Looks great, hopefully we can now just encourage Graham to come back since he (in my opinion) do have great knowledge to share!
And I noticed also all those nice degeneration resistor at the diff.pair and current mirrors, gain goes down but since everything achieve better linearity there's NO/LESSER NEED for big NFB!
And peace to ya all! 😎
/Michael
And I noticed also all those nice degeneration resistor at the diff.pair and current mirrors, gain goes down but since everything achieve better linearity there's NO/LESSER NEED for big NFB!
And peace to ya all! 😎
/Michael
Once upon a time, when this place had all of a half-dozen members or so, Jam and I used to mount our long-suffering steeds and joust with one another about resistors vs. current sources. I insisted that M'lady Resistor's charms were greater, while he, sadly, found it all to easy to resist-her, preferring that painted hussy Current Source. The cad!
Now, I see that not one, but two of our heavier hitters (Nelson and John) have weighed in with opinions similar to mine. Incidentally, John, I wasn't aware that current mirrors had the additional penalty of noise. Interesting. I was just going on sound quality and bandwidth observations.
Nelson's observation--whether in this thread or another, I don't remember--that some people may prefer the spectral balance of distortion that results from higher open loop gain (hence higher negative feedback) may be pertinent to Jam's preference, and also to Graham's. My personal feeling is that I got tired of the upper midrange and treble artifacts that result from high feedback back when I was selling audio. It may measure better, but it doesn't sound like real music.
Speaking only for myself, I use current sources (and by extension, current mirrors) only when nothing else will do. In fact, I'm arguing with myself at the moment as to whether to use a resistor or a current source in the balanced phono stage I've been toying with off and on. Usually I'm comfortable with the idea of a current source biasing a differential since it delivers nominal DC, but being a phono stage where noise is a particular nuisance, I'm back to considering a simple resistor. Radical concept, eh?
Grey
Now, I see that not one, but two of our heavier hitters (Nelson and John) have weighed in with opinions similar to mine. Incidentally, John, I wasn't aware that current mirrors had the additional penalty of noise. Interesting. I was just going on sound quality and bandwidth observations.
Nelson's observation--whether in this thread or another, I don't remember--that some people may prefer the spectral balance of distortion that results from higher open loop gain (hence higher negative feedback) may be pertinent to Jam's preference, and also to Graham's. My personal feeling is that I got tired of the upper midrange and treble artifacts that result from high feedback back when I was selling audio. It may measure better, but it doesn't sound like real music.
Speaking only for myself, I use current sources (and by extension, current mirrors) only when nothing else will do. In fact, I'm arguing with myself at the moment as to whether to use a resistor or a current source in the balanced phono stage I've been toying with off and on. Usually I'm comfortable with the idea of a current source biasing a differential since it delivers nominal DC, but being a phono stage where noise is a particular nuisance, I'm back to considering a simple resistor. Radical concept, eh?
Grey
Grey,
For the record, I am not a huge fan of current mirrors but do work very well in a folded cascode topology.
I am also not a big fan of large amounts of negative feedback( I use a lot of degeneration in my designs) for the same sonic reasons you mentioned, in fact there is a lot to be said for good designs with no global feedback. After all feedback is a subtractive thing and sometimes a bit more is removed than........
Now current sources are a different matter............😉 but they have to be used correctly as I have said before a current source can alo be a current limiter if not used correctly. hint ....hint..
Yes I do miss the old jousting days.........................
For the record, I am not a huge fan of current mirrors but do work very well in a folded cascode topology.
I am also not a big fan of large amounts of negative feedback( I use a lot of degeneration in my designs) for the same sonic reasons you mentioned, in fact there is a lot to be said for good designs with no global feedback. After all feedback is a subtractive thing and sometimes a bit more is removed than........
Now current sources are a different matter............😉 but they have to be used correctly as I have said before a current source can alo be a current limiter if not used correctly. hint ....hint..
Yes I do miss the old jousting days.........................
Attachments
How about for twenty dollars?
"After all feedback is a subtractive thing and sometimes a bit more is removed"
Emitter and source degeneration are negative feedback............
Come to think about it ......... I don't think I have ever seen an amplifier (with voltage and current gain, not just a follower) that did not have negative feedback somewhere in the circuit. I am reminded of that old joke whose punch line goes: "Now that we have established what you are, we are merely haggling over the price."
"After all feedback is a subtractive thing and sometimes a bit more is removed"
Emitter and source degeneration are negative feedback............
