Does cap voltage affect tone?

[Chuck H]

O.K., here you are (I think I picked the same caps as before):

Click image for larger version. Name: ecapimp.png Views: 17 Size: 10.5 KB ID: 49647

Measured values (using a PeakTech 2170 LCR/ESR meter with 4-wire Kelvin clamps):
(the numbers mean C@1kHz/ESR@1kHz/ESR@10kHz/ESR@100kHz)

black: 23.6µ/0.46/0.37/0.33 (22µ/450V, Siemens Long Life, 1992)
red: 31.6µ/0.93/0.0.77/0.0.71 (32µ/350V, 1967)
mag: 33.0µ/0.62/0.54/0.41 (32µ tantalum)
green:20.7µ/1.46/1.24/1.06 (22µ/35V)

Here's another

[Helmholtz]

I believe you may have confused a Crossover frequency and a POLE... not the same thing... The POLE is the -3dB point you may be referring to.

The definition of a filter pole seems to have changed over the years. In newer literature the +/-3dB frequency of a first order filter is actually called a (complex) pole. Nevertheless it's just a corner frequency followed by a 6dB/octave slope.

[Chuck H]

The definition of a filter pole seems to have changed over the years. In newer literature the -3dB frequency of a first order filter is actually called a (complex) pole. Nevertheless it's just a corner frequency followed by a -6dB/octave slope.

Indeed! It seems antagonistic to tell someone they're "confused" based on semantics. But then, that isn't the only antagonism in that post either. So if that was the goal, good job. Insinuating that Juan is using a "Web Calculator"?

Tiny concession at the end... How magnanimous. Where do I get in line to kiss the ring?

[Mike Sulzer]

By the same token, string to string beating can create all sorts of low frequencies, even as low as 1 Hz of even less if they are *almost* in tune, what matters is difference and you can get as close as you wish.

And no need for tubes or even *electronic* stuff in the path, Piano tuners for ages have counted beatings per second between double and triple strings hit by a single hammer key for a richer sound and consistency.

Beats are not power at low frequencies. Two close frequencies create an amplitude pattern that the ear-brain can detect. Creating power at the low frequencies requires a non-linearity. Without the non-linearity, a Fourier analysis would show no power at the low frequency. Beats can do various things in an amplifier. More complex beats than produced by a pair of sine waves might be sufficiently asymmetrical to produce "interesting" effects in an overdriven output stage, where the potential for output transformer saturation exists with even a small amount of power in the Hz range.

[Helmholtz]

A few thoughts regarding the claim that HV charge produces a higher C value by foil attraction in an E-cap.
While I could imagine such (minute) effect in a film/foil cap, I have problems with electrolytics. The reason is that squeezing the foils in an E-cap will NOT increase its capacitance (at least in theory). This is due to the construction/functional principle of an electrolytic: It consists of an anode foil that is fully covered by an insulating layer of aluminum oxide (Al2O3), a second plain aluminum foil and a paper layer soaked with a conductive soup (i.e. the electrolyte) between. Now, the plain foil is NOT the cathode of the capacitor. The real cathode is the electrolyte and the plain foil just serves to contact it. The dielectric is the extremely thin, extremely hard and incompressible (think emery or sapphire ) Al-oxide layer. In other words, the real capacitor is formed by the anode foil, the oxide layer and the electrolyte. As a consequence the C value will not change when the distance between the two foils changes as long as the oxide layer stays the same. Furthermore any attracting force caused by the charge does not act between the two foils but between the anode foil and the molecules (more exactly: ions or anions) of the electrolytic liquid. And I don't think these tiny molecules can effectively squeeze the hard oxide layer to a smaller thickness and thus change capacity.

I consulted some literature on electrolytics (actually data/application handbooks from the paper age) and found no indication of a voltage dependancy of capacitance. But what I found was that there are two standardized methods for C measurement. The first one evaluates the voltage to current ratio at 100/120Hz. This yields the AC capacitance and is the method used by most C-meters. I did not find any DC bias requirement for this measurement.
The second method uses the charge/discharge time and gives the DC capacitance. The latter is said to be typically 10% to 50% higher than the AC value. I could not find any details of the procedure.

The specified value is generally the AC capacitance.

