No!
Hi John,
You've said: 'The folded cascode connection is mainly trivial...............probably gives a little more gain.
With the greatest respect john, you are wrong....the folded cascode connection does NOT give a 'a little more gain' ....on the contrary, it contrives to reduce gain drasticaly......I trust running a quick simulation will make this self-evident.....
As for Dieck-man, words fail me.......
WTF are the 'moderators' when you need them?
P.S: Thanks DrG.....and others....
john curl said:
Hi John,
You've said: 'The folded cascode connection is mainly trivial...............probably gives a little more gain.
With the greatest respect john, you are wrong....the folded cascode connection does NOT give a 'a little more gain' ....on the contrary, it contrives to reduce gain drasticaly......I trust running a quick simulation will make this self-evident.....
As for Dieck-man, words fail me.......
WTF are the 'moderators' when you need them?
P.S: Thanks DrG.....and others....
Re: Symmetrical?
Elso:
I have and I would and I've used FETs at the inputs, too ... for years ...
Fred:
My first comp symm amp used 2SA872A/2SC1775 at input and 1st emitter follower, 2SA913/2SC1913 at the following voltage gain stages, emitter follower to output MOSFETs was also 2SA913/2SC1913 and outputs the ubiquitos 2SK134/2SJ49.
Rails were +/-42V, no load. Q points for semis were:
2SA872/2SC1775 diff stage and buffer - 2mA per device
2SA913/2SC1913 vgain stage - 25mA per device (yes, heat sinked!)
2SA913/2SC1913 output drvr - 15mA per device
Outputs - 100 to 250mA per device, depending on how I felt that day
Sorry, I don't remember resistor values, except when I first built the fool thing (~1982), I used the "largish" value of 220ohms for emitter degeneration in the diff stage. I was young then and had no idea (OK, just a little idea) of what I was doing.
mlloyd1
Elso:
I have and I would and I've used FETs at the inputs, too
... for years ...Fred:
My first comp symm amp used 2SA872A/2SC1775 at input and 1st emitter follower, 2SA913/2SC1913 at the following voltage gain stages, emitter follower to output MOSFETs was also 2SA913/2SC1913 and outputs the ubiquitos 2SK134/2SJ49.
Rails were +/-42V, no load. Q points for semis were:
2SA872/2SC1775 diff stage and buffer - 2mA per device
2SA913/2SC1913 vgain stage - 25mA per device (yes, heat sinked!)
2SA913/2SC1913 output drvr - 15mA per device
Outputs - 100 to 250mA per device, depending on how I felt that day
Sorry, I don't remember resistor values, except when I first built the fool thing (~1982), I used the "largish" value of 220ohms for emitter degeneration in the diff stage. I was young then and had no idea (OK, just a little idea) of what I was doing.
mlloyd1
Elso Kwak said:Hi Michael,
First I would ask myself if I would design a symmetrical , completely complementary, circuit or not. Then I would use FET's at the input.
The arrangement of the left schematic I have seen before. SAE? , Borbely???
Re: Re: Symmetrical?
Hi Elso,
Yes...Bongiorno...Borbely....etc...etc.....The circuit on the left looks symmetrical, but without some means to force equality of first stage currents, it's all form and little substance... ....
poor psrr, relatively low open-loop gain, (and therefore poor THD performance), even compared to the circuit on the right, which has no current mirror either,.......
The only comparatively succesfull attempt made in print to usefully improve this topology was by Linsley-Hood, as i have pointed out in another thread:
http://diyaudio.com/forums/showthread.php?postid=233008#post233008
The current mirror used in the later, while not quite forcing the diff. stage's currents to equality as precisely as in Self's design, does go a long way to aleviating its short comings.....
...and...i would steer well clear of FET's at the input.......see Self's book on the subject, with whom i am in total agreement in this regard.........cheers.
Elso Kwak said:Hi Michael,
First I would ask myself if I would design a symmetrical , completely complementary, circuit or not. Then I would use FET's at the input.
The arrangement of the left schematic I have seen before. SAE? , Borbely???
Hi Elso,
Yes...Bongiorno...Borbely....etc...etc.....The circuit on the left looks symmetrical, but without some means to force equality of first stage currents, it's all form and little substance...
....poor psrr, relatively low open-loop gain, (and therefore poor THD performance), even compared to the circuit on the right, which has no current mirror either,.......
The only comparatively succesfull attempt made in print to usefully improve this topology was by Linsley-Hood, as i have pointed out in another thread:
http://diyaudio.com/forums/showthread.php?postid=233008#post233008
The current mirror used in the later, while not quite forcing the diff. stage's currents to equality as precisely as in Self's design, does go a long way to aleviating its short comings.....
...and...i would steer well clear of FET's at the input.......see Self's book on the subject, with whom i am in total agreement in this regard......
...cheers.??????????????
"...and...i would steer well clear of FET's at the input."
And just about every good sounding amp I have heard uses a jfet and/or mosfet front end. Does Douglas Self really know more than Pass, Curl, Borbely, Carr, McCormak, and Marsh? I would think he would be the world's most famous audio designer in that case and have designed known industry benchmark amplifiers like Mr. Pass and Mr. Curl.....
