I like to build stuff up, but lack the type of test equipment necessary to measure these effects. I've used pspice just a little bit. But it seems like simulation really isn't the right tool for such subtle effects. I don't think that there are good enough models for cable microdiodes and DA/DH.
JF
JF
John,
For the record, yes, I have done some simulations of the type
you suggested. As expected and predicted by theory one does
get a non-null differential signal since we basically have two
different filters before the subtractor. I agree, however, that
the simulation helps to visualize the signals especially in the
case of assymetrical signals -- my brain is not good at doing
Fourier transforms.
So, before doing the simulation I wondered what we were
supposed to learn from it, and I still wonder. It shows that
a cap with DA behaves as a filter, which we already know
from the DA model. The model is linear so no new frequency
components enter and thus this cannot explain your cable
measurements.
BTW, I have not kept records of my simulations so I won't post
any graphs, but in my case all difference signals were more
than 120dB below the signal.
For the record, yes, I have done some simulations of the type
you suggested. As expected and predicted by theory one does
get a non-null differential signal since we basically have two
different filters before the subtractor. I agree, however, that
the simulation helps to visualize the signals especially in the
case of assymetrical signals -- my brain is not good at doing
Fourier transforms.
So, before doing the simulation I wondered what we were
supposed to learn from it, and I still wonder. It shows that
a cap with DA behaves as a filter, which we already know
from the DA model. The model is linear so no new frequency
components enter and thus this cannot explain your cable
measurements.
BTW, I have not kept records of my simulations so I won't post
any graphs, but in my case all difference signals were more
than 120dB below the signal.
john curl said:
Because, John, it doesn't take a Ph.D. to look at the DA model and figure out that those RC networks in parallel with the idealized capacitance aren't going to have any appreciable effect on frequency and phase response.
Your measurements are fine as far as they go. The "problem" is that they don't really quantify DA in terms that are relevant to audio signals.
As I've said before, seeing as DA is modeled using linear elements, then you need to look at the time domain. So the effects of DA will manifest themselves as changes in frequency and phase response.
You keep saying that sinusoidal measurements hide the efffects of DA.
No, they don't.
What they do is illustrate that the effects of DA under AC conditions are, for all intents and purposes as it relates to audio cables, non-existent.
se
Christer said:
If you want to get a more relevant picture of the effects of DA, instead of using John's subtractor circuit, just do a simple frequency and phase response plot using an idealized capacitor and one with the DA model components added.
se
Steve Eddy said:
Sure, and I have done such simulations earlier when DA has
been discussed elsewhere. I decided to
do what John suggested just so he couldn't complain
that nobody did it "his way".
Christer said:
Ah, gotcha. Thanks. 🙂
I think Bob Pease pretty much summed it up when he said:
Steve, lets' say a 100 ft stretch of speaker cable has 1000 pf. The soakage of a mylar cap of that size would be about 10 pF in series with 1 meg.
You drive this from an amplifer whose output impedance is lower than 1 ohm. In concept if you put this in a precision measuremnt system, you could measure perhaps as low as 0.01 dB of attenuation.
(YOU MUST AGREE - you certainly can't HEAR that kind of attenuation shift, 0.01 dB...)
But the soakage of 1 megohm, is below the 0.01dB by about a factor of 1000. If you want to wast your time by arguing with audio nuts about things that are 1000: 1 below the limits of MEASURABILITY - not to mention 100,000:1 below the limits of audibility --
-- BE MY GUEST, but don't waste MY time, OK?
Yeah, if you ran speaker cables not just 100 ft, but 10,000 ft. - and if you ran a rubber line cord that is perhaps 10 x worse than mylar - you might be able to measure it.
WHO CARES!
se
Hi,
In all frankness I wouldn't even expect it to measure the same anyway.
Cables are more than just DA and their capacitance is in shunt not in series as I assume you simulated it.
Cheers,😉
In all frankness I wouldn't even expect it to measure the same anyway.
Cables are more than just DA and their capacitance is in shunt not in series as I assume you simulated it.
Cheers,😉
Christer, et al, perhaps now we can discuss something. First: It is the TIME DISTORTION that happens bigtime with DA in caps. That is WHY you should use an asymmetrical pulse and VIEW the difference between the two caps with an oscilloscope or time domain print-out. Only the asymmetry will give you interesting results. The ratio can be perhaps 10,000 : 1.
