Bill, what of it? What is it that you believe my subjective listening report reveals? Unfortunately, I'm missing your intended point.
My point is that you claim more than Joe, but also have seemed to dismiss Scott's 20 year old measured evidence on a plausible mechanism. The fact that all in the 'believer' camp think no blind listening is required because its stereophile style day and night is bound to raise eyebrows.
Back before this thread was opened, Joe wrote this:
"Stuart, please, just give it a chance, that's all we are asking for. Scott has asked for "full disclosure" and that is what is coming. Everything will be testable."
The expectations were high, we are finally going to see some real results. As far as I can see there is nothing that wasn't already posted before.
Some see hostility, I only see well meant help.
"Stuart, please, just give it a chance, that's all we are asking for. Scott has asked for "full disclosure" and that is what is coming. Everything will be testable."
The expectations were high, we are finally going to see some real results. As far as I can see there is nothing that wasn't already posted before.
Some see hostility, I only see well meant help.
I thought Joe arrived at this mod by some process of thought, not random trials. Hence he must have been expecting something. I think he mentioned 'damping', although damping is not what you get from adding fairly pure reactance.Ken Newton said:Exactly because of circuit theory, I doubt that Joe was expecting to hear an significant increase in musicality from capacitively loading the DAC outputs by such an seemingly inconsequential degree.
The capacitive load is not inconsequential, although it may appear "seemingly inconsequential" to some. In the case of a typical opamp virtual ground at ultrasonic frequencies it is significant, although perhaps small due to the high degree of resistive damping already present.
In a previous thread I never managed to discover whether the 'observation' offered by means of a frequency response graph was the result of simulation or measurement. If simulation then we do have total and complete system knowledge, at least in principle. If the alleged effect is due to interaction between the added cap and the DAC output circuitry (as Joe seems to think) then a close examination of the DAC output stage and its model ought to throw some light on things.Observation of the phenomena tells that we don't yet have total and complete system knowlege of the total system, which is not at all the same as denying circuit theory.
In any case, we should at least use circuit theory to try to find a first approximation of what is going on. When I tried to do this for one of the simpler examples given by Joe my explanation was brushed aside instead of addressed and, if necessary, criticised.
OK. With that definition of soundstage we can postulate that the cap may work by reducing channel differences, perhaps particularly frequency-dependent channel differences. This is testable: by measurement, simulation, calculation and (possibly) listening.Coris said:The effect of this cap placed as known now, is to enlarge this space of the scene, and place even more precise the sound elements into this field of sounds. The location of the instruments become more precise, as the direction of the component sounds.
Since you haven't excluded a psychological causation, then there's alternative explanations which need to be addressed before coming to this conclusion.
Speaking only for myself, I have ruled out, at least to my own satisfaction, most psychological causations. For example, if I had expectation bias, it was that I would not hear any positive effect as I was rather skeptical beforehand. I've ruled out many other psychological causes as well, simply because the effect is always obvious, consistent and persists though various condition/environmental changes. This tends to eliminate human fatigue, stress and the like factors. Is this scientific, no, this is D.I.Y. not the A.E.S. All that said, I am still open to the possibility that some scientific lostening test might show otherwise. That, too, is part of keeping an open, yet skeptical, mind.
In general, I agree with the axiom that extraordinary reports require extraordinary evidence. But, what extraordinary thing has been reported here? I've seen no report of new physics circuit functioning, only a report of musically interesting perceptual result. The open search for the underlying physical circuit/system cause was one of the hoped for objectives of this thread, I believe.
My point is that you claim more than Joe, but also have seemed to dismiss Scott's 20 year old measured evidence on a plausible mechanism. The fact that all in the 'believer' camp think no blind listening is required because its stereophile style day and night is bound to raise eyebrows.
