i fear the moderators don´t like these kind of words....
do we wanna go back to feedback (or nonfeedback) now?
the circuits from Charles link looks not really low parts count and simple to me - compared to Nelsons preamp schematics.
do we wanna go back to feedback (or nonfeedback) now?
the circuits from Charles link looks not really low parts count and simple to me - compared to Nelsons preamp schematics.
From what I understood, this is the circuit of the Ayre v-3, and this should be a power amp, not a preamp (but you can drive it from a CDP).till said:
However, it is not clear to me how Q7 / Q8 are working. The gate is somehow attached between the output and rail - cascode? Sorry for the silly question, but I'm a very beginner....
Cheers, Tino
Charles Hansen said:
Ok.
Ok.
Ok.
Aren't you forgetting something? Better apply a load resistor too if you expect to get any output from this thing.
Ok.
Aren't you forgetting something? Like the load resistor again? The cathode isn't connected to ground anymore, remember? So why are you calcultaing the time constant based solely on source impedance?
Looking at the equivalent circuit for a triode, configured as a simple cathode follower with a load resistor, it appears to me that the gate to cathode capacitance's path to ground is through the parallel combination of the load resistor, plate resistance and the plate to cathode capacitance.
Yeah-yeah, sure-sure.
Ah, ok. Now the load resistor appears.
Wait a minute, you just said that the cathode voltage changed because of the voltage drop across the resistor due to the current flowing through it. Now you're saying that the cathode voltage changed because of the number of electrons there.
Which is it? Did it change because of the voltage drop across the resistor or did it change because there are more electrons on the grid?
Yet below you say it does change essentially instantaneously. Which is it? Does it change essentially instantaneously or not?
Not to mention the resistances that those capacitances are charging through.
How can it change any more instantaneously than the voltage across the cathode resistor which you just said above doesn't change instantaneously?
The current through the tube is a function of the voltage voltage between the grid and the cathode. Since the grid voltage is established by the input signal and the cathode voltage is established by the cathode resistor, and since you say the voltage across the cathode resistor doesn't change instantaneously, I fail to see how the current through the tube can change instantaneously in response to the voltage on the grid.
If that were the case, then we would also have to expect the tube itself to have a bandwidth on the order of 300 GHz.
I think you need to go back and think this through again.
se
An externally hosted image should be here but it was not working when we last tested it.
zinsula said:
From what I understood, this is the circuit of the Ayre v-3, and this should be a power amp, not a preamp (but you can drive it from a CDP).
However, it is not clear to me how Q7 / Q8 are working. The gate is somehow attached between the output and rail - cascode? Sorry for the silly question, but I'm a very beginner....
Cheers, Tino
I'm no expert myself, but the gates are at AC ground and the sources are driven by Q1 and Q2. Q7/8 drains drive the output stage. There's a symmetrical antiphase version at Q9 and Q10, but driven by the other side of the input diff amps. It sure looks like a relatively straightforward grounded gate stage.
Charles, you're heartily invited to tell me I'm full of it and explain it properly.
Rather than chew through it line by line, I was going to just go home this evening and stick some scope probes on my preamp to look at the actual signal transit time through the CF. But that would be cheating, wouldn't it?
Hi,
The only thing you'd probably learn from that is that the discharge time of a coupling cap is higher than the transfer time of electrons in a vacuum.
You would if it weren't for the inevitable interelectrode capacitances which would require trickery_it exists_ to neutralise.
I agree, you are both talking about different things here.
Cheers,😉
The only thing you'd probably learn from that is that the discharge time of a coupling cap is higher than the transfer time of electrons in a vacuum.
You would if it weren't for the inevitable interelectrode capacitances which would require trickery_it exists_ to neutralise.
I agree, you are both talking about different things here.
Cheers,😉
SY, I'm not sure what the experiment is. You are apparently going to look at the transit time through a cathode follower, but you are basing this on Steve Eddy's claim of 300 GHz bandwidth. I never said this -- he was completely mis-reading my post.
From what I understand, the transit time and bandwidth of a box of gain are unrelated. In a reductio ad absurdem, you could have an amp with an infinite bandwidth, but which could have a delay between input and output of years. Put in a narrow pulse, wait a year, and an identical pulse comes out without spread.
