AnalogPackrat
Well-known member
[quote author="Wavebourn"]"What is better: fat free cookies, or sugar free sousages?" [/quote]
Florida oranges, of course!
Florida oranges, of course!
To DC servo or not to DC servo,this is the question NEW INFO
- Thread starter chrissugar
- Start date
Help Support GroupDIY Audio Forum:
AP, haven't you heard?
75 percent of the citrus is wiped out across the nation.
Lighten up, man.
Or leave me alone.
75 percent of the citrus is wiped out across the nation.
Lighten up, man.
Or leave me alone.
Flatpicker
Well-known member
[quote author="bcarso"]Nice job Chris. I confess I think about things a lot more than I have the opportunity and time to listen to them, so I always like to hear some real results of careful tests, particularly when I trust the person and know they have at most a minimally skewed preconception.[/quote]
Took the words right from my mouth. Incidentally, I've been using the "Forssell" method of dc servo for quite some time, and it's always worked well for me!
Thanks again, Chris.
Took the words right from my mouth. Incidentally, I've been using the "Forssell" method of dc servo for quite some time, and it's always worked well for me!
Thanks again, Chris.
mikep
Well-known member
[quote author="Svart"]next up: voltage biased electrolytic caps?[/quote]
Ive been wondering about this too. I havent tried to listen for a difference in sound quality, but one thing I do know is that a cap with significant AC signal and a low-z load (line output or mixbus feed-on) will have a shorter lifetime if it is not biased. I suppose this is just ripple current. why don't cap manufacturers rate ripple current versus bias voltage? has anyone ever seen such a graph?
mike p
Ive been wondering about this too. I havent tried to listen for a difference in sound quality, but one thing I do know is that a cap with significant AC signal and a low-z load (line output or mixbus feed-on) will have a shorter lifetime if it is not biased. I suppose this is just ripple current. why don't cap manufacturers rate ripple current versus bias voltage? has anyone ever seen such a graph?
mike p
[quote author="mikep"][quote author="Svart"]next up: voltage biased electrolytic caps?[/quote]
Ive been wondering about this too. I havent tried to listen for a difference in sound quality, but one thing I do know is that a cap with significant AC signal and a low-z load (line output or mixbus feed-on) will have a shorter lifetime if it is not biased. I suppose this is just ripple current. why don't cap manufacturers rate ripple current versus bias voltage? has anyone ever seen such a graph?
mike p[/quote]
I am not familiar with that mechanism or any such qualification on ripple current specs. Of course an unbiased polar capacitor with any AC terminal voltage will be reverse biased during part of that waveform which isn't a valid operating mode (although perhaps often done).
In many high impedance blocking applications the pole frequency is set so low as to avoid significant terminal voltage and observation of nominal circuitry offsets will often provide a small DC bias.
For any capacitor listening test I would suggest a null test or variant on that so you can isolate just the deviation rather than try to parse out some very small part of a whole.
JR
Ive been wondering about this too. I havent tried to listen for a difference in sound quality, but one thing I do know is that a cap with significant AC signal and a low-z load (line output or mixbus feed-on) will have a shorter lifetime if it is not biased. I suppose this is just ripple current. why don't cap manufacturers rate ripple current versus bias voltage? has anyone ever seen such a graph?
mike p[/quote]
I am not familiar with that mechanism or any such qualification on ripple current specs. Of course an unbiased polar capacitor with any AC terminal voltage will be reverse biased during part of that waveform which isn't a valid operating mode (although perhaps often done).
In many high impedance blocking applications the pole frequency is set so low as to avoid significant terminal voltage and observation of nominal circuitry offsets will often provide a small DC bias.
For any capacitor listening test I would suggest a null test or variant on that so you can isolate just the deviation rather than try to parse out some very small part of a whole.
