. I see some posted in agreement that it is possible but then again only sims are shown. I d like to see someone post real life results where the issue is shown, BJT vs V fets in relation to UGLF. After gate resistors are used on V Fet to achieve stability, the effective FT is substancially decreased, maybe only a few Mhz higher than modern BJTs, nothing of note. BJT gm is that much higher than Vfets that it will still have the edge.
Its a diffrent story with RF circuits, even common V fet like IRF610 are very good divices for use in amplifiers in the range of 100 Mhz (maybe more, havent tried). Nowhere is it mentioned on the datasheets that these mosfets can be used for RF.
Last edited:
This is an interesting idea. Going to have to do some reading on this one. The end question is how much do you have to theoretically boost the ULGF to counter the lesser gm of V-FETs. Is it even possible? and if so, how likely are you to achieve it in reality?
If you pull out all the stops and go for distributed heavily biased drivers and a tight layout. What's the realistic maximum Ft that can be achieved?
Pretty fast. Here is one of my first adventures into a HEC mosfet amp. A prototype of a single end input, balanced bridge output power amp circuit I made up back in 2005......has it really been that long ago?
gosh
Thread here.
Post 23 shows the two outputs at 3Vp each phase (6Vp across load resistor), 1MHz.
Can be, but I think the error correction makes the non-linear operation of the mosfet as linear or more so than a typical BJT circuit. Plus mosfets are faster and can better handle the reactive impedance of a speaker. The original HEC circuit, the Amplified Diode, was developed by Dr Hawksford to be used to drive those great big Darlington power transistors that came out in the late 70's. Current gain for Darlington power transistors is far from linear. Bob ingeniously adapted that circuit to use mosfets. Here is the original paper.
gosh Thread here.Post 23 shows the two outputs at 3Vp each phase (6Vp across load resistor), 1MHz.
Can be, but I think the error correction makes the non-linear operation of the mosfet as linear or more so than a typical BJT circuit. Plus mosfets are faster and can better handle the reactive impedance of a speaker. The original HEC circuit, the Amplified Diode, was developed by Dr Hawksford to be used to drive those great big Darlington power transistors that came out in the late 70's. Current gain for Darlington power transistors is far from linear. Bob ingeniously adapted that circuit to use mosfets. Here is the original paper.
Last edited:
There are a number of fet driving bjt amps.... the fet chosen will be able to deliver high peak base current drive and not get very hot doing it.
Can we get an output which uses both fet and bjt? How would you do it?... yes the biasing has to be worked out but it would give the best of both worlds. Someone can figure out how to do it well, I would think. But then, no one has done it yet... in SIM?
Thx-RNMarsh
Can we get an output which uses both fet and bjt? How would you do it?... yes the biasing has to be worked out but it would give the best of both worlds. Someone can figure out how to do it well, I would think. But then, no one has done it yet... in SIM?
Thx-RNMarsh
Wavebourn's done that, and claims good results. IIRC, the details are buried somewhere in John Curl's thread and/or one of Bob Cordell's threads. Maybe best to PM him, to save 1/2 a day searching.
Regards - Godfrey
edit: Nevermind, found it.
Is that the sort of thing you were thinking of?Even mine?
Last edited:
As each technology has its own disadvantage in terms of performance, it would be interesting to know whether this BJT/FET solution:
a) Introduces the disadvantages from both worlds
or
b) Eliminates the disadvantages by some sort of mutual compensation (e.g. gm droop + gm doubling = straight line). If stable, this would be interesting!
I will try to find further info on this (preferably with performance comparisons) - any tips welcome!
a) Introduces the disadvantages from both worlds
or
b) Eliminates the disadvantages by some sort of mutual compensation (e.g. gm droop + gm doubling = straight line). If stable, this would be interesting!
I will try to find further info on this (preferably with performance comparisons) - any tips welcome!
