Does cap voltage affect tone?

[SoulFetish]

I don't think you guys are being fair to cerrem, his credentials seem good http://www.merrenaudio.com/home
I was hoping to get some insight into his take on things.
.

I think your right, ‘64. FWIW, I respect him, I like Merrin’s site, and even corresponded a few years back about maybe winding me a Transformer. I found him pleasant and would have no problem working with him on a project. I’m a little disappointed though, I think hes being unnecessarily defensive and his response to Fahey was really disrespectful and uncalled for.
We have people who come from many different backgrounds, with expertise in a variety of disciplines. if someone make an error or hasn’t yet learned something you might have already learned, don’t be an asshole. At some point you didn’t know before you learned. Pretty simple.
Anyways, im not one to hold a grudge.

[Helmholtz]

I think your right, ‘64. FWIW, I respect him, I like Merrin’s site, and even corresponded a few years back about maybe winding me a Transformer. I found him pleasant and would have no problem working with him on a project. I’m a little disappointed though, I think hes being unnecessarily defensive and his response to Fahey was really disrespectful and uncalled for.
We have people who come from many different backgrounds, with expertise in a variety of disciplines. if someone make an error or hasn’t yet learned something you might have already learned, don’t be an asshole. At some point you didn’t know before you learned. Pretty simple.
Anyways, im not one to hold a grudge.

The exchange of technical arguments de facto shows respect.

[SoulFetish]

The exchange of technical arguments de facto shows respect.

Great Point!

[nevetslab]

Cap voltage rating effects in the impedance & phase response on electrolytic caps

In reference to the original post asking if capacitor voltage rating affects tone, where in a circuit it is placed, as well as the dielectric type, it can. When the question posed of using a 1uF/10V electrolytic bypass cap on a cathode resistor, and instead, using a much higher voltage (and physical size) cap of 1uF/450V, would that sound any different?

Perhaps. Pursuing this thought with the aid of a 2-Ch FFT Analyzer, set up to plot impedance, as well as plotting phase response of a capacitor, I found a very interesting difference. In the examples that follow, I was using 22uF as the cap value, and selected several types:

Nichicon VX series 22uF/63V Axial Electrolytic
Nichicon VX series 22uF/100V Axial Electrolytic
Nichicon PW series 22uF/100V Radial Electrolytic
F & T 22uF/500V Axial Electrolytic
Multicap 10uF/200V Polypropylene Axial Film



Normally, in comparing impedance plots of caps, the main difference we see is where the cap approaches the ESR value of the cap, and seriously deviates from the ‘straight line’ slope that looks like a decreasing 6dB/Oct line in the impedance. This usually takes place above 10kHz. Looking at the phase response of the current thru the cap in the test circuit shows significant changes well within the audio range. The phase response can be thought of as a plot of Dissipation Factor, which IS a non-linearity attribute in capacitors.

In Helmholtz’ post # 27 on 7/08/18 where he set up his Vellman PCSU200 USB Scope/Generator, set to plot impedance, it made me want to go dig out my HP 3575A Gain/Phase Meter to set up with my Vellman USB scope/generator rig, and add the phase response to the impedance plots.

Prior to getting at that instrument, I pulled up my Wavetek 5830A 2-Ch FFT Analyzer, and set it up to display Impedance/Phase plots of the caps listed above. What really surprised me was the plot of the F & T 22uF/500V Axial Electrolytic. Normally used in high charging current applications in the power supply of tube amps, here in this setup, there’s only about 1mA of sine wave current flowing thru the cap. I didn’t bother with a constant current signal source, but just used a 1k resistor following the output of the Tracking Generator’s 1V RMS source to approximate the constant current in the test.

With that small amount of current flowing thru the F & T cap, it’s phase curve as well as it’s impedance curve was far worse than any of the small caps we’d normally be using as a cathode bypass part. Brand new cap. I haven’t done any other measurements on it yet, but it was surprising. In this instance, I’d expect one might hear the difference using that cap instead of any of the physically small 22uF caps, but it still has a resistor across it (cathode bypass cap), so hard to say.

Looking at the two Nichicon VX series 22uF axial lead caps, there IS a significant improvement in the linearity of the cap just in selecting a 100V vs 63V rating. In my normal component selection on caps, I’ll normally select a higher voltage rating than one closer to the applied voltage in circuit, just due to this improvement. Space constraints not withstanding, of course.

I also plotted a 10uF/200V polypropylene axial film cap, just to see how it appeared in the test setup. With it, the impedance plot is ruler flat, and the phase plot hugs the 90 deg reference line quite closely, showing very low non-linearity, typical of film caps vs electrolytic caps.



Somewhere I recall reading a test procedure for dielectric absorption that was done in a non-steady-state fashion, where after the test signal stopped, what was heard was the energy not given up by the cap, and ‘appearing’ out of time with the test signal. It’s that dielectric absorption characteristic that has the tendency to first, NOT accept ALL of the applied charge, and, what has been accepted, DOESN’T give it all back in time. There’s some small amount of energy that gets trapped in the dielectric, and has the effect of ‘time smearing’ and sloshing around inside the cap, so to speak. We’ve all seen voltage recovery in electrolytic caps, after fully discharging them, they still have residual charge that reappears, unless there is a load across it to keep it discharged.

