The Acrosound 20-20 has only an input cap. All the rest is direct coupled (12AX7 to EL84), with DC feedback and regular NFB. It is simple at first glance, but not easy to understand in details. There are drawbacks with DC design, but in terms of transient and details, that amp beats everything else I have.
OK, let's talk about the drawback of the DC amps. That will make us sleep a little better tonight.
The main one is that you don't have to spend $$ in good coupling caps.
Because of the nature of DC coupled design, the tubes are all DC linked, therefore:
- one tube failure usually takes down the rest with it (especially at the power stage).
- The performance is greatly impacted by the type/brand/gm/emission tubes used.
- NOS versus new tubes: some design won't work as well with new production tubes which is understandable as they were design with the tubes back then.
I am sure I am missing many others.
But IMO, these can be overcome is some ways, and DC designs are the best even if not always practical.
Because of the nature of DC coupled design, the tubes are all DC linked, therefore:
- one tube failure usually takes down the rest with it (especially at the power stage).
- The performance is greatly impacted by the type/brand/gm/emission tubes used.
- NOS versus new tubes: some design won't work as well with new production tubes which is understandable as they were design with the tubes back then.
I am sure I am missing many others.
But IMO, these can be overcome is some ways, and DC designs are the best even if not always practical.
the use of diodes or NE2 neon lamps helped, careful design with this issue in mind is possible..
where loss of the driver tube can happen, diodes between grid and cathode ensured no more than 0.6v of grid bias, in addition to the cathode resistors in the final tubes that can be used to blow open in such an event..
Maybe because some "audio guru" said it was a wonderful way to make things sound better.....I'm not sure but it might have been the same crack head who gave us the "optimized electron stream" stuff.
I never made it through all these pages. It started going the way of the "blameless tube amp" thread, so I decided to turn off the computer, crank up my "near blameless" SE amp and listen instead. It is a two stage design, and contains not one, but two coupling caps per channel. It can NOT POSSIBLY exhibit blocking distortion by design.
I started this thread by going from listening to unbiased Al-El's in speaker crossovers to a full range pair - which blew me away in how coherent they could sound.
I attributed this to the absence of a crossover with its high pass and low pass filters implemented using capacitors - in their worst nightmare; unbiased, with true AC current flow.
Perhaps it's unfair to compare cap performance in the highly biased state, to the 0V bias state; the 0V state is going to sound worse. So of course the change of removing the capacitor factor entirely could be pronounced in speakers, but not so much in amplifiers.
Ah, might as well make all 4 the same (due to previous damage) Looks like I've got some cap huntin' to do, be needin' 4 of 'em @ 0.22, preferably 600V for reliability. Thanks!
I attributed this to the absence of a crossover with its high pass and low pass filters implemented using capacitors - in their worst nightmare; unbiased, with true AC current flow.
Perhaps it's unfair to compare cap performance in the highly biased state, to the 0V bias state; the 0V state is going to sound worse. So of course the change of removing the capacitor factor entirely could be pronounced in speakers, but not so much in amplifiers.
Ah, might as well make all 4 the same (due to previous damage) Looks like I've got some cap huntin' to do, be needin' 4 of 'em @ 0.22, preferably 600V for reliability. Thanks!
This topic reminds me one case when one grandmother asked me to repair her TV, because "That small tube at the bottom of the back does not glow". I switched on the TV and indeed found that 6Р4П tube (twin pentode with weird number of legs) does not glow. First of all I thought that as usual, solder joint on it's socket got broken, during to constant expansion and shrinking, as always happened in Soviet TVs. I resoldered it, but it did not glow.
I measured filament voltage, it was absent. I measured 220V AC on the primary, it was absent. I measured the voltage before the power switch, it was absent. Both fuses were fine. Then I opened the power plug and found that one bolt got unscrewed and fell off.
Of course improperly designed crossovers, or improperly placed speakers can cause frequency response problems. And of course filters should contain capacitors and/or inductors. But it does not mean tat all of them are bad. The thread highlighted real problems that gave the start to the belief that "capacitors are bad", and we discussed the ways how to deal with such problems.
JBL used to use 2x9V battery to bias their speaker network capacitors.
JBL K2 S9500 crossover network
This is true - having had the 103Sigma for over a decade, and having made the tqwt cabinet for several friends and family.
I think it is not the cap involved in other speakers, rather that the units involved in lf and hf have a different threshold before they start to move (physically). The hf units are more responsive. And there could be a severe hysteresis effect in the lf unit.
Loaded with a higher volume level, this “defect” fades away, but at low level listening to the two or three way speakers becomes a nightmare. The sound becomes thin and top heavy.
Some speakers claim to have units that are matched it their dynamic behaviour. I have had the quad electrostat EL57 for two decades and the only one I know where there is a great match.
plasnu,
Thanks for the info on the: JBL K2 S9500 crossover network
Perhaps JBL decided that the 'cost and performance' of uni-polar electrolytics was better than using non-polar electrolytics. But uni-polar electrolytics would require DC bias. However, 18V bias would not take care of signals larger than 18V peak.
