MikeB said:
Hi Mike,
the bootstrapped cascade version used by Mark Levinson in they amplifiers. As I remember the reson was the lower distortion, which is same as Your results. They used it in the VAS stage as well.
The reference volteg created by diode string, bypassed with small film capacitors.
So it reduce the distortion, maybe not 1000 time lower, but it does!
sajti
Hi !
To show what i try to tell i attached fft-plots. The upper is the wrong cascoded (showing the effect), the lower the functional one.
Red = output
Blue = input (voltage at base of +input)
Green = feedback (voltage at base of -input)
Observe on the upper one that feedback and input shows the same signal, creating "no" error voltage. The input signal already contains the distortions, and the feedbacksignal shows the same distortions, yielding in openloop distortions + the distortions at input. The feedback is not functional here.
The lower one still has this effect, but reduced by -60db, and the feedback does its job giving thd of ~-130db.
It's likely that real world devices behave even worse.
The concept of feedback is still valid, but can be easily knocked out, especially for higher freqs.
I tried a lot to find flaws in the concept of feedback, except the mechanism of distorted distortions through the feedbackloop giving high order harmonics all i found was that feedback works if used correctly.
The next days i will try to build a really low distortive amp, i have the feeling that such amp sounds absolutely great...
Mike
To show what i try to tell i attached fft-plots. The upper is the wrong cascoded (showing the effect), the lower the functional one.
Red = output
Blue = input (voltage at base of +input)
Green = feedback (voltage at base of -input)
Observe on the upper one that feedback and input shows the same signal, creating "no" error voltage. The input signal already contains the distortions, and the feedbacksignal shows the same distortions, yielding in openloop distortions + the distortions at input. The feedback is not functional here.
The lower one still has this effect, but reduced by -60db, and the feedback does its job giving thd of ~-130db.
It's likely that real world devices behave even worse.
The concept of feedback is still valid, but can be easily knocked out, especially for higher freqs.
I tried a lot to find flaws in the concept of feedback, except the mechanism of distorted distortions through the feedbackloop giving high order harmonics all i found was that feedback works if used correctly.
The next days i will try to build a really low distortive amp, i have the feeling that such amp sounds absolutely great...
Mike
Attachments
Hi, Mike,
What is this design trying to achieve? http://www.diyaudio.com/forums/attachment.php?s=&postid=751151&stamp=1130173511
The differential's emitors is bootstrapped towards output node stage via R17/R3, what is the effect? (note that this R17/R3 has the same ratio as feedback divider R7/R6)
What is this design trying to achieve? http://www.diyaudio.com/forums/attachment.php?s=&postid=751151&stamp=1130173511
The differential's emitors is bootstrapped towards output node stage via R17/R3, what is the effect? (note that this R17/R3 has the same ratio as feedback divider R7/R6)
Hi, Mike,
Since you are exploring differential pair, I have something that I want to ask you.
I don't know if this have been done by someone. This configuration should have non-limiting current feeded to diff pair, transients should be better. This is done by a "Rush Cascode". A nice side-effect is that the CCS has no emitor tied to - rail (only collector), I think this would improve the PSRR feeded to diff pair.
Point D is only -2V2 from gnd (drop of 1 diode + 1 LED). The drop at R1 is about 1V. I make 2mA by putting 510 ohm for R1.
The interesting thing is this : Assume there is quite big signal entering point A, say +1V. If point A rise +1V, then more or less, point B will also rise +1V (neglecting VBE difference). If point B is rising 1V, then the drop at R1 will rise to 2V (that is twice the equilibrum), making current passing R1 (and T1-T2) multiply 2 times.
So, the current feeded to point A of differential pair is not steady at all.
If the current passing through C-E of T1 and T2 rises, then the work of B-E (current magnifying by base modulation) should happen less, T1 and T2 works more as "common base". If T1 and T2 works as common base, then the distortions of this diff pair can be much less (is it?
), because signal feeded to VAS is actually made by modulated emitors of T1-T2 more than modulated base of T1-T2. This can happen because the drop in R1 is very small (only 1V). If you make the drop at R1 approaching Rail supply, then R1 will act like ordinary Resistor CCS, and this effect will not take place (significantly).
