Well, back to 'work'.
It is still a mystery to me as to why some changes in audio design are as important as they are, at least to me (and my customer base).
There may not be any 'measurable' difference, but there is an audible difference. This has been happening for many decades to me, in fact, it often drives my design 'improvements' over the recent years. This might be better, (or different) passive parts, better grounding and transformer phasing, high speed diodes, better solder and circuit board materials, more 'ideal' passive connectors, and finally, better quality wire. Cleanliness in construction is another factor.
Now, a little bit of history:
Back when I started in electronics, over 50 years ago, I built kits with marginal solder, 'doorbell' wire that easily melted if overheated, and a soldering GUN that was huge, by today's standards. The kits worked, did what they did, OK, and worked for many years. In fact, I bet that if I got a couple of these kits out of storage, and carefully reformed the caps, they would work essentially as specified.
Yet, I would not build them like that, today, at least not audio designs. Why, is because of my accumulated experience with what works and what fails to meet expectations in audio.
Let me jump ahead 10 years, to 40 years ago, when I was assigned a project to make a solid state mixing board for the Grateful Dead. This was a combined effort with my colleague (a former boss, Ron Wickersham, still with Alembic Inc.)
I selected an 'advanced' IC that 'ran rings' around the popular 741, called an HA-911, a military part, made with dielectric isolation, an expensive process at the time.
The price was daunting, perhaps $5 each, or about an hour's pay for an engineer at the time. We got lucky and got 1000 rejects (tempco) in flat pack that cost us only 0.25 each. A challenge to test, and they had to be hard wired into the circuit, but still a good deal. In those days, we used coupling caps just about everywhere. We were used to it, and any DC offsets could be taken out. The problem was that we selected a 2.2uF CERAMIC CAP that Ron found surplus. These parts, at the time, were relatively expensive, but small and non-polar. We also used Tantalum caps for higher values, and Allen Bradley mil grade pots and baked carbon resistors throughout the entire unit. We built it, tested it for distortion, using my trusty Heathkit IM analyzer, able to resolve 0.005% distortion, looked for oscillations, etc, with my trusty 80 MHz Tek Scope, and released it for audition with the band.
They started to use it, and put away their vacuum tube mixers, except as back-up. Well, after awhile on the road, they damaged the solid state mixer, and had to revert to the tube mixers in reserve. They NEVER went back to solid state mixer, they could EASILY hear the difference between an OPEN LOOP TUBE MIXER and my solid state IC based mixer, and soon I was out of a job.
Now, what did I do wrong? For the time, NOTHING! Yet, the solid state mixer, 'sucked'. During my latent period of being on unemployment, I attempted to 're-think' what it took to make really good solid state electronics.
About one and 1/2 years later, in late 1972, I was re-hired by the GD to build 'The Wall of Sound' system. Now what? What could I do that was different and perhaps be successful? This is where the topologies that you see in the Levinson JC-2 were born, coupled with some a Matti Otala's ideas, of high open loop bandwidth and high slew rate were added, and guess what? SUCCESS! Yes, and not only with the GD, but with Mark Levinson, as well, then an up-in-coming maker of high end audio electronics.
And people wonder why I am so skeptical of traditional op amps. But it was not just the op amp that was the problem, it was ALSO the coupling caps, but I had not yet measured this problem with my test equipment. About 1 year later, in 1974, Tek showed me the way to measure caps, and ceramics, (except for a few small value and expensive ones) are really SCARY, when you actually see their non-linearity. more later
It is still a mystery to me as to why some changes in audio design are as important as they are, at least to me (and my customer base).
There may not be any 'measurable' difference, but there is an audible difference. This has been happening for many decades to me, in fact, it often drives my design 'improvements' over the recent years. This might be better, (or different) passive parts, better grounding and transformer phasing, high speed diodes, better solder and circuit board materials, more 'ideal' passive connectors, and finally, better quality wire. Cleanliness in construction is another factor.
Now, a little bit of history:
Back when I started in electronics, over 50 years ago, I built kits with marginal solder, 'doorbell' wire that easily melted if overheated, and a soldering GUN that was huge, by today's standards. The kits worked, did what they did, OK, and worked for many years. In fact, I bet that if I got a couple of these kits out of storage, and carefully reformed the caps, they would work essentially as specified.
Yet, I would not build them like that, today, at least not audio designs. Why, is because of my accumulated experience with what works and what fails to meet expectations in audio.
Let me jump ahead 10 years, to 40 years ago, when I was assigned a project to make a solid state mixing board for the Grateful Dead. This was a combined effort with my colleague (a former boss, Ron Wickersham, still with Alembic Inc.)
I selected an 'advanced' IC that 'ran rings' around the popular 741, called an HA-911, a military part, made with dielectric isolation, an expensive process at the time.
