Back to the original topic, you can get a similar effect
for a follower device if you bias it with an Aleph type
current source.
For example, you could take an N channel follower and
attach it to the negative rail through the Aleph source
made with a P channel Mosfet (and pnp control transistor)
or vice versa.
for a follower device if you bias it with an Aleph type
current source.
For example, you could take an N channel follower and
attach it to the negative rail through the Aleph source
made with a P channel Mosfet (and pnp control transistor)
or vice versa.
mikek describes it exactly; nothing new, but perhaps worth a shot in a lineamp or low power class-A application.
It is true that one could use other approaches to control the current source. I hadn't thought about using the Aleph configuration, becaue I was thinking more along the lines of trying to control Vgs rather than compensate for load impedance; I guess they accomplish nearly similar things...
It is true that one could use other approaches to control the current source. I hadn't thought about using the Aleph configuration, becaue I was thinking more along the lines of trying to control Vgs rather than compensate for load impedance; I guess they accomplish nearly similar things...
If stability issues can be solved, then linearity could perhaps also be improved by swapping the current source and the resistor. Whether the input device is FET or BJT wouldn't really matter much.
The whole idea is to get something like the circuit in this thread but with better linearity by virtue of the greater loop gain.
http://www.diyaudio.com/forums/showthread.php?threadid=2281
Soon I should be able to find the time to play with it in simulation... meanwhile any related ideas are more than welcome.
The whole idea is to get something like the circuit in this thread but with better linearity by virtue of the greater loop gain.
http://www.diyaudio.com/forums/showthread.php?threadid=2281
Soon I should be able to find the time to play with it in simulation... meanwhile any related ideas are more than welcome.
a current mirror ...
... would be a bit of a change from what I was thinking about, because it has no gain.
I want a high loop gain, and use the (current) feedback through the follower for linearization.
On the other hand, I may have to look at things like what you suggest, because of stability problems.
By the way, I think the circuit has a lot to do with the Sziklai. The PNP I drew in my circuit is a unity gain current buffer that the Sziklai doesn't have. The bottom device in my drawing corrsponds to the common source device in the Sziklai; it juast happens to be the same polarity as the folower, whereas it is opposite in the Sziklai; the current buffer allows it to have the same polarity. You can look at my drawing as a "folded Sziklai".
The PNP serves the same purpose as the capacitor in the circuit listed in the thread on jfet buffers.
-- mirlo
... would be a bit of a change from what I was thinking about, because it has no gain.
I want a high loop gain, and use the (current) feedback through the follower for linearization.
On the other hand, I may have to look at things like what you suggest, because of stability problems.
By the way, I think the circuit has a lot to do with the Sziklai. The PNP I drew in my circuit is a unity gain current buffer that the Sziklai doesn't have. The bottom device in my drawing corrsponds to the common source device in the Sziklai; it juast happens to be the same polarity as the folower, whereas it is opposite in the Sziklai; the current buffer allows it to have the same polarity. You can look at my drawing as a "folded Sziklai".
The PNP serves the same purpose as the capacitor in the circuit listed in the thread on jfet buffers.
-- mirlo
hi mirlo
looking at it from an AC point of view should further illuminate our appreciation of the two circuits.
The sziklai, (A.K.A complementary feedback pair), consists of a common-emitter stage driving a second, complementary-polarity common-emitter stage, with 100% shunt, (voltage)-derived, series, (voltage)-applied negative feedback.
At AC, the circuit you've described amounts to an emitter, (or indeed source), follower with a voltage controlled current sink.
Note that the pnp device, like all BJT's, (and contrary to almost universal opinion), is a voltage-operated device, even though a finite base current is required to remove base-charge as soon as collector-emitter current is established by the application of emitter- base voltage.
Therefore, far from possesing current-feedback, the circuit at AC remains resolutely, a single-transistor, 100% shunt, (voltage)-derived, series, (voltage)-applied negative feedback scheme; ie, a source- follower. The common base base stage, which should'nt affect stability directly, merely induces push-pull action, which enhances the followers current-sinking capabilities and provides for a significant degree of cancellation of even-order harmonics.
looking at it from an AC point of view should further illuminate our appreciation of the two circuits.
The sziklai, (A.K.A complementary feedback pair), consists of a common-emitter stage driving a second, complementary-polarity common-emitter stage, with 100% shunt, (voltage)-derived, series, (voltage)-applied negative feedback.
At AC, the circuit you've described amounts to an emitter, (or indeed source), follower with a voltage controlled current sink.