What I did was make an anecdotal listening report as accurately as my limited ability with language allowed. This is the nature of a non-scientific musical character listening report, isn't it? If your interpretation is that I made more 'claims' than Joe, I suppose, that's your prerogative. However, I'm still searching for the significance of the point you intended that this make? Perhaps, I'm still missing it?
As for the dismissal of any plausible mechanism, I most certainly have not done that. I've specifically been open to all manner of technical causes, such as even the percieved affect of the high frequency reponse roll-off simply being euphonic. Joe's further experiments with EQ'ing back to flat, which I have not yet performed, now strongly argue against that particular possibility. Athe most I've done along these lines is speculate that the cause may be time domain related, due to the seemingly non-linear manner in which the effect seems to subjectively manifest.
As an audio hobbyist, I explore many different avenues that may provide more musically satisfying sound reproduction in my home. Joe's discovery is merely one. It's not a religion. Not for myself, nor for Joe for that matter. The music is the only ends, all else is merely means.
Last edited:
OK. With that definition of soundstage we can postulate that the cap may work by reducing channel differences, perhaps particularly frequency-dependent channel differences. This is testable: by measurement, simulation, calculation and (possibly) listening.
Well, I thought at this possible explanation too. But... There is here about a cap placed over the differential phases of a stereo channel (DAC) output. So there are two caps (which are not exactly identical as parameters) which are placed over the differential outputs of the DAC for each of the stereo channels.
Actually there are here two different circuits (for each of the stereo channels) which are slightly different as parameters (two DAC outputs, two caps, two further processing circuits, and maybe slightly differences for the PCB layouts of these two channels circuits. All these are not at all just identical to explain a lowering of the differences, but the opposite. It is not possible technically for these channels to have identical parameters/characteristics.
The improvement it occur for each of the two stereo channels alone, and then these two channels improved in quality, it bring together an improved soundstage. This improvement for the soundstage it happen only when to apply these caps in mentioned places...
"Reducing channel differences, perhaps particularly frequency-dependent channel differences" it may not fit just well as explanation in this case... Don`t you think?
For me it is quite obvious that all about this cap is an increasing in precision/accuracy of the analogue signals outputted by the DAC chip. I can not see yet how and where this increasing in signals (parameters) precision occur. If this detail it will be found, then it may be quite easy to find a measurement procedure...
Last edited:
Some months ago a thread discussed at great length what is meant by 'soundstage'. There was no agreement. People who use the word all seem to know what they themselves mean by it, but others (who also use the word) disagreed. For judging things that makes it about as useful as the word 'nice'. In most cases 'soundstage' was used of music reproduction where there was no original sound to compare to, as it was all 'plugged in' and multi-tracked in a studio.
As we don't know what is meant by 'soundstage' we cannot know what is meant by 'improvement of soundstage'. Except "it was nice; now it is nicer".
https://en.wikipedia.org/wiki/Sound_stage
Let us assume for the moment that the issue is intermods from ultrasonics. It may not be that, but it is plausible. These intermods are likely to come in bursts which in time correspond to genuine program material. However, the amplitude of the intermods in each channel may depend on parameters which are poorly controlled between samples, such as opamp slew rate limits or first stage quiescent current flow. If the added cap reduces the ultrasonics significantly (plausible), and if the cap matching is significantly better than the relevant opamp parameter matching (plausible), then the bursts of intermods will be more similar between channels with the cap than without the cap.
On the other hand, if the cap increases ultrasonics (also plausible, if less likely) then it could be that the intermods output get limited at the same level for each channel instead of varying between channels (because without the cap they are below the limiting level).
This is all wild speculation, but intended as a sort of 'proof by construction' that adding a component which may not have a tight tolerance could still reduce channel differences if it affects some other parameter which has a very loose tolerance.
Note that the listener does not distinguish between genuine program material and these intermods. In either case when the performer does X the listener hears a burst of sound around position A. A small part of this sound is not in the program. If this varies between channels then it could have the effect of smearing the stereo image around A.
