In addition to the excellent Bateman articles, if you would like to read the original groundbreaking article on the subject of capacitor dielectric materials and the effect on audio circuitry, I suggest the following from 1980:
Picking Capacitors - Walter G. Jung and Richard Marsh
At least one and probably both of the authors "stop by" this board from time to time (probably sometimes thinking," O what have we done!")
Picking Capacitors - Walter G. Jung and Richard Marsh
At least one and probably both of the authors "stop by" this board from time to time (probably sometimes thinking," O what have we done!")
The links to the Cyril Bateman Articles in Electronics World Magazine move around from time to time so you will need to do a search:
Measuring Speaker Cables: 1 Cyril Bateman Dec 1996 p925
Measuring Speaker Cables: 2 Cyril Bateman Jan 1997 p52
Measuring Speaker Cables: 3 Cyril Bateman Feb 1997 p119
Capacitor Distortion: Part 1 Cyril Bateman July 2002 p12 S
Capacitor Distortion: Part 2 Cyril Bateman Sept 2002 p16 S
Capacitor Distortion: Part 3 Cyril Bateman Oct 2002 p12 S
Capacitor Distortion: Part 4 Cyril Bateman Nov 2002 p40 S
Capacitor Distortion: Part 5 Cyril Bateman Dec 2002 p44 S
Capacitor Distortion: Part 6 Cyril Bateman Jan 2003 p44 S
Capacitor Sounds II: Part 1
Real-Time Hardware Cyril Bateman July 2003 p36 S
Capacitor Sounds II: Part 2
Distortion v Time v Bias Cyril Bateman Aug 2003 p46 S
Capacitor Sounds II: Part 3
Distortion Meter Cyril Bateman Sept 2003 p46 S
Capacitor and Amplifier Distortions Cyril Bateman Nov 2003 p44 S
Measuring Speaker Cables: 1 Cyril Bateman Dec 1996 p925
Measuring Speaker Cables: 2 Cyril Bateman Jan 1997 p52
Measuring Speaker Cables: 3 Cyril Bateman Feb 1997 p119
Capacitor Distortion: Part 1 Cyril Bateman July 2002 p12 S
Capacitor Distortion: Part 2 Cyril Bateman Sept 2002 p16 S
Capacitor Distortion: Part 3 Cyril Bateman Oct 2002 p12 S
Capacitor Distortion: Part 4 Cyril Bateman Nov 2002 p40 S
Capacitor Distortion: Part 5 Cyril Bateman Dec 2002 p44 S
Capacitor Distortion: Part 6 Cyril Bateman Jan 2003 p44 S
Capacitor Sounds II: Part 1
Real-Time Hardware Cyril Bateman July 2003 p36 S
Capacitor Sounds II: Part 2
Distortion v Time v Bias Cyril Bateman Aug 2003 p46 S
Capacitor Sounds II: Part 3
Distortion Meter Cyril Bateman Sept 2003 p46 S
Capacitor and Amplifier Distortions Cyril Bateman Nov 2003 p44 S
Well it seems my foolishness has engendered some great material and resources from you folks. Thank-you so much! I have read the Jung and Marsh article before - influenced my decisions.
I'm going to dive into this head-on and get to the bottom of understanding the results I got with my silly-'scope measurements, and I'll report back. That's what I'm doing tonight. No wife/kids, just electronics...geez! Sigh.
I'm going to dive into this head-on and get to the bottom of understanding the results I got with my silly-'scope measurements, and I'll report back. That's what I'm doing tonight. No wife/kids, just electronics...geez! Sigh.
By the way DF96 - what terrific articles Bateman wrote! Here's the first one:
http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Cap_Sound_Part_1.pdf
Thanks,
Mike
http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Cap_Sound_Part_1.pdf
Thanks,
Mike
I've no idea what circuit you're using but you can't drive a large capacitor with a 10 kHz square (or any other) wave because it looks pretty much like a short circuit at that frequency. At best, you're probably burning in (burning up?) your signal generator.