[nickb]

A few thoughts regarding the claim that HV charge produces a higher C value by foil attraction in an E-cap.
While I could imagine such (minute) effect in a film/foil cap, I have problems with electrolytics. The reason is that squeezing the foils in an E-cap will NOT increase its capacitance (at least in theory). This is due to the construction/functional principle of an electrolytic: It consists of an anode foil that is fully covered by an insulating layer of aluminum oxide (Al2O3), a second plain aluminum foil and a paper layer soaked with a conductive soup (i.e. the electrolyte) between. Now, the plain foil is NOT the cathode of the capacitor. The real cathode is the electrolyte and the plain foil just serves to contact it. The dielectric is the extremely thin, extremely hard and incompressible (think emery or sapphire ) Al-oxide layer. In other words, the real capacitor is formed by the anode foil, the oxide layer and the electrolyte. As a consequence the C value will not change when the distance between the two foils changes as long as the oxide layer stays the same. Furthermore any attracting force caused by the charge does not act between the two foils but between the anode foil and the molecules (more exactly: ions or anions) of the electrolytic liquid. And I don't think these tiny molecules can effectively squeeze the hard oxide layer to a smaller thickness and thus change capacity.

I consulted some literature on electrolytics (actually data/application handbooks from the paper age) and found no indication of a voltage dependancy of capacitance. But what I found was that there are two standardized methods for C measurement. The first one evaluates the voltage to current ratio at 100/120Hz. This yields the AC capacitance and is the method used by most C-meters. I did not find any DC bias requirement for this measurement.
The second method uses the charge/discharge time and gives the DC capacitance. The latter is said to be typically 10% to 50% higher than the AC value. I could not find any details of the procedure.

The specified value is generally the AC capacitance.

For fun I did some guesstimating of the pressure exerted on the electrolyte. I could not find any good numbers for the dielectric constant or electrode spacing so it's a (barely) educated guess for those. Now keep in mind that the pressure to compress a liquid ( i.e electrolyte) is something like 1100 bar for a 5% volume change. I think you'll agree that the pressure inside is so tiny at 1.1 x 10e-9 as to make any accusations of incorrect guesses moot.

Here are the numbers for a 22uF cap at 250V

Capacitance (uF)	Volts	Charge	spacing (m)	Force (n)	Dielectric Cont	Area (m^2)	Pressure (n/m^2)	Pressure (Bar)
22	250	0.0055	1.00E-007	2.7225E-006	10	0.0248868778	0.000109395	1.09395E-009

As far as I am aware, ideas such as ESR zero derive more usually for SMPS design where the effect of ESR on loop stability and residual ripple are of interest. Not of any relevance to a tube guitar amp.

[Helmholtz]

As far as I am aware, ideas such as ESR zero derive more usually for SMPS design where the effect of ESR on loop stability and residual ripple are of interest. Not of any relevance to a tube guitar amp.

True.

[Helmholtz]

For fun I did some guesstimating of the pressure exerted on the electrolyte. I could not find any good numbers for the dielectric constant or electrode spacing so it's a (barely) educated guess for those. Now keep in mind that the pressure to compress a liquid ( i.e electrolyte) is something like 1100 bar for a 5% volume change. I think you'll agree that the pressure inside is so tiny at 1.1 x 10e-9 as to make any accusations of incorrect guesses moot.

Here are the numbers for a 22uF cap at 250V
Capacitance (uF) Volts Charge spacing (m) Force (n) Dielectric Cont Area (m^2) Pressure (n/m^2) Pressure (Bar)
22 250 0.0055 1.00E-007 2.7225E-006 10 0.0248868778 0.000109395 1.09395E-009

There is no attraction/pressure between the foils. The electrolyte is conductive and thus carries no field and a dielectric constant cannot be defined. Compressing the conductive electrolyte would not change anything. The dielectric is the Al-oxide and its thickness is around 1µm (=10E-6 m) having a dielectric constant of 9.5. So you would need the compressibility of the solid Al2O3.

[Helmholtz]

Here is the probably most comforting argument regarding the OP's question:

If HV e-caps would only show their spec'd C at rated voltage there would be the severe problem of measurement/verification. I have never heard of a C-meter that allows measurement with a bias of several hundred volts. (I have worked with some having an optional bias of a few volts though.)

Cap manufacturers need to make sure that their products measure within spec/datasheet on the meters used by the incoming components inspection guys of the high volume customers to avoid claims/complaints. If quality verification requires special test conditions they need to specify in the datasheets.

For a while I was the manager of a QC group of EEs at one of our production sites. We did things like second source component qualification/release, write component specs and define test conditions and gear for incoming inspection. Our component specs were part of the purchase/supply contract. HV e-caps we bought in millions and we even did real lifetime testing in our SMPS products. We did not use/specify a HV charge for C/ESR verification.

Now if actually some HV e-caps show a 20% increase in C (this is the maximum I would expect from my experience for a new cap, the reason probably being of electro-chemical nature), who cares?
And my measurements above show that the ESR of the HV types is fine even at zero DC.

[SoulFetish]

Sounds like you are clueless..... I could explain for 100 pages and you still wont get it... You don't even know what a Network Analyzer is... Look up a HP 3577 and read... It's an ANALOG signal sweep and it sweeps as low as 5 Hz ... not in the RF range... its not DIGITAL it measures impedance ..Been designing and building audio amps guitar amps for 45 years.. Also design capacitors for a major manufacturer....Stay out of my post and stay out of yours...