It would be good to see what kind of capacitors that JBL used in the crossovers that used 2 x 9V batteries. I am not sure we know without a visual inspection.
I had a pair of old JBL crossovers. Simple 6 dB/octave. The capacitor was a non-polar electrolytic. Unfortunately, one of the non-polar electrolytic capacitors had about 50% error versus its 4.7uF rating. Of course, I replaced the capacitor with the closest value I had, 5uF. 6.4% error is lots better than 50% error. Oh, did I mention that I used a modern film capacitor?
Thanks for the info on the: JBL K2 S9500 crossover network
Perhaps JBL decided that the 'cost and performance' of uni-polar electrolytics was better than using non-polar electrolytics. But uni-polar electrolytics would require DC bias. However, 18V bias would not take care of signals larger than 18V peak.
It would be good to see what kind of capacitors that JBL used in the crossovers that used 2 x 9V batteries. I am not sure we know without a visual inspection.
I had a pair of old JBL crossovers. Simple 6 dB/octave. The capacitor was a non-polar electrolytic. Unfortunately, one of the non-polar electrolytic capacitors had about 50% error versus its 4.7uF rating. Of course, I replaced the capacitor with the closest value I had, 5uF. 6.4% error is lots better than 50% error. Oh, did I mention that I used a modern film capacitor?
Last edited:
using standard polarized electros (back to back) in cross-over networks is not a good idea, even with a bias battery. Their capacitance tolerance is far too broad and is not particularly stable over time. -50% to +100% is not untypical.
Non-polarized electros specifically made for cross-over service are made to tighter tolerance and long life in cross-over service.
When I was about 20 (over 50 years ago) I made a couple of speaker boxes using the most expensive Philips drivers I could afford, and the ready-built cross-over network Philips sold for them at that time. I still have those boxes in use now. It uses non-polarize electros and they are still good. They may well produce a bit of distortion, but drivers of that vintage made about 5% THD midband at good volume settings anyway.
Non-polarized electros specifically made for cross-over service are made to tighter tolerance and long life in cross-over service.
When I was about 20 (over 50 years ago) I made a couple of speaker boxes using the most expensive Philips drivers I could afford, and the ready-built cross-over network Philips sold for them at that time. I still have those boxes in use now. It uses non-polarize electros and they are still good. They may well produce a bit of distortion, but drivers of that vintage made about 5% THD midband at good volume settings anyway.
"Capacitor distortion does not correlate with dissipation factor or ESR, but rather with dielectric absorption. Bateman's tests suggest that dielectric absorption is in fact the main cause of capacitor distortion."
https://www.diyaudio.com/forums/parts/151392-electrolytic-capacitors-120.html#post5478553
I wonder why bipolar caps distort half as much as polar caps. Is it because of H2 cancellation?
PS: Based on this chart, it looks like a much higher voltage capacitor (like computer cap) beats lower voltage Silmic or bipolar, if I'm correct...
https://www.diyaudio.com/forums/parts/151392-electrolytic-capacitors-120.html#post5478553
I wonder why bipolar caps distort half as much as polar caps. Is it because of H2 cancellation?
PS: Based on this chart, it looks like a much higher voltage capacitor (like computer cap) beats lower voltage Silmic or bipolar, if I'm correct...
Last edited:
Bateman's (Bateman was a capacitor manufacturing engineer) main deduction for all his tests was that distortion in electrolytics, any electrolytic, is so small it cannot be heard, regardless of presence or absence of bias. This view is maintained by Doug Self (an experinced professional audo designer) wrt coupling capacitors, but with the proviso that you make any electo used as coupling large enough that is signal volt drop is negligible over the whole audio band. It folows from Doug Self that if you use an electro that has significant signal drop in-band, as must be the case in cross-over networks, there will be distortion.
So, there is a discrepancy between the views of two expert professionals.
The graphs plotted by the contributor linked by Plasnu bear this out, predominantly 2nd harmonic at THD levels around 0.005%. You cannot hear that. Nobody can. And even if you could, loudspeakers are orders of magnitude worse, which would mask anything from capacitors.
Bateman did his measurements with a specially designed sensitive THD meter, on known good or new capacitors as I recall. What has been on my list of things to do for years is to repeat the tests on electrolytics that have been in service for some years, and do the tests with an intermod tester. I suspect that may give a different story, particularly with non-polars.
So, there is a discrepancy between the views of two expert professionals.
The graphs plotted by the contributor linked by Plasnu bear this out, predominantly 2nd harmonic at THD levels around 0.005%. You cannot hear that. Nobody can. And even if you could, loudspeakers are orders of magnitude worse, which would mask anything from capacitors.
Bateman did his measurements with a specially designed sensitive THD meter, on known good or new capacitors as I recall. What has been on my list of things to do for years is to repeat the tests on electrolytics that have been in service for some years, and do the tests with an intermod tester. I suspect that may give a different story, particularly with non-polars.
Keit, that's very reasonable, but then why do you think many people (including Nelson Pass in the above thread) confess their preference for Silmic, bipolar or high voltage caps. It seems to me that distortion figure is actually pretty much correlated to peoples preference even if it is extremely low... Any other meaningful measurement that I can find?