Ignoring the whole issue of whether burn-in does anything, caps in audio circuits should not have any audio voltage across them, unless used for eq like RIAA phono circuits. Coupling caps in particular need to be large enough so as not to have significant impedance at audio frequencies.
That means that burn-in by trying to impress a large audio signal across a coupling cap is a fundamentally flawed concept. At best you can limit the current and force some current back and forth. Possibly you should be measuring current? I mention this because understanding the current flow will be a huge educational experience. I have a simple test with one resistor that will show you the difference between electrolytics, though probably not the films you're using without some exotic equipment.
IMO, the boutique caps people talk about are often too small for the application. They get chosen due to availability of Teflon or other dielectrics in practical values, not necessarily what the circuit really needs. It's not surprising they sound different when people often compare different value caps in a region where they affect the response and phase.
Ignoring the whole issue of whether burn-in does anything, caps in audio circuits should not have any audio voltage across them, unless used for eq like RIAA phono circuits. Coupling caps in particular need to be large enough so as not to have significant impedance at audio frequencies.
That means that burn-in by trying to impress a large audio signal across a coupling cap is a fundamentally flawed concept. At best you can limit the current and force some current back and forth. Possibly you should be measuring current? I mention this because understanding the current flow will be a huge educational experience. I have a simple test with one resistor that will show you the difference between electrolytics, though probably not the films you're using without some exotic equipment.
IMO, the boutique caps people talk about are often too small for the application. They get chosen due to availability of Teflon or other dielectrics in practical values, not necessarily what the circuit really needs. It's not surprising they sound different when people often compare different value caps in a region where they affect the response and phase.
Hi Conrad, thanks for the informative post. I would be interested in the current test you mentioned. Can you post the specifics?
I've got my signal generator hooked up to a mic preamp, so I don't think I'm putting it at risk (it's going into a Green pre I built from PeterC at groupdiy.com).
I am curious then why many claim that their systems sound different after caps have "burned in" for a couple hundred hours...
Thanks,
Mike
I've got my signal generator hooked up to a mic preamp, so I don't think I'm putting it at risk (it's going into a Green pre I built from PeterC at groupdiy.com).
I am curious then why many claim that their systems sound different after caps have "burned in" for a couple hundred hours...
Thanks,
Mike
Last edited:
Here's a fun test that will teach you a lot about capacitors. I'll try to do it with words alone...
Get yourself 3 aluminum electrolytic capacitors, say 220 uF @ 6.3 volts, 220 uF @ 25 volts and 220 uF @ 100 volts.
Hook up a cap in series with a 2 kohm resistor. Hook your signal generator to the pair with ground at the cap end and signal at the resistor end. Hook your scope ground to that same ground end of the cap. Hook the probe to the junction of the resistor and cap- that should be the only thing connected there. The scope is just showing you the voltage across the cap.
Drive the mess with a bipolar square wave at a low frequency and about 20 volts peak to peak. Turn the gain of the scope all the way up. You'll see a series of spikes for each edge of the square wave.
Increase the frequency until the waveform becomes a triangle wave. Too high and the typical scope won't have enough gain to see a good signal. I actually do this with an extra preamp on the scope because we're looking at mV.
What you have is a simple integrator. When the square wave is positive the cap charges up. When the square wave is negative the cap discharges down. It's fairly linear because we're way down on the RC curve. IOW, the observed voltage is way smaller than the applied voltage.
Using the 6.3 volt cap, look at the triangle wave. Note that the peaks don't line up- there's a straight line portion. Compare the results between the three caps.
Tell me what you see and why you think it happens. Tell me about the linearity of the waveform sides. In theory this will work with film caps, but the differences are way too small to see directly. We can solve that too, but for now electrolytics are more fun.
Get yourself 3 aluminum electrolytic capacitors, say 220 uF @ 6.3 volts, 220 uF @ 25 volts and 220 uF @ 100 volts.
Hook up a cap in series with a 2 kohm resistor. Hook your signal generator to the pair with ground at the cap end and signal at the resistor end. Hook your scope ground to that same ground end of the cap. Hook the probe to the junction of the resistor and cap- that should be the only thing connected there. The scope is just showing you the voltage across the cap.