On the other hand, if the cap increases ultrasonics (also plausible, if less likely) then it could be that the intermods output get limited at the same level for each channel instead of varying between channels (because without the cap they are below the limiting level).
This is all wild speculation, but intended as a sort of 'proof by construction' that adding a component which may not have a tight tolerance could still reduce channel differences if it affects some other parameter which has a very loose tolerance.
Note that the listener does not distinguish between genuine program material and these intermods. In either case when the performer does X the listener hears a burst of sound around position A. A small part of this sound is not in the program. If this varies between channels then it could have the effect of smearing the stereo image around A.
Judging (let alone measuring - which means numbers) the "depth and richness" of an "imaginary" auditory experience would not be easy. Hence Wikipedia confirms that 'soundstage' is not a well-defined term.
DF96 said:...In any case, we should at least use circuit theory to try to find a first approximation of what is going on...
Yes, I suspect that we all would agree with that.
The fact that Joe reports the effect only occurs with SDM DACs strongly suggests to me that the cause may be noise floor modulation related. While the quantization noise floor is mostly ultrasonic for most audio SDM DACs, possibly intermodulation is producing an audible effect which increased capacitive loading is acting on in some perceived positive manner. That's just speculation though. I don't know how useful commonly accessible simulation tools would be in exploring such a complex possibility.
I implore you to conduct one of Joe's experiments first. Should you happen to find that you also perceive the described effect, wouldn't the search for technical causality become more interesting? At the least, having the physical experiment in front of you would enable instrumented circuit analysis. Just something to consider.
Gee, thanks Joe. 😡At least we have at last a positive suggestion...
I've been trying to investigate and measure this, too. Problem being that I used to have about 15 or more DACs on the shelf to test, but now find myself down to only two. Not getting good results.
Never thought to say "Send me some and I'll test them". That's a bold step on Stuart's part. He'd be the one to test them for sure, as he has better test facilitates than I, and can do a more rigorous testing procedure.
Upstaged by SY again.
Please don't forget
But please don't forget...
I have no doubt that the relative "improvement" is dramatic by adding the filter. What I doubt is whether the change is dramatic between putting the filter before the I/V and putting the filter after the I/V...
Rolling off top end is not new... I have done it in many ways... And so other people... But this hypothesis about changing the operation of the DAC in a positive way by loading it's output is a new concept...
I am confident enough, despite my bias against these kind of listening tests, that they seem to favour "sameness' for the reasons I gave on Blowtorch II thread...
But I am confident that the difference is so marked that even double-blind "sameness" can't mask it!
How is that for confidence.
But please don't forget...
I have no doubt that the relative "improvement" is dramatic by adding the filter. What I doubt is whether the change is dramatic between putting the filter before the I/V and putting the filter after the I/V...
Rolling off top end is not new... I have done it in many ways... And so other people... But this hypothesis about changing the operation of the DAC in a positive way by loading it's output is a new concept...
Absolutely Ken. My opinion too. Definitely this cap approach it should be tried by those who have an increased explanation potential. To build up a own theory about how this it may works, based on own appreciations, experiences and knowledge is essential for a real progress in this case.
The effect of this cap placed as known now, is to enlarge this space of the scene, and place even more precise the sound elements into this field of sounds.
Creating a speaker that disappears has been my obsession.
When the cone or dome of a tweeter is distorted because of peak in signal, the tweeter will produce mechanical sound. This sound will be perceived coming directly from the tweeter, thus reducing speaker ability to disappear. Any removal of sibilance or energy from the tweeter will normally improve the soundstage.
You may be right... We assume here that the reproducing chain/system is all right. At least the same system is used before and after mounting these caps in place...
I do not have whatsoever such issues/distortions. I can very well hear the high end sounds as all the rest, including very low basses, as coming from their places in the sound scene, and not from the speakers/tweeters. I may point here that the recording, and its quality is crucial for a good reproduced soundstage. All my speakers are a little bit over 100w, and I load it no more than around 10w as usually... The picks are for sure much lower than 100w.
