The Fuxtor
Addicted Member
Think maybe I will just order 2 Nichy Muse caps for each channel and just parallel them.... If it sucks, they are easily changed and swapped...
Thanks fellas,!
Thanks fellas,!
Coupling caps
- Thread starter The Fuxtor
- Start date
Don't forget to bypass with multiple smaller values like 10 uF, 1 uF, 0.1 uF, and 0.01 uF. This will dramatically lower the ESR and ESL, and improve the higher frequencies.
You're still going to have dielectric absorption issues, of course, but one fixes what one can.
You're still going to have dielectric absorption issues, of course, but one fixes what one can.
Probably overkill on an amp with transformer coupled drivers, a 1uF metallized film polyprop may be plenty, stacked onto a Muse elyticap.
Well, kills the ESR/ESL and DA is a funny property which tends to hammer the higher frequencies.
In any event, the cost is a few dollars per side, so why worry about it?
The 3000A is from 1966, the year I enlisted. Serious solid state hi fi was relatively new and very Space Age. This is quite an advance from my totl 1959 Sansui SM-80, all tube except the germanium transistor phono stage, very modern. My SM-80 was built when John Kennedy was elected to the Presidency, the predecessor to the 1000, predecessor to the 1000A, Sansui's last totl tube receiver.
I love this old stuff.
I love this old stuff.
I have one of these and purchased 6 x 1000uF bipolars which will be breadboarded and bolted sideways vs vertically like the original 1000uF cans. I haven't gotten to the project yet, but found the idea here:
http://audiokarma.org/forums/index.php?threads/sansui-3000a-speaker-prot-mod.647005/
http://audiokarma.org/forums/index.php?threads/sansui-3000a-speaker-prot-mod.647005/
Am I wrong here with thinking that 2 equal value caps in series gives half the value?
Such as 1000µF+1000µF(in series) = 500µF DP?
So a single Nichicon ES Muse 500µF + a jumper for the 2nd cap would work just fine?
Cheers,
James
Such as 1000µF+1000µF(in series) = 500µF DP?
So a single Nichicon ES Muse 500µF + a jumper for the 2nd cap would work just fine?
Cheers,
James
Yes, capacitors in series reduce the capacitance. If equal, halved. The shorthand for the reciprocal series equation with two values is:
Given C1 = C2, both of value C, the equation reduces to:
CTotal = (C1 x C2) / (C1 + C2)
Given C1 = C2, both of value C, the equation reduces to:
CTotal = (C^2) / (2 x C)
CTotal = 1/2 x C
CTotal = 1/2 x C
the OP's schematic shows 1000uf/35v and there's probably room for only one
if for no other reason than LS limits.
this presumes the output blocking (and is casually also a coupling) is bad.
(the only way I would think it was bad would be if it was open and no sound)
you could try back-to-back (haven't noticed whether these come in non-polar)
polymer caps which I've found to me smaller.
if there's room then 3000uf is 3 NPs in parallel, or 6 in back-to-back pairs of
2000uf each, you could mount these on a perfboard with pins that solder
into the original holes.
if for no other reason than LS limits.
this presumes the output blocking (and is casually also a coupling) is bad.
(the only way I would think it was bad would be if it was open and no sound)
you could try back-to-back (haven't noticed whether these come in non-polar)
polymer caps which I've found to me smaller.
if there's room then 3000uf is 3 NPs in parallel, or 6 in back-to-back pairs of
2000uf each, you could mount these on a perfboard with pins that solder
into the original holes.
Electrolytics poorly age and the properties can deteriorate short of total failure.
One area is ESR, which increases over time. Another is dielectric absorption which is high for older capacitors and can increase with age. DA causes electrons to migrate into the dielectric and then slowly migrate out when the field is reversed. This averages the signal being applied to the capacitor with its signal history. This is why A/D and D/A circuitry requires capacitors with low DA. Finally, there is leakage as the dielectric degrades, deteriorates, and disintegrates over time. This causes distortion but may not rise to the level of a full short which would blow the output stage.
Anyway, my point was that the properties may degrade the sound without causing failure via short or open.
Not sayin', just sayin'.
no problem with ecaps going bad. I have a bag full of replaced caps from one Hafler dh101 alone.
some folks assume that caps that go "bad" can be tested for the usual suspects but that bag
of mine has caps that whistle randomly and do NOT test bad.
I would assume that no sound is not the same as low..bad sound so my derivative deduction
is that they are open.
up to OP to clarify.
some folks assume that caps that go "bad" can be tested for the usual suspects but that bag
of mine has caps that whistle randomly and do NOT test bad.
