"The JAES article you're talking about, can it be the one about error correction? " I think it may be this one, but without making an 80 mile trip to the engineering library I can't be sure.
Search Results
A Topology to Linearize Miller-Effect Compensated Amplifiers
Vol. 32, Number 6 pp. 430 (1984)
Author: Steven J. Gunderson
Abstract: An additional topology to Miller-effect compensated amplifiers, which reduces high-frequency distortion, is described. The addition is implementable in discrete or integrated circuits, with little or no additional complexity required in designs incorpor
" The AudioXpress link tells me nothing more than there are pcb's available. Somebody who can fill in for me?" The Lang amplifier is a cross coupled dual diff input amplifier with the interstage we are discussing. It is designed for two pair of TO3 lateral MOSFET outputs. If the Semelabs Buz900D/905D were used it could deliver as much as 50W class A or 200W class AB.
Search Results
A Topology to Linearize Miller-Effect Compensated Amplifiers
Vol. 32, Number 6 pp. 430 (1984)
Author: Steven J. Gunderson
Abstract: An additional topology to Miller-effect compensated amplifiers, which reduces high-frequency distortion, is described. The addition is implementable in discrete or integrated circuits, with little or no additional complexity required in designs incorpor
" The AudioXpress link tells me nothing more than there are pcb's available. Somebody who can fill in for me?" The Lang amplifier is a cross coupled dual diff input amplifier with the interstage we are discussing. It is designed for two pair of TO3 lateral MOSFET outputs. If the Semelabs Buz900D/905D were used it could deliver as much as 50W class A or 200W class AB.
hi djk...
the gunderson jaes article is also available as us patent #4511857 from
www.uspto.gov
....you will need to install free tiff viewer to download the material.....
the gunderson jaes article is also available as us patent #4511857 from
www.uspto.gov
....you will need to install free tiff viewer to download the material.....
I use "Patent Grabber" for Macintosh which can download tiff or PDF's! Send me a message if anyone wants pdf's. I'm sure there is a freeware or shareware for Windows too. Check http://www.versiontracker.com
I notice though that this patent is a little bit to old for pdf.
I notice though that this patent is a little bit to old for pdf.
Re: the electrical performance is astonishing and I'm sure the musical performance also
I just wan't to tell that i have tried this out this weekend.. I had rebuild one of my old amps which my little brother had broken by a party!! Dammnnn!!!! My dad wanted it.
Okay first i repaired and test it then i removed the Servo and modified the feedback loop as in the circuit for the schematic uploaded by Per.
Only one trimmer in the circuit for idlecurrent -- dc offset at the output +/- 5 - 10mV without matching of the BJT (Sorry volks! No Fet's at all)....
Now this amp sounds better than my own
...... snif snif I wan't IT BACK ... And he is only laughing!
I have listen to it for some hours this weekend. Maybe i will post the schematic ... Maybe i don't .. You never know!?
Sonny
HarryHaller said:
I just wan't to tell that i have tried this out this weekend.. I had rebuild one of my old amps which my little brother had broken by a party!! Dammnnn!!!! My dad wanted it.
Okay first i repaired and test it then i removed the Servo and modified the feedback loop as in the circuit for the schematic uploaded by Per.
Only one trimmer in the circuit for idlecurrent -- dc offset at the output +/- 5 - 10mV without matching of the BJT (Sorry volks! No Fet's at all)....
Now this amp sounds better than my own
...... snif snif I wan't IT BACK ... And he is only laughing!I have listen to it for some hours this weekend. Maybe i will post the schematic ... Maybe i don't .. You never know!?
Sonny
an interesting quote...
"Separately, there are a few areas in even the best designs where cost considerations are evident, but the conscientious audio designer makes sure that to the best of his knowledge, they don't impose sonic strictures. But the more expensive approach does not always bring improvement -- audible or otherwise. Certainly improvement does not necessarily follow from increasing complexity. More likely, the reverse is true. Progress is made when you scientifically systematize and quantify noted effects. This industry has been besieged by a number of unsubstantiated hypotheses (and its share of sales malarkey) of late.
"We need more scientific methodology so there is less 'caveat emptor', even if the snake oil has been largely reserved for those who can (or wish to) afford it. Objective double-blind testing has eradicated some long-held audio myths -- to my ears at least. Not that all amps, or preamps, sound alike -- but a lot of good ones are not necessarily distinguishable. Sure, use a more expensive part if it really does sound better, but don't waste a lot of peoples' money if you can't prove it. The best designs evolve from those individuals and companies who maintain a healthy skepticism for unsupported postulates, but are quick to grasp the provable achievements. The real art is bringing the greatest good (music) to the greatest number."
by the late bob tucker of Dynaco
"Separately, there are a few areas in even the best designs where cost considerations are evident, but the conscientious audio designer makes sure that to the best of his knowledge, they don't impose sonic strictures. But the more expensive approach does not always bring improvement -- audible or otherwise. Certainly improvement does not necessarily follow from increasing complexity. More likely, the reverse is true. Progress is made when you scientifically systematize and quantify noted effects. This industry has been besieged by a number of unsubstantiated hypotheses (and its share of sales malarkey) of late.
