Very happy to see too many good people together sharing know how
And in a peacefull way
I like that!
very nice to see that
Carlos
And in a peacefull way
I like that!
very nice to see that
Carlos
Thanks JCX,
Appreciate the correction, 17dB ain't so bad. I won't make that error again!
I have built SS Class AB amps using VASs with and without heatsinks, and sonically, I cannot hear a difference. But clamping a TO220 heatsink to the VAS sure brings down its temperature, which must improve reliability considerably.
So, my own conclusion is that brake fade in audio amps is certainly a possibility, but in practice is difficult to detect, at least in the VAS.
Now, the input stage diff pair has virtually unchanging Vce across both devices. It might vary 100mV up and down from Vcc, but this is trivial when the dissipations are calculated. The current flow, however, will deviate rather more, but surely no more than +/-10% max. So here I see no issue with thermal distortion at all; the principal distortion must come from the Vbe variations with greater and lesser current flow due to differential action.
This could be eliminated by running each collector arm of a diff pair into a current sink, and then attaching the base of an emitter follower from the collector of each arm with its emitter to rail. Each emitter follower collector takes its current from the emitter of its respective 'parent' transistor. This requires emitter degeneration on the diff pair, but then means that the diff pair at all times has constant Vbe, which removes the Vbe distortion completely. Padamieki, you might like this approach; I've not tried it yet, perhaps one day. It may sound better......
Cheers,
Hugh
Appreciate the correction, 17dB ain't so bad. I won't make that error again!
I have built SS Class AB amps using VASs with and without heatsinks, and sonically, I cannot hear a difference. But clamping a TO220 heatsink to the VAS sure brings down its temperature, which must improve reliability considerably.
So, my own conclusion is that brake fade in audio amps is certainly a possibility, but in practice is difficult to detect, at least in the VAS.
Now, the input stage diff pair has virtually unchanging Vce across both devices. It might vary 100mV up and down from Vcc, but this is trivial when the dissipations are calculated. The current flow, however, will deviate rather more, but surely no more than +/-10% max. So here I see no issue with thermal distortion at all; the principal distortion must come from the Vbe variations with greater and lesser current flow due to differential action.
This could be eliminated by running each collector arm of a diff pair into a current sink, and then attaching the base of an emitter follower from the collector of each arm with its emitter to rail. Each emitter follower collector takes its current from the emitter of its respective 'parent' transistor. This requires emitter degeneration on the diff pair, but then means that the diff pair at all times has constant Vbe, which removes the Vbe distortion completely. Padamieki, you might like this approach; I've not tried it yet, perhaps one day. It may sound better......
Cheers,
Hugh
Just so i get this straight the diff pair drives the Vas as a current source rather than a voltage source.
The question i ask myself then is what is the output impedance of the diff pair since the vas input impedance varies unlinearily and with temperature.
Does anyone have any figures for it, yes i am lazy😀 .
The question i ask myself then is what is the output impedance of the diff pair since the vas input impedance varies unlinearily and with temperature.
Does anyone have any figures for it, yes i am lazy😀 .
Hi Sy,
When I take time to make a respectful and relevent posting it is not nice to subsequently find at a later date that it has been removed from the contextual flow without any stated reason as to why, or a personal message that this has happened.
__________________________________________________
Hi Hjelm,
Yes the differential stage drives a bipolar VAS stage as a differential voltage to current converter. Also, any VAS input current flow requirement can phase linearly develop series potentials across the differential and mirror transistor base-emitter junctions plus the emitter resistors.
The VAS input impedance varies with frequency due especially to Miller Effect within the VAS transistor, thus the 'base' current requirement will be non-linear with varying amplitude and frequency. This is why some incorrectly stabilised designs can tend towards oscillation when the the VAS voltage is high.
Getting back to where Mikeks was attempting to tell me that I was wrong citing known phase distortion. It is the 'Blameless' design that is wrong because it uses a Miller Effect *inducing* stabilising capacitor, and this has nothing to do phase distortion.