Drive the mess with a bipolar square wave at a low frequency and about 20 volts peak to peak. Turn the gain of the scope all the way up. You'll see a series of spikes for each edge of the square wave.
Increase the frequency until the waveform becomes a triangle wave. Too high and the typical scope won't have enough gain to see a good signal. I actually do this with an extra preamp on the scope because we're looking at mV.
What you have is a simple integrator. When the square wave is positive the cap charges up. When the square wave is negative the cap discharges down. It's fairly linear because we're way down on the RC curve. IOW, the observed voltage is way smaller than the applied voltage.
Using the 6.3 volt cap, look at the triangle wave. Note that the peaks don't line up- there's a straight line portion. Compare the results between the three caps.
Tell me what you see and why you think it happens. Tell me about the linearity of the waveform sides. In theory this will work with film caps, but the differences are way too small to see directly. We can solve that too, but for now electrolytics are more fun.
Last edited:
I think they call that the "psychoacoustic effect," or in other words, you'll find what you're looking for, facts be damned.
Thanks Conrad, I'll see if I have the necessary values and give it a shot. I appreciate your wisdom. I'll report back findings if I'm able to do the tests.
OK this really getting to me. In an effort to redeem myself in the eyes of my peers, I have undertaken to calculate what is going on in my setup. The calculated values are quite different than my measured values.
I've got a 1.006uF cap - who cares the brand - that value vetted by my Fluke 87V.
I've got a 100R resistor, 1%, that value vetted exactly on the Fluke.
I've got a 10kHz square wave, value again vetted on the Fluke and also on the digital sig gen.
I've got a 10.12v Voltage coming from my mic preamp, again value vetted on the Fluke.
Please check my calculations.
X(c) (reactive capacitance of capacitor) = 1/(2 x pi x freq x capacitance) = 1/(2 x 3.14 x 10^-4 x 10^-6 = 15.9R
Z (total impedance) = sqrt(R^2 + Xc^2) = sqrt(100^2 + (15.9)^2) = 101.3R
I = E/Z
I = 10.12v/101.3R = .0999A = .1A
Vc (Voltage across the cap) is given by Vc = Xc x I = 15.9R x .1A = 1.59V
For the resistor, we know V=IR. Therefore Vr = .1A x 100R = 10v
OK. Keep in mind I'm measuring across the cap, cap leg to ground. On my Fluke 87V I get the following:
Vc = 0.47v instead of 1.59v (yes I have the meter set to AC V!)
I - I can't seem to measure this at all. Yes I broke the circuit completely and put the meter in series with the circuit, but can't get a reading on my Fluke. Dunno.
Vr (voltage across resistor) measured to 3.16v.
I'm not an idiot - but I must be missing some obvious or stupid things. Can anyone share any thoughts as to why my measurements would be so difft from the calculated values?
In case you're thinking maybe since I'm driving a square wave the results would be difft, I tried setting the sig gen to sine wave instead of square. For the Vr I got 2.93v instead of the 3.16v...still way off from the calculated 10v! Sigh. Thx for the eyes of wisdom.
I've got a 1.006uF cap - who cares the brand - that value vetted by my Fluke 87V.
I've got a 100R resistor, 1%, that value vetted exactly on the Fluke.
I've got a 10kHz square wave, value again vetted on the Fluke and also on the digital sig gen.
I've got a 10.12v Voltage coming from my mic preamp, again value vetted on the Fluke.
Please check my calculations.
X(c) (reactive capacitance of capacitor) = 1/(2 x pi x freq x capacitance) = 1/(2 x 3.14 x 10^-4 x 10^-6 = 15.9R
Z (total impedance) = sqrt(R^2 + Xc^2) = sqrt(100^2 + (15.9)^2) = 101.3R
I = E/Z
I = 10.12v/101.3R = .0999A = .1A
Vc (Voltage across the cap) is given by Vc = Xc x I = 15.9R x .1A = 1.59V
For the resistor, we know V=IR. Therefore Vr = .1A x 100R = 10v
OK. Keep in mind I'm measuring across the cap, cap leg to ground. On my Fluke 87V I get the following:
Vc = 0.47v instead of 1.59v (yes I have the meter set to AC V!)
