'Relevance' is a new metric in this discussion, of course it is not relevant. But that wasn't the point I was making. Talk to Mark about that.
Vacuboy, so would you think talking about (or otherwise invoking) "regulator linearity" makes any sense?
I am sure there is also an effect of the moon tidal forces on capacitor distortions, since the moon gravity distorts the capacitor geometry, therefore the capacitance is modulated, leading to distortions.
I don’t understand why a Hilbert transform should be used at all.
When computing the FFT of a ‘real’ time domain signal, the result is a complex frequency domain signal.
After setting all complex values above 22,500 hz to zero, all that has to be done is to compute the inverse FFT resulting again in a ‘real’ time domain signal.
Hans
Last edited:
Because the Hilbert transform delivers the signal envelope (that is, the envelope is the analytic signal amplitude).
Another way to look at this paradox would be to explain how a sequence of linear operators (addition, Hilbert transform, inverse Hilbert transform, everything in the time domain) can lead to a non linear behavior (that is, distortions).
I find other published parameters of the LT1963 clearly better than 7805, no question. Noise figure however, is kind of comparable.
From datasheet: "Output voltage noise is typically 40nV/√Hz over this frequency bandwidth for the LT1963 (adjustable version). For higher output voltages (generated by using a resistor divider), the output voltage noise will be gained up accordingly. This results in RMS noise over the 10Hz to 100kHz bandwidth of 14μVRMS for the LT1963 increasing to 38μVRMS for the LT1963-3.3."
I noted without the resistor divider the adjustable version will output 1.2V hence at 5V output noise is expected to be ~ 55μVRMS. Seems LT1963 has crap noise performance as you said, unfortunately it is the only one Markw4 said he tried in comparison to the 7805 in the AK4499 eval board. Makes one wonder what actually went on.
Last edited:
Ok. Got it, thank you.
Hilbert was used twice for the original and the filtered time domain signals to get their envelopes.
I missed that in Scott’s explanation.
Hans
I think what Scott did is this:
- Band-limit the original signal by FFT to 22050 Hz.
- Use Hilbert to convert the original and band-limited signals to analytic form and take the (complex) magnitude to get the envelopes.
- Difference the envelopes.
Some commercial dac designers say it does make an audible difference. I am not a commercial dac designer, nor the first to think of it. The post I linked earlier (#20) to refers to the Linear Audio article by John Walton, "A comparative overview of power supply regulator designs with listening tests." Apparently, John thought there seemed to be some correlation between listening test results and measured regulator THD.
You can poo-poo it all if you want, doesn't bother me.
However, if you start attacking me personally in a way that is permissible to the mods, they I may choose to ignore or to return the favor.
Have a nice day 🙂
Last edited:
Link or whereabouts?
But you are still giving him treats by responding. 🙁
Poo-poo-ing is attack, no?
As I've warned.
There is another case involving dacs that I wrote about in another forum thread.
It involved an ADM7150. It didn't sound very good for AVCC, but I was able to make it sound quite a bit better by increasing the voltage drop across it by a couple of volts, and by adding a 33-ohm resistor to ground thus increasing current draw by 100ma. Such things often improve regulator performance. The data sheets are often suggestive, say, showing that PSRR varies with voltage drop.
Never sounded as good as using an AD797 opamp buffer AVCC supply though.
I will not go into trying to explain what 'good' or 'better' sounds like in this context. Suggest trying it, if curious.
Last edited:
You too.
I have the Linear Audio Volume 4 with the John Walton article you quoted. Can you point me where exactly (page and paragraph, please) does he invoke the "regulator THD" measurements and correlation with the listening test results? Since I have only the paper version, I can't do a text search.
Thank you.
John Walton AKA Jackinnj? Linear Audio | your tech audio resource
For some reason I could download the Pdf for free but wont post that link until I have checked with Jan that it should be possible. The testing was with a preamp with 40dB PSRR. I see no mention of regulator distortion though. I'll have to re-read.
Bottom line...Bybee is rubbish 😀
Edit: scanned pdf, no THD mention for the regulator
For some reason I could download the Pdf for free but wont post that link until I have checked with Jan that it should be possible. The testing was with a preamp with 40dB PSRR. I see no mention of regulator distortion though. I'll have to re-read.
Bottom line...Bybee is rubbish 😀
Edit: scanned pdf, no THD mention for the regulator
That's what I thought, just another made up argument to defend the indefensible. Poo-poo.
For some reason I thought one of the papers used fMRI, but yeah, both PET/CT and fMRI are (usually, there's some pretty incredible work going into new radio-tracers) going to be looking at spatial metabolic activity.
I work on the hardware side of a cancer diagnostics lab, in a way trying to design a screening technology for first regular testing before we have to escalate to imaging/biopsy for diagnosis. So, yeah, this stuff is pretty interesting to me.
And, welp, conversation moved on. Whatevers...
Don't think that bit made it to the article, but the post I linked to https://www.diyaudio.com/forums/pow...ofit-upgrade-317-based-reg-2.html#post5642205
...lists properties of super regulators as follows:
The post seems to refer back to his study on regulators with listening tests for context, including something about whether Jan would be interested in follow up work.
Thus, I take it he wasn't sure enough about it to stick his neck out in a formal publication, most people aren't. But, he seems to think it could use some more investigation based on suggestive correlation with measurements/listening.
Given that reading of it, I think it comports reasonably well with what some others have observed.
No one seems to have exactly pinned it down. Not surprising in that it isn't the only variable in play circuits that humans listen to. As he suggested, more work should probably be done on it, presumably by people who like to run studies to publish.
Last edited:
- Status
- Not open for further replies.