Re: V3proto
I built it fully complementary because that is the way I think it is best. I think fully complementary is better than CCS or plain resistor or bootstrap. If you want to change it to non-complementary, that is up to you. You should build it the way you think it is best.
lumanauw said:
I built it fully complementary because that is the way I think it is best. I think fully complementary is better than CCS or plain resistor or bootstrap. If you want to change it to non-complementary, that is up to you. You should build it the way you think it is best.
fdegrove said:
Yup.
You still have that stopwatch?
*tick-tick-tick-tick-tick...*
Mmmm. But an ideal model wouldn't include the parasitic capacitances that he mentioned now would it?
Picoseconds...that's a pretty short distance in time, isn't it?
Mmmm. What was the name of that woman who was the first female Admiral in the US Navy? She was big on computing technology and used to carry little lenghts of wires around with her to demonstrate what a nanosecond was, what a picosecond was, etc.
She'd appear on talk shows, whip out a little length of wire and say "That's a picosecond."
She was a pretty cool old broad.
se
SY said:
Very good, Gracehopper. When you can take the pebbles from my hand...
Thanks!
Geez, attachments of graphic images have to be downloaded now?
*sigh*
An externally hosted image should be here but it was not working when we last tested it.
We're all getting old....
I first met Grace Hopper in '75 at a computer conference. She was 'only' a commodore back then. We were both sitting in the back of a lecture room, and I asked her if she mind if I smoked, to which she replied 'not at all darling' and promptly whipped out her pack of Lucky Strikes....
An amazing women; Congress had to authorize her military retention on a yearly basis as even back then she was about 20 years past mandatory retirement.
http://www.google.com/search?q=grace+hopper&sourceid=mozilla-search&start=0&start=0
I first met Grace Hopper in '75 at a computer conference. She was 'only' a commodore back then. We were both sitting in the back of a lecture room, and I asked her if she mind if I smoked, to which she replied 'not at all darling' and promptly whipped out her pack of Lucky Strikes....
An amazing women; Congress had to authorize her military retention on a yearly basis as even back then she was about 20 years past mandatory retirement.
http://www.google.com/search?q=grace+hopper&sourceid=mozilla-search&start=0&start=0
Steve Eddy said:
Mmmm. What was the name of that woman who was the first female Admiral in the US Navy? She was big on computing technology and used to carry little lenghts of wires around with her to demonstrate what a nanosecond was, what a picosecond was, etc.
She'd appear on talk shows, whip out a little length of wire and say "That's a picosecond."
She was a pretty cool old broad.
se
No, it was nanoseconds she carried around and handed out.
I have one of her nanosecond wires somewhere since she
held a lecture here in town maybe 25 years ago. She also
had a microsecond but she only showed that one. She had
a whole bunch of nanoseconds to hand out, though, and
claimed she had stolen the wire from a phone repairman
in Pentagon. She also told how she had spent at least half
an hour in swedish customs trying to explain nanoseconds
and microseconds, and add an admirals hat on an elderly
greyhaired lady to that and I can understand why they
were puzzled.
The story behind these wires was that a long long time ago
a fellow programmer sometimes used to tell her that she
could save a microsecond here and there in a program, and
she used to answer "so what, I don't know how much a
microsecond is anyway. Can you show me one?" The fellow
couldn't, but eventually Hopper herself realized it could
be visualized by cables of approriate length.
She must be one of the funniest lecturers I've ever heard,
and probably the funniest and most unorthodox admiral
ever in the US Navy.
OK, I spent a little time measuring stuff where I sorta knew the outcome in advance. Bottom line: there's a difference of many orders of magnitude between delay time in the transit time sense and time constants from RC charging. The former are REALLY small: I can measure down to about 4-5 ns with my crude setup and the delay through a common cathode voltage gain stage feeding a cathode follower is well below that. The risetimes are another thing, but they're just ordinary poles of the sort well-accounted-for in standard control theory. I'm still not sure what the philosophical objections to feedback loops are.