Hugh was quite correct in quoting a 10% figure for differential current variation with the Blameless design. 10% flows through the Miller connected VAS C.dom, and only 0.01% of that into the Darlington, 0V collector connected VAS driver base !!!!!!
Mikeks, the 10% current that flows through the 100pF capacitor is ninety degrees leading wrt. the input signal. The Vbe and series resistor voltage drops are no longer in phase with the waveform being amplified. To lead current charge the *stabilising* capacitor means that a leading differential input voltage must suddenly develop, this means that a global NFB error must also develop at the output terminal.
When loudspeaker back EMF leads the amplified waveform a lowly biased bipolar class-B output stage *will* near instantaneously reverse commutate through a portion of its static bias to develop the differential input stage voltage error that is necessary to charge the C.dom. This will not show up with steady sinewave passive resistor testing !!!!!
I suggested a method whereby you could check this for yourself.
I recently read one of Hugh's postings recommending the use of non-Miller connected VAS C.dom stabilisation, but it did not seem to hit home. Nelson Pass was recommending this over twenty years ago.
Destroyer knows what he heard and he had the guts to say so before effecting a permanent cure.
____________________________________________________
Hi Sam9,
You obviously had a chance to read my last posting, but I was peeved that it disappeared. Allow me to state my understanding based upon some common amplifier simulations, which I have not saved.
Miller C.dom plus Mosfet. Mosfets have higher bias voltages than bipolar and a lower natural output impedance when VAS driven, thus the output stage reverse commutation is much less and more of a simple phase change for a given dynamic loudspeaker back EMF potential. The error is not sharply discontinuous and only arises after much higher output levels.
Stability can be more of a problem with Mosfets, and there is a risk for the amplifier's output terminal effectively becoming inductive at higher audio frequencies without this showing up in THD figures. This could lead to tweeter waveform distortion.
Miller C.dom plus class-A. Class-A output stages have low output impedance and discontiuities do not arise until after natural current flows have been exceeded, thus the Miller C.dom induced affectation is limited to a phase change effect.
Cheers ............ Graham.
When I take time to make a respectful and relevent posting it is not nice to subsequently find at a later date that it has been removed from the contextual flow without any stated reason as to why, or a personal message that this has happened.
__________________________________________________
Hi Hjelm,
Yes the differential stage drives a bipolar VAS stage as a differential voltage to current converter. Also, any VAS input current flow requirement can phase linearly develop series potentials across the differential and mirror transistor base-emitter junctions plus the emitter resistors.
The VAS input impedance varies with frequency due especially to Miller Effect within the VAS transistor, thus the 'base' current requirement will be non-linear with varying amplitude and frequency. This is why some incorrectly stabilised designs can tend towards oscillation when the the VAS voltage is high.
Getting back to where Mikeks was attempting to tell me that I was wrong citing known phase distortion. It is the 'Blameless' design that is wrong because it uses a Miller Effect *inducing* stabilising capacitor, and this has nothing to do phase distortion.
Hugh was quite correct in quoting a 10% figure for differential current variation with the Blameless design. 10% flows through the Miller connected VAS C.dom, and only 0.01% of that into the Darlington, 0V collector connected VAS driver base !!!!!!
Mikeks, the 10% current that flows through the 100pF capacitor is ninety degrees leading wrt. the input signal. The Vbe and series resistor voltage drops are no longer in phase with the waveform being amplified. To lead current charge the *stabilising* capacitor means that a leading differential input voltage must suddenly develop, this means that a global NFB error must also develop at the output terminal.
When loudspeaker back EMF leads the amplified waveform a lowly biased bipolar class-B output stage *will* near instantaneously reverse commutate through a portion of its static bias to develop the differential input stage voltage error that is necessary to charge the C.dom. This will not show up with steady sinewave passive resistor testing !!!!!
I suggested a method whereby you could check this for yourself.
I recently read one of Hugh's postings recommending the use of non-Miller connected VAS C.dom stabilisation, but it did not seem to hit home. Nelson Pass was recommending this over twenty years ago.