Come to think about it ......... I don't think I have ever seen an amplifier (with voltage and current gain, not just a follower) that did not have negative feedback somewhere in the circuit. I am reminded of that old joke whose punch line goes: "Now that we have established what you are, we are merely haggling over the price."
We all have a tendency to make things more complex than they need to be.......................
I have seen a lot of designs that are overly complex, current mirrors on top of cascodes and current sources and the list goes on. It might be a great excersise in design but the question to be asked is how does it sound and not how low can I get the distortion or how good it looks on the simulator.
I wish more people would listen to Mr.Pass (and sometimes Fred ) he has dropped a lot of hints that have been ignored. A prime example adding a current mirror to the differential of the Aleph. One member posted that the addition of a mirror to the differential reduced the distortion by a factor of ten and claimed he had improved it, but it sure hurt the sound. Mr. Pass had said so himself, that being said I am sure there are other ways to improve the Aleph...........and I will leave that to the Master.
Regards,
Jam
I have seen a lot of designs that are overly complex, current mirrors on top of cascodes and current sources and the list goes on. It might be a great excersise in design but the question to be asked is how does it sound and not how low can I get the distortion or how good it looks on the simulator.
I wish more people would listen to Mr.Pass (and sometimes Fred ) he has dropped a lot of hints that have been ignored. A prime example adding a current mirror to the differential of the Aleph. One member posted that the addition of a mirror to the differential reduced the distortion by a factor of ten and claimed he had improved it, but it sure hurt the sound. Mr. Pass had said so himself, that being said I am sure there are other ways to improve the Aleph...........and I will leave that to the Master.
Regards,
Jam
post #168 simulations
Attached are simulations on the circuit in post #168 using LTSpice. This is my cascoded input and cascoded current sources embellishment of Graham's earlier circuit.
TOP plot: 20KHz - 50mv intput 450mv output into 1 ohm load
BOTTOM plot: 500KHz - 50mv input 450mv output into 1 ohm
Low noise and low distortion at very low power is one of my main goals.
I will probably build both circuits to determine how the extra complexity for cascodes affects the sound. Circuit comments are appreciated to help get the best design using the best transistors before I order parts.
JCX...I'm still working on the cascoded JFETs, and would appreciate suggestions for transistor numbers that cascode well. I will use resistor loads for the JFET input stage and +/- 12 volt to 15 volt power supplies.
Attached are simulations on the circuit in post #168 using LTSpice. This is my cascoded input and cascoded current sources embellishment of Graham's earlier circuit.
TOP plot: 20KHz - 50mv intput 450mv output into 1 ohm load
BOTTOM plot: 500KHz - 50mv input 450mv output into 1 ohm
Low noise and low distortion at very low power is one of my main goals.
I will probably build both circuits to determine how the extra complexity for cascodes affects the sound. Circuit comments are appreciated to help get the best design using the best transistors before I order parts.
JCX...I'm still working on the cascoded JFETs, and would appreciate suggestions for transistor numbers that cascode well. I will use resistor loads for the JFET input stage and +/- 12 volt to 15 volt power supplies.
Attachments
Jam, we experienced designers can only 'hint' at things. If we came on too strong, we would wind up in 'texas' or even worse. It is frustrating, sometimes. 
Mr. Curl,
You are right, most of us forget that experience is the best teacher . Well, Texas is not that bad and with your and Mr.Pass's credentials I doubt you will if you will be sent anywhere that far south.
I have seen a lot of budding designers turn a silk purse into a sows ear, but then again they might end up working for Mark Levinson........... 😀
You are right, most of us forget that experience is the best teacher . Well, Texas is not that bad and with your and Mr.Pass's credentials I doubt you will if you will be sent anywhere that far south.