I would assume that no sound is not the same as low..bad sound so my derivative deduction
is that they are open.
up to OP to clarify.
I don't believe that electrolytic caps have "rectification" or that ESR per se is a source of distortion. By definition, distortion is a non linearity in the transfer function- that is either the voltage or current out is not equal to a constant times the voltage or current in, and that certainly is true if you examine the transfer function of an electrolytic even if no rectification occurs.
Dielectric absorption is arguably not a source of harmonic distortion, it is, however a source of "time domain distortion" as in effect it amounts to some portion of the signal being delayed and released at a later time, but in the case of coupling caps only for frequencies near the cut off frequency. At frequencies significantly below this there is no signal passing through to care about, and at frequencies significantly above this there is no voltage across the cap to worry about. At these elevated frequencies ESR becomes an issue, but if the input resistance is say, 6 orders of magnitude higher than the ESR (say 100k versus 0.1 ohms)- then who cares? Well, at least until frequencies approaching the SRF of the caps.
As you approach the SRF you need to add/use films in parallel (or instead of) as electrolytic caps have very low self-resonant frequencies (SRF) due to their large self inductances. SRFs can be 10s to 100s of kHz- and that can effect the high frequency response. Shunting this with a cap which is say, an order of magnitude smaller and which has at least an order of magnitude higher SRF (1MHz and above) and a low ESR can minimize the effect.
You can (and I do) model and simulate with these effects in place and can demonstrate changes in frequency response, step response and distortion for amplifiers that correspond quite well to measured results.
The rectification effect of the oxide layer in electrolytic capacitors is well known, at least among electrical engineers, well researched, and well accepted; it fully explains why an electrolytic capacitor cannot be used in the signal path of a non-linear circuits when rectification matters, and, in part, why the device has polarity.
You may research this topic in scholarly journals or even modern capacitor buying guides if you disbelieve it. The effects were known prior to the 1920s, and it is commonly mentioned in any modern manufacturer's guide. It truly is not in dispute.
As far as ESR, adding additional impedance to a circuit which alters the voltage changes the circuits characteristics. The effect can be demonstrated using a speaker crossover where old motor-run PIO capacitors, from the 1950s, having high ESR are replaced with modern films.
The effect of ESR in dramatically reducing Q can be seen in high-speed computer motherboards, and I lump cell phones in with such devices. Low-ESR capacitors, or parallel circuits of even high-ESR capacitors, create a tank circuit with such high Q that ringing occurs from the rapid dumping of switching change. Paralleling capacitors for decoupling purposes often reduces the ESR to the point that ringing can be a significant issue. (The issue is somewhat complicated by the fact that the ESL of the capacitors can create mutually resonant circuits.)
From the micro effects are the macro effects built.
I have not addressed the effects of AC signals on the lifespan of electrolytic capacitors, or the generally limited lifespan of electrolytic capacitors as a whole, but the short lifespan also has ramifications for using such capacitors for coupling purposes.
Dielectric absorption aka "soakage" performs signal averaging. Again, this is well-known in the analog circuit world. It also explains why a high-voltage capacitor, even with a properly-connected shorting bar across its terminals, may retain (and recover after removal of the shorting bar) sufficient charge—after days, months, or even years—to deliver a dangerous shock.
Uh, no, it isn't oscillation at all.
What Bob is describing is well known and arises in a few ways.
The typical fashion is piezo compression because of dielectric stress. As the capacitor is stressed and relaxes the internals vibrate, creating sound. Magnetostriction effects may also contribute, just like in transformers.
The worst way it is vaporization of the internals, either because the ESR causes ohmic heating or because of internal leakage, both boiling off the electrolyte. Install an electrolytic capacitor backwards and the whistling effect is pronounced, at least until the capacitor explodes. The sound does not long last, but the effects of being next to an exploding capacitor, particularly of a large size, may lifelong remain.
Capacitors can readily oscillate. The ESL and capacitance of each capacitor may combine with other capacitors to form tank circuits. This is a well-known problem in computer motherboards, particularly at high speed. It is one of the reasons why mixing capacitors of different values and different ESRs is a tricky design practice.
Well, I'm actually an electrical engineer- a PhD actually, and a senior fellow (emeritus) with a well known IC design company (Analog Devices).
I am aware of the asymmetric current flow properties of metal-oxide interfaces, they were, after all, the first crude semiconductors, but I can find no evidence of the effect as being a significant contributor to the non-linearity of bipolar electrolytic caps. Please provide a reference so that I might become more informed.
