Yes you did:
"my suggestion of two R-Core transformers seems to give the best possible solution (save, of course, batteries with filtered output)."
jan
has anyone tried chokes on output of the supplies or Johns floating common mode regulators? i bought most of the parts for the latter to give it a try, some of the circuits hes playing with have pretty low PSRR so i presume they would suffice
You are correct, it is my mistake.
I'd like to rephrase my quoted statement.
Since I didn't compare batteries with filtered output to my suggested PSU, I cannot say if one of them is better and which one is better.
I agree. In cases like this, it is difficult to judge what will be better without all-out building and measurement. But we can make an educated guess.
In the case of the battery, the only thing to worry about is the battery noise which is known and can be tamed relatively easy.
With a floating mains supply we see that there are many more issues than with the battery, like capacitance to the mains, which probably is not linear, and ingress of mains noise and impulses through the supply.
All these can be solved, it's just a matter of engineering, but it stands to reason that it will cost effort in time and parts to do it.
So it seems a safe bet that a battery supply is at least easier and less problem-prone. Couple that to a few posts of people who tried and reported that the battery supply is cleaner, the decision seems easy.
But, this is diy so only you decide what you want to build. That's personal preference.
jan
I agree with most of what you said.
The drawback I personally find in batteries is the need to replace, or charge, them -- which may occur in the middle of listening session. It can be overcome by pre-planning, yet I find it inconvenient.
As for the report given here about battery supply being cleaner, no specifics were mentioned about the mains supply it was compared to, so there is no way to tell. Batteries are definitely cleaner from most mains supply. If they are cleaner than all mains supply is yet to be determined, by experience. (Experience includes both measurements and listening evaluations).
I determined that the floating 24V SMPs were the source of my noise problem.
I have done a test with regulated linear supplies with much lower stray capacitance between the mains and low voltage side and the noise is now below what I can measure on both the unbalanced SEN half and the balanced output. If I turn my 100W tube amp up to full volume I can still hear a little noise from the new PSUs if I put my ear within about 6 inches of the tweeter of my 88dB speakers. Normal listening volume is 30dB below this where it is inaudible.
I also grounded the current drain PSU and the SEN end of the shields of the output cables.
Listening test tomorrow.
The new PSUs are nothing special - 20RMB each I think (about $3). They are 12V. I also have some unregulated 18V ones to try.
I did another measurement with the sound card. Noise is down below the measurement floor but distortion results don't look that great. 2nd harmonic is about -40dB and 3rd about -70dB some 4th and 5th, nothing else. Much worse than I was expecting. I'm not sure if I'm measuring right or if my servo has messed up the circuit or perhaps it's just not balanced right. Another possibility is that the SEN input impedence is too high for the DAC but I have used quads of jfets so I think it should be about 8R.
Maybe time to try removing the servo.
I have done a test with regulated linear supplies with much lower stray capacitance between the mains and low voltage side and the noise is now below what I can measure on both the unbalanced SEN half and the balanced output. If I turn my 100W tube amp up to full volume I can still hear a little noise from the new PSUs if I put my ear within about 6 inches of the tweeter of my 88dB speakers. Normal listening volume is 30dB below this where it is inaudible.
I also grounded the current drain PSU and the SEN end of the shields of the output cables.
Listening test tomorrow.
The new PSUs are nothing special - 20RMB each I think (about $3). They are 12V. I also have some unregulated 18V ones to try.
I did another measurement with the sound card. Noise is down below the measurement floor but distortion results don't look that great. 2nd harmonic is about -40dB and 3rd about -70dB some 4th and 5th, nothing else. Much worse than I was expecting. I'm not sure if I'm measuring right or if my servo has messed up the circuit or perhaps it's just not balanced right. Another possibility is that the SEN input impedence is too high for the DAC but I have used quads of jfets so I think it should be about 8R.
Maybe time to try removing the servo.
> The drawback I personally find in batteries is the need to replace, or charge, them -- which may occur in the middle of listening session.
If you choose AAA size NiMH with 800mAH capacity, it will run for 20 hours.