Mixing the two time constants together, whether in a long explanation or in quick comments, is the source of the confusion and rancor, IMO. One is very, very short and a constant delay, the other is rather longer and has exponential characteristics.
Mixing the two time constants together, whether in a long explanation or in quick comments, is the source of the confusion and rancor, IMO. One is very, very short and a constant delay, the other is rather longer and has exponential characteristics.
SY said:
What exactly were you measuring?
Does 4-5ns represent the delay between the input signal and the feedback signal across the cathode resistor (which is the feedback element)?
The philosophical objections I've seen most often are that feedback is always "late" (i.e. that it can't correct for what's happening "now") and that it produces higher order harmonics.
se
Steve, I'm injecting (and monitoring) a square wave into the grid of a common-cathode stage, this stage is direct-coupled to a cathode follower, and then I look at the output of the cathode follower. Looking at leading and trailing edges, the point at which the output signal starts to change is less than 2 ns after the corresponding point of the input signal- i.e., I can resolve this to 2 ns and any difference is below my resolution limit.
Now, the SLOPE of that change is indeed different between input and output, but the time point at which the change begins is identical within measurement resolution.
Zero surprise here, since feedback WORKS when it is done in conformance to proper and well-understood criteria.
Now, the SLOPE of that change is indeed different between input and output, but the time point at which the change begins is identical within measurement resolution.
Zero surprise here, since feedback WORKS when it is done in conformance to proper and well-understood criteria.
Hi,
FWIW, you can have the best circuit in the world (which one is that anyway) and if you apply too high levels of FB the timbre of the instruments will have subtly changed, spatial clues will have gone down the drain too.
From that perspective, yes FB CAN be a bad thing.
How much is too much and how you'd measure what you'd be hearing, I don't know but timesmear seems to play a role here.
Cheers,
FWIW, you can have the best circuit in the world (which one is that anyway) and if you apply too high levels of FB the timbre of the instruments will have subtly changed, spatial clues will have gone down the drain too.
From that perspective, yes FB CAN be a bad thing.
How much is too much and how you'd measure what you'd be hearing, I don't know but timesmear seems to play a role here.
Cheers,
fdegrove said:FWIW, you can have the best circuit in the world (which one is that anyway) and if you apply too high levels of FB the timbre of the instruments will have subtly changed, spatial clues will have gone down the drain too.
From that perspective, yes FB CAN be a bad thing.
Cheers,
how abou the reverse: take the best circuit in the world, reduce the level of feedback in it. What will happen to the timbre and spatial clues? Will you guarante an improvement? if you cannot, wouldn't you have to conclude that LESS FB can be a bad thing, based on the logic above?
What does that mean for this discussion?
I thought anything CAN be a bad thing.
Charles,
I have several question about your design.
1. R3,4,5,6 in the gate of differential pairs (1K5) + mosfet's input capacitance (TO-220, about 1n7) will produce lowpass filter. Is this on purpose? To limit bandwith? What is your main goal putting 1K5 in the gate of differential transistor?
2. R53,54,55,56 is connected to output. It gives somekind of feedback, modulating the drop in R61,62,63,64, instead if we put those R to ground (fix the drop in R61,62,63,64).
Somehow personally I think those R can be to ground to give constant drop at the folded cascode. But you are experienced designer. What is your reason, putting those R relatively to output, so the folded cascode is not at a fixed value?
I have several question about your design.
1. R3,4,5,6 in the gate of differential pairs (1K5) + mosfet's input capacitance (TO-220, about 1n7) will produce lowpass filter. Is this on purpose? To limit bandwith? What is your main goal putting 1K5 in the gate of differential transistor?
2. R53,54,55,56 is connected to output. It gives somekind of feedback, modulating the drop in R61,62,63,64, instead if we put those R to ground (fix the drop in R61,62,63,64).
Somehow personally I think those R can be to ground to give constant drop at the folded cascode. But you are experienced designer. What is your reason, putting those R relatively to output, so the folded cascode is not at a fixed value?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.