JR
mikep
Well-known member
[quote author="JohnRoberts"][
I am not familiar with that mechanism [/quote]
Ive never heard anyone else talk about it. This is something I more or less discovered myself, working as a tech and servicing lots of equipment. It seems the first caps to go, or the first ones to show increasing ESR, are often the unbiased, highly loaded ones. there are other factors too, of course. I guess I should qualify it as a theory.
I am not familiar with that mechanism [/quote]
Ive never heard anyone else talk about it. This is something I more or less discovered myself, working as a tech and servicing lots of equipment. It seems the first caps to go, or the first ones to show increasing ESR, are often the unbiased, highly loaded ones. there are other factors too, of course. I guess I should qualify it as a theory.
[quote author="mikep"][quote author="JohnRoberts"][
I am not familiar with that mechanism [/quote]
Ive never heard anyone else talk about it. This is something I more or less discovered myself, working as a tech and servicing lots of equipment. It seems the first caps to go, or the first ones to show increasing ESR, are often the unbiased, highly loaded ones. there are other factors too, of course. I guess I should qualify it as a theory.[/quote]
Perhaps I wasn't clear. I am not aware of any mechanism where bias voltage affects ripple current. I am aware of deterioration from reverse polarity which is what occurs when significant AC terminal voltage occurs at zero DC bias.
No new theory required IMO.
JR
I am not familiar with that mechanism [/quote]
Ive never heard anyone else talk about it. This is something I more or less discovered myself, working as a tech and servicing lots of equipment. It seems the first caps to go, or the first ones to show increasing ESR, are often the unbiased, highly loaded ones. there are other factors too, of course. I guess I should qualify it as a theory.[/quote]
Perhaps I wasn't clear. I am not aware of any mechanism where bias voltage affects ripple current. I am aware of deterioration from reverse polarity which is what occurs when significant AC terminal voltage occurs at zero DC bias.
No new theory required IMO.
JR
G
Guest
Guest
Mediatechnology;
distortions of electrolytics are more audible when they are used in input of microphone amps. It means that it is hard to measure their specific using ordinary measurement procedures that have the more errors the less is the signal. However it sounds as a heresy, but the best measuring tool in audio still is a listening test, however it does not have neither digital, nor analog display that is independent on the listener.
distortions of electrolytics are more audible when they are used in input of microphone amps. It means that it is hard to measure their specific using ordinary measurement procedures that have the more errors the less is the signal. However it sounds as a heresy, but the best measuring tool in audio still is a listening test, however it does not have neither digital, nor analog display that is independent on the listener.
mikep
Well-known member
Wayne, you pretty much ennunciated what I was tryng to say. I think J.R. is also saying the same thing. a polarized cap with even 1 mV foreward DC bias into a high-z load will NEVER reverse bias regardless of the AC signal amplitude.
[quote author="mediatechnology"]But I don't think a mix summing buss coupler would see enough current to give temp rise/ripple current problems. Speaker crossovers - yes.[/quote]
One example I can think of off the top of my head is in a Neve 83010 module from the 51 series consoles. Aux busses are driven by TL07x configured as non-inverting followers (or low gain non-inverting depending on an optional resistor). this leads to pretty close to zero offset, of uncertain polarity. so it doesn't need much loading to reverse bias the 100u/4V cap with high signal levels. the caps are terminated with 10k in parallel with the feed on resistors, which IIRC are 6k8. that is not exactly a high-z load. in the heat of a tracking session proper gain staging can get ignored especially on the cue mix, which isn't going to tape. many engineers will run those sends near full-up on most channels.
in any case, those caps start to fail before anything else in the module, even though they are in one of the coolest, uncramped locations.
mike p
[quote author="mediatechnology"]But I don't think a mix summing buss coupler would see enough current to give temp rise/ripple current problems. Speaker crossovers - yes.[/quote]
One example I can think of off the top of my head is in a Neve 83010 module from the 51 series consoles. Aux busses are driven by TL07x configured as non-inverting followers (or low gain non-inverting depending on an optional resistor). this leads to pretty close to zero offset, of uncertain polarity. so it doesn't need much loading to reverse bias the 100u/4V cap with high signal levels. the caps are terminated with 10k in parallel with the feed on resistors, which IIRC are 6k8. that is not exactly a high-z load. in the heat of a tracking session proper gain staging can get ignored especially on the cue mix, which isn't going to tape. many engineers will run those sends near full-up on most channels.