If you do an AC analysis, rather than a transient analysis in your simulation, you will see only tiny phase differences, BUT when you do a transient analysis in the time domain, you can see up to 10% deviation in the waveform, and this is just from 1 cap! Think about 10 caps in the audio chain !!!
Does this matter? We think so.
We have known about the seeming limited effect of DA as implied by the model. This is why I ignored DA for almost 10 years. Almost 30 years ago, a fellow engineer did his masters thesis at UCB on DA. He sent me the relevant papers and his writeup on the effect, so I could have been on top of this back in 1974, when I was working with Mark Levinson and in 1977 when I designed the Symmetry 'Transient-Perfect' electronic crossover, but I thought then that mylar caps were just great! They were cheap, small, reliable, and they had no non-linear distortion that I could measure at the time, down to 0.001% IM.
Still, people did hear them, and we traced it to DA in the following years. This brought out, in our newer designs, direct coupling and servos to control offset about 20 years ago.
The person who did the first computer simulation that I know about was Scott Wurcer of Analog Devices. He had access to a good simulation program and I still have his printout. Unfortunately, I have never been successful in directly attaching anything to this website or any others, in general, so I will decline to do so at this time.
If you do an AC analysis, rather than a transient analysis in your simulation, you will see only tiny phase differences, BUT when you do a transient analysis in the time domain, you can see up to 10% deviation in the waveform, and this is just from 1 cap! Think about 10 caps in the audio chain !!!
Does this matter? We think so.
We have known about the seeming limited effect of DA as implied by the model. This is why I ignored DA for almost 10 years. Almost 30 years ago, a fellow engineer did his masters thesis at UCB on DA. He sent me the relevant papers and his writeup on the effect, so I could have been on top of this back in 1974, when I was working with Mark Levinson and in 1977 when I designed the Symmetry 'Transient-Perfect' electronic crossover, but I thought then that mylar caps were just great! They were cheap, small, reliable, and they had no non-linear distortion that I could measure at the time, down to 0.001% IM.
Still, people did hear them, and we traced it to DA in the following years. This brought out, in our newer designs, direct coupling and servos to control offset about 20 years ago.
The person who did the first computer simulation that I know about was Scott Wurcer of Analog Devices. He had access to a good simulation program and I still have his printout. Unfortunately, I have never been successful in directly attaching anything to this website or any others, in general, so I will decline to do so at this time.
fdegrove said:
Measure the same as the simulations you mean? No, most
certainly not, since we don't have ideal components in real
life, not even ideal capacitors with DA.
I just did what John suggested. It is his job to explain what
it has to do with his cable measurements. I didn't understand
that before simulating, and I don't understand it after.
Since John seems to agree with the usual model for DA (he
suggested using it for the simulation), it is obvious that DA
couldn't explain his measurements, but that was obvious
already without the simulations.
John,
First, you suddenly used the term time distorsion. Do you
use this a synonym to linear distorsion as you used before,
or else, how do you define it? I ask this just to avoid further
confusion.
I suppose you suggested the subtractor model for simulation
since it has some relevance to feedback loops around an
amp stage. However, don't forget that typically we will not
have two caps with the same value on the two inputs, especially
where one of them is ideal and the other one has DA. Different
value caps or no cap at all on one input is likely to swamp the
effect of DA, I would guess. It does
hint us, though, that it might make sense to do some thinking
about trying to "match" capacitors. Perhaps it may even be
sensible to introduce a cap on an otherwise DC coupled input
just to get a better cancelling effect. Although I am not so
sure any real music will have signals assymetric enough to
cause any errors of the magnitude you suggested (I may be
wrong there, of course), I admit there may be some relevance
to what you say about cumulative effects from many amp
stages. I would have to think about, or simulate, how they
add up, though. Can't quite figure that out in my head, not
right now at least.
Finally, I presume, this simulation experiment was meant to
illustrate the effects of DA, rather than explain your cable
measurements, which AFAICS must be due to some other
phenomenon (assuming they are not just some kind of
measurement error, which is still an open question until they
can be repeated by others).