What do you think of this? Is this better or worse than ordinary CCS'ed diff pair?
Since you are exploring differential pair, I have something that I want to ask you.
I don't know if this have been done by someone. This configuration should have non-limiting current feeded to diff pair, transients should be better. This is done by a "Rush Cascode". A nice side-effect is that the CCS has no emitor tied to - rail (only collector), I think this would improve the PSRR feeded to diff pair.
Point D is only -2V2 from gnd (drop of 1 diode + 1 LED). The drop at R1 is about 1V. I make 2mA by putting 510 ohm for R1.
The interesting thing is this : Assume there is quite big signal entering point A, say +1V. If point A rise +1V, then more or less, point B will also rise +1V (neglecting VBE difference). If point B is rising 1V, then the drop at R1 will rise to 2V (that is twice the equilibrum), making current passing R1 (and T1-T2) multiply 2 times.
So, the current feeded to point A of differential pair is not steady at all.
If the current passing through C-E of T1 and T2 rises, then the work of B-E (current magnifying by base modulation) should happen less, T1 and T2 works more as "common base". If T1 and T2 works as common base, then the distortions of this diff pair can be much less (is it?
), because signal feeded to VAS is actually made by modulated emitors of T1-T2 more than modulated base of T1-T2. This can happen because the drop in R1 is very small (only 1V). If you make the drop at R1 approaching Rail supply, then R1 will act like ordinary Resistor CCS, and this effect will not take place (significantly).What do you think of this? Is this better or worse than ordinary CCS'ed diff pair?
Attachments
To combine with your idea...
I think of this "common base" approach is because when we enter signal to base of a transistor, and take the output via emitor or collector, it will be distorted due to non-linear internal cap of B-E/B-C. But if we use the same transistor as "common base", signal entering via emitor and output via collector, I think it will be better than signal entering to base. I'm not sure, though....
The modulated voltage drop of R2 (which is needed to drive the VAS) comes from T1 modulated in common base, I hope....
What is your opinion about this?
I think of this "common base" approach is because when we enter signal to base of a transistor, and take the output via emitor or collector, it will be distorted due to non-linear internal cap of B-E/B-C. But if we use the same transistor as "common base", signal entering via emitor and output via collector, I think it will be better than signal entering to base. I'm not sure, though....
The modulated voltage drop of R2 (which is needed to drive the VAS) comes from T1 modulated in common base, I hope....
What is your opinion about this?
Attachments
Hi lumanauw, principally a nice idea, but, you throw away completely any CMRR. The circuit would seize to act as diffamp (only amplifying the difference between its inputs). Instead you get the difference plus the input, rendering the feedback unfunctional.
Remember, the diffamp ideally only puts out the difference, independent of the voltage at point B.
Also, point B sees large amounts of 2nd harmonics, you would amplify these.
If you want to use feedback, keep the ccs.
Mike
Remember, the diffamp ideally only puts out the difference, independent of the voltage at point B.
Also, point B sees large amounts of 2nd harmonics, you would amplify these.
If you want to use feedback, keep the ccs.
Mike
ilimzn
ilimzn,
I need you help regarding amp schematic you sent me 6-8 months ago(with irfp240/9140).I placed mje340 and 350 on spot of 2n5551 and they are getting very hot.Why?any suggestion?
Send me email on: test8894@yahoo.com
Regards,
Zoran
ilimzn,
I need you help regarding amp schematic you sent me 6-8 months ago(with irfp240/9140).I placed mje340 and 350 on spot of 2n5551 and they are getting very hot.Why?any suggestion?
Send me email on: test8894@yahoo.com
Regards,
Zoran
@Zox - For some reason I never got your email or read this question. The MJE parts should be hot, but not so hot that you can't hold them with your fingers.
@MikeB, looking at the schematics on the first page, C3 is rather superfluous as the collector of the conncted transistor sees nearly no voltage swing.
@MikeB, looking at the schematics on the first page, C3 is rather superfluous as the collector of the conncted transistor sees nearly no voltage swing.
Eva said:........the worst modulation by far is the one coming from power supply ripple (the one from real power supplies, not from PSpice voltage sources
Not really...Just use a simple RC filter...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.