"We need more scientific methodology so there is less 'caveat emptor', even if the snake oil has been largely reserved for those who can (or wish to) afford it. Objective double-blind testing has eradicated some long-held audio myths -- to my ears at least. Not that all amps, or preamps, sound alike -- but a lot of good ones are not necessarily distinguishable. Sure, use a more expensive part if it really does sound better, but don't waste a lot of peoples' money if you can't prove it. The best designs evolve from those individuals and companies who maintain a healthy skepticism for unsupported postulates, but are quick to grasp the provable achievements. The real art is bringing the greatest good (music) to the greatest number."
by the late bob tucker of Dynaco
OK, some time has gone since I started this thread. I have dived into the circuit and I think I understand it but can anyone tell me the pros and cons between
a convential diff input stage followed by a ordinary high gain stage followed by an emitter follower
vs.
a connvential diff input stage plus a folded cascode (serving as a high gain stage) followed be an emitter follower.
Is there a significant advantage of this folded cascode?
a convential diff input stage followed by a ordinary high gain stage followed by an emitter follower
vs.
a connvential diff input stage plus a folded cascode (serving as a high gain stage) followed be an emitter follower.
Is there a significant advantage of this folded cascode?
Peranders:
Thanks for not giving up on this thread; I'd like to learn something out of it also
I always though that the big benefit to using a folded cascode was when you needed to do both level shifting and isolation (as in the conventional cascode) in a single stage. I had suspected that the conventional cascode resulted in better overall performance because in the folded cascode you loose a little current in the input to the cascode.
I'm willing to learn from the experienced gurus and experts; only one of my texts even mentions folded cascodes. Maybe I'll wander over to newsgroup sci.electronics and ask guru Win Hill and report back what he says; that should get things going a bit
mlloyd1
Thanks for not giving up on this thread; I'd like to learn something out of it also
I always though that the big benefit to using a folded cascode was when you needed to do both level shifting and isolation (as in the conventional cascode) in a single stage. I had suspected that the conventional cascode resulted in better overall performance because in the folded cascode you loose a little current in the input to the cascode.
I'm willing to learn from the experienced gurus and experts; only one of my texts even mentions folded cascodes. Maybe I'll wander over to newsgroup sci.electronics and ask guru Win Hill and report back what he says; that should get things going a bit
mlloyd1
OK, as I mentioned earlier, I submitted a question to the
sci.electronics.design newsgroup to ask about folded cascodes.
For those of you using a low bandwidth connection (the group
generates almost as much traffic as we do here, even without
H.H. gratuitous images <--- JOKE WARNING! ), here is the text of the
replies, led off by my original question.
The first response is from Win Hill, one of the authors of "The Art
of Electronics".
Let's get this party started .... I mean resumed ......
mlloyd1
..............
Subject: Re: folded cascode vs. conventional cascode (Win, are you out there?)
Date: Mon, 03 Jun 2002 13:28:48 GMT
Michael <mlloyd1@NO-SPAM-PLEASEenteract.com> wrote:
>OK, I'll confess. I a wee bit of an old fart. None of my texts talk about
>folded cascodes. As near as I can tell, the advantage to using a folded
>cascode over a conventional cascode is the combination of doing both level
>shifting AND (the usual benefit) "isolation" in a single stage.
>
>For the sake of discussion, I'm talking DC to 500KHz tops - I'll pass on
>light amplification for my audio, I pick up enough noise as it is.
>
>Just in case it isn't clear enough, my primary question is: Are there any
>technical performance reasons (linearity, bandwidth, etc) to prefer a
>folded cascode over a conventional cascode to be used with, say, a JFET
>diff amp in the input stage of an audio power amp? I know the Jeff Rowland
>power amps from some years ago did something along this line, but he seems
>to have shifted to those IC power amp thingies now (LM3886, etc). Gotta
>drive that cost down, don't cha know.
____________
From: Win Hill <whill@picovolt.com>
>
> ... my primary question is: Are there any technical
> performance reasons (linearity, bandwidth, etc) to prefer
> a folded cascode over a conventional cascode ...
Both circuits are cascodes, so that's a draw. The folded
cascode provides level shifting as you said and that's a
plus. There is one issue associated with a carelessly-
designed folded-cascode circuit, and that's excess noise.