Last edited:
Plasnu,
I can't answer that. But I can suggest some possibilities. One is that if what you can hear with your ears does not corelate with a measurement with instruments, then you are not using the right instruments. And there is such a thing as tunnel vision, which is very human characteristic that even the best scientists are subject to. We don't test for things we don't know to test for. Audio folk love to talk about and test non-linear distortion - be it THD or harmonic spectrums. But its actually intermod our ears are sensitive to. It happens that, USUALLY, intermod corelates with harmonic distortion. But NOT ALWAYS. You can have low harmonic dist and much higher intermod.
And there is also the "Volkswagen effect" a human characteristic where we see what we want to see, and ignore glaring defects.
There is the phenomena of "observer bias". If you do a measurement of something, or you know of a measurement, you tend to think your ears hear what you measured. That's why the medical profession devised the double blind test. Which they very seldom actually do - but that's another story.
It has long been my professional experience that if you get two very good practitioners who make contradictory claims, then there are complicating factors the neither have mentioned, often because they hadn't realized the importance of these factors, or they just assumed that the other knows or should know about them. You know the old saying: Don't assume - it makes an ***(donkey) of U and ME. Put the same two very good practitioners in the same room and let them debate and demonstrate to their heart's content, and more often than not, they end up agreeing with each other. And they and anyone within earshot learns something.
This doesn't work with religion though.
Something that was highlighted in the English magazine Wireless World decades ago: Sometimes an audio system defect, be it a bump in a frequency response, or a particular source of non-linear distortion, can apparently suit a particular person.
A contributor to this magazine knew a musician friend, and "knew" that musician's stereo speakers were not very good. He decided to build really good speakers for his friend. When finished, he took the speakers to friend's home and they had a listen. Friend said "no good".
the WW contributor was stunned - he had selected drivers known to have low distortion. He did all the right calculations. He tested the result and verified a very good flat frequency response.
"Why no good" he asked his friend.
Friend said "with my old speakers, I can tell the difference between this recording of a Stradivarius and that recording of a Yamaha." They both had another listen, and both agreed - on the old speakers, BOTH could tell them apart, but on the new speakers, neither could.
Why was it so? It turned out that the old speakers had prominent peaks coinciding with certain harmonics or overtones produced by the Stradivarius. If said friend was not a violinist, he would have a completely different view.
In answer to your question: In short, danged if I know. But there will be a reason somewhere.
I can't answer that. But I can suggest some possibilities. One is that if what you can hear with your ears does not corelate with a measurement with instruments, then you are not using the right instruments. And there is such a thing as tunnel vision, which is very human characteristic that even the best scientists are subject to. We don't test for things we don't know to test for. Audio folk love to talk about and test non-linear distortion - be it THD or harmonic spectrums. But its actually intermod our ears are sensitive to. It happens that, USUALLY, intermod corelates with harmonic distortion. But NOT ALWAYS. You can have low harmonic dist and much higher intermod.
And there is also the "Volkswagen effect" a human characteristic where we see what we want to see, and ignore glaring defects.
There is the phenomena of "observer bias". If you do a measurement of something, or you know of a measurement, you tend to think your ears hear what you measured. That's why the medical profession devised the double blind test. Which they very seldom actually do - but that's another story.
It has long been my professional experience that if you get two very good practitioners who make contradictory claims, then there are complicating factors the neither have mentioned, often because they hadn't realized the importance of these factors, or they just assumed that the other knows or should know about them. You know the old saying: Don't assume - it makes an ***(donkey) of U and ME. Put the same two very good practitioners in the same room and let them debate and demonstrate to their heart's content, and more often than not, they end up agreeing with each other. And they and anyone within earshot learns something.
This doesn't work with religion though.
Something that was highlighted in the English magazine Wireless World decades ago: Sometimes an audio system defect, be it a bump in a frequency response, or a particular source of non-linear distortion, can apparently suit a particular person.
A contributor to this magazine knew a musician friend, and "knew" that musician's stereo speakers were not very good. He decided to build really good speakers for his friend. When finished, he took the speakers to friend's home and they had a listen. Friend said "no good".
the WW contributor was stunned - he had selected drivers known to have low distortion. He did all the right calculations. He tested the result and verified a very good flat frequency response.
"Why no good" he asked his friend.
Friend said "with my old speakers, I can tell the difference between this recording of a Stradivarius and that recording of a Yamaha." They both had another listen, and both agreed - on the old speakers, BOTH could tell them apart, but on the new speakers, neither could.
Why was it so? It turned out that the old speakers had prominent peaks coinciding with certain harmonics or overtones produced by the Stradivarius. If said friend was not a violinist, he would have a completely different view.
In answer to your question: In short, danged if I know. But there will be a reason somewhere.
Last edited:
May I add a little experience from my experiments. Example Nichicon KZ 1000uf 25v vs Kz 1000uf 50v use in a regulated 5v supply. Immediately you can hear that the 50v sounds much better. Is this due to lower esr/dielectric absorption or ????? Strangely I've found that in general higher voltage caps of same brand value sounds better.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.