I have already spiced and optimized lineups suggestion,
removing the 10k and upping the emitter resistor on the CCS
to get 1.7ma in each differential.
What I found to be the case is that doing this decreases OLG in the LTP , so I replaced the 10k with 1K and decreased the emitter degen's to compensate.
Here is the unity gain curve of the FA1...
What is funny, is that C7 ,which is removed above..(15pf on the NFB , adds a
pronounced hump (like a 2 pole response) on the amp.
what is the use of it??? I have removed it from the "real world" circuit, no oscillation.
Another thing that is surprising is that balance can be achieved in the LPT, CM ,and VAS by simply manipulating open loop gain.
(without the diode)....
BTW , I now use 2SC2682 for my CCS (120V/100ma device)
Just playin' with a monster OP-AMP....
OS
removing the 10k and upping the emitter resistor on the CCS
to get 1.7ma in each differential.
What I found to be the case is that doing this decreases OLG in the LTP , so I replaced the 10k with 1K and decreased the emitter degen's to compensate.
Here is the unity gain curve of the FA1...
An externally hosted image should be here but it was not working when we last tested it.
What is funny, is that C7 ,which is removed above..(15pf on the NFB , adds a
pronounced hump (like a 2 pole response) on the amp.
what is the use of it??? I have removed it from the "real world" circuit, no oscillation.
Another thing that is surprising is that balance can be achieved in the LPT, CM ,and VAS by simply manipulating open loop gain.
(without the diode)....
BTW , I now use 2SC2682 for my CCS (120V/100ma device)
Just playin' with a monster OP-AMP....
OS
"Why I posted that schematic, is because Douglas Self refered to it,
as to show, that copycats of his schematics, whithout their own knowledge,
will copy even when the original designer made a mistake.
And in this case D. Self called his including of that resistor:
a mistake (bad move)."
Lineup
To be fair to Silicon Chip magazine, that resistor has been there in many of their designs long before D.S. achieved fame with his publications.
The SC ULD2 was referred to in the article as being based on their 20W Class A design of 2007.
To quote from that 2007 article : "Some of the changes in the design (referring to their 15W Class A design from May 1998)
are based on ideas and circuits published by the noted audio designer Douglas Self and outlined in a number of his books (available from the SILICON CHIP Bookshop)
I think that your words are a little harsh, as Douglas Self has been clearly credited by the magazine.
SandyK
as to show, that copycats of his schematics, whithout their own knowledge,
will copy even when the original designer made a mistake.
And in this case D. Self called his including of that resistor:
a mistake (bad move)."
Lineup
To be fair to Silicon Chip magazine, that resistor has been there in many of their designs long before D.S. achieved fame with his publications.
The SC ULD2 was referred to in the article as being based on their 20W Class A design of 2007.
To quote from that 2007 article : "Some of the changes in the design (referring to their 15W Class A design from May 1998)
are based on ideas and circuits published by the noted audio designer Douglas Self and outlined in a number of his books (available from the SILICON CHIP Bookshop)
I think that your words are a little harsh, as Douglas Self has been clearly credited by the magazine.
SandyK
Yeah... sometimes a little bit harsh...me too do those things
Sometimes our passion disturbs our education.
Lineup was bothered because i have posted my car..hehehe..old folk!... now he will post a beautifull text filled with colours saying he is not old.
Lineup is a nice folk... has bad moments alike i have and others have Sandy... also he has the "banishment forum"... you enter there and will be banished..ahahahahah!
The problem is that all of us, including Lineup, are becoming a little bit old... and patience is reduced a lot those days.
A hug for you two fellows... we love audio.... that's the wonderfull thing....hehehe.. a very terrible thing too... we turn biased...you see your passion by this Aussie Magazine..not even yours!
regards,
Carlos
Sometimes our passion disturbs our education.
Lineup was bothered because i have posted my car..hehehe..old folk!... now he will post a beautifull text filled with colours saying he is not old.
Lineup is a nice folk... has bad moments alike i have and others have Sandy... also he has the "banishment forum"... you enter there and will be banished..ahahahahah!
The problem is that all of us, including Lineup, are becoming a little bit old... and patience is reduced a lot those days.
A hug for you two fellows... we love audio.... that's the wonderfull thing....hehehe.. a very terrible thing too... we turn biased...you see your passion by this Aussie Magazine..not even yours!
regards,
Carlos
I wish you could be less personal, Carlos.