The price was daunting, perhaps $5 each, or about an hour's pay for an engineer at the time. We got lucky and got 1000 rejects (tempco) in flat pack that cost us only 0.25 each. A challenge to test, and they had to be hard wired into the circuit, but still a good deal. In those days, we used coupling caps just about everywhere. We were used to it, and any DC offsets could be taken out. The problem was that we selected a 2.2uF CERAMIC CAP that Ron found surplus. These parts, at the time, were relatively expensive, but small and non-polar. We also used Tantalum caps for higher values, and Allen Bradley mil grade pots and baked carbon resistors throughout the entire unit. We built it, tested it for distortion, using my trusty Heathkit IM analyzer, able to resolve 0.005% distortion, looked for oscillations, etc, with my trusty 80 MHz Tek Scope, and released it for audition with the band.
They started to use it, and put away their vacuum tube mixers, except as back-up. Well, after awhile on the road, they damaged the solid state mixer, and had to revert to the tube mixers in reserve. They NEVER went back to solid state mixer, they could EASILY hear the difference between an OPEN LOOP TUBE MIXER and my solid state IC based mixer, and soon I was out of a job.
Now, what did I do wrong? For the time, NOTHING! Yet, the solid state mixer, 'sucked'. During my latent period of being on unemployment, I attempted to 're-think' what it took to make really good solid state electronics.
About one and 1/2 years later, in late 1972, I was re-hired by the GD to build 'The Wall of Sound' system. Now what? What could I do that was different and perhaps be successful? This is where the topologies that you see in the Levinson JC-2 were born, coupled with some a Matti Otala's ideas, of high open loop bandwidth and high slew rate were added, and guess what? SUCCESS! Yes, and not only with the GD, but with Mark Levinson, as well, then an up-in-coming maker of high end audio electronics.
And people wonder why I am so skeptical of traditional op amps. But it was not just the op amp that was the problem, it was ALSO the coupling caps, but I had not yet measured this problem with my test equipment. About 1 year later, in 1974, Tek showed me the way to measure caps, and ceramics, (except for a few small value and expensive ones) are really SCARY, when you actually see their non-linearity. more later
Attachments
My experience with live rock music is not hi-fidelity formost. We hosted an early Aerosmith concert, 11+.
The GD, with Ron Wickersham and Bear (Owsley) used the MX-10's, bypassing the second stage in some way. This is how they mixed the vocals for a few years. My job was to replace the MX-10's with something, solid state, that sounded just as good, yet was more reliable on the road, the tubes being sensitive to vibration damage. We did our master recording direct, from a splitter in the mike lines, and direct boxes.
Not all rock and roll is just overload and distortion. One should come to know this, if objectivity is to be maintained.
Not all rock and roll is just overload and distortion. One should come to know this, if objectivity is to be maintained.
Show me a serious PA speaker with less than several percent distortion at contractualy specified SPL's. That is, the PA shall provide 120dB at X feet the fidelity is an N/A.
Scott, why do you have to be so nasty and condescending? It really doesn't suit you...
I had a similar bad experience with ceramic capacitors in a completely different setting. I had built what everyone called "the Leach amp", which was a fully complementary-symmetry design by Dr. W. Marshall Leach, featured as a construction article in Audio magazine around 1976. It was the second good sounding solid-state amp I had heard, after the Ampzilla (I didn't hear JC's JC-3 for several more years), but unlike the Ampzilla, I could just barely afford to build it.
They sold bare boards and a group of us did a joint buy to get the semiconductors in order to meet the $25 minimum from Allied. I picked up metal oxide resistors as money permitted, and I scored big time on the small-value capacitors -- I got a bunch of surplus aerospace grade ceramics! Cool!
The problem was that my amp never sounded all that great. I wanted to like it, but never did. After owning it for a couple of years, I removed the low-pass input filter designed to eliminate any TIM. There was a series resistor of 150 ohms or so and a 0.01 uF ceramic for a corner frequency around 100 kHz.
OMG! The sound I had remembered in a friend's amp was suddenly there! Fantastic!
At the time I thought that it proved we could hear past 20 kHz. A few years later, JC and WJ did their ground-breaking article on dielectric absorption and I began to have my doubts about being able to hear up to 100 kHz...
Of course, JC and WJ were right (again), and I learned one of many lessons I was to learn from JC.
At no time were any PA's used nor harmed during any of that construction project or listening tests.
What is interesting is that musicians, if they really care, can tell you more about their sound than you will be able to easily discern. This is why I had to listen to THEM and not to some textbook telling me that low slew rate is OK or that 100dB of feedback is ideal.
When you have 50 large loudspeakers, made with high quality, like JBL, strong speaker cabinets that don't rattle, and line arrays that focus the sound, much like a horn, then you can have low distortion at normal levels. Maybe even, very low distortion. The worst that I ever measured, AND we removed them from service, was a midrange horn that had 3% second harmonic distortion at 2W power input or about 115dB output at 1 meter/horn. Too much distortion, it was better to replace them with 100 or so JBL quality 5'' speakers in an array.