I am aware of the physics of the films employed in capacitors- including electrolytics, and the forming and failure of the dielectric films and why they need to be biased appropriately. However, this does not appear to be a problem with so called bipolar dielectrics, at least according to any manufacturers data that I can find. Again, please provide a reference so that my education can proceed.
This is a case of sophistry. Capacitors cannot oscillate, there is no "gain" mechanism to cause it to happen. Conservation of energy prohibits it.
They can, as you say, act as tank circuits which need to be stimulated by an energy source of some kind. To illustrate this you can have a near infinite Q element, such as a crystal and you can place a probe across it and it will NEVER oscillate on its own. It has to be given the correct excitation i.e. to satisfy the required conditions for oscillation in order to produce an output.
As far as magnetostriction is concerned. Well, at one point I designed magnetostrictive thin film heads for disk drive read channels (and the read channels themselves) so I am well aware of the effect.
Magnetostrictive effects are well known in ferromagnetic materials, and there are instances where tuned circuits have been built using this effect in inductors and piezo electric effects in capacitors, but I am unaware of the effect in electrolytic caps.
I would be very interested to know what the magnetostrictive material in an electrolytic cap is. I can find, after an admittedly cursory search, no reference to such an effect so I would appreciate you providing a reference as I'm always interested in learning new things.
I am aware of the piezo electric effects that occur in capacitors, but I was unaware that it occurred due to voltage stress in modern electrolytic caps, but it's certainly plausible.
Again, ESR can be modelled as a complex, frequency and time dependent resistor. It does not have any voltage dependence that I am aware of and I can find no mention of such an effect in the literature, and the changes over time are ageing effects and can be modelled as a worst case value with the attendant frequency dependent element added. This does not cause distortion, by definition. There is no non-linearity in the transfer function of a device induced by its ESR.
Yes, I am aware of the "soakage effect" and the physics that results in its occurrence. Historically it is modelled by a series of parallel RC elements or C elements with added delays. These represent the process of charge release.
The effect is one that is modelled as a totally linear system. It will affect the impulse response of the system when compared to an ideal cap, but as there are no voltage dependent non-linear elements in the system, by definition, there cannot be any harmonic distortion, or for that matter intermodulation distortion of any kind.
Sigh. I forsee yet another debacle about capacitors, ending in a smoking ruin of locked thread. I am putting down the capacitors, and backing away.
For those interested in technical discussions on the subject, Cyril Bateman's excellent work on capacitors is a good source for actual measurements of capacitors in audio, and he understood how to do proper measurements using low-distortion audio sources. (So it is not analogous to Pons and Fleischmann calorimetry mistakes.) Bateman truly was The Man when it came to capacitors in audio.
As far as oscillation, I engaged in no sophistry and your strawman is yours, not mine, and the fact that you set that strawman ablaze means that you must extinguish it, not me.
I very clearly described a tank circuit formed from capacitors of various ESR and ESL on motherboards obviously being stimulated by dumping of charge in digital logic. I never said capacitors would oscillate on their own; that's just silly. (snorts) No no no. What I wrote was a short comment which should not need qualification for an EE. Do I really need to each time detailed write, as I elsewhere did, long explanations like this:
The OP wrote, "Is it acceptable to use 2 polar caps tied together". A significant difference exists between polar, NPE, bi-polar, and solid-body electrolytics and the different types cannot be together lumped. The typical capacitor used in the signal path is a garden-variety polar electrolytic, with poor properties for signal-path use. Bi-polar has, IIRC, about half the distortion of polar or NPE. But these are not commonly used in the signal path. Most amplifiers, particularly tubes, are loaded with ordinary BaTiO3 ceramics and ordinary polar electrolytics in the signal path, and these have undesirable properties.
For those interested in technical discussions on the subject, Cyril Bateman's excellent work on capacitors is a good source for actual measurements of capacitors in audio, and he understood how to do proper measurements using low-distortion audio sources. (So it is not analogous to Pons and Fleischmann calorimetry mistakes.) Bateman truly was The Man when it came to capacitors in audio.
As far as oscillation, I engaged in no sophistry and your strawman is yours, not mine, and the fact that you set that strawman ablaze means that you must extinguish it, not me.
I very clearly described a tank circuit formed from capacitors of various ESR and ESL on motherboards obviously being stimulated by dumping of charge in digital logic. I never said capacitors would oscillate on their own; that's just silly. (snorts) No no no. What I wrote was a short comment which should not need qualification for an EE. Do I really need to each time detailed write, as I elsewhere did, long explanations like this:
Here's an example I previously discussed. In computer motherboards the ESR losses, again a parasitic impedance of the capacitor, will damp out high-frequency spikes from transistor switching. Each time a transistor cycles it must dump its charge; the frequency of that dumping is a square-wave oscillator. Because motherboards operate at the hundreds of MHz or GHz, this creates a square-wave at that frequency. But ESR losses in the capacitors then converts the energy of the oscillation into heat, and thus damp out the oscillation which would otherwise pass through the ground plane.