Because the words you say have not a nice spirit, in the bottom of your words.
I just think that topic about Current mirrors
should not be Destroyed by a lot of off topic stuff. Like discussing Cars.
You have your own threads for this. Where the moderators does not mind off topic.
---------
Ostripper
When you reduce LTP Current, you lose some gain. And also Bandwidth = speed.
To get much gain + bandwidth we increase currents.
But at same time increase Current will increase noise.
A good compromise is often 0.5 - 2 mA in each LTP transistor
Regards
Lineup
Because the words you say have not a nice spirit, in the bottom of your words.
I just think that topic about Current mirrors
should not be Destroyed by a lot of off topic stuff. Like discussing Cars.
You have your own threads for this. Where the moderators does not mind off topic.
---------
Ostripper
When you reduce LTP Current, you lose some gain. And also Bandwidth = speed.
To get much gain + bandwidth we increase currents.
But at same time increase Current will increase noise.
A good compromise is often 0.5 - 2 mA in each LTP transistor
Regards
Lineup
I too, search for knowledge, the magical sound of the bootstrap
is next on my "hitlist". I will model the original DX amp to
steal the magic sound (not the circuit) of that little cap and
put it into the VAS of frugal 2 .
The unity gain plot is very interesting, every amp has its own,
like a "fingerprint". DX amps ,frugal, Quasi, all different. What
happens at HF affects how an amp behaves in the audible
spectrum and how reliable /stable it is...
Yes current for speed, but emitter degen +- resistors drastically affects OL gain in both current mirror and LTP.
working on it now...
OS
yes, you are doing good progress
now you have good tools to evaluate before you start building for real
emitter resistors are both for good and for bit of sacrifice
-large value improves linearity and reduces distortion
but takes away a bit of gain
-smaller values or no emitter resistor gives maximal gain,
but will give higher distortion
... there are tradeoffs
current level is both for good and for bit of sacrifice
-large currents gives higher gain + bandwidth
but also higher current noise in transistor
-small currents give low noise
but not so high gain and bit lower bandwidth, slewrate
... there are tradeoffs
no global feedback, or lower global negative feedback
often have to use larger emitter resistor values to reduce distortion
and make use of less gain in each transistor
... so instead comes the need for more gain stages, more transistors
in order to get the wanted total gain
in a global feedback amplifier we can not use too much gain in input stage
because this stage will determine so much of the overall quality via the feedback
this is why often only like x10 to x25 is used in input stage
.. even if these transistor can have very high gain capacity
using resistors to reduce the gain to a reasonable level in input
gives high linearity and low distortion in the important input stage
but at the same time we want not too little gain in input
because first stage have a great influence of the total performance
while for global feedback amps the VAS stage can have very high gain with not very linear operation
gain here can be several times 100 or even max of the VAS transistors possibility
This is the art of making good amplifiers.
Making the right decisions. About what we can scarifice a bit of.
To get more of the wanted quality.
now you have good tools to evaluate before you start building for real
emitter resistors are both for good and for bit of sacrifice
-large value improves linearity and reduces distortion
but takes away a bit of gain
-smaller values or no emitter resistor gives maximal gain,
but will give higher distortion
... there are tradeoffs
current level is both for good and for bit of sacrifice
-large currents gives higher gain + bandwidth
but also higher current noise in transistor
-small currents give low noise
but not so high gain and bit lower bandwidth, slewrate
... there are tradeoffs
no global feedback, or lower global negative feedback
often have to use larger emitter resistor values to reduce distortion
and make use of less gain in each transistor
... so instead comes the need for more gain stages, more transistors
in order to get the wanted total gain
in a global feedback amplifier we can not use too much gain in input stage
because this stage will determine so much of the overall quality via the feedback
this is why often only like x10 to x25 is used in input stage
.. even if these transistor can have very high gain capacity
using resistors to reduce the gain to a reasonable level in input
gives high linearity and low distortion in the important input stage
but at the same time we want not too little gain in input
because first stage have a great influence of the total performance
while for global feedback amps the VAS stage can have very high gain with not very linear operation
gain here can be several times 100 or even max of the VAS transistors possibility
This is the art of making good amplifiers.Making the right decisions. About what we can scarifice a bit of.
To get more of the wanted quality.
