Can´t be easier. The voltage reference can be remote since no fluctuating current is drawn. Only EMI to consider and that can be shorted by a cap at the opamp input.
I am using a distributed opamp setup¨to generate virtual earths in an active filter (Sallen-Key configuration sources and sinks appreciable currents to ground). I think for this use it is warrented; to make it useful to generate rail voltages, you would need to have dismal PSRR in the serviced circuit in order to justify the expense imo.
Actually it is better if the rails not only do not track but are of unequal voltages for preamp level circuitry.
I've looked around but not carefully enough to get a solid answer. That's the entire dynamic range. But it's a moving band of sensitivity that requires a noticeable time to adjust, much in the same way as eyesight (try going from a bright room to staring at the stars... it takes 5-10 minutes to fully acclimate). I don't know the literature about how long it takes nor the spot dynamic range, but given the difficulty many of us have with Bill's sousa band experiment (with attendant masking), we're talking closer to 60 db spot range.
So that would certainly relax the 150 dB range you're suggesting.
I see a whole lot of opinions from the critics, but is this getting us anywhere?
One of the 'joys' of engineering is to do a 'great' design, something better than what is normally done. Sometimes this means that you are 'getting away' with something, by using skill and experience, and you make a better product at lowest cost. Other times, it is to make the best design possible, hang the cost, but get it competitive with the outside world, and perhaps WIN a listening comparison with other products. Sometimes, we paint ourselves into a corner, by concentrating on one part of a design and find that we have something that is vulnerable to the outside world, through the power supply, etc, and we have to put extra attention to the power supplies to keep things OK. In any case, it can be good engineering, rather than mediocre engineering. That is the goal.
One of the 'joys' of engineering is to do a 'great' design, something better than what is normally done. Sometimes this means that you are 'getting away' with something, by using skill and experience, and you make a better product at lowest cost. Other times, it is to make the best design possible, hang the cost, but get it competitive with the outside world, and perhaps WIN a listening comparison with other products. Sometimes, we paint ourselves into a corner, by concentrating on one part of a design and find that we have something that is vulnerable to the outside world, through the power supply, etc, and we have to put extra attention to the power supplies to keep things OK. In any case, it can be good engineering, rather than mediocre engineering. That is the goal.
You are correct - in my haste I did not check the drawing.
Thank you for pointing that out.
"The PSSR issue is really at what level is power supply noise an issue. As the range of human hearing without damage is around 140 dB a design goal could be 150 dB to allow some margin for other things such as noise and distortion that also contaminate the signal. "
Naah - that's absolute nuts. Sorry to be blunt.
Take the 'systems' view to bring a rational approach to you design goals (apologies SY). Further, not many systems have real world DR's of 90 dB considering the source material, ampifer, speakers etc. The lowly LP is 60 dB and that's a good pressing and CD is 90 dB on a good day.
Naah - that's absolute nuts. Sorry to be blunt.
Take the 'systems' view to bring a rational approach to you design goals (apologies SY). Further, not many systems have real world DR's of 90 dB considering the source material, ampifer, speakers etc. The lowly LP is 60 dB and that's a good pressing and CD is 90 dB on a good day.
There are many different users. In a stadium empty someone may indeed put their ear up to a loudspeaker and expect it to be quiet. Outside noise levels can run around 40 dBa so that requires 10 dBa. Then when the crowd cranks up you want a peak level of 115 dBa minimum at the crowd which can be 200' to 800' away. No trouble meeting this apparent 140 dB S/N! But I suspect you are missing the tricks to getting there in a stadium. (Hint noise drops when you parallel sources. See http://ashly.com/wp-content/uploads/2016/03/nX-multi-mode-amplifiers.pdf for specs and figure there are 400 channels of amplifiers used!)
But the 150 dB is a target. It is not met by a single chain. Which is why I said "For the perfect digital CD standard my arithmetic says 94 dB to the output from a nominal 5% ripple. Some from the regulators and some from the circuit under power. "
Derfy, I have always modeled the ears as a 70 dB dynamic range with a 70 dB level adjuster. But unless you are riding the volume control you will always have users mismatching levels, so the system needs more than 70 dB. At 150 dB you are essentially error proof. Except for the idiot who connects his earphones to the 1000W power amplifier output. Just call him very very very very loudly after that.
Last edited:
Yes. Not many people know or like tight, clean sound with very little artifacts.
Jan
Right, I wasn't insinuating that 60-70 dB is a target SNR, just that 150 dB is, well, a little on the deep end. Let's settle for 100 db SNR and call it good?
I think Gerald Stanley argues for at least an amplifier's output noise to be 115dB below full output, for someone in the vicinity not to notice hiss from a moderately efficient loudspeaker.
When I was doing a switchmode amp for a now-defunct company we were getting about 30uV rms unweighted at the output for a 100W into 4. I have forgotten the exact gain, but we could not get our hands on a digital source that was not dominating the overall system noise performance, something like 100uV rms at the amp output. I think that was with a Bel Canto DAC borrowed from Andrew Jones. I had to keep reminding the putative head of the effort that the noise he was hearing out of one of Andrew's good-sized speakers was not from the amplifier.