Note that the pnp device, like all BJT's, (and contrary to almost universal opinion), is a voltage-operated device, even though a finite base current is required to remove base-charge as soon as collector-emitter current is established by the application of emitter- base voltage.
Therefore, far from possesing current-feedback, the circuit at AC remains resolutely, a single-transistor, 100% shunt, (voltage)-derived, series, (voltage)-applied negative feedback scheme; ie, a source- follower. The common base base stage, which should'nt affect stability directly, merely induces push-pull action, which enhances the followers current-sinking capabilities and provides for a significant degree of cancellation of even-order harmonics.
Thanks mikek. "Current feedback" may be the wrong term ...
... but I still think they are similar circuits.
Assuming I did the analysis correctly, and made the right approximation, the loop gain of the folded thing is similar to the loop gain you would get if you loaded a CFP Sziklai with a current source instead of a resistor, and their stability characteristics should be similar.
Since even resistor loaded Sziklai circuits are sometimes reputed to have stability issues (maybe whoever ran into those problems wasn't doing a careful design, who knows) the extra high impedance node makes it worse.
I don't understand why you think the feedback will work primarily on even order harmonics -- it should work on any nonlinearity in essentially the same way; it doesn't symmetrize the circuit.
If I wanted to remove the rest of the even order harmonics, I might try mirroring the polarities and running the mirrored version in parallel. But I'm not sure of the benefit of that if the circuit reduces distortion enough as it is.
... but I still think they are similar circuits.
Assuming I did the analysis correctly, and made the right approximation, the loop gain of the folded thing is similar to the loop gain you would get if you loaded a CFP Sziklai with a current source instead of a resistor, and their stability characteristics should be similar.
Since even resistor loaded Sziklai circuits are sometimes reputed to have stability issues (maybe whoever ran into those problems wasn't doing a careful design, who knows) the extra high impedance node makes it worse.
I don't understand why you think the feedback will work primarily on even order harmonics -- it should work on any nonlinearity in essentially the same way; it doesn't symmetrize the circuit.
If I wanted to remove the rest of the even order harmonics, I might try mirroring the polarities and running the mirrored version in parallel. But I'm not sure of the benefit of that if the circuit reduces distortion enough as it is.
Attachments
wait a minute ... actually ...
... the circuits (both of them, really) are simple current-feedback op-amps wired as a voltage followers.
A current feedback op-amp has a (source or emitter) follower between the noninverting and inverting inputs, and integrates the current in that follower on the capacitance of a high impedance node for gain, and uses the integration result to drive an output stage.
That's pretty much exactly what these things do.
The only difference, I suppose, is that a traditional current feedback op-amp would have another (source or emitter) follower where these circuits use current source outputs.
This is a good realization, because it means there is a lot of literature, app notes, and other BS out there written that might help with stability issues.
As usual, there is nothing new under God's old sky.
I'll draw a picture to clarify this if anyone indicates interest.
I would also be interested in any feedback, (current or otherwise), on my other sketched ideas ... look up "BLS doodle" and "triode region"
-- mirlo
... the circuits (both of them, really) are simple current-feedback op-amps wired as a voltage followers.
A current feedback op-amp has a (source or emitter) follower between the noninverting and inverting inputs, and integrates the current in that follower on the capacitance of a high impedance node for gain, and uses the integration result to drive an output stage.
That's pretty much exactly what these things do.
The only difference, I suppose, is that a traditional current feedback op-amp would have another (source or emitter) follower where these circuits use current source outputs.
This is a good realization, because it means there is a lot of literature, app notes, and other BS out there written that might help with stability issues.
As usual, there is nothing new under God's old sky.
I'll draw a picture to clarify this if anyone indicates interest.
I would also be interested in any feedback, (current or otherwise), on my other sketched ideas ... look up "BLS doodle" and "triode region"
-- mirlo
Hi mirlo...
Hope having nice weekend.....
sziklai:
100% series-voltage feedback, applied across two series connected common-emitter stages, with the shunt-derived feedback signal from the second stage returned, (in series), to the emitter of the first stage.
Your 'white' solid state cascode:
100% series-voltage feedback, applied across ONE stage.
The common base stage, far from being a negative feedback loop, (and contrary to Self, www.dself.demon.co.uk/discrete.htm), infact constitutes a positive feedback loop.
I have used 'signal-tracing' in the following figures to illustrate the later, which shows that a low-swinging signal coupled through the capacitor, ( output 1), or the common-base stage, (output 2), merely adds to the high-swinging signal at the source of the FET caused by the positive-swinging voltage input to it's gate.....In otherwords we have a positive voltage feedback loop.....which is what is required to induce push-pull action.....