Don't be a cunt.

[cerrem]

Here is the probably most comforting argument regarding the OP's question:

If HV e-caps would only show their spec'd C at rated voltage there would be the severe problem of measurement/verification. I have never heard of a C-meter that allows measurement with a bias of several hundred volts. (I have worked with some having an optional bias of a few volts though.)

Cap manufacturers need to make sure that their products measure within spec/datasheet on the meters used by the incoming components inspection guys of the high volume customers to avoid claims/complaints. If quality verification requires special test conditions they need to specify in the datasheets.

For a while I was the manager of a QC group of EEs at one of our production sites. We did things like second source component qualification/release, write component specs and define test conditions and gear for incoming inspection. Our component specs were part of the purchase/supply contract. HV e-caps we bought in millions and we even did real lifetime testing in our SMPS products. We did not use/specify a HV charge for C/ESR verification.

Now if actually some HV e-caps show a 20% increase in C (this is the maximum I would expect from my experience for a new cap, the reason probably being of electro-chemical nature), who cares?
And my measurements above show that the ESR of the HV types is fine even at zero DC.

There were a few meters that allowed HV bias on caps during measurement... A slight modification to the unit would allow this... For example, old Sprague cap testers... One went to 1500V and a later unit TO-6A from the 1960's went to 600V ... This DC voltage was intended for leakage testing...while the Capacitance test applied 35V ac.... With a simple mod to protect the front end, this would allow to use the DC voltage simultaneous with the C test... The main LCR meter I use most often is the HP 4263B ...since the 3577 is a pain in the A$$ to break out for day to day bench measurements....bought it at surplus for a little over $200 and it looks like it was hardly used... The 4263B has some front end protection to allow you to measure a charged cap.... Would need to look at the data sheet to check the limiting test voltage depending on capacitance...they provide a quick formula...
As for capacitor variance....I did mention that depending on the di-electric material and the cap construction and charge density, the electric field can cause a force for the plate attraction to slightly deform the di-electric material.... it is a small effect but has occurred in some caps at higher voltages..paper in wax, paper in oil..some films have slight compression, modulus of elasticity ....Glass caps have practically no effect... These issues were noted at Maxwell R&D Labs during cap design and manufacturing.....
Probably the most noted effects of cap variance would be Temperature and Applied voltage.... The di-electric "constant" is not really constant ...since this is not perfect or linear...
The manufacturers of the dielectric material, such as DuPont, will have data sheets showing the di-electric constant vs temperature.... for example TEFLON will show a negative slope , ie as temperature increases dielectric constant decreases ...for TEFLON the slope is not that steep....figure 2.1 to 1.9 over 20C to 100C .... Other materials can have a steeper slope... There is also a change in di-electric constant withe respect to applied voltage.... The electric field produced polarizes the material depending on charge density and film thickness...construction, package size.. Ceramic caps can such as X5R would be one of the worst offender for voltage gradient vs di-electric constant .....also poor temperature vs di-electric constant...

[printer2]

Manage to find it. I'll just do a copy and past.

I wondered how much I could see if I scope two different capacitors with the same microfarad ratting but different voltage ratings. I wanted to use standard Fender values but the lowest value I had was 25V in 22uF, I also have 22uF at 350V. Now being that 25V is much larger than the voltage generated across a 1.5k resistor with a 12AX7 and a 100k plate resistor I went searching through some of my unused audio equipment and sacrificed one and came up with a 22uF at 6.3V. Now this is not really a scientific test, I had a circuit that I was breadboarding and I just swapped out the capacitors I had and used the 6.3V and the 350V. I switched between the two using a SPDT switch and took some picture of the scope. The pictures are of an 80 Hz signal, the first is the generator at the input jack, the next is the 6.3V and the third is the 350V. I am really surprised at the results as I would have thought the 350V cap would look terrible. Test setup/capacitors

Capacitor Bypass TestSignal Generator



6.3V capacitor



350V capacitor

[Enzo]

And that brings us home.

We can bat back and forth esoteric electronic phenomena, and tiny effects of this and that. And we can speak of specialized testing gear or modifying existing gear to hunt down vanishing changes. But context is everything. The OP wondered if it would change things in his guitar amp if he used a high voltage 20uf cap instead of a 25v 20uf cap for a cathode bypass. He wasn't looking for a parts per million effect. The scope tends to agree with me that he'd never hear the difference.