"The point here is, if its completely impossible to run this design without oscillation, then the THd+N graph caould not possibly have been obtained legitithemately........the whole thing stinks.....makes one wonder....are those THD curves in his book worth the paper they are printed on????"
Is that the same Douglas Self refered above in the " Douglas Self op. amp. problem.." thread...... or are there two of them?
http://www.diyaudio.com/forums/showthread.php?postid=232374#post232374
Like Hugh, I can't seen to keep up with all your post.
"...and...i would steer well clear of FET's at the input."
And just about every good sounding amp I have heard uses a jfet and/or mosfet front end. Does Douglas Self really know more than Pass, Curl, Borbely, Carr, McCormak, and Marsh? I would think he would be the world's most famous audio designer in that case and have designed known industry benchmark amplifiers like Mr. Pass and Mr. Curl.....
"The point here is, if its completely impossible to run this design without oscillation, then the THd+N graph caould not possibly have been obtained legitithemately........the whole thing stinks.....makes one wonder....are those THD curves in his book worth the paper they are printed on????"
Is that the same Douglas Self refered above in the " Douglas Self op. amp. problem.." thread...... or are there two of them?
http://www.diyaudio.com/forums/showthread.php?postid=232374#post232374
Like Hugh, I can't seen to keep up with all your post.
There can be no sound engineering basis for FET's of whatever complexion, at the front-end of an audio-frequency power amplifier, unless envisaged source impedance exceeds 10K in magnitude....
....And there is no earthly reason why this should ever be the case in competently designed source equipment....
..Of course...neo-homeopathic...holistic...subjective notions in this respect may be discounted with gusto....
....And there is no earthly reason why this should ever be the case in competently designed source equipment....
..Of course...neo-homeopathic...holistic...subjective notions in this respect may be discounted with gusto....
Wow, I don't read the solid state stuff fast enough. Too absorbed with hollow state...
IME, there are a lot of good-sounding amps out there with bipolars, FETs, and tubes in the front end. Each has its own advantages. I think that calling one categorically "better" is rather limiting. To Fred's point, I think one could name some pretty fine maps with bipolar diff amps in the front end: Bryston, GAS, Adcom, and Crown are four with which I have personal experience. IIRC, the old Robertson, Bedini, and Iverson amps all had bipolar input stages, and their reputations with the golden-ear community are... golden.
Less name-calling and more analysis would be helpful for us ss novices- and I'm not directing that solely at you, Fred. Not at all.
IME, there are a lot of good-sounding amps out there with bipolars, FETs, and tubes in the front end. Each has its own advantages. I think that calling one categorically "better" is rather limiting. To Fred's point, I think one could name some pretty fine maps with bipolar diff amps in the front end: Bryston, GAS, Adcom, and Crown are four with which I have personal experience. IIRC, the old Robertson, Bedini, and Iverson amps all had bipolar input stages, and their reputations with the golden-ear community are... golden.
Less name-calling and more analysis would be helpful for us ss novices- and I'm not directing that solely at you, Fred. Not at all.
The Great Jfet Conspiracy!
"unless envisaged source impedance exceeds 10K in magnitude...."
Oh you mean like volume controls, microphone elements, RC filter networks...
Isn't remarkable that Burr Brown and Analog Devices can sell so many Jfet input op amps with so few applications for them.
Who could possibly be buying them and what would they use them for?
"unless envisaged source impedance exceeds 10K in magnitude...."
Oh you mean like volume controls, microphone elements, RC filter networks...
Isn't remarkable that Burr Brown and Analog Devices can sell so many Jfet input op amps with so few applications for them.
Who could possibly be buying them and what would they use them for?
Re: No reason?????
If transconductance is the issue, then it is one very...very good reason to discount FET's at the input stage of a power amp.......it's at least an order of magnitude lower than that of a BJT,...which translates to an order of magnitude greater, and perforce unnecessary non-linearity, as a result of reduced open-loop gain, and therefore feedback factor......
Jocko Homo said:
If transconductance is the issue, then it is one very...very good reason to discount FET's at the input stage of a power amp.......it's at least an order of magnitude lower than that of a BJT,...which translates to an order of magnitude greater, and perforce unnecessary non-linearity, as a result of reduced open-loop gain, and therefore feedback factor......
Less name-calling? I am all for that!
I don't remember the name calling ....... just naming some of the most talented and well regarded audio designers know to most of the high end community.
Douglas Self's name never seems to come up, except on this forum and in Wireless World. And yes I have read many of his articles. There must be SOME reason for the increasing popularity of jfets and mosfets in many of the new amplifier designs. I don't guess it could anything to do with how they sound. What could that possibly have to do with amplifier design?
I don't remember the name calling ....... just naming some of the most talented and well regarded audio designers know to most of the high end community.
Douglas Self's name never seems to come up, except on this forum and in Wireless World. And yes I have read many of his articles. There must be SOME reason for the increasing popularity of jfets and mosfets in many of the new amplifier designs. I don't guess it could anything to do with how they sound. What could that possibly have to do with amplifier design?