Consider the output current of a conventional cascode,
i = gm Vin
By contrast, the output from a folded cascode is
i = I_bias - gm Vin,
where I_bias = V_bias / R and R is the pullup resistor.
This can be rearranged to reflect the current inversion.
- i = gm (Vin + en) - V_bias / R
Normally one ignores the 2nd term, calling it a dc bias.
We can expand these equations to include noise sources,
i = gm (Vin + en)
- i = gm (Vin + en) - (V_bias + Vn) / R
where en is the input stage's voltage noise and Vn is the
noise in the path, including the folded cascode transistor.
It's customary to reflect all noise back to the input so we
can compare it with the signal to determine SNR.
- i = gm [ Vin + en - V_bias/(R gm) - Vn/(R gm) ]
We've defined V_bias as the quiet part of the bias voltage,
so we can eliminate that. We have two noise components,
en and Vn / gm R
To get a handle on these we can take en as the Johnson noise
in the input transistor's re = 1/gm = kT/qIc so we see that
it's related to our standing input-stage bias current Ic.
The bias resistor R is also related to Ic. The bias current
is generally 2 Ic, so that it can reflect a full Ic onward,
and it's generally derived from a voltage reference V_bias,
so R = V_bias / 2 Ic
Let's take some typical values. We'll pick Ic = 1.0mA so
that gm is 40mmho, and we'll pick en = 8nV for our input
stage's voltage noise density. We'll say our bias voltage
is 2.5V, so our bias resistor is 1.25k and the dimensionless
gm * R = 50. This says that the folded-cascode bias-current
zener noise will be reflected back to the input divided by 50,
so we want our zener bias-voltage noise to be much less than
8 * 50 = 400nV, let's say 150nV. Sadly many zeners are MUCH
noisier than that. So are most voltage references, such as
an LM385-2.5, at 800nV per root Hertz. So are power-supply
voltages, if one divides a portion for use as a reference.
This means that a careless designer who fails to properly
filter the voltage for the folded-cascode bias current has
a noisier amplifier as a result.
What about the voltage-noise of our folded cascode transistor?
It's also reduced by gm * R or 50x, which is another way of
saying it's reduced by the ratio re/R, which makes good sense.
We're happy to be able to ignore it entirely.
OK, Michael, I hope this has been helpful. Even though the
noise voltages discussed above are very small, smaller than
the signal input to a power amplifier, one could say it's a
consideration, and I had to come up with something!
___
Thanks, /.-.\
((( ))
- Win \\\//
\\\
Winfield Hill //\\\
Rowland Institute for Science /// \\\
Cambridge, MA \/ \/
______________
From: "Kevin Aylward" <kevin@anasoft.co.uk>
A key, or _the_ key, thing about a folded cascode is that you can get it
to run at a lower supply voltage then if it was non folded. Ideally you
try and bias the cascode so that only about 300mm or so is across the
current sources feeding the collectors of the differential pair. A
disadvantage is that the input offset is usually at least twice as much.
This is because two currents are subtracted from each other etc...
Kevin Aylward
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
______________
From: fpm@u.washington.edu (Frank Miles)
A folded cascode (if I understand your terminology correctly) may introduce
greater additional distortion products due to Vbe variation in the CB stage.
For old 'scope designers, its added current consumption could be a problem,
though this would depend on what power supplies were present. As someone
else already pointed out, the folded cascode might fit inside power supplies
better than a normal cascode.
-frank
sci.electronics.design newsgroup to ask about folded cascodes.
For those of you using a low bandwidth connection (the group
generates almost as much traffic as we do here, even without
H.H. gratuitous images <--- JOKE WARNING!
), here is the text of thereplies, led off by my original question.
The first response is from Win Hill, one of the authors of "The Art
of Electronics".
Let's get this party started .... I mean resumed ......
mlloyd1
..............
Subject: Re: folded cascode vs. conventional cascode (Win, are you out there?)
Date: Mon, 03 Jun 2002 13:28:48 GMT
Michael <mlloyd1@NO-SPAM-PLEASEenteract.com> wrote:
>OK, I'll confess. I a wee bit of an old fart. None of my texts talk about
>folded cascodes. As near as I can tell, the advantage to using a folded
>cascode over a conventional cascode is the combination of doing both level
>shifting AND (the usual benefit) "isolation" in a single stage.