First, you suddenly used the term time distorsion. Do you
use this a synonym to linear distorsion as you used before,
or else, how do you define it? I ask this just to avoid further
confusion.
I suppose you suggested the subtractor model for simulation
since it has some relevance to feedback loops around an
amp stage. However, don't forget that typically we will not
have two caps with the same value on the two inputs, especially
where one of them is ideal and the other one has DA. Different
value caps or no cap at all on one input is likely to swamp the
effect of DA, I would guess. It does
hint us, though, that it might make sense to do some thinking
about trying to "match" capacitors. Perhaps it may even be
sensible to introduce a cap on an otherwise DC coupled input
just to get a better cancelling effect. Although I am not so
sure any real music will have signals assymetric enough to
cause any errors of the magnitude you suggested (I may be
wrong there, of course), I admit there may be some relevance
to what you say about cumulative effects from many amp
stages. I would have to think about, or simulate, how they
add up, though. Can't quite figure that out in my head, not
right now at least.
Finally, I presume, this simulation experiment was meant to
illustrate the effects of DA, rather than explain your cable
measurements, which AFAICS must be due to some other
phenomenon (assuming they are not just some kind of
measurement error, which is still an open question until they
can be repeated by others).
Hi,
No, not quite. I meant don't expect cable sims and caps to measure the same.
Which is why I distinguished between measurements od a cap in series and a cap in shunt, which is how a cable's capicitance is modeled.
It's not the first time John mentions this, I think.
Linear distortion, as some call is still a distortion whether it's in the time domain or elsewhere. Frequency changes are called linear distortions too.
Try to keep them way outside of the audioband and if you can you'll hear the difference, I'm sure of it.
Why not drop this silly SE assumption about having to prove everything that is claimed by the claimant and do some research on your own?
That's the best way to find out and convince yourself IMHO.
Assuming you need convincing that is...
Cheers,😉
No, not quite. I meant don't expect cable sims and caps to measure the same.
Which is why I distinguished between measurements od a cap in series and a cap in shunt, which is how a cable's capicitance is modeled.
It's not the first time John mentions this, I think.
Linear distortion, as some call is still a distortion whether it's in the time domain or elsewhere. Frequency changes are called linear distortions too.
Try to keep them way outside of the audioband and if you can you'll hear the difference, I'm sure of it.
Why not drop this silly SE assumption about having to prove everything that is claimed by the claimant and do some research on your own?
That's the best way to find out and convince yourself IMHO.
Assuming you need convincing that is...
Cheers,😉
fdegrove said:
Quite agree, bu I didn't understand quite what John
wanted use to learn from the simulation and whether it
was supposed to be relevant to capacitors or cables or both.
My question was if he used time distorsion as synonumous
with linear distorsion, or if it was a new term he threw in.
Since linear distorsion, as I understand is, is deviations from
flat/desired frequency and phase response we obviously
want it well outside the audio band.
So you are claiming the whole interantional research community
in basically all disciplines is silly? Really Frank, I know you
better than to expect such a stupid comment from you.
It is not something Steve
cooked up, it is standard and reasonable practice. That does
not exclude one might still do some thinking and research on
other peoples claims. I have often done that when reading
seemingly unsubstantiated claims by other reasearchers. I
have even published scientific journal and conference articles
proving claims in other peoples publication to be wrong. That
might take months or even years of work and it sure would
be simpler to ask them to prove the claims themselves. They
wouldn't usually try to prove themselves wrong, of course,
and even less publish such proofs.
I don't mind doing some more thinking myself about these things, but then
I need to know what John's point is and what we are supposed to find out.
Now, time to go to bed and read Coetzee.
Hi,
Ideally, we don't want any of it. When it does occur, and it will, we want it way outside audibility.
Now what's audible or not is open for debate.
Me, I want it way out into the hundreds of thousands of Hertz because these distortions have repercussions that can show up into the audio band proper due to phase rotations for instance.
Well, this is not a university, neither is it a scientific forum, is it?
So, I reckon that it's worthwhile for someone with a vested interest in the audio bizz such as SE to sit up and listen, do his homework and than talk and report back not the other way around.
That would be stupid indeed.
What say you, Christer?
Is that common sense? Or is the jury still out?
Cheers,😉
Ideally, we don't want any of it. When it does occur, and it will, we want it way outside audibility.