When I was doing a switchmode amp for a now-defunct company we were getting about 30uV rms unweighted at the output for a 100W into 4. I have forgotten the exact gain, but we could not get our hands on a digital source that was not dominating the overall system noise performance, something like 100uV rms at the amp output. I think that was with a Bel Canto DAC borrowed from Andrew Jones. I had to keep reminding the putative head of the effort that the noise he was hearing out of one of Andrew's good-sized speakers was not from the amplifier.
Derfy, as I keep mentioning 94 should do it. I don't know why you keep bring up 42?
Brad,
The current Crown audio power amplifiers are around 102, the Ashly's I use start above that and the big ones spec' at 114 and typical is 118.
Yes I have had a consultant stick his head in a live horn loaded loudspeaker that was capable at 1M to do 144 dB to check for hum and noise! So he could have experienced 150 dB or more!
That must have been tempting...
Many, many years ago, I was at a Hawkwind gig -- very loud. At the end it became apparent that someone had nodded off with his head slumped into a bass bin! A swift kick from a roadie brought him round... God knows if he ever heard anything ever again!
Good morning gents ----
Well that was fun.... SY is correct but wrong, as usual. Missed the point for the more narrow tech call.
Flushing out the issues with a spec called PSRR and what is actually required is the question and just a number like PSRR does not indicate any superiority at all. Rather the PS and amp together plus speaker effec and acceptable distortion level plus noise of room et al factor into what is needed. So we have a range from 60dB to 100dB. And, as was mentioned, this can be taken care of in both supply and amplifier.
So, when someone says your circuit Only has xxdB PSR, it doesnt mean much of anything without the PS noise numbers as well. The total and not one in isolation. And, of course the entire system is really the number you end up with.
THx-RNMarsh
Well that was fun.... SY is correct but wrong, as usual. Missed the point for the more narrow tech call.
Flushing out the issues with a spec called PSRR and what is actually required is the question and just a number like PSRR does not indicate any superiority at all. Rather the PS and amp together plus speaker effec and acceptable distortion level plus noise of room et al factor into what is needed. So we have a range from 60dB to 100dB. And, as was mentioned, this can be taken care of in both supply and amplifier.
So, when someone says your circuit Only has xxdB PSR, it doesnt mean much of anything without the PS noise numbers as well. The total and not one in isolation. And, of course the entire system is really the number you end up with.
THx-RNMarsh
Well, >100 dB ref full output is achievable. What I don't think you get is 90 dB (say) at 1 W output (~2.8 V into 8 ohms) and then your full entitlement at full output (say 60 V out) of >120 dB because ripple shoots up, IR issues in the ground wiring arise etc. Once you start delivering power into the load, life gets a bit more difficult.
(time for AFEC)
(time for AFEC
)I'm glad someone at Ashly knows what they are doing. Some years ago I was testing one of their DSP-based EQ boxes and the noise performance was poor. When I told Marshall Buck about this he doubted my noise measurements, but then he was hard-wired not to believe anything that didn't have the imprimatur of Audio Precision, for among other reasons because he was selling used ones
I pointed out that I had been making noise measurements when AP was just a gleam in Hofer and Cabot's eyes, when they worked for Tektronix. Buck was characteristically unmoved.
Well, >100 dB ref full output is achievable. What I don't think you get is 90 dB (say) at 1 W output (~2.8 V into 8 ohms) and then your full entitlement at full output (say 60 V out) of >120 dB because ripple shoots up, IR issues in the ground wiring arise etc. Once you start delivering power into the load, life gets a bit more difficult.
(time for AFEC
As Derfy pointed out in actual use 70 - 90 dB is often enough. With a crowd in the stadium 45 may actually be fine. So high power issues not so much.
However today's project is a heterodyne filter for a 4 channel switching amplifier. Stadium empty means the beat frequencies are an issue.
Then there was the time I mentioned to a manufacturer that I only measured 40 dB S/N. They disagreed they got 60. I was using 0 dbm as the input they were using clipping. Here I thought we were talking about dynamic range which is noise to clipping. Ah well.
I'm glad someone at Ashly knows what they are doing. Some years ago I was testing one of their DSP-based EQ boxes and the noise performance was poor. When I told Marshall Buck about this he doubted my noise measurements, but then he was hard-wired not to believe anything that didn't have the imprimatur of Audio Precision, for among other reasons because he was selling used ones
I pointed out that I had been making noise measurements when AP was just a gleam in Hofer and Cabot's eyes, when they worked for Tektronix. Buck was characteristically unmoved.
They bought an AP! Also have one of Ron Schafer's students working for them.
But the what do I know? I was using Klark Teknik's audio delay version 1 in large churches and that was only 12 bits. Other folks found it didn't have enough dynamic range.
- Status
- Not open for further replies.