I - I can't seem to measure this at all. Yes I broke the circuit completely and put the meter in series with the circuit, but can't get a reading on my Fluke. Dunno.
Vr (voltage across resistor) measured to 3.16v.
I'm not an idiot - but I must be missing some obvious or stupid things. Can anyone share any thoughts as to why my measurements would be so difft from the calculated values?
In case you're thinking maybe since I'm driving a square wave the results would be difft, I tried setting the sig gen to sine wave instead of square. For the Vr I got 2.93v instead of the 3.16v...still way off from the calculated 10v! Sigh. Thx for the eyes of wisdom.
1. DMMs don't always work too well at higher frequencies. Some do, many don't.
2. DMMs almost never give the correct reading for a waveform other than a sine wave. In most cases what they actually measure is the average value of full-wave-rectified AC but then they display what would be the corresponding RMS value for a sine wave (by multiplying by approximately 1.1). This calculation is only correct for a sine wave so for all other waveshapes you get a wrong result.
3. Your calculation applies to the fundamental 10kHz only. You need to repeat for all the harmonics in the square wave: 30kHz, 50kHz etc. then add up the results taking account of phase. Not easy. Just treat your calculation as an approximation, or alternatively do a time domain calculation by looking at the charging and discharging of the cap when fed by an approximate square wave of current.
2. DMMs almost never give the correct reading for a waveform other than a sine wave. In most cases what they actually measure is the average value of full-wave-rectified AC but then they display what would be the corresponding RMS value for a sine wave (by multiplying by approximately 1.1). This calculation is only correct for a sine wave so for all other waveshapes you get a wrong result.
3. Your calculation applies to the fundamental 10kHz only. You need to repeat for all the harmonics in the square wave: 30kHz, 50kHz etc. then add up the results taking account of phase. Not easy. Just treat your calculation as an approximation, or alternatively do a time domain calculation by looking at the charging and discharging of the cap when fed by an approximate square wave of current.
Wishfull thinking?
Marketing hyperbole?
They're messing with you. Sort of like telling somebody they need a muffler bearing (exhaust belt, blinker fluid, etc.) and sending them to an auto parts store to get one.
All the basic calcs only work with sine waves. If you can't get current readings, check the meter fuse and where the leads are plugged in. Somewhere up there I saw 0.1 amps. Few signal generators and no mic preamps I know of will deliver that kind of current. You may be operating at impedances far below what your sources can cleanly (or without burning up) deliver. Remember, a good opamp usually delivers about 20 mA with the wind to its back on a good day, though there are some that certainly can do more.
OK after an entire evening of googling and experimenting, I am pleased to report some of my findings match what I expect.
First, I was concerned by Conrad's comment that I may have fried my Green Pre. So after learning about input/output impedances, I settled on using a 600R resistor for the caps - not because of the 600R typical loading but because it would give me a preamp output below 20mA which was in the right range of not blowing up my pre.
When I redid the calculations, as posted above but with a .01uF cap and 600R resistor, I found I should get:
3.53v across the resistor. I measured 3.7v
9.36v across the capacitor. I measured 9.03v.
I will continue to investigate and post, but I'm sure this is eye-rollingly simple stuff to you guys who know electronics well. But it's an education for me, and maybe someone benefits from the sharing. Next up will be the .1uF and the 1uF.
First, I was concerned by Conrad's comment that I may have fried my Green Pre. So after learning about input/output impedances, I settled on using a 600R resistor for the caps - not because of the 600R typical loading but because it would give me a preamp output below 20mA which was in the right range of not blowing up my pre.
When I redid the calculations, as posted above but with a .01uF cap and 600R resistor, I found I should get:
3.53v across the resistor. I measured 3.7v
9.36v across the capacitor. I measured 9.03v.
I will continue to investigate and post, but I'm sure this is eye-rollingly simple stuff to you guys who know electronics well. But it's an education for me, and maybe someone benefits from the sharing. Next up will be the .1uF and the 1uF.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.