in any case, those caps start to fail before anything else in the module, even though they are in one of the coolest, uncramped locations.
mike p
[quote author="mikep"]
One example I can think of off the top of my head is in a Neve 83010 module from the 51 series consoles. Aux busses are driven by TL07x configured as non-inverting followers (or low gain non-inverting depending on an optional resistor). this leads to pretty close to zero offset, of uncertain polarity. so it doesn't need much loading to reverse bias the 100u/4V cap with high signal levels. the caps are terminated with 10k in parallel with the feed on resistors, which IIRC are 6k8. that is not exactly a high-z load. in the heat of a tracking session proper gain staging can get ignored especially on the cue mix, which isn't going to tape. many engineers will run those sends near full-up on most channels.
in any case, those caps start to fail before anything else in the module, even though they are in one of the coolest, uncramped locations.
mike p[/quote]
That still strikes me as a fairly easy load for an electrolytic cap and small reverse bias into that impedance will not result in much current/heat.
Sometimes a rash of component failures in a given circuit location may just be a bad batch of components, and not from in-circuit stress. I've seen all kinds of issues with electrolytic caps ranging from integrity of electrolyte seals, to quality of internal lead attachment swage to foil.
You might be able to determine more about failure mode by dissecting the part. I learned more than I ever wanted to know about capacitor manufacturing from chasing down sundry problems at my old job. We even had an xray machine so you could look inside stuff nondestructively, but I never found that of much help with capacitors.
JR
PS: I've never been much of a fan of very low voltage caps. I had some low voltage high capacitance blocking caps fail open in my consumer CD player after maybe 10 years of service. I doubt they were over stressed, just made too cheap for their own good.
One example I can think of off the top of my head is in a Neve 83010 module from the 51 series consoles. Aux busses are driven by TL07x configured as non-inverting followers (or low gain non-inverting depending on an optional resistor). this leads to pretty close to zero offset, of uncertain polarity. so it doesn't need much loading to reverse bias the 100u/4V cap with high signal levels. the caps are terminated with 10k in parallel with the feed on resistors, which IIRC are 6k8. that is not exactly a high-z load. in the heat of a tracking session proper gain staging can get ignored especially on the cue mix, which isn't going to tape. many engineers will run those sends near full-up on most channels.
in any case, those caps start to fail before anything else in the module, even though they are in one of the coolest, uncramped locations.
mike p[/quote]
That still strikes me as a fairly easy load for an electrolytic cap and small reverse bias into that impedance will not result in much current/heat.
Sometimes a rash of component failures in a given circuit location may just be a bad batch of components, and not from in-circuit stress. I've seen all kinds of issues with electrolytic caps ranging from integrity of electrolyte seals, to quality of internal lead attachment swage to foil.
You might be able to determine more about failure mode by dissecting the part. I learned more than I ever wanted to know about capacitor manufacturing from chasing down sundry problems at my old job. We even had an xray machine so you could look inside stuff nondestructively, but I never found that of much help with capacitors.
JR
PS: I've never been much of a fan of very low voltage caps. I had some low voltage high capacitance blocking caps fail open in my consumer CD player after maybe 10 years of service. I doubt they were over stressed, just made too cheap for their own good.
mikep
Well-known member
[quote author="mediatechnology"]mikep what brand, sleeve color and date code were they? [/quote]
Ive got some modules right here... this one was recapped (probably about 10 years ago) with black sleeve panasonic 105 degree 100u/25V. code looks like "G072". I know there were some bad (leaky) panasonic caps a few years back that look alot like these, but I don't see goop anywhere and most are still functioning ok. they are the same caps fitted all over the module, failling first in the afformentioned location...
some of the channels have different caps and it is the same story. loosing capacitance, ESR going up up up. in the top of the module there IS a heat issue and things are getting bad there too.