Re: Re: No reason?????
In my view, it's not so much the amount of the transconductance at this stage but the quality (linearity) of it that's important. Remember that, in a typical noninverting opamp configuration, only half of the differential input stage falls inside the feedback loop. The other half needs to be intrinsically very linear if it's going to effectively linearize the rest of the amplifier.mikek said:
That's ten to you in Rio Linda
"its at least an order of magnitude lower than that of a BJT"
I think if you will go compare a 2SJ147 or 2SK170 at 3mA to a typical small signal BJT the BJT is about 2 to 3 times greater, not an order of magnitude. Why do so many amplifier designs use emitter degeneration resistors if transconductance is everything?
Local feedback instead of doing all the negative feedback globally? Na... that couldn't be it...........
"its at least an order of magnitude lower than that of a BJT"
I think if you will go compare a 2SJ147 or 2SK170 at 3mA to a typical small signal BJT the BJT is about 2 to 3 times greater, not an order of magnitude. Why do so many amplifier designs use emitter degeneration resistors if transconductance is everything?
Local feedback instead of doing all the negative feedback globally? Na... that couldn't be it...........
Re: Re: Re: No reason?????
Completely untrue...i fear.....the transistor to which the negative feedback network is returned merely acts as a very linear unity-gain buffer to said network...., with the subtraction of feedback from input signal effectively taking place at the (+) input transistors base-emitter, (or gate-source), as the case may be.....
It may be instructive to compare the diff. stage with series, (voltage)-derived, series (voltage) applied negative feedback, with its equivalent, single common-emitter, (or source), input stage, (from which the basic diff. stage derives).....with similarly applied, and derived feedback....This should provide insight into why this is the case.
Joe Berry said:
Completely untrue...i fear.....the transistor to which the negative feedback network is returned merely acts as a very linear unity-gain buffer to said network...., with the subtraction of feedback from input signal effectively taking place at the (+) input transistors base-emitter, (or gate-source), as the case may be.....
It may be instructive to compare the diff. stage with series, (voltage)-derived, series (voltage) applied negative feedback, with its equivalent, single common-emitter, (or source), input stage, (from which the basic diff. stage derives).....with similarly applied, and derived feedback....This should provide insight into why this is the case.
Re: That's ten to you in Rio Linda
Using degeneration allows you trade input stage trans-admittance for increased linearity over a broader differential input range.
This can be done to a greater extent with BJT's than FET's, without adversely reducing input stage trans-admittance gain to such an extent, that the later stages of the amplifier, particularly the output stage, are left exposed, as it were, by the decline in open-loop gain inevitably engendered by this strategy.
An FET, of whatever persuasion, if degenerated to the same extent as its BJT counterpart, would needlessly subject subsequent amplifier stages to a feedback factor reduced by up to an order of magnitude.
Moreover, there is no way, with current technology, that an FET-based diff. stage, even degenerated to the same extent as its BJT counterpart, can posses the same intrinsic linearity as an all-BJT diff. stage.
The defence rests.
Fred Dieckmann said:
Using degeneration allows you trade input stage trans-admittance for increased linearity over a broader differential input range.
This can be done to a greater extent with BJT's than FET's, without adversely reducing input stage trans-admittance gain to such an extent, that the later stages of the amplifier, particularly the output stage, are left exposed, as it were, by the decline in open-loop gain inevitably engendered by this strategy.
An FET, of whatever persuasion, if degenerated to the same extent as its BJT counterpart, would needlessly subject subsequent amplifier stages to a feedback factor reduced by up to an order of magnitude.
Moreover, there is no way, with current technology, that an FET-based diff. stage, even degenerated to the same extent as its BJT counterpart, can posses the same intrinsic linearity as an all-BJT diff. stage.
The defence rests.
Re: Less name-calling? I am all for that!
I don't have the statistics to say that FETs are increasing (or not) in popularity (do you count by number of designs, units sold, or what?), but there are some solid reasons to use them. I can say the same thing about bipolars. From my admittedly limited experience, the greater transconductance of bipolars is beneficial in reducing input stage distortion. One trade off is input impedance, where FETs excel. Another is bias curent and current noise where, again, FETs get the nod.
As usual, the starting point has to be a definition of design goals and priorities- two people may disagree strongly about what technology is the best because they're aiming at different goals.
FWIW, I've found Self's articles to be very educational and readable for laymen like me.
Fred Dieckmann said:
I don't have the statistics to say that FETs are increasing (or not) in popularity (do you count by number of designs, units sold, or what?), but there are some solid reasons to use them. I can say the same thing about bipolars. From my admittedly limited experience, the greater transconductance of bipolars is beneficial in reducing input stage distortion. One trade off is input impedance, where FETs excel. Another is bias curent and current noise where, again, FETs get the nod.
As usual, the starting point has to be a definition of design goals and priorities- two people may disagree strongly about what technology is the best because they're aiming at different goals.
FWIW, I've found Self's articles to be very educational and readable for laymen like me.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.