When dozens or hundreds of decoupling capacitors are placed in parallel on a motherboard the system's ESR drops to a negligible level and the losses vanish with it. This means the Q of the circuit increases. (Again, lower losses mean higher Q and higher losses means lower Q.) The square-wave in a high-Q circuit will not be damped by the parasitic impedance. As we know, large filter capacitors poorly remove high-frequency noise or even ripple current at mains harmonics. This is why small decoupling capacitors are used. As a result the square wave now bounces back and forth through the capacitors creating standing waves in the ground plane. The energy has no place to go, as the LC perfectly stores and releases energy without losses.
But the problem can become worse than this.
The common approach to decoupling adds capacitors of different sizes. The dissimilar capacitors now have different ESL. At high frequencies that ESL can create LC tank circuits with other capacitors where the capacitance from one device interacts with the parasitic inductance (ESL) of another device, or even with with the parasitic inductance of the PCB wiring or device leads. This goes for the AM band and low MHz, even.
The solution is adding high-ESR capacitors—or resistors in series with the capacitors—to ruin the Q by converting unwanted energy into heat. By adding impedance to the filtering capacitors the losses sufficiently increase that the circuit's Q drops and the oscillator is ruined. Losses overcome energy and the circuit is damped. That impedance is inherent in the capacitor's ESR. It has just enough. Think of the flip side, where a designer deliberately created an oscillator based upon ideal properties only to discover that the real-world impedance consumes just enough energy to make it not function.
So the properties of the ideal capacitor created through bypassing (multiple decoupling capacitors in parallel) are adjusted by adding impedance to ensure losses necessary to damp any undesirable oscillation or ringing.
When dozens or hundreds of decoupling capacitors are placed in parallel on a motherboard the system's ESR drops to a negligible level and the losses vanish with it. This means the Q of the circuit increases. (Again, lower losses mean higher Q and higher losses means lower Q.) The square-wave in a high-Q circuit will not be damped by the parasitic impedance. As we know, large filter capacitors poorly remove high-frequency noise or even ripple current at mains harmonics. This is why small decoupling capacitors are used. As a result the square wave now bounces back and forth through the capacitors creating standing waves in the ground plane. The energy has no place to go, as the LC perfectly stores and releases energy without losses.
But the problem can become worse than this.
The common approach to decoupling adds capacitors of different sizes. The dissimilar capacitors now have different ESL. At high frequencies that ESL can create LC tank circuits with other capacitors where the capacitance from one device interacts with the parasitic inductance (ESL) of another device, or even with with the parasitic inductance of the PCB wiring or device leads. This goes for the AM band and low MHz, even.
The solution is adding high-ESR capacitors—or resistors in series with the capacitors—to ruin the Q by converting unwanted energy into heat. By adding impedance to the filtering capacitors the losses sufficiently increase that the circuit's Q drops and the oscillator is ruined. Losses overcome energy and the circuit is damped. That impedance is inherent in the capacitor's ESR. It has just enough. Think of the flip side, where a designer deliberately created an oscillator based upon ideal properties only to discover that the real-world impedance consumes just enough energy to make it not function.
So the properties of the ideal capacitor created through bypassing (multiple decoupling capacitors in parallel) are adjusted by adding impedance to ensure losses necessary to damp any undesirable oscillation or ringing.
The OP wrote, "Is it acceptable to use 2 polar caps tied together". A significant difference exists between polar, NPE, bi-polar, and solid-body electrolytics and the different types cannot be together lumped. The typical capacitor used in the signal path is a garden-variety polar electrolytic, with poor properties for signal-path use. Bi-polar has, IIRC, about half the distortion of polar or NPE. But these are not commonly used in the signal path. Most amplifiers, particularly tubes, are loaded with ordinary BaTiO3 ceramics and ordinary polar electrolytics in the signal path, and these have undesirable properties.
agree totally with @Retrovert
sorry - this whistling was random, could occur at startup or after minutes/hours, and gave you the initial
impression that it was a volume control issue of "crap on the tracks" (soon to be trademarked). was
not in the input circuits of the phono stages nor in the power supply.
done twice with my DH101 of 40 years, and with several other units (see my guides on upgrading
the DH101, Dh100, DH110, etc)
and the caps are themselves 20+ years old in one case, and 40 in the other. with continuous
use since the late 1970s and there's no way anyone (except those offering 6 figure bribes)
can convince me they are still good even using today's measurements like ripple current,
ESR, etc. Plus any measurements that will come into play in the next 30 minutes/days/years.
and on my meters they test good but anything else like active frequency in-circuit testing
will probably need unmolested examples from the 1970s. if they're still good. maybe used
and NOS suffer the same fate like Chinese 1000 year eggs after 40 years.