Well, quite a OT post/answer...
I do not have whatsoever such issues/distortions. I can very well hear the high end sounds as all the rest, including very low basses, as coming from their places in the sound scene, and not from the speakers/tweeters. I may point here that the recording, and its quality is crucial for a good reproduced soundstage. All my speakers are a little bit over 100w, and I load it no more than around 10w as usually... The picks are for sure much lower than 100w.
Well, quite a OT post/answer...
Last edited:
Never thought to say "Send me some and I'll test them"...
Upstaged by SY again.
Sorry Pano, if only it was that simple. No, you have not been upstaged.
Note that I did cover this:
I don't want to give a quick response right now, in view of Stuart's post I think it has at least some merit, but the logistics may be the problem.
So I did manage to quickly cover myself. No upstaging yet. Not quite the solution yet.
Because logistics is the problem, not the solution. It might have been possible and have opened a way, if we were leaving in rather closer proximity to each other. I live Down Under and supplying Stuart with two identical DACs is a problem. It's also likely to be very expensive. Not to mention shipping from Oz is a bummer.
But I did want to say, that the proposal at least had merit and the thought positive, but... still there are some obstacles.
Also, I am just expressing a hope that we can turn the thread around in a more positive direction.
So if anybody knows how we can go ahead with Stuart's proposal or down that road, I am at least willing to listen.
Cheers, Joe
Absolutely Ken. My opinion too. Definitely this cap approach it should be tried by those who have an increased explanation potential...
Yes, this is about getting motivated.
Nothing will ever get done without motivation... NOTHING!
Not even getting out of bed. 😀
Cheers, Joe
There is a guy I know who has built my Elsinore MK5 (current is Mk6) loudspeakers, and he put it this way.
Yeah DAC/filters can be quite controversial for some reason.
Have you measured what the capacitor placement is doing to the signal? It is pretty easy to create CDs with 1-bit impulse wave files, or periodic pink noise that ARTA or other measurement programs can capture and analyze.
I recently did just that with one of Pioneer's Legato Link players as I was curious exactly what the "secret sauce" was. Turned out to be a linear phase filter with fairly slow roll-off...about -0.25dB down at 10Khz, -4.5dB at 20Khz. That really cleans up the ringing/pre-ringing on impulse and tone burst signals.
Best Regards,
Xxxxx
Does this give any of you guys any ideas?
-
Yeah DAC/filters can be quite controversial for some reason.
Have you measured what the capacitor placement is doing to the signal? It is pretty easy to create CDs with 1-bit impulse wave files, or periodic pink noise that ARTA or other measurement programs can capture and analyze.
I recently did just that with one of Pioneer's Legato Link players as I was curious exactly what the "secret sauce" was. Turned out to be a linear phase filter with fairly slow roll-off...about -0.25dB down at 10Khz, -4.5dB at 20Khz. That really cleans up the ringing/pre-ringing on impulse and tone burst signals.
Best Regards,
Xxxxx
Does this give any of you guys any ideas?
-
If you use mail (insured, of course) and you're not sending boat anchors, it's not at all expensive. Priority mail insured for 20 pounds (slightly under 10kg) is $83.
I'll sweeten the deal (easy for me to say because I'll get paid for the article, regardless of the outcome 😀): you pay postage here, I'll pay the return postage.
edit- I will restate what anyone who knows me already is aware of: I disclose everything about my test setup, controls, and error bars so that anyone who wants can try to replicate.
I'll sweeten the deal (easy for me to say because I'll get paid for the article, regardless of the outcome 😀): you pay postage here, I'll pay the return postage.
edit- I will restate what anyone who knows me already is aware of: I disclose everything about my test setup, controls, and error bars so that anyone who wants can try to replicate.
- Status
- Not open for further replies.
- Home
- Source & Line
- Digital Source
- Practical Implementations of Alternative Post-DAC Filtering