Misconceptions about feedback annoy me. But I do understand where they come from, because the topic isn't completely intuitive, and math is hard.
There is no recursion, no obstruction. The fake audio media keeps crying about ECHOES, ECHOES, ECHOES. Make Amplifiers Great Again.
The Tektronix 7104 did 1 GHz with 7A29 amplifier plug-ins. So did the really old 519 scope, but using direct access to traveling wave deflectors. There were Tek scan conversion scopes that do at least 17 GHz back 20+ years ago. ( Scan Converter is a silicon target analog storage scope with a scanning electron camera that reads out the charge pattern off that. ) And as far back as the 60s there was a commercially available scope that did multi GHz using a 6+ foot long CRT with traveling wave deflectors, but no amplifiers. I'm sure if you look at LLNL or recent Tek stuff they have even faster ones, maybe not commercially available.
I have an old Tek 7912 scan converter that does 750 Mhz using the 7A29 plug-in and 1 GHz with direct access plug-in. Got it for $100 at a surplus store back in the 70s. (I think no one knew what it was.) It's a quite nice storage scope actually, using a Tek 602 readout monitor or bigger Video monitor. The 7912A version was like 1 GHz with the amplifier plug-in.
1) Tek 519
2) Tek R7912
3) Tek 7104
Make amplifiers great again?
Use no global negative feedback.
Input buffer, step up trafo, output buffer, classA.
No more grifters please.
Much thanks for this. Now I'm finally getting what you're saying. With a small enough granularity we could see below the equations, which assume zero transit (propagation) time, to the primordial. In the audio world this would be esoteric, but at higher speeds and/or larger propagation times might become relevant. So, even the equations including recursion are approximations. The real world is messy, and gets messier the closer you look. Very cool.
All good fortune,
Chris
How quick some people forget:
I remember 2 of the fastest analog scopes, they both had -3 dB bandwidths of 1 GHz.
The rise time was aproximately 350nsec.
One of those scopes measured the signal of a nuclear blast, and made a record of it on very high sensitivity Film.
You needed that kind of bandwidth, and the special phosphor on the CRT.
Of course, the scope and Film were behind thick walls of concrete, lead, or both. You do not want to fog the film, do you?
500 MHz scopes were kind of slow. You just need to purchase another brand to get 1GHz analog bandwidtn.
I remember 2 of the fastest analog scopes, they both had -3 dB bandwidths of 1 GHz.
The rise time was aproximately 350nsec.
One of those scopes measured the signal of a nuclear blast, and made a record of it on very high sensitivity Film.
You needed that kind of bandwidth, and the special phosphor on the CRT.
Of course, the scope and Film were behind thick walls of concrete, lead, or both. You do not want to fog the film, do you?
500 MHz scopes were kind of slow. You just need to purchase another brand to get 1GHz analog bandwidtn.
Last edited:
Whether of it's of any practical consequence is a completely separate matter but the big chunk of stored energy in the magnetized iron called an OPT could reasonably be suspected of have some element of propagation delay.
Forget to close the blast doors?
Ouch!
We are not talking about your insecure home that has the garage door wide open.
Ouch!
We are not talking about your insecure home that has the garage door wide open.
I got the impression that it was just the bad old days and stuff happens. But that probably happened. Also, sometimes yields were almost an order of magnitude better than expected. Double ouch.
All good fortune,
Chris
All good fortune,
Chris
rdf,
Yes!
And not only the laminations . . .
Just forget about the laminations.
Start with a typical output transformer (either a single ended or a push pull output transformer).
Take those laminations out, but disturb nothing else.
Send a 20kHz signal to the amplifier input.
The output tubes will not mind, at 20kHz, there is plenty of primary inductance (plenty of primary inductive reactance).
Then . . .
Consider the intrinsic distributed capacitance that is across the primary or primaries. . . . Lag.