I have seen a lot of budding designers turn a silk purse into a sows ear, but then again they might end up working for Mark Levinson........... 😀
LineSource,
engaging Mr Curl or Hanson on the subject will be much more useful, I've only used discrete fets a few times in buffers and current sources, nothing as complex as your complementary cascoded matched diff pair; search here and on AudioAsylum for the few gems they have let trickle out
some info is at the Vishay web site, they've bought out Siliconix which has several must read fet app notes
most of the fet diff pairs I use come with dozens of supporting transistors and have thousands of hours of engineering by people with years of experience, working on advanced semiconductor processes that discrete fet users can only dream about (except for noise however, why hasn’t any manufacturer taken another run at low noise, the AD743/5 is still the best I know of)
by starting at the other end (the output stage), and using a very good op amp front end you could be listening now while contemplating "the best topology" to replace the op amp
engaging Mr Curl or Hanson on the subject will be much more useful, I've only used discrete fets a few times in buffers and current sources, nothing as complex as your complementary cascoded matched diff pair; search here and on AudioAsylum for the few gems they have let trickle out
some info is at the Vishay web site, they've bought out Siliconix which has several must read fet app notes
most of the fet diff pairs I use come with dozens of supporting transistors and have thousands of hours of engineering by people with years of experience, working on advanced semiconductor processes that discrete fet users can only dream about (except for noise however, why hasn’t any manufacturer taken another run at low noise, the AD743/5 is still the best I know of)
by starting at the other end (the output stage), and using a very good op amp front end you could be listening now while contemplating "the best topology" to replace the op amp
BETTER AMPLIFIER TOPOLIGIES??
I would appreciate sketches of any other amplifier topologies which would deliver very low noise and distortion at low power into a 1 ohm resistive load.
A complementary differential topology using cascoded JFETs would be very helpful. I am currently working with N cascode of 2SK389 input 2SK246 cascode, and P cascode of 2SJ109 input 2SJ103 cascode, and would appreciate other pairing recommendations which have higher transconductance and Id current in order to use closer to 10ma class A bias. Going from JFETS to bipolars in the VAS stage with low distortion presents a problem unless the bias is high.
Here is an update of one design I am considering.
I would appreciate sketches of any other amplifier topologies which would deliver very low noise and distortion at low power into a 1 ohm resistive load.
A complementary differential topology using cascoded JFETs would be very helpful. I am currently working with N cascode of 2SK389 input 2SK246 cascode, and P cascode of 2SJ109 input 2SJ103 cascode, and would appreciate other pairing recommendations which have higher transconductance and Id current in order to use closer to 10ma class A bias. Going from JFETS to bipolars in the VAS stage with low distortion presents a problem unless the bias is high.
Here is an update of one design I am considering.
Attachments
john curl said:Jam, we experienced designers can only 'hint' at things. If we came on too strong, we would wind up in 'texas' or even worse. It is frustrating, sometimes.
Not true, certainly as far as I am concerned. You may post anything here as long as it isn't rude, or infringes another's IP.
Anyway, the sinbin sometimes brings out the best in people, look at Fred for instance 🙂
/ducks and covers....
I've been to Texas several times. It's a wonderful place with
friendly cheerful people and lots and lots of space. I recommend
going in the spring when it's green and not too hot.
Also I highly recommend a movie about Texas called True Stories.
😎
friendly cheerful people and lots and lots of space. I recommend
going in the spring when it's green and not too hot.
Also I highly recommend a movie about Texas called True Stories.
😎
The stars at night are big and bright ..........
"Jam, we experienced designers can only 'hint' at things. If we came on too strong, we would wind up in 'texas' or even worse. " - J Curl
Down here with us amatuer designers.............. The great thing is when you are already in Texas you can come on as strong as you like.
You and Nelson keep dropping those 'hints'. There have been some really good one's that sneak into the post about a related subject.
They usually escape many people's notice but are very often much more interesting than the main subject. 
Deep in the heart of Texas,
Fred
"Jam, we experienced designers can only 'hint' at things. If we came on too strong, we would wind up in 'texas' or even worse. " - J Curl
Down here with us amatuer designers.............. The great thing is when you are already in Texas you can come on as strong as you like.
You and Nelson keep dropping those 'hints'. There have been some really good one's that sneak into the post about a related subject.
They usually escape many people's notice but are very often much more interesting than the main subject. Deep in the heart of Texas,
Fred
Hints
John, I believe the problem for most experienced designers is not telling there trade secrets but keeping up appearences as a SOTA designer. But I appreciate your hints.john curl said:Jam, we experienced designers can only 'hint' at things. If we came on too strong, we would wind up in 'texas' or even worse. It is frustrating, sometimes.
Elso, that is true. But Nelson: You are known as an easy going good humored guy, but I am not, as you know. 😉
Indeed Fred,
I liked NP's comment about his use of CCS........
You should know there is always one better than Texas for the naysayers; Downunder, where no-one is ever listening.......
Cheers,
Hugh
I liked NP's comment about his use of CCS........
You should know there is always one better than Texas for the naysayers; Downunder, where no-one is ever listening.......
Cheers,
Hugh
- Status
- Not open for further replies.