Destroyer knows what he heard and he had the guts to say so before effecting a permanent cure.
____________________________________________________
Hi Sam9,
You obviously had a chance to read my last posting, but I was peeved that it disappeared. Allow me to state my understanding based upon some common amplifier simulations, which I have not saved.
Miller C.dom plus Mosfet. Mosfets have higher bias voltages than bipolar and a lower natural output impedance when VAS driven, thus the output stage reverse commutation is much less and more of a simple phase change for a given dynamic loudspeaker back EMF potential. The error is not sharply discontinuous and only arises after much higher output levels.
Stability can be more of a problem with Mosfets, and there is a risk for the amplifier's output terminal effectively becoming inductive at higher audio frequencies without this showing up in THD figures. This could lead to tweeter waveform distortion.
Miller C.dom plus class-A. Class-A output stages have low output impedance and discontiuities do not arise until after natural current flows have been exceeded, thus the Miller C.dom induced affectation is limited to a phase change effect.
Cheers ............ Graham.
Another angle on this
You can model a node in a circuit as either a voltage source with a series impedance or a current source with parallel impedance.
With transistors, it is usually easiest to model the collector node as a current source with parallel Z. In the case of the diff pair the parallel Z is composed mainly of a non-linear resistance due to the variation of Ic with Vbe and a non-linear capacitance due to junction capacitance. This current source feeds the base of the "VAS" (which isn't a voltage amp at all it is really just a current amp - misleading name) which often has a resistor across it - this resistance is approximately in parallel with the diff transistor Z. The resistor is typically around about 1k.
The diff transistor Z resistance is typically >50k-ohms and the capacitance < 10pF. The capacitance can be increased by the beta of the transistor via Miller effect however, this is complex to calculate. Since the "VAS" BE resistor is usually much smaller by comparison, say 1k, you can get a good model by assuming the diff transistor is an ideal current source in parallel with the BE resistor.
You can model a node in a circuit as either a voltage source with a series impedance or a current source with parallel impedance.
With transistors, it is usually easiest to model the collector node as a current source with parallel Z. In the case of the diff pair the parallel Z is composed mainly of a non-linear resistance due to the variation of Ic with Vbe and a non-linear capacitance due to junction capacitance. This current source feeds the base of the "VAS" (which isn't a voltage amp at all it is really just a current amp - misleading name) which often has a resistor across it - this resistance is approximately in parallel with the diff transistor Z. The resistor is typically around about 1k.
The diff transistor Z resistance is typically >50k-ohms and the capacitance < 10pF. The capacitance can be increased by the beta of the transistor via Miller effect however, this is complex to calculate. Since the "VAS" BE resistor is usually much smaller by comparison, say 1k, you can get a good model by assuming the diff transistor is an ideal current source in parallel with the BE resistor.
Hi traderbam,
Yes. But ! The Blameless VAS stage is Darlington and does not have a 1k B-E resistor. I would not doubt the VASs which have this component could sound better whilst measuring worse.
Cheers ........... Graham.
Yes. But ! The Blameless VAS stage is Darlington and does not have a 1k B-E resistor. I would not doubt the VASs which have this component could sound better whilst measuring worse.
Cheers ........... Graham.
Hi Graham,
Yes, the "VAS" is sometimes a darlington. I think I'm losing track of which "blameless" amp we are discussing in this thread. The one on Self's site (via link on page 1) is a simple CE bipolar with 2k2 BE resistor.
Yes, the "VAS" is sometimes a darlington. I think I'm losing track of which "blameless" amp we are discussing in this thread. The one on Self's site (via link on page 1) is a simple CE bipolar with 2k2 BE resistor.
I am very honored to see Trader, Graham, Aksa and many others here
Thank you very much, to spend your time with this subject.
Carlos
Thank you very much, to spend your time with this subject.
Carlos
Padamiecki (from page 10)
If I can make head or tail of the schematic, it looks a little like a transconductance amp - it essentially produces a current output, and the voltage you see at the output is entirely subject to the value of the load resistance.
(Sorry - first post here - hope this helps...)