But, back to reality…..Enzo said it best: Context is Everything.

[Enzo]

Thanks for all that effort. Interesting.

When testing those caps, did we have a 1500 ohm resistor in parallel in all cases? That being a very common cathode resistor value.

I am not wrapping my head around how these differences would make sound different, and to what extent, in a circuit.

[nevetslab]

In the test circuit, the cap under test was only bridged by the input impedance of the Diff Amp of the analyzer....I think 100k but might be 1M. Cap was fed via 1k resistor from a 1V swept sine signal, one of many available sources on the menu. I haven't done further tests to see how the RC circuit reacts. I'll have a look at that too. I tend to think the cathode resistor would swamp out what we see on just the cap's characteristics.

[Enzo]

My thinking too, that whatever subtle cap differences are tiny when shunted by the resistor. I imagine something like changes the rolloff from 9Hz to 8.67Hz. Or something.

[Helmholtz]

Cap was fed via 1k resistor from a 1V swept sine signal, one of many available sources on the menu

In this setup the caps are effectively shunted by the 1k series resistor + generator output impedance (probably 50 Ohms).

I wonder if the F&T cap needs forming. Its curve looks more like a cap with a low value (leakage) resistor in parallel. It doesn't show the typical capacitive -6dB/octace or -10dB/decade slope like all my measurements did. Would be interesting to know exact C and ESR values.
Also I have found that the performance of somewhat degraded e-caps instantly improves, when measured with a DC bias of 2V.

Some info:
https://www.dfrsolutions.com/hubfs/R...=1528744207492

[nevetslab]

I haven't set up my gear to measure ESR, but with the GenRad 1617A Capacitance Bridge on the bench, I've got the F & T cap used in the impedance/phase plots inserted, with 400VDC bias applied. 22.4uF & DF of 0.07, 120Hz generator voltage applied about 2V. In reading the leakage, with 400VDC applied, it first showed around 6mA leakage, but has been slowly falling, now around 1.1mA. I'll let it form for awhile.

After I had let the cap form for 40 min, I discharged it, removed it from the GR 1617A and moved it over to my GenRad 1658 Digibridge to get two sets of readings, that @ 120Hz, showing 21.9uF Cs & 21.77 Cp with DF of 0.0768. @ 1kHz, they changed to 21.05uF Cs, DF of 0.4212, and 17.88uF Cp, DF of .4204. The change in DF with respect to frequency does track the increase in phase response on the plot.

In the GenRad 1617A manual, a calculation for ESR is given as D/(2*pi*F*Cs). For 120Hz, using the numbers on the 1658 Digibridge, I get:
0.0768/(2*3.14*21.9uF*120Hz) = 4.651 ohms, and at 1khz, I get :0.4204/(2*3.14*21.05uF*1kHz) = 3.18 ohms.

I’ll compare that with the values I get from my ESR measurement set-up. It seems mighty high of an ESR value, though the resonance shelf it leveled off at on the plots against the smaller 22uF caps I plotted, it does seem reasonable.

Looking at the best of the 22uF caps I ran…the VX series 22uF/100V, using the numbers from the GenRad 1658 bridge, I get 21.09uF Cs, 0.0194DF @ 120Hz & 19.66uF Cs, 0.0595DF @ 1khz. Using the ESR calculation, I get:
0.0194/(2*3.14*21.09uF*120Hz) = 1.22 ohms, and at 1kHz, I get:
0.0595/(2*3.14*19.66uF*1kHz) = 0.482 ohms

If I had set the frequency span lower, say 0-5khz instead of 0-50khz, we would have see more of the -10dB/decade slope up to where it departs heading towards the ESR and internal resonance. I had merely grabbed what I thought would have been a good 'typical' example of a high voltage rated 22uF cap in the test set up, and we saw a head-scratcher result.

On the plots I ran, it was set up to show trend behavior, and wasn't calibrated with some known resistor plots to give scale in the impedance domain.

I've added some pages from Nichicon's Aluminum Electrolytic Tech Notes, and a page from the GenRad 1617A Capacitance Bridge, showing relationships:

Nichicon_TechNotes_Al_Electrolytics-1.pdf
Nichicon_TechNotes_Al_Electrolytics-2.pdf
GR1617-relationships.pdf

I had made the comment about the phase plot providing information on Dissipation Factor vs Frequency, where DF = Tan of Theta x 100 (%). As the frequency increases, the phase angle increases, which results in the increase of DF. I found on Nichicon's VX series data sheet, the Tan of Theta steadily decreases with voltage rating until 100V, and above 100V, it increases again.

How important is DF to the sound of capacitors? Well, it is a non-linear characteristic in capacitors. In the case of the cap being a cathode resistor bypass cap, probably not much. If, on the other hand, it was a signal coupling cap from a cathode follower circuit, probably more so. Usually we're choosing an appropriately rated film cap there instead of an electrolytic. One reason is their losses/non-linearities are much lower.

If it were a power supply cap, parts with high DF over time will tend toward failure from increase in self-heating of the ESR in the cap.