How often do you listen to your system for more than 20 hours ?
The problem of batteries is HF noise and some small internal resistance (something like 50mohm per cell). This will be solved by the simple LC.
Of course you can also add a regulator after the battery. Then the circuit see the reg and not the battery.
Patrick
If you choose AAA size NiMH with 800mAH capacity, it will run for 20 hours.
How often do you listen to your system for more than 20 hours ?
The problem of batteries is HF noise and some small internal resistance (something like 50mohm per cell). This will be solved by the simple LC.
Of course you can also add a regulator after the battery. Then the circuit see the reg and not the battery.
Patrick
Sometimes you see solutions with an ON/OFF switch that connects the battery to the charger in the OFF position.
BTW You can get cell phone type LiPo batteries of say 3.7V at 700mAH for just a few dollars, and these have no memory effect and are rechargable almost indefinitely. I have used them and they are very quiet, for a battery that is.
BTW A simple test is to draw some current and connect the battery terminal to a headphone or the input of an amp through a capacitor.
This way you can quickly compare battery noise figures by just listening, not just the level but also the 'character'.
jan
BTW You can get cell phone type LiPo batteries of say 3.7V at 700mAH for just a few dollars, and these have no memory effect and are rechargable almost indefinitely. I have used them and they are very quiet, for a battery that is.
BTW A simple test is to draw some current and connect the battery terminal to a headphone or the input of an amp through a capacitor.
This way you can quickly compare battery noise figures by just listening, not just the level but also the 'character'.
jan
Being the electrochemical devices they are, I'm fairly confident that if you look hard enough when you pull current you will also see low-frequency noise, which will be very sample-dependent.
But the nice thing about this circuit and extensions of it (cascoded devices for example) is that it has rather excellent normal-mode power supply rejection. However, common-mode noise (particularly evil SMPS crud and mains HF noise) is where the off-line power supply needs rather excellent common-mode rejection, and over a very broad bandwidth.
It seems to me that when the I/V stage uses a floating supply, the DAC itself needs also a floating supply. Or am I wrong here? If I'm correct, the total consumption needs be taken into account.
I think the SEN circuit has good normal-mode PSRR and a balanced SEN has good common-mode PSRR for correlated common mode noise in the two power supplies of the balanced halves.
My problem I think was that there was no correlation in the SMPS switching noise between the two different PSUs on the two balanced halves and a component of this noise is common mode so it got through.
A single SMPS using an output transformer with two output windings for the two floating supplies of a balanced pair might be a lot better since the switching noise would be correlated and would cancel out.
No doubt with lots of work one could get an SMPS to function adequately, but going linear or with batteries has got to be a lot easier. This is not a project where we need to be green
No doubt with lots of work one could get an SMPS to function adequately, but going linear or with batteries has got to be a lot easier. This is not a project where we need to be green
I was wondering if it would be possible to get SEN onto a computer sound card and also started thinking about mobile devices. I don't think linear or multiple batteries is a realistic option in either case.
SEN PCM1794 measurement
I have got some measurements from an ADC with balanced inputs. I was comparing my power supply options for the floating supplies.
FWIW, here are the results. I had to dial down the signal from the DAC to -30dB to avoid clipping in the ADC. I was mainly interested in noise which shouldn't be affected except that the SNR should be 30dB better than shown. The distortion results are probably not representative though and they are still way higher than I was expecting.
I have got some measurements from an ADC with balanced inputs. I was comparing my power supply options for the floating supplies.
FWIW, here are the results. I had to dial down the signal from the DAC to -30dB to avoid clipping in the ADC. I was mainly interested in noise which shouldn't be affected except that the SNR should be 30dB better than shown. The distortion results are probably not representative though and they are still way higher than I was expecting.
Attachments
I believe your results. I'm not that confident in mine.
Spice indicates that the distortion is quite sensitive to getting the operating point right - I don't think I'm there yet.
There are also quite a lot of differences in the circuit I'm measuring and I'm using a real DAC chip which I know doesn't like driving anything but a virtual ground.