It is the push-pull action, and not the positive feedback per-se, which provides for cancellation of even-order artifacts. Which is why i suggested the use of a wilson mirror for the current sink, to enhance the efficacy of the push-pull action.
Any tendency to local parasitic oscillation can be tamed by including small resistor, (of the order of 100R), in series with the FET's gate....
Hope having nice weekend.....
sziklai:
100% series-voltage feedback, applied across two series connected common-emitter stages, with the shunt-derived feedback signal from the second stage returned, (in series), to the emitter of the first stage.
Your 'white' solid state cascode:
100% series-voltage feedback, applied across ONE stage.
The common base stage, far from being a negative feedback loop, (and contrary to Self, www.dself.demon.co.uk/discrete.htm), infact constitutes a positive feedback loop.
I have used 'signal-tracing' in the following figures to illustrate the later, which shows that a low-swinging signal coupled through the capacitor, ( output 1), or the common-base stage, (output 2), merely adds to the high-swinging signal at the source of the FET caused by the positive-swinging voltage input to it's gate.....In otherwords we have a positive voltage feedback loop.....which is what is required to induce push-pull action.....
It is the push-pull action, and not the positive feedback per-se, which provides for cancellation of even-order artifacts. Which is why i suggested the use of a wilson mirror for the current sink, to enhance the efficacy of the push-pull action.
Any tendency to local parasitic oscillation can be tamed by including small resistor, (of the order of 100R), in series with the FET's gate....
It really is negative feedback. To determine the sense of the feedback, you look at the signals in the loop, don't worry about the input.
I'll look for your article, mikek. Unfortunately on this side of the pond access to Electronics World is not as easy as it should be. And sadly that magazine is getting thinner and thinner... perhaps because of forums like this. Does EW even have a website??
-- mirlo
I'll look for your article, mikek. Unfortunately on this side of the pond access to Electronics World is not as easy as it should be. And sadly that magazine is getting thinner and thinner... perhaps because of forums like this. Does EW even have a website??
-- mirlo
signal tracing..
Hi mirlo
Please download 'supplementary class notes' from
http://users.ece.gatech.edu/~mleach/ece4435/
these should clarify why 'signal tracing' can be extremely usefull in establishing whether feedback is positive, (or otherwise).......in a linear circuit.
I however made an incorrect statement earlier on for which i apologise...ie:
Which is why i suggested the use of a wilson mirror for the current sink, to enhance the efficacy of the push-pull action.
The above is incorrect, because the use of a currrent mirror for the current sink would result in negative feedback, and not the desired positive feedback, which is what is required to induce push-pull action....
This can be explained as follows:
Input voltage to FET gate goes high, causing an increase in drain current. This causes an equivalent fall in collector current of the cascode BJT, which is reflected by the current mirror, causing an attendant fall in the source,(and of course drain), current of the FET. This cancels the increase in drain currrent initially caused by the positive-going input signal in the first place....This Mirlo, is negative feedback.
However this does not affect the circuit you've described as it, (quite correctly), does'nt use a current mirror for the FET current sink.
chiao
Hi mirlo
Please download 'supplementary class notes' from
http://users.ece.gatech.edu/~mleach/ece4435/
these should clarify why 'signal tracing' can be extremely usefull in establishing whether feedback is positive, (or otherwise).......in a linear circuit.
I however made an incorrect statement earlier on for which i apologise...ie:
Which is why i suggested the use of a wilson mirror for the current sink, to enhance the efficacy of the push-pull action.
The above is incorrect, because the use of a currrent mirror for the current sink would result in negative feedback, and not the desired positive feedback, which is what is required to induce push-pull action....
This can be explained as follows:
Input voltage to FET gate goes high, causing an increase in drain current. This causes an equivalent fall in collector current of the cascode BJT, which is reflected by the current mirror, causing an attendant fall in the source,(and of course drain), current of the FET. This cancels the increase in drain currrent initially caused by the positive-going input signal in the first place....This Mirlo, is negative feedback.
However this does not affect the circuit you've described as it, (quite correctly), does'nt use a current mirror for the FET current sink.
chiao
publication...
Mirlo,
Alas Electronics world is getting more disappointing by the minute, (to my mind anyway)......the editorial errors in some issues simply reek of a lack of commitment, which is frankly rather disheartening!!
Perhaps i'll have to look at having my next work published in Audio Electronics.
Mirlo,
Alas Electronics world is getting more disappointing by the minute, (to my mind anyway)......the editorial errors in some issues simply reek of a lack of commitment, which is frankly rather disheartening!!
Perhaps i'll have to look at having my next work published in Audio Electronics.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.