Using LT (but not believing in it fully) to augment the "real world"
can really make for good sounding amplification (and reliability).
Not linear, huh?? Model my amps, model everyone else's amps,
model "secret" amps that are IP (don't ask ).
Since we listen to the VAS, (OPS just buffers, except for CFP's)
this seems to be a very important aspect to good sound.
With frugal 1 tweaking the VAS (linearizing) is just gain/current/
compensation..
[red is output, blue/ green are VAS devices.]
I will soon update all schema's of frugal 1 threadwide to reflect
on a very good amp with absolutely no errata. (kicks butt
)
I had to play around with the bootstrap aka..BS "mystery", and hope to see the "magic" . A while back AKSA explained to me the
theory but I just had to know more. The bootstrap's natural "rolloff" in gain at HF is indeed interesting..
[This is the frugalamp with a bootstrapped current source]
After seeing this on my sim I pulled out the MJE340 and in
went 2 3K3's and a 100uf cap. The character of the amp's sound
changed , especially in the bass department.
Due to the very good design of this amp the highs were still
clear as a bell with no noise or turn-on thump.
All this was interesting, but I still wanted the better slew and
control of an actively loaded gain section.
[Again , Red is output - green/blue are VAS devices.)
That's why the FA2 (above) seems to be the "best of both worlds" . All the fuss about current mirrors, "secret diodes",
It's HOW the input stage is loaded that makes all the difference
What is cool about the differential VAS is all the options
you have in compensating it. Cdom is tricky ,(I've seen the
"pole zero effect" ) but "doable". The FA 2 is very similar
to Mike B's "symasym" or ampslab "BI240" with the exception
of the 3'rd compensation mechanism on the "back side" of the
VAS [c10/R16]. These components define both the
current and unity gain point of the amp.. .
[THE NEW FRUGALAMP2]
Another good byproduct of this topology is the balance
of the load on the LPT's with the added "gift"
of being able to tailor the amps response.
I'm going to build it, since I still can't/won't truely rely on a sim for
gauging the actual musicality or reliability.
OS
can really make for good sounding amplification (and reliability).
Not linear, huh?? Model my amps, model everyone else's amps,
model "secret" amps that are IP (don't ask
).Since we listen to the VAS, (OPS just buffers, except for CFP's)
this seems to be a very important aspect to good sound.
With frugal 1 tweaking the VAS (linearizing) is just gain/current/
compensation..
An externally hosted image should be here but it was not working when we last tested it.
[red is output, blue/ green are VAS devices.]
I will soon update all schema's of frugal 1 threadwide to reflect
on a very good amp with absolutely no errata. (kicks butt
)I had to play around with the bootstrap aka..BS "mystery", and hope to see the "magic" . A while back AKSA explained to me the
theory but I just had to know more. The bootstrap's natural "rolloff" in gain at HF is indeed interesting..
An externally hosted image should be here but it was not working when we last tested it.
[This is the frugalamp with a bootstrapped current source]
After seeing this on my sim I pulled out the MJE340 and in
went 2 3K3's and a 100uf cap. The character of the amp's sound
changed , especially in the bass department.
Due to the very good design of this amp the highs were still
clear as a bell with no noise or turn-on thump.
All this was interesting, but I still wanted the better slew and
control of an actively loaded gain section.
An externally hosted image should be here but it was not working when we last tested it.
[Again , Red is output - green/blue are VAS devices.)
That's why the FA2 (above) seems to be the "best of both worlds" . All the fuss about current mirrors, "secret diodes",
It's HOW the input stage is loaded that makes all the difference
What is cool about the differential VAS is all the options
you have in compensating it. Cdom is tricky ,(I've seen the
"pole zero effect"
) but "doable". The FA 2 is very similarto Mike B's "symasym" or ampslab "BI240" with the exception
of the 3'rd compensation mechanism on the "back side" of the
VAS [c10/R16]. These components define both the
current and unity gain point of the amp..
.[THE NEW FRUGALAMP2]
An externally hosted image should be here but it was not working when we last tested it.
Another good byproduct of this topology is the balance
of the load on the LPT's with the added "gift"
of being able to tailor the amps response.
I'm going to build it, since I still can't/won't truely rely on a sim for
gauging the actual musicality or reliability.
OS
Don't say that , I,m still the garbage collector. where do you think
I get the parts..