I still have them in case any one wishes to invest thousands (hours, dollars, measurement tools)
in trying to isolate/diagnose it. Free and I pay postage but only on the condition hundreds/thousands
of hours be put into the diagnoses and the preliminary theory be followed.
payback - someone comes up with a practical solution (many have told me I was wrong), and no
one has put theory into practice. I will back the owner/developer for that billion dollar IPO
( I know Sandhill road like the back of my hand). ALL, I mean all the major design houses,
can now save billions from testing, prototypes, voicing (millions in equipment, design engineering,
parts selection, listening) - the ROI would be measured in hours.
this is not the first time I have offered these "bad" capacitors" to someone but you have to
check through my thousand+ postings to see who they are. these caps are not obvious -
they're simply in direct line of the sound chain, and to everyone's satisfaction may very
well test and work well as bypasses in some low end power supply. and if they have zobels
then whistling may not find its way out.
not interested in hearing about I'm wrong and should go back and replace those Nichicon Muse
ES green bi-polar caps with 40+ YO cheapies. and further some other theory about random
whistles. next thing you know the deniers come and declare I'm hearing things.
I have enough voices in my head to know the differences between ear-worms (many in foreign
languages but not dead languages), back ground tunes, and requiems from the greats composers
and no religious/violent/political messages.
so let's not invest hours over 50 cent caps that takes minutes to replace. build me that Sound Quality
tool that I can put in front of speakers, in circuits, out-of-circuits, at phono cartridge outputs, that
puts a number on SQ. that relative/absolute number will stop the my-audio-is-better-than-yours
argument we see everywhere.
sorry - this whistling was random, could occur at startup or after minutes/hours, and gave you the initial
impression that it was a volume control issue of "crap on the tracks" (soon to be trademarked). was
not in the input circuits of the phono stages nor in the power supply.
done twice with my DH101 of 40 years, and with several other units (see my guides on upgrading
the DH101, Dh100, DH110, etc)
and the caps are themselves 20+ years old in one case, and 40 in the other. with continuous
use since the late 1970s and there's no way anyone (except those offering 6 figure bribes)
can convince me they are still good even using today's measurements like ripple current,
ESR, etc. Plus any measurements that will come into play in the next 30 minutes/days/years.
and on my meters they test good but anything else like active frequency in-circuit testing
will probably need unmolested examples from the 1970s. if they're still good. maybe used
and NOS suffer the same fate like Chinese 1000 year eggs after 40 years.
I still have them in case any one wishes to invest thousands (hours, dollars, measurement tools)
in trying to isolate/diagnose it. Free and I pay postage but only on the condition hundreds/thousands
of hours be put into the diagnoses and the preliminary theory be followed.
payback - someone comes up with a practical solution (many have told me I was wrong), and no
one has put theory into practice. I will back the owner/developer for that billion dollar IPO
( I know Sandhill road like the back of my hand). ALL, I mean all the major design houses,
can now save billions from testing, prototypes, voicing (millions in equipment, design engineering,
parts selection, listening) - the ROI would be measured in hours.
this is not the first time I have offered these "bad" capacitors" to someone but you have to
check through my thousand+ postings to see who they are. these caps are not obvious -
they're simply in direct line of the sound chain, and to everyone's satisfaction may very
well test and work well as bypasses in some low end power supply. and if they have zobels
then whistling may not find its way out.
not interested in hearing about I'm wrong and should go back and replace those Nichicon Muse
ES green bi-polar caps with 40+ YO cheapies. and further some other theory about random
whistles. next thing you know the deniers come and declare I'm hearing things.
I have enough voices in my head to know the differences between ear-worms (many in foreign
languages but not dead languages), back ground tunes, and requiems from the greats composers
and no religious/violent/political messages.
so let's not invest hours over 50 cent caps that takes minutes to replace. build me that Sound Quality
tool that I can put in front of speakers, in circuits, out-of-circuits, at phono cartridge outputs, that
puts a number on SQ. that relative/absolute number will stop the my-audio-is-better-than-yours
argument we see everywhere.
