Mis-application is not my problem. There are lots of things that contribute to PSRR, this circuit is hardly an op-amp with gain and nothing else. Start with AC input impedance mis-match and work from there if there is a problem. The DS clearly shows 80-90dB at 10kHz PSRR, figure 12.
Ed's claim is that this is only true at G = 1, and deteriorates at higher gains. That seems surprising to me- in the SSM and THAT devices that I've been using, it goes in the opposite direction, higher gains have higher PSR. But PSR versus gain isn't given in the 797 sheet, so I was hoping that you could point to some data.
Yeah the impedance mismatch at the inputs of the 797 in the AS019 Jensen circuit are substantial and CMRR is closely related to PSRR for most opamps. There is probably one gain setting for a given source impedance to the left of the input transformer that works best, and that's almost certainly not at highest gains.
Last edited:
SY asks a good question does PSSR go up or down with gain?
He has shown units that go up, my anecdotal experience is down, but I'll have full curves in a bit. Now from a practical sense -150 db noise re 1 volt is pushing things in many environments, calculating numbers in the -200 dB range is an academic view most likely to fail in real world applications.
I was looking at the guaranteed spec of -120 dB (depending on version etc.) PSSR and then the derating from the typical curve. So while you should (and most likely will) have 80 dB PSSR at a gain of one it is not guaranteed.
Now as to the Jensen circuit, it pops up on this website often. Now I suspect it started as an NE5532 circuit and some one just "updated" the part number.
That is correct you are not responsible for misuse of the AD797. So I'll blame John for it as that is the practice here.
But the issue was who would use the AD797 in a circuit that had such poor performance? This is just one example of where it is not properly applied.
Around here we built a bunch of different mic preamp designs. It was done as a teaching example for the new guys. They picked some designs, I added a few. One example presented on multiple websites clearly was not even functional! Others were 'classic" designs picked for their "sound qualities."
We tried a bunch of different transformers from four different manufacturers. The circuits ran the gamut from germanium bipolar junction transistor through FET to IC. (Yes I spent a few grand on parts for this!)
It quickly became clear that batteries provided the simplest clean power supply. Now batteries have limits not just on capacity but also sometimes can have bubble noise.
Some of the best sounding mic preamps we tried have poor PSSR.
So since I have been working on my power supply article it seemed worthwhile to build an AC supply that would work in all cases as well as or better than batteries. The other option would be to redesign the preamps to have the same sound performance but better PSSR. Quiet power supplies seemed easier and a more universal solution.
ES
Yes. I wonder over why better supplies aren't used. Although it is laudable to make a circuit as high-rejection as feasible, it's also not that difficult to make very clean supplies, except perhaps for the details of common-mode mains noise.
Batteries are almost ideal for their isolation, although even that is not perfect --- the physically larger they are, the more they have capacitance to the rest of the universe. If a few have excess noise, then a very quiet regulator can be provided, and will be a good idea anyway to stabilize things as the batteries discharge.
I'm not sure about the importance of PSRR in a real circuit .
If you have a properly designed power supply, its ripple+noise should be under 80dB (as Mr. Self shows in his book), so... provided the OA is able to mantain a PSRR about 30dB over the entire audio band (quite common), the OA output "ripple" should be under 1uV... mostly below the noise floor...
abraxalito, this whole scenario you formed from thin air over there about SD, is entirely in your mind, you know that? its just your own rhetoric you like to repeat every now and then and you have superimposed it onto text of the test, as well as misunderstanding the test perhaps, but youre right, no point going there about the test, it may be another case of you framing the same thing i'm saying slightly differently, so it appears different enough to continue argument, while being quite baffling.
are you talking about a different test? that would explain the confusion, but do you think you could leave out the religious commentary? especially given last I knew, you were using cheap NOS? dacs with x7r or worse, ceramic cap based active filters and DSP performed on ARM dev boards; so yes perhaps you are right not to worry about low jitter clocking
regarding 100ns and SD dacs (not nS .. nano Seimens?) he said nothing of the sort and almost nobody mentioned what dacs they were using, in either of the 2 threads, except for Steve and someone else with an ASUS and its not even clear that guy did the test.
it was designed to be the significant enough for people to notice, nothing to do with dac type, its not impacted by dac type, each different dac will add its own additional intrinsic amount regardless
DACs are much better now, 4 years later, but to remind you of the actual words.
what type of dacs people were using in the majority, you can only have made up …
I myself havent taken the test, also i'm running synchronous clocking on my ESS and digital filtering on my mac, of the apodising type. so neither the async, or much of the internal filter is in use
are you talking about a different test? that would explain the confusion, but do you think you could leave out the religious commentary? especially given last I knew, you were using cheap NOS? dacs with x7r or worse, ceramic cap based active filters and DSP performed on ARM dev boards; so yes perhaps you are right not to worry about low jitter clocking
regarding 100ns and SD dacs (not nS .. nano Seimens?
