We certainly need to know how the various mfr test/measure their PSRR to compare apples to apples. And just as certainly how much PSRR is needed in our use?
The question for audio is how much psrr is needed? Now line level circuits will all be powered via voltage regulators with low noise and ripple output. So really, only PA need to be questioned IMO.
So, I just took a commercial PA off the shelf and measured the ac ripple and noise on each rail (true rms wide band) which uses +/- 10,000mfd/60vdc after rectifier... non-regulated PS..... the ripple and noise was 35mV NL.
So with 40db psrr we would have 0.35mV ripple and noise at the speaker terminals? is that inaudible? If not, how much more psrr is required for inaudible?
THx-RNMarsh
Last edited:
And here's an interesting link it this regard:
Dr. Diamond MD - Human Stress Provoked By Digitized Recordings
Fire on!
And here's an interesting link it this regard:
Dr. Diamond MD - Human Stress Provoked By Digitized Recordings
Fire on!
Watch out SY! All due to the digital delays at Carnegie Hall, if you must go up to St. Paul and try a goober burger.
20 Log (.00035 x 1.414 / 60) = -102 dB re full power. So with 30 dB of headroom (the maximum anyone should need) -72 re average level. With 42 dB of dynamic range as allowed for in the MIDI standard that still leaves -30 dB of noise below signal. Throw in the Fletcher Munson sensitivity curves at 120 Hertz and that should be just below anyone's threshold.
Now putting your ear near a horn loaded loudspeaker in a quiet room with the amplifier input shorted you may be able to hear it.
As P= E**2/R or 1.5e-8 watts into 8 ohms or -78 dBW. At 92 dB/W @ 1M = 14 dBa SPL. Hearing threshold at 120 Hertz is around 20 dB SPL. So should meet the threshold of hearing at 1/2 M.
"Now putting your ear near a horn loaded loudspeaker in a quiet room with the amplifier input shorted you may be able to hear it."
Ed,
Wouldn't that still have a gain component to it, the level that the amplifier was set or would the ps ripple be the same at any level with the input shorted?
Ed,
Wouldn't that still have a gain component to it, the level that the amplifier was set or would the ps ripple be the same at any level with the input shorted?
Normally there is additional filtering on the input stages, so output ripple should be primarily from the power supply and less effected by gain, other than reduction of global feedback by forward gain.
.35 mV is probably not a problem for most systems. I have measured quite a few power amps and if you get around 500 uV or less you will get very few complaints. 200 uV or less is pretty easy to do and makes me wonder why some commercial products are well into the mV range and these tend to mate with horn type stuff. But these are designed not engineered.
Well, it doesn't take much to do better than this typical PS example I measured. But, I am making the point that extreame PSRR is not needed. If 40db takes care of most typical systems, surely >50-60dB is over the top.
Though many CFA topologies do not have benefit of high nfb, such as found in VFA, the CFA PSRR is more than sufficient in practice.
THx-RNMarsh
Though many CFA topologies do not have benefit of high nfb, such as found in VFA, the CFA PSRR is more than sufficient in practice.
THx-RNMarsh
Last edited:
I'd offer my services but I'm too busy being a genius.So.... What, say, 5 people would you want to add to a list of vital geniuses...?
I need a drink to consider it...
Drink? That is for those who need a social crutch, like geniuses. I think I'll join ya.
No, just listened to demo material and watched how digital editing worked. I went home and kicked my 2 track Ampex 351 several times.
that's new, care to expand? So PSRR is not that important, how convenient for rescuing your CFA pet
.
50-60db "over the top" ??
Maybe if the amp is used as a glorified virtual ground and does not
need to be low distortion at higher levels.
That does sort of negate any sort of "audiophile" CFA's , huh ?
I "crippled" my newest CFA to <60db. 60hz THD rose X10 (50-100ppm) .
Allowing it's standard -100db .... I was back down to 5 .
IMD takes a "hit" at low rejection , as well.
Is there any audible advantage to a low PSRR circuit that increases in
distortion because of this ?
I always try for that 100db , not really hard to mix CFA and VFA "tricks".