Ive got some modules right here... this one was recapped (probably about 10 years ago) with black sleeve panasonic 105 degree 100u/25V. code looks like "G072". I know there were some bad (leaky) panasonic caps a few years back that look alot like these, but I don't see goop anywhere and most are still functioning ok. they are the same caps fitted all over the module, failling first in the afformentioned location...
some of the channels have different caps and it is the same story. loosing capacitance, ESR going up up up. in the top of the module there IS a heat issue and things are getting bad there too.
radiance
Well-known member
[quote author="chrissugar"].....Also I think that direct, capacitorless coupling (with DC servo circuit) has the cleanest audio because there are no artifacts induced by the DC servo circuit and there is the advantage of eliminating a coupling cap.
[/quote]
I don't get this.... :? Capacitorless coupling WITH DC SERVO CIRCUIT has no artifacts induced by the DC servo circuit???
[/quote]
I don't get this.... :? Capacitorless coupling WITH DC SERVO CIRCUIT has no artifacts induced by the DC servo circuit???
clintrubber
Well-known member
Common misconception unfortunately. Various more elaborate explanations around here IIRIC, but the short story is that the servo-circuit enables a 'better' cap to be used (since it can be a lower value), so less artefacts from the cap. And if all is done well then the servo adds little or at least less nasties than it helps to avoid.
Regards,
Peter
Regards,
Peter
radiance
Well-known member
Thanks Peter. Yeah, spend all afternoon reading about 800+ search results on the word "servo" LOL. Still don't REALLY grasp how it's actually working :evil: :evil: :evil:
What I did learn is that most people here seem to dislike the sound of DC servo circuits and that the listening test in this post prove that a good quality cap is a safer bet.
I think I'll have to learn a programme like spice to really get these things...
What I did learn is that most people here seem to dislike the sound of DC servo circuits and that the listening test in this post prove that a good quality cap is a safer bet.
I think I'll have to learn a programme like spice to really get these things...
jdbakker
Well-known member
[quote author="radiance"]What I did learn is [...] that the listening test in this post prove that a good quality cap is a safer bet.[/quote]
How did you get that from this thread? Apart from the quote you posted earlier, Chris wrote on page 2:
[quote author="chrissugar"]For perfectly transparent sound (mastering,orchestral recording) the DC coupled with DC servo is the way to go, but for non critical application high quality polyprop or elco (Panasonic, ELNA, Nichicon) can produce very good results.[/quote]
Both results would seem to indicate that a properly implemented servo, with a good cap, is closer to 'a piece of wire' than capacitor coupling.
[quote author="radiance"]I think I'll have to learn a programme like spice to really get these things...[/quote]
SPICE isn't very good at giving an idea of the 'sound' of a circuit, especially when idealized models are used for critical components (and it can be quite hard to get accurate models).
JDB.
How did you get that from this thread? Apart from the quote you posted earlier, Chris wrote on page 2:
[quote author="chrissugar"]For perfectly transparent sound (mastering,orchestral recording) the DC coupled with DC servo is the way to go, but for non critical application high quality polyprop or elco (Panasonic, ELNA, Nichicon) can produce very good results.[/quote]
Both results would seem to indicate that a properly implemented servo, with a good cap, is closer to 'a piece of wire' than capacitor coupling.
[quote author="radiance"]I think I'll have to learn a programme like spice to really get these things...[/quote]
SPICE isn't very good at giving an idea of the 'sound' of a circuit, especially when idealized models are used for critical components (and it can be quite hard to get accurate models).
JDB.
bcarso
Well-known member
[quote author="radiance"]. Still don't REALLY grasp how it's actually working :evil: :evil: :evil: ...