Now what's audible or not is open for debate.
Me, I want it way out into the hundreds of thousands of Hertz because these distortions have repercussions that can show up into the audio band proper due to phase rotations for instance.
Well, this is not a university, neither is it a scientific forum, is it?
So, I reckon that it's worthwhile for someone with a vested interest in the audio bizz such as SE to sit up and listen, do his homework and than talk and report back not the other way around.
That would be stupid indeed.
What say you, Christer?
Is that common sense? Or is the jury still out?
Cheers,😉
There was an EW/WW article years ago entitled "Gain From Passives".
Anybody have a copy ?.
Eric.
Anybody have a copy ?.
Eric.
JF, you did good! Thanks for the references about voltage variable resistors. That is worth serious consideration when it comes to micro-diodes. It is really the same thing, at least in principle.
Christer, if you want to doubt things, so be it. I measure what I measure. However, for some of you, just think about how we improve audio quality.
Do we just think it through, from the knowledge we have gained by taking a course or reading a book? Or do we try things, try to put aside our initial opinion, and if we find that something that we try works for us, we might find out how it works for others, then if it also seems to work for others of like mind and interest in improving things, we might also try to find a physical reason for why something works. At least we might make a hypothesis as to why, and change it, if we find another direction that fits the situation better. This is how I learn new concepts. It makes me a successful audio designer.
Now what about 'education'. I have been accused as an 'education snob'. Let's get real, I come from a working class background, and I only first attended college to become a TV repair technician. Once in college, a whole new world was opened up to me. I went from tech ed., to engineering, to physics, and in real employment, back to electronics engineering. It took me years of after-hours classes to catch up and formally become a senior electronics engineer. I am not an academic, but I both learn from and teach academics. If I need a Ph.D., I hire one. Sometimes they hire me. This is the pursuit of understanding, without self limiting. Works for me.
Christer, if you want to doubt things, so be it. I measure what I measure. However, for some of you, just think about how we improve audio quality.
Do we just think it through, from the knowledge we have gained by taking a course or reading a book? Or do we try things, try to put aside our initial opinion, and if we find that something that we try works for us, we might find out how it works for others, then if it also seems to work for others of like mind and interest in improving things, we might also try to find a physical reason for why something works. At least we might make a hypothesis as to why, and change it, if we find another direction that fits the situation better. This is how I learn new concepts. It makes me a successful audio designer.
Now what about 'education'. I have been accused as an 'education snob'. Let's get real, I come from a working class background, and I only first attended college to become a TV repair technician. Once in college, a whole new world was opened up to me. I went from tech ed., to engineering, to physics, and in real employment, back to electronics engineering. It took me years of after-hours classes to catch up and formally become a senior electronics engineer. I am not an academic, but I both learn from and teach academics. If I need a Ph.D., I hire one. Sometimes they hire me. This is the pursuit of understanding, without self limiting. Works for me.
Hi,
Yes, John...Their behaviour is anologous to microdiodes but I doubt they are in real life.
I've given hints already.
More than I intended to really.
A suggestion stemming from our local university: repeat the same for an enamelled solid core wire and compare the results.
Remember what I told you Steve? Hint #1?
Admiration all over from me, John...some of us pushed that even further and I'm sure you'd be the last to say stop.
Cheers and with the utmost respect to all of you,😉
Yes, John...Their behaviour is anologous to microdiodes but I doubt they are in real life.
I've given hints already.
More than I intended to really.
A suggestion stemming from our local university: repeat the same for an enamelled solid core wire and compare the results.
Remember what I told you Steve? Hint #1?
Admiration all over from me, John...some of us pushed that even further and I'm sure you'd be the last to say stop.
Cheers and with the utmost respect to all of you,😉
fdegrove said:
Okay, but it could have been how copper versus silver react to the insulator. That seems ruled out.
Thanks for the reply.
JF
john curl said:
Well, I'm having a really good day today. I'll have to look more carefully at information on intergranular boundaries. (Intergranular boundaries have not been discussed much at diyAudio. Unless, there is yet another term for it.)
And there is a better article on Varistors here:
http://www.littelfuse.com/PDFs/AppNotes/an9767.pdf
Thanks!
JF
- Status
- Not open for further replies.