And so it goes. What is really important, is I could easily go into hi end audio design, after working with the GD, with the SAME electronics. Yes, most of what gets built here as a simple comp. diff. input. was designed for the GD more than 40 years ago, because Crowns sounded lousy, and tube amps were becoming unreliable.
When you have 50 large loudspeakers, made with high quality, like JBL, strong speaker cabinets that don't rattle, and line arrays that focus the sound, much like a horn, then you can have low distortion at normal levels. Maybe even, very low distortion. The worst that I ever measured, AND we removed them from service, was a midrange horn that had 3% second harmonic distortion at 2W power input or about 115dB output at 1 meter/horn. Too much distortion, it was better to replace them with 100 or so JBL quality 5'' speakers in an array.
And so it goes. What is really important, is I could easily go into hi end audio design, after working with the GD, with the SAME electronics. Yes, most of what gets built here as a simple comp. diff. input. was designed for the GD more than 40 years ago, because Crowns sounded lousy, and tube amps were becoming unreliable.
some ceramic formulations have large voltage coefficient of capacitance giving nonlinear C with varying signal level and consequent distortion
there is no need to invoke "dielectric absorption" - which has a good fitting linear model and results in milli to micro dB frequency response variations in such a low impedance source low pass filter application
frequency dependent dielectric loss is another effect that can be larger than dielectric absorbtion
I have searched in materials physics texts as well as precision instrumentation design literature for dielectric absorption models and correlation with nonlinear behavior in continuous filter circuits and I really can't see why audiophiles are so certain that dielectric absorption is "THE" explaining capacitor audio quality metric
there is no need to invoke "dielectric absorption" - which has a good fitting linear model and results in milli to micro dB frequency response variations in such a low impedance source low pass filter application
frequency dependent dielectric loss is another effect that can be larger than dielectric absorbtion
I have searched in materials physics texts as well as precision instrumentation design literature for dielectric absorption models and correlation with nonlinear behavior in continuous filter circuits and I really can't see why audiophiles are so certain that dielectric absorption is "THE" explaining capacitor audio quality metric
Last edited:
Hi Charles,
I read the post by Scott as a live sound by band type joke. I think along those same lines after doing live sound and servicing road gear. That's opposed to what can happen in a recording studio. Even there, a wailing amp in a warehouse with a mic can be used to "get that sound". It was amusing, not nasty or condescending.
The amps built by most in the 70's suffered from many horrors. Between carbon comp resistors, ceramic caps and the unbelievable - tantalum capacitors, it's amazing anyone sold anything. Some manufacturers were wise though. One I know of used NPO ceramics, mica and wet tantalum capacitors. They sounded really clean compared to the other stuff out there. I'm thinking of a model sold as early as 1969, so some engineers were aware of what to stay away from in the way of parts. That's well before anyone was writing about DA or any non-linearity in the audio business.
The worst place for capacitors that behave badly are in high impedance circuits that see a large amplitude signal across them. Obvious to many, but not clear to all.
Hi John,
-Chris
I read the post by Scott as a live sound by band type joke. I think along those same lines after doing live sound and servicing road gear. That's opposed to what can happen in a recording studio. Even there, a wailing amp in a warehouse with a mic can be used to "get that sound". It was amusing, not nasty or condescending.
The amps built by most in the 70's suffered from many horrors. Between carbon comp resistors, ceramic caps and the unbelievable - tantalum capacitors, it's amazing anyone sold anything. Some manufacturers were wise though. One I know of used NPO ceramics, mica and wet tantalum capacitors. They sounded really clean compared to the other stuff out there. I'm thinking of a model sold as early as 1969, so some engineers were aware of what to stay away from in the way of parts. That's well before anyone was writing about DA or any non-linearity in the audio business.
The worst place for capacitors that behave badly are in high impedance circuits that see a large amplitude signal across them. Obvious to many, but not clear to all.
Hi John,
While that is certainly true, most concerts do not attempt to produce high quality sound at all. It's volume and whether or not you can feel the bass. Being deaf afterwards is one sure sign that it was loud. Perhaps the cleanest sound I heard was in Supertramp concerts. Yet, I was deaf walking out.
-Chris
Yes that is true. I doubt that the 0.01 uF ceramics I was using were NPO, despite the fact that they were aerospace qualified. They were about the size of the common Wima 5mm lead spacing boxes -- perhaps 8mm x 8mm x 2.5mm. As I recall they were 100 volt parts, but that was 30+ years ago... It is certainly a strong possibility that they exhibited other non-linearities besides DA.
Across the audio band I doubt there would be much variation at all.