In my simulation, again simulation, the L-FET driver and BJT output has the advantage of it can't drive the outputs to the rail, so the clipping behavior is very benign. No rail stick. Of course, a boosted VAS and driver would overcome that. The disadvantage is less overhead. 2sk1056 was the part I am looking at.
Yes. IMO, this could have the best of both worlds. But too bad it was in the social column and not a DIY build.... would love to see real details for a working amp because you never read about this approach in books.
THx-RNMarsh
It merely requires reading sales brochures.
http://wegavision.pytalhost.com/Accuphase/accuphaseE-206p/Accuphase-E-206-3.jpg
(E305, P102, etcetera)
http://wegavision.pytalhost.com/Accuphase/accuphaseE-206p/Accuphase-E-206-3.jpg
(E305, P102, etcetera)
I disagree strongly with this assertion.
Of course there are always exceptions. There are those who manage to design pretty bad MOSFET amplifiers; at the same time, there are plenty of designers that manage to make BJT designs perform poorly.
Generalizations are generally wrong, but here I am making a generalization
.Cheers,
Bob
Yes, there are some designers who wreck the ft of MOSFETs with too-large gate stopper resistors, but this need not be so. Among other things, gate Zobel networks enable good MOSFET stability with smaller gate stopper resistors. The key is in understanding how oscillators can be formed with MOSFETs and how to damp the associated resonances. That being said, I generally don't try to push ULGF of MOSFET designs much higher than those for BJT designs.
Cheers,
Bob
Just trying to model a HEXFET output version of my REASONABLE project. I am having terrible trouble with stability in clipping or square wave input. Playing in the model with gate stoppers seemed to not have the effect they did on the laterals. 1 to 500 Ohms. Bouncing around, it does seem like it is around the output gate the problem originates, but I found it sensitive to the IPS current and to the choice of a feedback pair for the IPS current source. Could an example of a gate Zobel be shown and what parameters should be used to calculate it? I am attempting to use the IRF 140's with lateral drivers.
I also noticed improvements when adding a cap from the center of the IPS current mirror to the rail. Yet another way to reduce system bandwidth I guess.
I also wonder how much I am chasing calculated anomalies that are not real. Until I build something, it is a guess.
I also noticed improvements when adding a cap from the center of the IPS current mirror to the rail. Yet another way to reduce system bandwidth I guess.
I also wonder how much I am chasing calculated anomalies that are not real. Until I build something, it is a guess.
Why not? It's a given that layout becomes more critical but if you can push the output stage pole as high as possible you can then push the ULGF up and probably get better performance.
Is it fair to compare BJTs to MOSFETs without optimising the design to take advantage of the potential strengths MOSFETs?
The only question for me is whether it is a worth while exercise and it would be less problematic to just use BJTs.
Hi tvrgeek,
I described the vertical MOSFET parasitic oscillator topologies and the gate Zobel technique in my MOSFET power amplifier with error correction paper, published in the JAES in about 1984, and it is available on my web site at CordellAudio.com - Home.
SPICE simulations will generally not pick up these issues, as they leave out many of the inductive parasitics. The HEXFETs are very fast devices, and it is easy for (any) fast device to find a way to oscillate in a circuit. One concern is that fast devices may oscillate at many tens of MHz, and often in a burst fashion, so the parasitic bursts may sometimes go un-noticed and the resulting bad sound will be blamed on all sorts of things, including the HEXFET itself.
Laterals are more tame because they are not as fast, partly due to their internal distributed gate resistance. I think I said something about the speed of laterals in my MOSFET amp cited above. You can't get into as much trouble with them. Often one sees fairly large-value gate stopper resistors used with them, in the 220-470 ohm range. Their gm being lower also makes them more tame.
Beyond a certain value, gate stoppers alone may not stabilize a vertical MOSFET, since there are parasitic capacitances on the gate side of the gate stopper that may bypass the gate stopper in creating an oscillator topology. Damping of the gate circuit, right at the gate pin, is important.
Cheers,
Bob
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.