I think I'll have to learn a programme like spice to really get these things...[/quote]
You might check out Self's book, which Google excerpts a relevant bit of here
Playing with servo circuits in Spice can be illuminating, but I agree with jdb that you won't find out much about purported audio quality of capacitors that way.
I think I'll have to learn a programme like spice to really get these things...[/quote]
You might check out Self's book, which Google excerpts a relevant bit of here
Playing with servo circuits in Spice can be illuminating, but I agree with jdb that you won't find out much about purported audio quality of capacitors that way.
Svart
Well-known member
I think I'll hop on this train and throw out a couple of my own opinions.
I've done some very non-scientific listening tests between servo'd stages and those using caps. I've tried a ton of different caps and I DO have to side with the caps-have-a-sound crowd although I think it's much less of an issue than most think. I eventually found my way to bipolar 'lytic caps and I really don't hear any difference between servos and good bipolar 'lytics, even after multiple stages. In fact, once I went bipolar, I stopped auditioning caps altogether and haven't looked back.
I have a question. Maybe it's a statement as well.
I wonder if there is some kind of euphonic attribute that servos impart on the audio?
I've done some very non-scientific listening tests between servo'd stages and those using caps. I've tried a ton of different caps and I DO have to side with the caps-have-a-sound crowd although I think it's much less of an issue than most think. I eventually found my way to bipolar 'lytic caps and I really don't hear any difference between servos and good bipolar 'lytics, even after multiple stages. In fact, once I went bipolar, I stopped auditioning caps altogether and haven't looked back.
I have a question. Maybe it's a statement as well.
I wonder if there is some kind of euphonic attribute that servos impart on the audio?
> Still don't REALLY grasp how it's actually working
Amplifer, plus small-cap, works like a large cap.
The flaws are the sum of the cap-flaws (may be smaller cuz the cap is smaller), plus the amplifier flaws, divided by the usual loss in the error-injection path (but large loss means more amplification needed, hence more amplification flaw).
It is different. Because coupling caps tend to be large enough to have maybe-audible flaws, and servo caps may be smaller, and amplifiers may be fairly benign, and there is little (not "no") audio in the servo path, it may be better, or perhaps just different enough to amuse the ear.
I do suspect, if you can reliably blind-test a difference, then one implementation is "bad". Also that many stated opinions would not survive blind A/B testing. The ear-ego path is a treacherous thing.
Amplifer, plus small-cap, works like a large cap.
The flaws are the sum of the cap-flaws (may be smaller cuz the cap is smaller), plus the amplifier flaws, divided by the usual loss in the error-injection path (but large loss means more amplification needed, hence more amplification flaw).
It is different. Because coupling caps tend to be large enough to have maybe-audible flaws, and servo caps may be smaller, and amplifiers may be fairly benign, and there is little (not "no") audio in the servo path, it may be better, or perhaps just different enough to amuse the ear.
I do suspect, if you can reliably blind-test a difference, then one implementation is "bad". Also that many stated opinions would not survive blind A/B testing. The ear-ego path is a treacherous thing.
radiance
Well-known member
[quote author="jdbakker"]
SPICE isn't very good at giving an idea of the 'sound' of a circuit, especially when idealized models are used for critical components (and it can be quite hard to get accurate models).
JDB.[/quote]
No, I want to learn Spice for understanding....not for judging sound...
Svart: Bipolar is not the same as "non polarised" right?
SPICE isn't very good at giving an idea of the 'sound' of a circuit, especially when idealized models are used for critical components (and it can be quite hard to get accurate models).
JDB.[/quote]
No, I want to learn Spice for understanding....not for judging sound...
Svart: Bipolar is not the same as "non polarised" right?
Sleeper
Well-known member
it IS the same. BiPolars are usually marked NP or at least the nichicons I have are.
I'm with Svart, I like the NPs.
They are a bit larger than polarised caps so that can be PITA or just not work at all if you are using them in a recap.
Kelly