The philosophical (as opposed to pragmatic) arguments I see against the use of loop feedback are based on its retrospective nature. I honestly don't know offhand how retrospective a CF's feedback is, so looking at front-to-back time will give me a measure of that. Since, as you point out, one of the input ports is connected directly to the output, much in the manner of an opamp connected as a unity-gain buffer (hard connection between output and inverting input), the transit time will give me an empirical notion as to the relative signal delay between the ports.
So... do the FETs I was talking about before work the way I was speculating? I'm trying to learn something about transistor amps and it's always nice to get it right from the horse's mouth, as it were.
The philosophical (as opposed to pragmatic) arguments I see against the use of loop feedback are based on its retrospective nature. I honestly don't know offhand how retrospective a CF's feedback is, so looking at front-to-back time will give me a measure of that. Since, as you point out, one of the input ports is connected directly to the output, much in the manner of an opamp connected as a unity-gain buffer (hard connection between output and inverting input), the transit time will give me an empirical notion as to the relative signal delay between the ports.
So... do the FETs I was talking about before work the way I was speculating? I'm trying to learn something about transistor amps and it's always nice to get it right from the horse's mouth, as it were.
SY said:
Yeah, I was thinking in terms of feedback which in hindsight isn't a bandwidth issue.
If there were no delay between the input signal and the feedback signal, then the feedback signal would never be in phase with the signal and the feedback would never become positive.
In any case, I don't see how the feeback could possibly be delayed by only a few picoseconds as Charles claims unless there were no parasitic capacitance anywhere in the circuit.
se
Charles's explanation looks pretty plausible to me, though I'm not fully following the analogies. Some experiment will give me a bit more insight, and that's tonight's project.
V3proto
Charles,
I have quite the same thinking, but rather different.
Can you call this X-amp? No ground involved in speaker outputs.
My idea is like this, please comment.
The front end is still folded cascode, driving half power amp, resulting in X amp like your V3proto. But I wanted to use only single differential, so dont use Q2-Q4, and replace Q8 and Q10 with CCS. Which is better, replacing Q8-Q10 with CCS or plain resistor (or bootstrap maybe?) if I wanted single differential like I told you above?
Charles,
I have quite the same thinking, but rather different.
Can you call this X-amp? No ground involved in speaker outputs.
My idea is like this, please comment.
The front end is still folded cascode, driving half power amp, resulting in X amp like your V3proto. But I wanted to use only single differential, so dont use Q2-Q4, and replace Q8 and Q10 with CCS. Which is better, replacing Q8-Q10 with CCS or plain resistor (or bootstrap maybe?) if I wanted single differential like I told you above?
Maybe there is some "mechanism" of delay involved in audio amplifier+feedback path. I remember Nelson Pass said something about "Hall of mirror" effect, if the gain stages in audio power amp is too long. It makes the feedback less effective. Is this caused by time or capacitance delay, maybe inherent in every transistor?
fdegrove said:
Well, if your circuit were infinitely small... 🙂
Not saying you can remove time from the equation. Just that I don't believe the time involved in the cases we've been talking about here is on the order of a few picoseconds as Charles has claimed.
se
Hi,
Eh, eh, eh....That wouldn't work either...no matter how small it is there will allways have to be a distance between input and output.
To bridge a distance takes? Time.
To charge and discharge a capacitance takes times, to charge a space takes time....even in a vacuum.
To make one triode different from another takes a distance between the elements...and so on.
You still have that stopwatch?
Seriously, I think Charles was talking from an ideal model POV. Not necessarily a real situation.
Picoseconds...that's a pretty short distance in time, isn't it?
Cheers,😉
P.S. Slowly reaching a state of homeostasis....ZZZZzzzzzzzzz.
Eh, eh, eh....That wouldn't work either...no matter how small it is there will allways have to be a distance between input and output.
To bridge a distance takes? Time.
To charge and discharge a capacitance takes times, to charge a space takes time....even in a vacuum.
To make one triode different from another takes a distance between the elements...and so on.
You still have that stopwatch?
Seriously, I think Charles was talking from an ideal model POV. Not necessarily a real situation.
Picoseconds...that's a pretty short distance in time, isn't it?
Cheers,😉
P.S. Slowly reaching a state of homeostasis....ZZZZzzzzzzzzz.
- Status
- Not open for further replies.