Consider the intrinsic leakage inductance (and consequential inductive reactance) from the primary or primaries, to the secondary or secondaries. . . . Lag.
Yes there is delay.
I hope so, unless we are talking about a new physics.
Yes!
And not only the laminations . . .
Just forget about the laminations.
Start with a typical output transformer (either a single ended or a push pull output transformer).
Take those laminations out, but disturb nothing else.
Send a 20kHz signal to the amplifier input.
The output tubes will not mind, at 20kHz, there is plenty of primary inductance (plenty of primary inductive reactance).
Then . . .
Consider the intrinsic distributed capacitance that is across the primary or primaries. . . . Lag.
Consider the intrinsic leakage inductance (and consequential inductive reactance) from the primary or primaries, to the secondary or secondaries. . . . Lag.
Yes there is delay.
I hope so, unless we are talking about a new physics.
Note that these are issues of phase and not of delay, as rdf was proposing.
All good fortune,
Chris
All good fortune,
Chris
Phase?
Or
Delay?
They are not the same thing, but they are related.
1. Send a 1kHz sine wave into a 10uSec delay line. The input to output relative phase is 10usec/1000usec x 360 degrees = 3.6 degrees phase lag.
2. Send a 10kHz sine wave into a second identical 10uSec delay line. The input to output relative phase is 10usec/100usec x 360 degrees = 36 degrees phase lag.
What ever you want to call them, phase and delay are related proportionally to the frequency (inversely or directly, depending on which way you are looking at it).
Start both the 1kHz sine wave, and the 10kHz sine wave, turn them on at exactly the same time, and with both at 0 degrees, and 0 Volts.
With both sine wave signal amplitudes of 1 Volt peak, after 10usec both delay lines output 0V (remember, 10usec After the start, we finally see the begining.
But 10usec beyond that (20usec after the start), the delay lines put out:
1. 1 kHz signal: Sine of 3.6 degrees x 1 Volt = 0.0627 Volts
2. 10kHz signal: Sine of 36 degrees x 1 Volt = 0.5878 Volts
About 63 mV versus about 588 mV, that is a Big difference.
That has been an important factor of all that I have said in this thread.
I hope that clears up my earlier statements in this thread.
Or
Delay?
They are not the same thing, but they are related.
1. Send a 1kHz sine wave into a 10uSec delay line. The input to output relative phase is 10usec/1000usec x 360 degrees = 3.6 degrees phase lag.
2. Send a 10kHz sine wave into a second identical 10uSec delay line. The input to output relative phase is 10usec/100usec x 360 degrees = 36 degrees phase lag.
What ever you want to call them, phase and delay are related proportionally to the frequency (inversely or directly, depending on which way you are looking at it).
Start both the 1kHz sine wave, and the 10kHz sine wave, turn them on at exactly the same time, and with both at 0 degrees, and 0 Volts.
With both sine wave signal amplitudes of 1 Volt peak, after 10usec both delay lines output 0V (remember, 10usec After the start, we finally see the begining.
But 10usec beyond that (20usec after the start), the delay lines put out:
1. 1 kHz signal: Sine of 3.6 degrees x 1 Volt = 0.0627 Volts
2. 10kHz signal: Sine of 36 degrees x 1 Volt = 0.5878 Volts
About 63 mV versus about 588 mV, that is a Big difference.
That has been an important factor of all that I have said in this thread.
I hope that clears up my earlier statements in this thread.
Last edited:
Looks like my reference to a 17 GHz Scan Converter scope is way off, it was 6.5 GHz back in the 70s. Actually it was a French design that Tektronix licensed. The TEK model 7250.
