IMO it is all about unhappy use of BJTs as input diff pair for audio. This BJT topology is fine for DC circuits.
In case of BJTs I prefer another kind of input stage - CFB topology. Almost no SR limitation and almost no danger of input overload by fast signals. For me the only good sounding BJT opamp is AD844 - CFB topology with SR of 2500V/us and transimpedance OLG corner somewhere between 12 - 18 kHz.
In case of BJTs I prefer another kind of input stage - CFB topology. Almost no SR limitation and almost no danger of input overload by fast signals. For me the only good sounding BJT opamp is AD844 - CFB topology with SR of 2500V/us and transimpedance OLG corner somewhere between 12 - 18 kHz.
Thank you, John.
You can make an excellent preamp based on this topology, please look here:
http://www.diyaudio.com/forums/showthread.php?postid=945127#post945127
You can make an excellent preamp based on this topology, please look here:
http://www.diyaudio.com/forums/showthread.php?postid=945127#post945127
traderbam said:
Perhaps off topic, but the question being, if you are using a FET and especially a MOSFET in an input stage (here 40V Vds is easy), why limit the current to 1mA, assuming you can meet power dissipation criteria? Unlike BJTs, increasing the current brings practically no disadvantages, and has the advantage of increasing gm, as well as advantages when designing the next stage. Note that I am skillfully avoiding the issue of a flattening gm curve with increasing Id, so we don't end up discussing the relative merits of order vs amount of nonlinearity
Oh I have. It takes time. I have also found a spice model from Zetex that was rather flattering compared with the 10 parts I bench tested: the actual gm was one third of the model, consistent across all the parts. Parts and model were within datasheet spec. Getting a gm of around 6mS for MOS and J FETs at mA currents is not a problem. Getting 20mS without using 10s of mA may be a problem. I'll keep looking...
> So does biting the cascode bullet help us out?
Yes. It is not so difficult.
> Is there a FET that has 20mS gm at a milliamp or so Id, with Vdg max as low as you like and significantly less capacitance?
Not that I know of.
Why so worry about capacitance. Just drive them with low enough resistance. Nelson uses even IRF9610 for the front end of his Aleph-X which has hugh capacitance. I did not hear any complaints.
Patrick
Yes. It is not so difficult.
> Is there a FET that has 20mS gm at a milliamp or so Id, with Vdg max as low as you like and significantly less capacitance?
Not that I know of.
Why so worry about capacitance. Just drive them with low enough resistance. Nelson uses even IRF9610 for the front end of his Aleph-X which has hugh capacitance. I did not hear any complaints.
Patrick
I don't think it is so simple. The advantages of FETs were more linear gm curve, lower noise, and higher input resistance. The disadvantage is lower gm and higher C. If you have to add another stage to mitigate the disadvantages, if you care about them, then what happens to the those original advantages?
Some designs will work better (sound better) with bipolar inputs.
Some designs will work better (sound better) with bipolar inputs.
Hi,
As suitable FET's seem to be hard to find, expensive or faked I having been looking at the op-amp based design at the following link :- http://www.wnaudio.com/amos100.pdf
This design is in fact very similar to the circuits put forward in Walt Jung's book.
As suitable FET's seem to be hard to find, expensive or faked I having been looking at the op-amp based design at the following link :- http://www.wnaudio.com/amos100.pdf
This design is in fact very similar to the circuits put forward in Walt Jung's book.
PMA said:That's what I have been always voting for - drive it from buffered stage with minimum output impedance.
BTW - IRF9610 sound wonderful as an input stage. People overlook MOSFETs at this position, the only explanation for me is that they have never tried. Many posts and not much work ....
I was wondering about this myself. I prefer the smaller MOSFETs myself, though the IRF610 and the odd IRF710 are on the top end of what I'd use for inputs. The IRFD series and the older 2N7000, BS107 and BS170 and some Zetex parts worked well for me...
john curl said:
BTW, many say Toshiba have discontinued the 2SK170, but that surprises me since they are still available at my local distributor that has discontinued many other japanese semicondictor for reasons of delivery problems, and they still have no warning about these JFETs being about to disappear, which they usually have long before they take them out of stock. I do understan some US companies has taken up production of these devices, but my supplier still sell the Toshibas.
Hi,
From the Leach amp pages...
"Because FET parameters are so unpredictable and FETs are more susceptible to flicker noise, I prefer the BJTs with emitter resistors." http://users.ece.gatech.edu/~mleach/lowtim/bckgrnd.html
Is the unpredictability a problem inherent in the manufacturing of FET's ?
http://users.ece.gatech.edu/~mleach/ece4391/noise_b.pdf
From the Leach amp pages...
"Because FET parameters are so unpredictable and FETs are more susceptible to flicker noise, I prefer the BJTs with emitter resistors." http://users.ece.gatech.edu/~mleach/lowtim/bckgrnd.html
Is the unpredictability a problem inherent in the manufacturing of FET's ?
http://users.ece.gatech.edu/~mleach/ece4391/noise_b.pdf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.