I think I have an idea of why you design with MOS-FETs for outputs in your later designs. You've explained this in a paper, if I remember.
However, amplifier designers have many different views on devices. Curl and McCormack, plus many others, believe in the J-FET input, MOS-FET driver and bipolar output topology. Others, like David Belles, believe that bipolars are better devices for input and driver stages, with MOS-FETs as outputs. Levinson uses bipolars in every stage.
So, you seem to be alone in your choice of using MOS-FETs for every stage. I'm sure you have good reasons. Still, I would think different devices might excel better in one amplification stage than another, but you obviously made a decision to not use J-FETs or bipolars in any stage.
Share with us how you came to believe this to be the best topology.
However, amplifier designers have many different views on devices. Curl and McCormack, plus many others, believe in the J-FET input, MOS-FET driver and bipolar output topology. Others, like David Belles, believe that bipolars are better devices for input and driver stages, with MOS-FETs as outputs. Levinson uses bipolars in every stage.
So, you seem to be alone in your choice of using MOS-FETs for every stage. I'm sure you have good reasons. Still, I would think different devices might excel better in one amplification stage than another, but you obviously made a decision to not use J-FETs or bipolars in any stage.
Share with us how you came to believe this to be the best topology.
Well, I'm sorry to destroy your illusions, but Nelson is not the strict believer in Mosfets as you seem to think.
For a quite long time now this is no longer true. It is well known here that Nelson Pass uses Jfet inputs in all his amps and he uses bipolars for cascodes. Only the output stage still consists of Mosfets and I guess his reply to why he's doing so, is something along 'because it allows simple designs'.
All the best, Hannes
For a quite long time now this is no longer true. It is well known here that Nelson Pass uses Jfet inputs in all his amps and he uses bipolars for cascodes. Only the output stage still consists of Mosfets and I guess his reply to why he's doing so, is something along 'because it allows simple designs'.
All the best, Hannes
This is one of those questions that either takes a really, really long answer or a really short one. The short answer, as has already been noted, is that Nelson doesn't always use MOSFETs in every circuit position; it certainly isn't the case at present.
The longer answer...hmmm...don't seem to have time for the longer answer, so I will attempt the impossible and try for a mid-length answer.
If you were to look at Nelson's earlier designs, you'd find that a lot of them were all bipolar. Then you'd see JFETs here and there. Then you'd see MOSFETs. Most recently he used power JFETs in one of the First Watt pieces. It's not so much a question of whether a device fits into the FET or bipolar category, it's a question of what characteristics the device has--gain, voltage limitations, current limitations, heat dissipation, input impedance, capacitance, and on and on throughout the entire list of specifications. When you sit down to design a circuit, you choose from a menu of available parts. Some people like steak, some like seafood, and some are vegetarians. Those inclinations tend to show up in what people choose from the menu.
When I was younger, I'd listen to one designer hold forth on why his design was better...and there'd always be this portion where he'd say that bipolars were clearly better than FETs because of this and that. And for a week afterwards, I'd go around thinking that bipolars were better. Then I'd hear another guy telling how FETs were better. And, sure enough, for a week I'd go around repeating what he said. And then a week for tubes, and finally back to bipolars...'round and 'round in a circle, with no end in sight.
So who's right?
The answer is that all gain devices are pretty pathetic. You pick a set of compromises that you think you can work with and you do the best you can.
Grey
The longer answer...hmmm...don't seem to have time for the longer answer, so I will attempt the impossible and try for a mid-length answer.
If you were to look at Nelson's earlier designs, you'd find that a lot of them were all bipolar. Then you'd see JFETs here and there. Then you'd see MOSFETs. Most recently he used power JFETs in one of the First Watt pieces. It's not so much a question of whether a device fits into the FET or bipolar category, it's a question of what characteristics the device has--gain, voltage limitations, current limitations, heat dissipation, input impedance, capacitance, and on and on throughout the entire list of specifications. When you sit down to design a circuit, you choose from a menu of available parts. Some people like steak, some like seafood, and some are vegetarians. Those inclinations tend to show up in what people choose from the menu.
When I was younger, I'd listen to one designer hold forth on why his design was better...and there'd always be this portion where he'd say that bipolars were clearly better than FETs because of this and that. And for a week afterwards, I'd go around thinking that bipolars were better. Then I'd hear another guy telling how FETs were better. And, sure enough, for a week I'd go around repeating what he said. And then a week for tubes, and finally back to bipolars...'round and 'round in a circle, with no end in sight.
So who's right?
The answer is that all gain devices are pretty pathetic. You pick a set of compromises that you think you can work with and you do the best you can.
Grey
Here in the technical vastness of the future we have a myriad
of gain devices to choose from, all offering different disadvantages.
The decision to use Mosfets on the input of the Alephs is an
interesting one - they were chosen because I wanted to have
a very simple 2 stage circuit. Mosfets are noisier and generally
have less linearity that JFETs, but they offer a higher
transconductance and higher current/dissipation.
Currently we use bipolar devices for cascodes on JFET front ends.
They don't measure better but they sound better and their
bandwidth corroesponds to our needs. Recently we re-evaluated
bipolars against our current Mosfets in front end level shifters.
This time we get the same measurements, but our panel doesn't
like the sound.
Live in the Future! It's just starting now....
of gain devices to choose from, all offering different disadvantages.
The decision to use Mosfets on the input of the Alephs is an
interesting one - they were chosen because I wanted to have
a very simple 2 stage circuit. Mosfets are noisier and generally
have less linearity that JFETs, but they offer a higher
transconductance and higher current/dissipation.
Currently we use bipolar devices for cascodes on JFET front ends.
They don't measure better but they sound better and their
bandwidth corroesponds to our needs. Recently we re-evaluated
bipolars against our current Mosfets in front end level shifters.
This time we get the same measurements, but our panel doesn't
like the sound.
Live in the Future! It's just starting now....
Nelson Pass said:.... Mosfets are noisier and generally
have less linearity that JFETs, but they offer a higher
transconductance and higher current/dissipation.
Currently we use bipolar devices for cascodes on JFET front ends.
They don't measure better but they sound better and their
bandwidth corroesponds to our needs. Recently we re-evaluated
bipolars against our current Mosfets in front end level shifters.
This time we get the same measurements, but our panel doesn't
like the sound.
Live in the Future! It's just starting now....
Mr. Pass, javascript:smilie('
')You stated that you currently use bipolar devices for cascodes on JFET front ends. What amps are you talking about? The X and XA series? Or are you talking about First Watt stuff?
You also state that you recently re-evaluated bipolars against MOSFETs as "front end level shifters," and the MOSFETs sound better. What are front end level shifters?
So, what is the compliment of devices on the X and XA series?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.