John are these actual regulators (using - gasp - feedback) or multiple cap multipliers?
Jan
Absolutely, and I sincerely thank you for calling me on that. I clearly meant "raises the question" and incorrectly said "begs the question", which is a very different thing. No excuses.
I don't see what that has to do with what you wrote. You were talking about your own ability to hear, or detect, so-called differences which you already said were at the limit of your ability of discrimination. I asked why you would care about differences that you can barely, if at all, hear, or can only hear in sighted evaluations.
Okay, I will try to clarify. I was talking about more than one thing in my original post, and I will try to address a portion of here.
Going back to the example of undithered 16/44 audio, I said that some people can hear the quantizing noise, and some can't.
But, what about people for whom hearing the quantizing noise is near their threshold of discrimination? In that case, it might turn out they can hear it sometimes and not other times, and so we would need to use statistical analysis to determine how reliably they can hear the noise, since a simple yes/no answer wouldn't work in that case.
If we ask them to try to discriminate multiple times, each time would be considered a trial. In order to be confident that the statistics we produce are reasonably reliable, one complete test would need to consist of many trials.
Even at that, since we are talking about what is essentially human subject medical testing, to be even more sure of the reliability of our results, we should probably replicate the whole test a few times.
The point is, the more certain we need or want to be about our human subject's ability to discriminate, the more time and effort we will need to invest.
In the case of JC and his products, it might not be unreasonable to suppose that some of his customers have better hearing than he does. Therefore, he would probably be wise to do whatever he needs to do to assure that his products satisfy as many customers as possible. To that end, he takes measurements (as he should), and also performs listening tests (as he should). To save time working on issues that are near his personal limit of discrimination, he uses sighted listening. To me, I would say if it works for him, and apparently it does, then its fine with me for him to work that way.
By analogy, I would also say that someone wanting to sell CDs should dither the audio, even if they cannot hear dither noise themselves, or if hearing dither noise is near the limits of their discrimination. Now, for CDs this is completely uncontroversial, and in my last post I simply wanted to use an uncontroversial example to illustrate that one point.
Going back to my original post, I also talked about my experience trying to hear things near my threshold of discrimination. As it happens, I can hear quantizing noise on undithered 16/44 audio. But, there are more difficult hearing discrimination problems I have encountered, and I was trying to describe the subjective experience of something that may be unfamiliar to some people. When trying to describe something that may be unfamiliar, it would seem to take more words to describe some of the nuance, as compared to what would be appropriate or necessary to describe some shared experience. For example, what if you wanted to describe the mental experience of seeing the color yellow to someone who had never seen it? Not that easy, probably.
Last edited:
See my post 90803 , it is only exercise to show how easy is to change 1A bridge with four diodes without any damaging of PCB. With PTH PCB simply cut off the bridge leads and use desoldering braid or use ZD 915 cheap properly made des. gun. Very useful tool.In fact every serious builders should have it.
John Curl should have more capable technicians to speed up diode assembly.
BTW, Vendetta phono stage high.rez. pictures show the worst soldered joints I have ever seen in commercial audio equipment. They would not pass IPC standards.
In my Borbely 419 phono or XOno clone I do not hear the difference between standard bridge and soft recovery diodes or various snubbers. Both are high grade phono stages. XOno is modified with Jung SR , Mr. Didden version with a huge PSRR.At the same time XOno circuit have very high power supply rej . ratio which filters diode noise.
For the sake of completeness, I probably should have mentioned that I was referring to discrimination when listening on my main playback system.
Whether I could discriminate to the same degree using my laptop soundcard, or cellphone, as two examples, would be highly doubtful.
Nezbleu, in my admittedly limited experience with shunt compared to other regulators, the differences can sometimes be pretty subtle. It will be best noticed on extremely demanding LP material, with much high and low frequency content which pushes the sound to extremes.
Even so, at times it will make you wonder where the difference comes from, is it by design concept or by some of the componens used, which implies a perhaps better result with the same but using better components. Normally, for a phono stage, I'd use a BD239 or some such 30W device, which should by reason be more than enough, yet I found that if I use a say 2SC5200 (150W) device, the bass lines come across with more precision and clarity delivering the same 10...20 mA of current. Just as I found that by using MOSFETs I miss out on some punch and raw power. To be sure, these were subtle differences you had to look for using a very well known program material (e.g. Billy Cobham's "Quadratnt four").