I think the main reason is the articles written by JC and WJ. These showed excellent correlation between DA and subjective sound quality. This has been further corroborated in my own personal experience with PCB materials.
Perhaps that was how Scott intended it. That was not how I read it. More importantly is how John read it. Unfortunately a double standard applies to his posts, so he is not always free to speak his mind.
And (obviously) by the fact that he stays here despite the insults (whether this was intended as one or not), John has become rather callused to these insults. But it doesn't mean that they don't hurt. He has earned much more respect than he is often given.
At this point in the circuit, the source impedance was Zo + 150 ohms, so say 200 ohms. The signal level was perhaps 2 Vrms at clipping or 100 mV during typical music playback levels. So I don't think this explains the dramatic subjective change in sound quality.
I suspect that the nonlinear distortion caused by the ceramic cap was the problem. NPO? No, I don't think so. I was even able to capture it with SMPTE distortion 400Hz, and published it in my IEEE paper given in 1978. I used a .01uF ceramic and 600 ohms to get the measurement in that paper. Two tone CCIF IM would be even better, but I did not need to do that test to get results. The distortion in typical large value ceramic caps of .01 uF or so or larger is very bad, virtually the worst I have ever seen in a cap, and the DA is not low, either, but comparable to Tantalum or Aluminum. DA is not a NASTY sort of distortion, just an inaccurate change in the signal.
Hi Charles,
Ahh, back to things technical. A comfortable spot for me.
So, the magnitude of effects possible from this capacitor can't explain the sound change that you heard. Then, what else was changed? After all, I only indicated a badly behaved capacitance. In other words, all faults possible in a normally functioning part. Is it possible your ceramic capacitor had been faulty in some way? A failed part. The only other explanation (being fixed to an explainable shift in performance from one part not capable of such a change) might be physiological in nature. I am not suggesting you are hearing things, but this is a large problem for people experimenting in any field.
What conclusions did you come up with that satisfied your need to know?
-Chris
I don't have any way to know how anyone takes any comment, but it was clear you took the comment as a very negative one. Life is easier if you do not assume the worst in people all of the time.
I disagree. In fact, I strongly disagree with that statement. While I refuse to discuss any details concerning moderation activities that concern John, I will tell you that what is assumed is far different than the truth. Remember that in any disagreement, there is more than one side. It is very common for people to feel like they are unjustly treated, but that is their own viewpoint.
Granted, and that is a concern. However, John bites. I have very often tried to be of some help, but this is clearly not something that has ever been accepted.
respect is a give and take thing. You can't claim respect, nor can you ignore how you treat others if this is important to you. There are excellent role models here as members. Compare their actions, then come and talk to me. Still, there are extended olive branches.
Ahh, back to things technical. A comfortable spot for me.
I didn't claim that that situation was the only one that would affect a signal, but it surely is a perfect setup for signal corruption. From your professional history, you are well aware of that.
So, the magnitude of effects possible from this capacitor can't explain the sound change that you heard. Then, what else was changed? After all, I only indicated a badly behaved capacitance. In other words, all faults possible in a normally functioning part. Is it possible your ceramic capacitor had been faulty in some way? A failed part. The only other explanation (being fixed to an explainable shift in performance from one part not capable of such a change) might be physiological in nature. I am not suggesting you are hearing things, but this is a large problem for people experimenting in any field.
What conclusions did you come up with that satisfied your need to know?
-Chris
Whatever it was, it was enough to cure me of ever using ceramics again. Another well respected designer told me that he found NPO ceramics to sound good in small values (~100 pF range). So I tried them again. They weren't nearly as bad as the other ceramic caps I had used before, but still audibly inferior to a cheap Wima PP.
I try to avoid ceramics even for digital circuits, but that is mostly superstition. I don't remember doing any listening tests to confirm that they harm the sound in this position. But I refuse to use X7R or Y5V anywhere in any product I make. (Well maybe in the battery bypass for a remote handset.) Those things are so awful that they don't deserve to be called capacitors. Even electrolytics sound better than that, and that's not saying a whole lot...
Last edited:
Not a failed part. This was a stereo amp. Virtually impossible that both units failed in identical locations is both channels while no other problems occurred whatsoever.
I went over to Dr. Leach's website (RIP) and found that my memory of the values was very far off (surprise!). The series resistor was 2k00 not 200R and the capacitor was 390 pF, not 0.01 uF.
I don't know how much the capacitor technology has changed since then, but virtually all of the 390 pF capacitors even close to the size of the units I used are NPO today. Whatever they were, they sure were nasty. As soon as I removed them, the amp went from being competent but boring and sub-part to nearly spectacular. I was literally stunned at the change.
None. I don't have a need to know, except that what works and what doesn't. I gave that up years ago, as it was just getting in my way and slowing me down.
- Status
- Not open for further replies.