>
>For the sake of discussion, I'm talking DC to 500KHz tops - I'll pass on
>light amplification for my audio, I pick up enough noise as it is
.>
>Just in case it isn't clear enough, my primary question is: Are there any
>technical performance reasons (linearity, bandwidth, etc) to prefer a
>folded cascode over a conventional cascode to be used with, say, a JFET
>diff amp in the input stage of an audio power amp? I know the Jeff Rowland
>power amps from some years ago did something along this line, but he seems
>to have shifted to those IC power amp thingies now (LM3886, etc). Gotta
>drive that cost down, don't cha know.
____________
From: Win Hill <whill@picovolt.com>
>
> ... my primary question is: Are there any technical
> performance reasons (linearity, bandwidth, etc) to prefer
> a folded cascode over a conventional cascode ...
Both circuits are cascodes, so that's a draw. The folded
cascode provides level shifting as you said and that's a
plus. There is one issue associated with a carelessly-
designed folded-cascode circuit, and that's excess noise.
Consider the output current of a conventional cascode,
i = gm Vin
By contrast, the output from a folded cascode is
i = I_bias - gm Vin,
where I_bias = V_bias / R and R is the pullup resistor.
This can be rearranged to reflect the current inversion.
- i = gm (Vin + en) - V_bias / R
Normally one ignores the 2nd term, calling it a dc bias.
We can expand these equations to include noise sources,
i = gm (Vin + en)
- i = gm (Vin + en) - (V_bias + Vn) / R
where en is the input stage's voltage noise and Vn is the
noise in the path, including the folded cascode transistor.
It's customary to reflect all noise back to the input so we
can compare it with the signal to determine SNR.
- i = gm [ Vin + en - V_bias/(R gm) - Vn/(R gm) ]
We've defined V_bias as the quiet part of the bias voltage,
so we can eliminate that. We have two noise components,
en and Vn / gm R
To get a handle on these we can take en as the Johnson noise
in the input transistor's re = 1/gm = kT/qIc so we see that
it's related to our standing input-stage bias current Ic.
The bias resistor R is also related to Ic. The bias current
is generally 2 Ic, so that it can reflect a full Ic onward,
and it's generally derived from a voltage reference V_bias,
so R = V_bias / 2 Ic
Let's take some typical values. We'll pick Ic = 1.0mA so
that gm is 40mmho, and we'll pick en = 8nV for our input
stage's voltage noise density. We'll say our bias voltage
is 2.5V, so our bias resistor is 1.25k and the dimensionless
gm * R = 50. This says that the folded-cascode bias-current
zener noise will be reflected back to the input divided by 50,
so we want our zener bias-voltage noise to be much less than
8 * 50 = 400nV, let's say 150nV. Sadly many zeners are MUCH
noisier than that. So are most voltage references, such as
an LM385-2.5, at 800nV per root Hertz. So are power-supply
voltages, if one divides a portion for use as a reference.
This means that a careless designer who fails to properly
filter the voltage for the folded-cascode bias current has
a noisier amplifier as a result.
What about the voltage-noise of our folded cascode transistor?
It's also reduced by gm * R or 50x, which is another way of
saying it's reduced by the ratio re/R, which makes good sense.
We're happy to be able to ignore it entirely.
OK, Michael, I hope this has been helpful. Even though the
noise voltages discussed above are very small, smaller than
the signal input to a power amplifier, one could say it's a
consideration, and I had to come up with something!
___
Thanks, /.-.\
((( ))
- Win \\\//
\\\
Winfield Hill //\\\
Rowland Institute for Science /// \\\
Cambridge, MA \/ \/
______________
From: "Kevin Aylward" <kevin@anasoft.co.uk>
A key, or _the_ key, thing about a folded cascode is that you can get it
to run at a lower supply voltage then if it was non folded. Ideally you
try and bias the cascode so that only about 300mm or so is across the
current sources feeding the collectors of the differential pair. A
disadvantage is that the input offset is usually at least twice as much.
This is because two currents are subtracted from each other etc...
Kevin Aylward
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
______________
From: fpm@u.washington.edu (Frank Miles)
A folded cascode (if I understand your terminology correctly) may introduce
greater additional distortion products due to Vbe variation in the CB stage.
For old 'scope designers, its added current consumption could be a problem,
though this would depend on what power supplies were present. As someone
else already pointed out, the folded cascode might fit inside power supplies
better than a normal cascode.
-frank
Thank you for your work to hunt an answer in this matter. I prefer first an explaination based on facts to start with then it may be useful to know what it meens in real life.
To summerize: Folded cascode isn't so great after all, more like a matter of taste. I think problems with unbalance can be a real problem if we talk distortion but the japaneze guy had acheived really low distortion (if we talk THD).
To summerize: Folded cascode isn't so great after all, more like a matter of taste. I think problems with unbalance can be a real problem if we talk distortion but the japaneze guy had acheived really low distortion (if we talk THD).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.