True, in the 60-es we didn't woryy about things like that, but the the quality of our sources were not as high as in the late 70-es, by which time our phono stages had improved quite a lot as well, as had our speakers. And decades have gone by, we have trained our ears for more by default.
Even so, at times it will make you wonder where the difference comes from, is it by design concept or by some of the componens used, which implies a perhaps better result with the same but using better components. Normally, for a phono stage, I'd use a BD239 or some such 30W device, which should by reason be more than enough, yet I found that if I use a say 2SC5200 (150W) device, the bass lines come across with more precision and clarity delivering the same 10...20 mA of current. Just as I found that by using MOSFETs I miss out on some punch and raw power. To be sure, these were subtle differences you had to look for using a very well known program material (e.g. Billy Cobham's "Quadratnt four").
True, in the 60-es we didn't woryy about things like that, but the the quality of our sources were not as high as in the late 70-es, by which time our phono stages had improved quite a lot as well, as had our speakers. And decades have gone by, we have trained our ears for more by default.
Derfy,
The first stage is typically a series regulator. That makes sure the following stages don't get over voltage. It does reduce noise but has decreasing gain with increasing frequency. It also has a non-linear output impedance. The second stage is typically a capacitance multipier. Some folks would put this first but in two stage designs it usually has a more linear output impedance and lowers noise a bit more. The final stage would be a shunt regulator. This not only provides a symmetric low impedance it also gives you clearly defined current paths.
The issue (not for the simplistic and often snide folks) is more than stable supply voltage, but rather keeping noise, both conducted and radiated away from the circuitry.
Hi Ed,
Most manufacturers of really good test equipment get this right, so their instruments can have amazing low noise floors. Of course, the various stages of an instrument (or audio equipment) have differing levels of immunity to power supply noise, hinting the more sensitive stages to have lower levels of noise - limiting the current requirements and making low noise regulation somewhat easier to design. Different stages should be isolated from each other as well. In other words, no matter how the design is accomplished, more effort in this direction is required if you can hear the differences that other diodes can make. The design isn't complete until this is the case.
Anyway, the working concept of a very low noise gain stage always includes the use of a power supply that pretty much completely isolates that circuit from the outside world, or internally generated issues. In other words, the power supply itself must be designed to be low noise.
-Chris
I think we would both agree on whatever form a really good power supply design would be. There isn't only one way to do this of course. But, the primary job of a power supply in any device is to supply the correct voltages at whatever operating current the circuit requires. The second and equally as important job is to isolate the circuit from outside noise, whether created outside the case, or by it's own rectification and regulation processes. The more susceptible a circuit is to noise, the better the power supply has to be, it's that simple.
Most manufacturers of really good test equipment get this right, so their instruments can have amazing low noise floors. Of course, the various stages of an instrument (or audio equipment) have differing levels of immunity to power supply noise, hinting the more sensitive stages to have lower levels of noise - limiting the current requirements and making low noise regulation somewhat easier to design. Different stages should be isolated from each other as well. In other words, no matter how the design is accomplished, more effort in this direction is required if you can hear the differences that other diodes can make. The design isn't complete until this is the case.
I can't imagine what you mean by that statement. If you mean that the highest quality gain stages are benefited by having low PSSR, that would seem to suggest that the circuit will also have higher distortion, as in a class A stage similar in concept to old tube technology products. From that standpoint, I would have to disagree with you unless it is desireable to have that level and type of distortion. But, then we have departed from the "piece of straight wire with gain" concept.
Anyway, the working concept of a very low noise gain stage always includes the use of a power supply that pretty much completely isolates that circuit from the outside world, or internally generated issues. In other words, the power supply itself must be designed to be low noise.
-Chris
Hi John,
One way to reduce rectifier noise is to use sane sizes of capacitance in the filter section. A 120 Hz ripple is much easier to knock down to tiny levels as compared to the RF energy created by using filter capacitors that are massively oversized. The best way to avoid diode switching noise is to avoid the situation that creates it in the first place. This concept is very effective and one I use when improving the performance of sound equipment designed after the early 70's (when engineers knew how to design a power supply). Cutting the size of the input capacitor really helps with this if it has been oversized. Even I know this, without being an engineer with papers to prove it. This is proved through many years of observations.