[Helmholtz]

There were a few meters that allowed HV bias on caps during measurement... A slight modification to the unit would allow this... For example, old Sprague cap testers... One went to 1500V and a later unit TO-6A from the 1960's went to 600V ... This DC voltage was intended for leakage testing...while the Capacitance test applied 35V ac.... With a simple mod to protect the front end, this would allow to use the DC voltage simultaneous with the C test... The main LCR meter I use most often is the HP 4263B ...since the 3577 is a pain in the A$$ to break out for day to day bench measurements....bought it at surplus for a little over $200 and it looks like it was hardly used... The 4263B has some front end protection to allow you to measure a charged cap.... Would need to look at the data sheet to check the limiting test voltage depending on capacitance...they provide a quick formula...

Thanks for your feedback.

I am interested in measurement principles and gear stuff and would love to see the input wiring of a C-meter allowing for HV bias, but I would be even more interested in actual measured values for low/no and HV bias. So if you have some or can produce them, please show.

I thought about wiring a high µF-high voltage cap in series with the meter's input. Something like 2000µ/350V or higher together with some charging resistor and protection diodes for the input. The C-value of the series cap needs to be around 100 times higher than the cap under test to avoid too low measuring results. Otherwise the reading would have to be corrected by some factor taking care of the series C circuit.

The operation manual of the HP 4263B states that the max. bias voltage allowed is 2.5V. To my understanding the higher protection voltage does not mean that it allows measurement with charged caps. In fact it is explicitely stated that charged caps must not be connected to the input terminals.

And please let's not mix e-caps with other types.

[J M Fahey]

There were a few meters that allowed HV bias on caps during measurement...

Exact Brand and Model please.
And copypaste the specific lines of the Manual stating so, your word is not enough. You lost credibility.

A slight modification to the unit would allow this...

If you need to modify something it means it does not do so on its own, and to boot most probably is not even intended (or possible) todo so.

For example, old Sprague cap testers... One went to 1500V and a later unit TO-6A from the 1960's went to 600V ... This DC voltage was intended for leakage testing...

Which is completely different to what you state.

while the Capacitance test applied 35V ac....

With NO HV bias.
Thanks for recognizing your intentional error.

With a simple mod to protect the front end, this would allow to use the DC voltage simultaneous with the C test...

To avoid endless retyping: >>If you need to modify something it means it does not do so on its own, and to boot most probably is not even intended (or possible) todo so.<<

The main LCR meter I use most often is the HP 4263B ...since the 3577 is a pain in the A$$ to break out for day to day bench measurements....bought it at surplus for a little over $200 and it looks like it was hardly used... The 4263B has some front end protection to allow you to measure a charged cap....

So now you are switching away from the test instrument you bragged about and you dropped the HV label too, replacing it with an undefined "charged" label.

Would need to look at the data sheet to check the limiting test voltage depending on capacitance...they provide a quick formula...

Yes, indeed, you would have to look at the datasheet, but more important paste here the lines stating they apply HV to the capacitor under test. Remembering you lost credibility so we need to see some substance backing your statements.

As for capacitor variance....I did mention that depending on the di-electric material and the cap construction and charge density, the electric field can cause a force for the plate attraction to slightly deform the di-electric material....

Who cares? We are not talking about that but E-cap capacitance variation with applied voltage.

it is a small effect but has occurred in some caps at higher voltages..paper in wax, paper in oil..some films have slight compression, modulus of elasticity ....Glass caps have practically no effect... These issues were noted at Maxwell R&D Labs during cap design and manufacturing.....

Who cares? Not the subject here.
Open another thread for them if you wish.

Probably the most noted effects of cap variance would be Temperature and Applied voltage....

Now you digress into Temperature.

The di-electric "constant" is not really constant ...since this is not perfect or linear...

You are switching from an audible phenomenon, what worried the OP, to a vague "philosophical" realm.
Butterfly in China indeed.

The manufacturers of the dielectric material, such as DuPont, will have data sheets showing the di-electric constant vs temperature.... for example TEFLON will show a negative slope , ie as temperature increases dielectric constant decreases ...for TEFLON the slope is not that steep....figure 2.1 to 1.9 over 20C to 100C .... Other materials can have a steeper slope...

Absolute waste of bandwidth and server space, we are not talking about that in this thread, you are just trying to deviate course for your own purposes.

There is also a change in di-electric constant withe respect to applied voltage.... The electric field produced polarizes the material depending on charge density and film thickness...construction, package size..

Please focus on E-Caps and provide specific data.

Ceramic caps can such as X5R would be one of the worst offender for voltage gradient vs di-electric constant .....also poor temperature vs di-electric constant...

Not talking Ceramics here.

I suggest you add these instead.

Very unrelated of course but at least what you probably dream about:



Hey!!! I bet you can connect a Million of these in parallel and use them as a cathode bypass!!!

A couple days ago a Forum member wanted to connect different cathode caps and worried about wiring, this way he could vary them remotely and without clicks or pops, think about the possibilities !!!