) he said nothing of the sort and almost nobody mentioned what dacs they were using, in either of the 2 threads, except for Steve and someone else with an ASUS and its not even clear that guy did the test.it was designed to be the significant enough for people to notice, nothing to do with dac type, its not impacted by dac type, each different dac will add its own additional intrinsic amount regardless
DACs are much better now, 4 years later, but to remind you of the actual words.
what type of dacs people were using in the majority, you can only have made up …
I myself havent taken the test, also i'm running synchronous clocking on my ESS and digital filtering on my mac, of the apodising type. so neither the async, or much of the internal filter is in use
Last edited:
the AD797 has an option that many op amps don't - you can connect the external decompensation C to the - rail to improve PSRR at high gains
a 2-pole variation on that theme: http://www.diyaudio.com/forums/solid-state/123613-class-biasing-ad797-6.html#post1525757
a 2-pole variation on that theme: http://www.diyaudio.com/forums/solid-state/123613-class-biasing-ad797-6.html#post1525757
In doing my power supply article I found that there are a number of audiophile design facts that I was unable to verify. Some others stood up well. Then there were a few new tricks. This is my microphone preamp power supply.
Note the transformer secondary reversal, that really got rid of even more noise than is shown in the attached AP plots. Right now the noise measured from rail to rail is around -150 dB re 1 volt. My design goal is 160 dB, so I may play a bit more.
ES
Attachments
Because the damping resistor (15K) is in parallel! You might need to take a closer look to see how the same resistor gets the transformer's capacitor and the diodes'!
Attached is the difference in output noise from the secondaries in and out of phase. The baseline is with the power supply unplugged.
Attachments
No, I was previously blissfully unaware it was entirely in my mind I must admit. This looks like an argument from authority though - do you have any explanations for where I'm overlooking stuff? Or any evidence that (for example) what Bob Adams says about S-D DACs sensitivity to jitter is entirely in his mind?
are you talking about a different test? that would explain the confusion, but do you think you could leave out the religious commentary? especially given last I knew, you were using cheap NOS? dacs with x7r or worse, ceramic cap based active filters and DSP performed on ARM dev boards; so yes perhaps you are right not to worry about low jitter clocking
Whither the 'religious' commentary? Chapter and verse if you please otherwise I'll conclude that's entirely in your own mind. What's wrong with ceramic cap filters (using NP0s there btw)?
Thanks for quoting his actual words - they clearly show his misunderstanding that audibility is largely only a function of the magnitude of the jitter (he's aware of the secondary aspect of its spectrum) rather than of the architecture of the DAC. That's something you've yet to grasp perchance?
Simple two transistor regulator. I used nearly the same circuit (scaled to higher voltages) in my MC preamp, His Master's Noise. Very inexpensive, works quite well.
Of possible interest, I tried doing the jitter comparison before reading the thread, and finding the posted results. And, I got the clean version right, but, the ranking of jitter I got perfectly in reverse order!! In other words, I ranked 0ns, 100, 30, 10, 10.
Now, this was using the PC monitors, with on-board relatively cheap S-D chip, whatever it is. Subjectively, there were slight differences between the samples, which I ranked on the basis that the worst jitter would sound the least natural, would be the edgiest, would have the most 'mechanical' quality to it. I did this in two go's, as in did a ranking, then left it for a few hours, came back and confirmed wther I was happy with that order, which I was. Yet, apart from the 0ns sample I got it completely back to front!
So, I'm wondering whether possibly the S-D characteristics being discussed are contributing to this ...
aaahh so now it doesnt matter that your claims were something you inserted into the commentary? if authority means reading the entire two threads and not seeing sigma delta mentioned by him anywhere at all and now discovering its unlikely you even read the thread, then sure, I guess reading the thread does give a certain authority, when apparently you didnt.
you are treating it like its my position, or i'm endorsing the whole thing? you dont remember my objection earlier? Jesus. get real mate, I didnt link it, or endorse it as an accurate test/model for all jitter we hear, just mentioned it was curious it wasnt audible to a certain person. this argument is boring, i'll leave you to argue with yourself, its becoming familiar
btw youve mentioned using X7r for signal coupling in this very thread, its not my fault if your claims were inaccurate
you are treating it like its my position, or i'm endorsing the whole thing? you dont remember my objection earlier? Jesus. get real mate, I didnt link it, or endorse it as an accurate test/model for all jitter we hear, just mentioned it was curious it wasnt audible to a certain person. this argument is boring, i'll leave you to argue with yourself, its becoming familiar
btw youve mentioned using X7r for signal coupling in this very thread, its not my fault if your claims were inaccurate
Last edited:
- Status
- Not open for further replies.