-Chris
Perhaps it's time to use restricted frequency range EI core transformers, or use restricted frequency range toroidal transformers (more recent development). The use of an electrostatic screen in the transformers would also be indicated. As you point out, once you can block AC mains borne noise, you have to avoid creating it yourself in the design.
One way to reduce rectifier noise is to use sane sizes of capacitance in the filter section. A 120 Hz ripple is much easier to knock down to tiny levels as compared to the RF energy created by using filter capacitors that are massively oversized. The best way to avoid diode switching noise is to avoid the situation that creates it in the first place. This concept is very effective and one I use when improving the performance of sound equipment designed after the early 70's (when engineers knew how to design a power supply). Cutting the size of the input capacitor really helps with this if it has been oversized. Even I know this, without being an engineer with papers to prove it. This is proved through many years of observations.
-Chris
Hi Jan,
I like to refer to test instruments because they deal often with the same frequency ranges we do in audio, but with that equipment the performance is what matters and what the engineering team is graded on. The only one thing that separates test equipment from most audio is cost. However, when the audio equipment pricing falls within the same range as test equipment, you can compare test equipment to audio equipment on an equal footing for power supply performance. I'm talking of course about good test equipment, not the B&K's or Leader of the world. More like Tektronix, HP, Boonton and similar instruments. What is the #1 place to improve the performance of an old instrument? Power supplies, with current technology. The actual signal handling circuits require a great deal more thought and engineering talent to improve without incurring negative side effects.
-Chris
It was, and it wasn't. Earlier amplification wasn't low noise to the point where better power supplies were demanded. When they were, carefully designed active regulation was employed to solve that problem. Again, test instrumentation is an excellent place to watch for power supply improvements.
I like to refer to test instruments because they deal often with the same frequency ranges we do in audio, but with that equipment the performance is what matters and what the engineering team is graded on. The only one thing that separates test equipment from most audio is cost. However, when the audio equipment pricing falls within the same range as test equipment, you can compare test equipment to audio equipment on an equal footing for power supply performance. I'm talking of course about good test equipment, not the B&K's or Leader of the world. More like Tektronix, HP, Boonton and similar instruments. What is the #1 place to improve the performance of an old instrument? Power supplies, with current technology. The actual signal handling circuits require a great deal more thought and engineering talent to improve without incurring negative side effects.
-Chris
Wow...
Ed I was merely thinking about John's continued and professed insistence on 60-ies circuits and 70-ies parts. No insults intended to him or you.
Did you not see the context of that post and the smily at the end?
Jan
The first active stage also attenuates 2xfmains ripple by 60-90dB, allowing the subsequent stages to operate in a much cleaner environment.
I'm sure this is not what Mr. Simon had in mind, but be aware that there is a general relationship between the CMRR, PSRR and the gain in an amplifier. See A general relationship between amplifier parameters, and its application to PSRR improvement - IEEE Xplore Document
Sorry, I don't have a free link, but the bottom line is: under certain constraints,
1/(CMRR) + 1/(PSRR+) + 1/(PSRR-) = (1/Adiff) * Zload/(Zload+Zout)
which means that everything else being constant, if PSRR goes down, CMRR goes up.
This is not to say that Mr. Curl's standard gain stage is today anything but a 60's audio boutique joke in terms of performance.
Jan,
Never point a toy gun at an officer.
Some fellow named Freud if he were still around might have had an interesting chat with you. Jung for JC.
Assuming the performance goal is zero distortion? What if the goal is listening pleasure?
Our statements aren't mutually exclusive. 😀 And, yes, staging your regulators does provide opportunities to optimize for different aspects allow you to put a low-power shunt regulator right where it's needed. But, as such, you're relying on that shunt regulator to largely drive the PSU's performance to that load.
You'll find me in the group calling that a 2-regulator system. Likewise, with the number of superlative regulator IC's (some of which our own Jack Walton brought to light), which even include some high efficiency switching (allowing for good staging of varying voltages for different parts of the circuit), it's hard to see a 3-regulator system being considered a "good thing" versus a strange solution.
Not to mention starting out with a good, well shielded, SMPS out front if superlative performance is the goal.
- Status
- Not open for further replies.