Alan
If I can make head or tail of the schematic, it looks a little like a transconductance amp - it essentially produces a current output, and the voltage you see at the output is entirely subject to the value of the load resistance.
(Sorry - first post here - hope this helps...)
Alan
Traderbam:
In your example four posts up:
If you have a current source don't you have to multiply the BE VAS resistor with the beta of the VAS transistor?
And when you say there is no influence of thermal kind in the VAS doesen't a varying Vbe(t) of the simple VAS transistor result in a strangely varying input impedance of the vas and thereby load the input diff pair with a difficult load, timevarying strangely?
I may be totally wrong here and i am sorry to be slightly off topic, not discussing the blameless topology exactly.
In your example four posts up:
If you have a current source don't you have to multiply the BE VAS resistor with the beta of the VAS transistor?
And when you say there is no influence of thermal kind in the VAS doesen't a varying Vbe(t) of the simple VAS transistor result in a strangely varying input impedance of the vas and thereby load the input diff pair with a difficult load, timevarying strangely?
I may be totally wrong here and i am sorry to be slightly off topic, not discussing the blameless topology exactly.
No. What are you thinking of?
Did I say that? I think I suggested the output transistors are bigger "memory distortion" culprits than the VAS.
Yes. The BE impedance thus appears in parallel with the 1k resistor, using the example. The (small sgnal) BE impedance will be approximately 250-ohms for a VAS with Ic = 10mA and beta=100 at 25C. The beta will vary a bit with temperature and Vce.
I must humbly beg to apologise to you 
.
You said it was parallel to the be not a emitter degenerator.
Not too familiar with VAS stage so i am sorry i misunderstood.
.You said it was parallel to the be not a emitter degenerator.
Not too familiar with VAS stage so i am sorry i misunderstood.
This thread was splitted!, have a complementary, dedicated subjective perception
THE YARDSTICK FOR PERCEPTION is the name of resultant split.
There, friends are discussing something related to human perception in relation to audibility.
Of course, all of you welcome, not only to share your technical ideas here, but to visit the thread's brother in "Everything else"
There, another reality....new ideas, some psychological, perceptual, philosophical and all human social sciences giving their ideas.... very interesting too.
I thank you very much, to give me so good atention, and to explore this matter.
regards,
Carlos
THE YARDSTICK FOR PERCEPTION is the name of resultant split.
There, friends are discussing something related to human perception in relation to audibility.
Of course, all of you welcome, not only to share your technical ideas here, but to visit the thread's brother in "Everything else"
There, another reality....new ideas, some psychological, perceptual, philosophical and all human social sciences giving their ideas.... very interesting too.
I thank you very much, to give me so good atention, and to explore this matter.
regards,
Carlos
If I can make head or tail of the schematic, it looks a little like a transconductance amp - it essentially produces a current output, and the voltage you see at the output is entirely subject to the value of the load resistance.
(Sorry - first post here - hope this helps...)
Alan [/B][/QUOTE]
yes, exactly it is a U to I converter!
(Sorry - first post here - hope this helps...)
Alan [/B][/QUOTE]
yes, exactly it is a U to I converter!
padamiecki said:
Which circuit is this referring to? Transconductance amplifiers use as feedback a sample of the output current rather than voltage, which I didn't notice in the schematics on this thread.
I really put two .12 caliber cartridges in my shot gun!
And i had the spectacular pleasure, to desintegrated that thing!
Carlos
And i had the spectacular pleasure, to desintegrated that thing!
Carlos
Prune said:
EH?? A transconductance amplifier inherently gives a current output for a voltage input. That's not the same thing as a normal voltage output amplifier used with output current sensing to achieve a similar thing.
Such a connection does give a controlled output current, but it isn't a transconductance amplifier in the strict sense of the term.
Yes, i gave it two shots... the damn circuit
Was transformed in sub-extract of bandit horse gazes evaporated.
Carlos
Was transformed in sub-extract of bandit horse gazes evaporated.
Carlos
- Status
- Not open for further replies.