You have me beat in the Layout department, Andrew T. whups
me in math, and I just can't mention why I can't be like glen...
I simmed 'em all (pokemon), just seems to be a logical correlation
between the sims and the final demeanor of the finished
product...
OS
I get the parts..
You have me beat in the Layout department, Andrew T. whups
me in math, and I just can't mention why I can't be like glen...
I simmed 'em all (pokemon), just seems to be a logical correlation
between the sims and the final demeanor of the finished
product...
OS
ostripper said:Using LT (but not believing in it fully) to augment the "real world"
can really make for good sounding amplification (and reliability).
Not linear, huh?? Model my amps, model everyone else's amps,
model "secret" amps that are IP (don't ask
Since we listen to the VAS, (OPS just buffers, except for CFP's)
this seems to be a very important aspect to good sound.
I'm going to build it, since I still can't/won't truely rely on a sim for
gauging the actual musicality or reliability.
OS
OS,
Have you done any linearity sims on Ltspice?
How does your amp go WRT THD and harmonic spectra at 20kHz /
full power / 4 ohms?
cheers
Terry
ostripper.
It is not very unusual to see VAS like in attachement.
In fact, I guess you have seen plenty of them.
How linear is this VAS?
If it is linear, then it would work well as a no feedback standlone amplifier.
But .. at least me, should not want to use it as my amplifier.
Regards
Lineup
Attachments
Yes I have, it seems to be very device dependant , it was
mentioned that I was running my MJE350/40's "too hot"
at 12ma so I set them at 8mA ,it simmed perfectly.
upon making the changes to the real amp I noticed slight
VAS distortion on the picoscope (could hear it too, golden ears).
cranked up the CCS to increase the bias a tad..11ma is
the "sweet spot" for a 350/40 VAS as shown on the frugal1.
I still want more, having built a real "symasym' on a veroboard
I like the sound and it's sim is impressive. I want at least
125W so I'll have to design its "big brother"..
OS
anatech said:
anatech
No not really what?
Seems like you do not agree with me. But you are not clear here.
My original thought was, that VAS stages can sometimes be very un-linear .. in global feedback amplifiers.
In contrast to the input stage,
where we can not temper with linearity too much.
Even if using plenty of global feedback.
Man, you have no idea ,but I'm happy now
Visited rhe CMthread, and added the buffered widlar to my VAS...WOW
most linear thing you could see ..rock solid..
here is schema..
An externally hosted image should be here but it was not working when we last tested it.
All at reduced current , which was my goal...now I can build the dang thing..
AKSA said:
Wrong again. Unlikely that anyone is likely to pay attention, but running a pair of undegenerated MJE350's (even from resistive LTP loads as low as 1.2k) without EF buffers will significantly load the LTP reducing OLG, decreasing OL linearity and the phase shift will actually be worse due to the fact that the miller capacitance of Q4 is working against a relatively high drive impedance (R5) of 1.2k (Q5 doesn't matter because it is cascoded by Q6)
However, for a definative confirmation on the performance evils of EF VAS buffers, have a look at Bob Cordell's amp for example - 2MHz unity loop gain frequency, stable slewing 300V/us and 0.0006% THD-20.
Frugalamp
"Wrong again. Unlikely that anyone is likely to pay attention, but running a pair of undegenerated MJE350's (even from resistive LTP loads as low as 1.2k) without EF buffers will significantly load the LTP reducing OLG,"
IMO, there are far too many designs,(even some of the highly thought of designs in DIYAudio) that heavily load the preceding stage. Some even use low HFE parallel devices at the output of an opamp. Many heavily load the LTP as well.
SandyK
"Wrong again. Unlikely that anyone is likely to pay attention, but running a pair of undegenerated MJE350's (even from resistive LTP loads as low as 1.2k) without EF buffers will significantly load the LTP reducing OLG,"
IMO, there are far too many designs,(even some of the highly thought of designs in DIYAudio) that heavily load the preceding stage. Some even use low HFE parallel devices at the output of an opamp. Many heavily load the LTP as well.
SandyK
This is just round 1 in development, the 350/40's are gone,
being replaced by 2SA1381/2SC3503 higher beta devices
and Cdom is gong down to 47pf. I will post all soon.
this is the best of all the popular AB amps Ive simmed so far
even beating the symasym.
stay tuned..
Ps - I need more models...(devices above)
OS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.