Accuphase does as well ... (my new one- below) is similar but simpler.
OS
Attachments
So PSRR is not that important, how convenient for rescuing your CFA pet
yes. very convenient.
-RM
I suspect you affected a lot more than just PSRR if the distortion increased.
Nothing wrong with over-the-top nor 100dB psrr, --- just dont need that much psrr for inaudible ripple and noise coming thru from the PS.
THx-RNMarsh
Last edited:
Presumably this was in a quiescent condition, no music playing and no load connected. So measure again with the amp pumping out audio into a real speaker as load.
a bit off topic
I had to share this. I've been working with a new client who has ambitious plans for a line of audio products. The skill levels of the participants are wildly variable. One of the products requires a MM phono preamp.
As I do generally, beyond simulations I build breadboards, typically on ground plane and pad-per-hole protoboard material, and use through-hole components where available. In this case a number of parts had to be SMD, so adapter sockets were used. Rather than incur the extra series noise of dual JFETs I elected to use a d.c. servo in the input stage, and rather than a series feedback configuration to determine gain, I'm running with a voltage follower with a bootstrapped cascode followed by a noninverting stage with gain.
After a number of layout blunders on the part of the novice person a rev 2 board was gotten to operate to the point I could get it into my lab and see how well it performed. Noise wasn't too much worse than the breadboard, but (1) there were microphonics at a worrisome level and (2) the distortion with level was disappointing---out of noise things just never got very good, although many would have accepted it.
After trying a number of things and determining that the input stages were the primary source of the distortion, and removing the loading on its output, I was getting essentially nowhere. Then I looked at the BOM again.
A capacitor that I specified as C0G on the schematic appeared as X7R. Eeeeeeek! Fortunately I had some proper parts. Substituting the right dielectric reduced the THD out of noise to nearly the Ap residual, for me around 3ppm. Before the cap change the THD at that input level was 48dB worse! And now the microphonics are drastically reduced as well.
So all in all a very good day!
I had to share this. I've been working with a new client who has ambitious plans for a line of audio products. The skill levels of the participants are wildly variable. One of the products requires a MM phono preamp.
As I do generally, beyond simulations I build breadboards, typically on ground plane and pad-per-hole protoboard material, and use through-hole components where available. In this case a number of parts had to be SMD, so adapter sockets were used. Rather than incur the extra series noise of dual JFETs I elected to use a d.c. servo in the input stage, and rather than a series feedback configuration to determine gain, I'm running with a voltage follower with a bootstrapped cascode followed by a noninverting stage with gain.
After a number of layout blunders on the part of the novice person a rev 2 board was gotten to operate to the point I could get it into my lab and see how well it performed. Noise wasn't too much worse than the breadboard, but (1) there were microphonics at a worrisome level and (2) the distortion with level was disappointing---out of noise things just never got very good, although many would have accepted it.
After trying a number of things and determining that the input stages were the primary source of the distortion, and removing the loading on its output, I was getting essentially nowhere. Then I looked at the BOM again.
A capacitor that I specified as C0G on the schematic appeared as X7R. Eeeeeeek! Fortunately I had some proper parts. Substituting the right dielectric reduced the THD out of noise to nearly the Ap residual, for me around 3ppm. Before the cap change the THD at that input level was 48dB worse! And now the microphonics are drastically reduced as well.
So all in all a very good day!
Any signal that would cause rail pumping in an amp (assuming a modicum of a psu investment) would also be perceptually masked. Cuz big signals are loud. Likewise, we'd expect to see 50/60 Hz spurs and their harmonics to show up big time in a imd stress test.
Sticking your head in your speaker without an input signal is probably the closest perceptual test for psrr (and input/noise hash) you'll easily find. Does not come recommended to stay in said position, especially if you're helping Ed test an installation. ;-)
Sticking your head in your speaker without an input signal is probably the closest perceptual test for psrr (and input/noise hash) you'll easily find. Does not come recommended to stay in said position, especially if you're helping Ed test an installation. ;-)
Last edited:
- Status
- Not open for further replies.