I did find some references to Lockheed modifying the Tek R7912 to do 3 GHz and later 3.5GHz in the late 70s for use by LLNL for fusion diagnostics. Supposedly "easy" mods, but this comes from microwave technologists. Some comments about fixing the back-plane transmission line termination and the plug-in amplifier. The R7912 CRT itself was rated at 2.4 GHz by Tektronix. The standard 7A19 plug-in amplifier limited its bandwidth to 600 MHz, hence the better performance of 750 MHz when using the later 7A29 plug-in amplifier developed for the Tek 7104 scope. There was a later R7912HB (high bandwidth) model which probably incorporated Lockheed's mods.
Some good info on Tek scopes here:
https://w140.com/tekwiki/wiki/7912
The rise time of the 7A29 plug-in is 320 picoSec, fast enough to see a few inches of wire propagation delay. The 7B10 and 7B15 time base plug-ins go down to 2 ns/div with a 10x magnifier button available (so 200 ps / div then)
I did find some references to Lockheed modifying the Tek R7912 to do 3 GHz and later 3.5GHz in the late 70s for use by LLNL for fusion diagnostics. Supposedly "easy" mods, but this comes from microwave technologists. Some comments about fixing the back-plane transmission line termination and the plug-in amplifier. The R7912 CRT itself was rated at 2.4 GHz by Tektronix. The standard 7A19 plug-in amplifier limited its bandwidth to 600 MHz, hence the better performance of 750 MHz when using the later 7A29 plug-in amplifier developed for the Tek 7104 scope. There was a later R7912HB (high bandwidth) model which probably incorporated Lockheed's mods.
Some good info on Tek scopes here:
https://w140.com/tekwiki/wiki/7912
The rise time of the 7A29 plug-in is 320 picoSec, fast enough to see a few inches of wire propagation delay. The 7B10 and 7B15 time base plug-ins go down to 2 ns/div with a 10x magnifier button available (so 200 ps / div then)
Last edited:
Do you mean the IgNobel prize? Or the Finnish... er... Danish... er... whatever Nobel committee?
No, Nobel. What he says flies directly in the face of everything we know about physics and electronics, overturning that is definitely Nobel material.
But since he doesn't back it up at all, I'll file it under pie in the sky ;-)
Jan
But since he doesn't back it up at all, I'll file it under pie in the sky ;-)
Jan
Wrong again. These signals are delayed, not phase shifted.
The 'relative phase' statement is meaningless. The signals are not phase shifted, they are delayed, by your own words.
Jan
Exactly, I think we're starting to get together on the two different concepts.
Delay would be an arrival time function, as in 60Hz / 180 degree = 1/120 sec. difference in arrival time through the amp loop. As you can see with this example, using any kind of logic knowing the speed of electricity, that this will not be seen on a capture scope. But you would see a simultaneous arrival and display of an inverted signal in sync with the original input signal.
Phase shifted signals have their affect due to the instantaneous voltage value difference when you finally combine them at a cathode or other node. So if you apply a 1 degree NFB you only get (without going into RMS calcs.) 1/180 of the affect of a true 180 degree NFB, or a 90 degree would be a 50% affectiveness... I'm probably over simplifying that to make the point. You gotta look at all the points of phase flip flops and count them and know what to expect. Ain't that part of the fun? I think some folks who just don't want to take the time to study how it works are afraid of misapplying it and so they just fall into the group that condemns it and try to say it's not what it's advertised to be. Trying to rewrite and disinform to rationalize their own aversion, ... et al, voice common opinions of non-FB amp SQ as somehow superior.... yadda...
It seems almost universally expressed by the flea-W, triode proponents who have no power to throw away in trade for better distortion figures or just want to say that non-FB triode SQ is PQ to them. OK, that's all.
Last edited:
Negative feedback got a bad reputation because it was used as a band-aid for poor linearity. Feedback can't make up for a design that started out bad.
Ed
Ed
So the medicine is bad because the disease is at fault? I think the logic of that shows the positve value of FB. Better to put the real blame where it belongs, right? FB can make up for a weak design.
That's kind of like saying car paint is bad because the company didn't make the car out of stainless... just to meet a price point.
Last edited: