Would you kindly state your references? I'm not doubting your ability to summarize the papers, but I always try to get back to the original source when possible.
Thanks,
~Tom
Tom - no probs - it's an interesting topic - and my forum posting is not peer reviewed before replying and sometimes a little rushed so please bear with the inconsistancies.
The Service Life of Large Aluminum Electrolytic Capacitors: Effects of Construction and Application. J.L. Stevens, J.S. Shaffer, and J.T. Vandenham. 2002/
This paper presents C and ESR and leakage variation results for different bias voltage conditions at 95C conditions. This is the only modern paper I've seen that raises the topic of different bias voltage conditions.
The Chemstry of failure of aluminium electrolytic capacitors. R.S.Alwitt & R.G.Hills. 1965.
Provides general chemistry appreciation of typical (but probably now far surpassed) cap technology with respect to capacitance variation, but may not reflect modern cap characteristics.
Online Monitoring Method and Electrical Parameter Ageing Laws of Aluminium Electrolytic Capacitors Used in UPS. K. Abdennadher et al. 2009.
This paper indicates the typical C and ESR trends over service life.
Simple Experimental Techniques to Characterize Capacitors in a Wide Range of Frequencies and Temperatures. A.M. R. Amaral & A.J.Marques Cardoso. 2010.
This paper indicates the % change in C that can occur with temp - which is an influence that may be part of the change observed in audible character during the burn-in procedure that people have used.
Aardvark, I think your post is more about the engineering competance of a design, rather than the influence of conditions and time on a particular part (albeit possibly a part that wouldn't be selected by most engineers, but has for some reason been selected by others).
Ciao, Tim
The Service Life of Large Aluminum Electrolytic Capacitors: Effects of Construction and Application. J.L. Stevens, J.S. Shaffer, and J.T. Vandenham. 2002/
This paper presents C and ESR and leakage variation results for different bias voltage conditions at 95C conditions. This is the only modern paper I've seen that raises the topic of different bias voltage conditions.
The Chemstry of failure of aluminium electrolytic capacitors. R.S.Alwitt & R.G.Hills. 1965.
Provides general chemistry appreciation of typical (but probably now far surpassed) cap technology with respect to capacitance variation, but may not reflect modern cap characteristics.
Online Monitoring Method and Electrical Parameter Ageing Laws of Aluminium Electrolytic Capacitors Used in UPS. K. Abdennadher et al. 2009.
This paper indicates the typical C and ESR trends over service life.
Simple Experimental Techniques to Characterize Capacitors in a Wide Range of Frequencies and Temperatures. A.M. R. Amaral & A.J.Marques Cardoso. 2010.
This paper indicates the % change in C that can occur with temp - which is an influence that may be part of the change observed in audible character during the burn-in procedure that people have used.
Aardvark, I think your post is more about the engineering competance of a design, rather than the influence of conditions and time on a particular part (albeit possibly a part that wouldn't be selected by most engineers, but has for some reason been selected by others).
Ciao, Tim
I think the characteristics of burn in are supposed to be:
Improved low frequency
More clarity of imaging and less hash.
Sound shift to "2nd harmonic feel". <<Sounds like.
Dielectric definitely creates a different "sound" PTFE / PVC / paper / polypropylene / Teflon. So would a change in dielectric properties within the cap create a difference? First thoughts are ESR.
Regards
M. Gregg
Improved low frequency
More clarity of imaging and less hash.
Sound shift to "2nd harmonic feel". <<Sounds like.
Dielectric definitely creates a different "sound" PTFE / PVC / paper / polypropylene / Teflon. So would a change in dielectric properties within the cap create a difference? First thoughts are ESR.
Regards
M. Gregg
Just a thought,
If we take two polypropylene caps of the same make e.g. LCR standard 100nF and their audio grade version of the same value again 100nF. The audio grade is physically bigger and seems to have less "edge" on high frequency also more detail is evident what is different if it s 100nF ? If we can hear the difference in the caps then is our equipment fauty somehow?
Regards
M. Gregg
If we take two polypropylene caps of the same make e.g. LCR standard 100nF and their audio grade version of the same value again 100nF. The audio grade is physically bigger and seems to have less "edge" on high frequency also more detail is evident what is different if it s 100nF ? If we can hear the difference in the caps then is our equipment fauty somehow?
Regards
M. Gregg
Last edited:
Although it depends on the exact construction of the cap (which dielectric is used, etc) manufacturers are very much aware of the changes in values from manufacturing, to transit time to the end user. There is a standard, IEC-384-9, which basically says capacitors should be rated based on the value expected at 1000 hours (approx 6 weeks) from manufacture date. The as-manufactured values are often different if measured immediately than what the cap is expected to be, or labeled to be, in the end user's hands.
Most, of not all, manufacturers subject capacitors to a de-aging process to account for this, for example, if capacitors are shipped from factory stock some time after manufacture. The end user can also subject a capacitor to a de-aging process explicitly or inadvertently (eg some soldering operations) if for some reason you want to revert the device to the original parameters.
See:
http://www.digikey.com/Web Export/Supplier Content/TDK_445/PDF/TDK_cap.pdf?redirected=1
and:
http://www.johansondielectrics.com/technical-notes/general/ceramic-capacitor-aging-made-simple.html
(almost every capacitor manufacturer will have similar documents in their reference material library, if you choose to seek out a specific brand's take on the matter).
Last edited:
Hello M Gregg,
You are an EE type of guy. I agree. A ESR delta will change the in circuit time constant with resulting frequency and phase response changes.
Leaving the same capacitor in the circuit and adjusting a resistor value may have a similar sonic effect as rolling capacitors in and out.
DT
All just for fun!
These last 2 post are only your opinion, and as such are worth very little. Dose any one have any real data? I see no response from the real EEs ( where a bad cap can kill someone ) which makes me think this cap burn in is in the same category as other audiophool time wasters. What electrical parameters ( you know, the ones that actually make a difference ) change with burn in? How much do they change? At least do a real blind listening teat.
Opinions are like a** h***s, eveyones got one.
Opinions are like a** h***s, eveyones got one.
I had read the abstract of this paper a while ago, but just now found the actual paper inself. It covers test results of some interesting in-circuit conditions primarily related to avionics and their affect on electrolytic capacitor aging:
http://www.vuse.vanderbilt.edu/~kou..._Kulkarni_Biswas_Koustoukos_Celaya_Goebel.pdf
It's a brief, easy read, and doesn't directly address every question raised in the OP's question and comments so far, but there are some good references listed which might prove useful for further reading.
http://www.vuse.vanderbilt.edu/~kou..._Kulkarni_Biswas_Koustoukos_Celaya_Goebel.pdf
It's a brief, easy read, and doesn't directly address every question raised in the OP's question and comments so far, but there are some good references listed which might prove useful for further reading.
Last edited:
Good, some real data. I guess the first thing is to recognize "burn in " as initial aging. Capacitors decrease value with age. (the rate depends on type of cap, dielectric, temp etc) The first ten hours being most critical. Since the manufacturers know this, they make the initial capacitance higher than its suppossed to be so that after 1000hrs (a number they picked out of a hat) it will be within spec. It sounds like most caps will be in spec after 10 to 100 hours but there covering there butts. The thing DIYs have to know is that any heat process (soldering, glue curing) will "de-age" the cap and increase its tolerance.
So we know that caps do "burn in" but please call it what it is ; aging (it continues after "burn in".
It effects there capacitance, which decreases with age. Can you hear it? I doubt it. (just my opinion). Heres why. The aging graph in the paper shows a worst case cap change of about 8% after 1000hrs. Most caps off the shelf have less tolerance than that. That would change a LPF from 1k to 1.1k. I doubt anyone here would hear that kind of change.
Last edited:
Thanks for the links Johnny2Bad. The Experimental Studies on Electrolytics paper shows ESR and Capacitance aging variation results equivalent to what has been identified by a few other authors, and has recently been put to practical usage in smps for generating maintenance/service alarms and health status on capacitor life.
My view of phrarod's request was more related to variation in electrolytic capacitor performance in the first few hundred hours of operation. Which is a topic that is exemplified in your excellent links to ceramic dielectric caps, however phrarod's interest was with electrolytic caps - quite a different beast.
I may have been a bit short on support for discounting ESR variation as a siginificant contributor to audible variation with electrolytic use in cathode bypass application. I think most here will appreciate that if the level of capacitance is such as to cause the filter corner to a very low frequency, then variation in a typically very low ESR per se would most likely have negligible effect, as per M.Greggs view.
My view of phrarod's request was more related to variation in electrolytic capacitor performance in the first few hundred hours of operation. Which is a topic that is exemplified in your excellent links to ceramic dielectric caps, however phrarod's interest was with electrolytic caps - quite a different beast.
I may have been a bit short on support for discounting ESR variation as a siginificant contributor to audible variation with electrolytic use in cathode bypass application. I think most here will appreciate that if the level of capacitance is such as to cause the filter corner to a very low frequency, then variation in a typically very low ESR per se would most likely have negligible effect, as per M.Greggs view.
Last edited:
Hello All,
Where opinion ends and speculation begins, or food for thought!
Considering the cathode bypass capacitor the only resistance between the capacitor and the cathode is the ESR of the capacitor. That ESR changes with frequency. Any delta between brands or even the plus minus of “identical” capacitors is huge by percentage and likely has sonic effects. Folks claim to hear differences all day data or not.
Try this, I call it the little equalizer. Place a 5 or 10 Ohm resistor in series with your cathode bypass capacitor. Besides adding just a little degenerative feedback it does smooth out variations of the ESR across the bandwidth and variations between capacitors.
DT
All just for fun!
Where opinion ends and speculation begins, or food for thought!
Considering the cathode bypass capacitor the only resistance between the capacitor and the cathode is the ESR of the capacitor. That ESR changes with frequency. Any delta between brands or even the plus minus of “identical” capacitors is huge by percentage and likely has sonic effects. Folks claim to hear differences all day data or not.
Try this, I call it the little equalizer. Place a 5 or 10 Ohm resistor in series with your cathode bypass capacitor. Besides adding just a little degenerative feedback it does smooth out variations of the ESR across the bandwidth and variations between capacitors.
DT
All just for fun!
What effect does the input/output impedance of the cathode have, the degen resistor? I dont know tubes. If someone can show me an equivalent cct (tube as impedances and voltage gain elements) I can show you how much ESR affects the response. Or someone can use a simulator. I bet the ESR makes little difference. (though i have lost bets before, thats why I prefer sims to thought experiments)
cbdb, a very useful source of detailed explanation of typical amp operation is obtained in Merlin's books, and he has provided some sections on his website - the following link is the most relevant to start an assessment of the impact of ESR on stage response for a simple triode stage.
http://www.freewebs.com/valvewizard1/Common_Gain_Stage.pdf
Ciao, Tim
http://www.freewebs.com/valvewizard1/Common_Gain_Stage.pdf
Ciao, Tim
Hello,
Berkeley Spice only goes so far. Spice teaches theory and will confirm the results of your thought experiments. Even with the best models a 10.00 uf capacitor is a 10.00 uf capacitor. The golden ears among us would readily report a difference that Spice would be blind to.
Modeling a tube it is much the same a modeling a JFET circuit things like grid, gate or drain, plate. Same Same. Effects of ESR will be tiny for both. Real world hearing is more sensitive.
Real world Jensen Paper in Oil costs more and sound better than metalized polypropylene just ask.
Please do simulate a small value resistor in series with that brand X capacitor it is an equalizer for varying value ESR in capacitor. Does your generic capacitor model include a provision for ESR?
DT
All just for fun!
If ESR is much much smaller than 1/gm then it will make no difference. If it makes no difference then changes in it will make no difference.
If ESR is small compared to 1/gm (but bigger than much much smaller) then it might make a small difference, but changes in ESR (due to ageing etc.) will probably be smaller still. If a change in ESR is much much smaller than 1/gm then it will make no difference.
If you want to hear capacitors, this is how to do it:
- avoid negative feedback, or have just enough to cause problems (i.e. not enough to solve problems)
- use an electrolytic or other non-linear capacitor late in the circuit to set the LF rolloff (so ensure that the non-linear cap has signal voltage across it, at least at LF)
- similarly, try to ensure that the HF rolloff is set by a non-linear capacitor late in the circuit
- use very low ESR caps and bypass them (this may create resonances)
If you follow these suggestion you will have an amp which is very sensitive to 'capacitor rolling' and you can tell yourself and your friends that this is due to its fine discrimination.
If ESR is small compared to 1/gm (but bigger than much much smaller) then it might make a small difference, but changes in ESR (due to ageing etc.) will probably be smaller still. If a change in ESR is much much smaller than 1/gm then it will make no difference.
If you want to hear capacitors, this is how to do it:
- avoid negative feedback, or have just enough to cause problems (i.e. not enough to solve problems)
- use an electrolytic or other non-linear capacitor late in the circuit to set the LF rolloff (so ensure that the non-linear cap has signal voltage across it, at least at LF)
- similarly, try to ensure that the HF rolloff is set by a non-linear capacitor late in the circuit
- use very low ESR caps and bypass them (this may create resonances)
If you follow these suggestion you will have an amp which is very sensitive to 'capacitor rolling' and you can tell yourself and your friends that this is due to its fine discrimination.
This is interesting,
Because it is possible to make an amplifier with very high spec technically i.e. measures well or runs well on simulation that can sound awful! I can see your point about being sensitive to "capacitor rolling", however you do not say if with this sensitivity you get an increase or decrease in low level signal sensitivity (detail)? This is aside from some instability. Some people prefer no feedback because they say it masks low level detail < I am still undecided on this one! However dielectric does seem to effect low level detail.
Regards
M. Gregg
Because it is possible to make an amplifier with very high spec technically i.e. measures well or runs well on simulation that can sound awful! I can see your point about being sensitive to "capacitor rolling", however you do not say if with this sensitivity you get an increase or decrease in low level signal sensitivity (detail)? This is aside from some instability. Some people prefer no feedback because they say it masks low level detail < I am still undecided on this one! However dielectric does seem to effect low level detail.
Regards
M. Gregg
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.