Working on a regulated supply to the VAS in my next amp using a small 317/337 regulating +-36V and a constant 125mA.
To measure I am using a Scarlett G3 with a differential, floating ground probe built like instructed on this forum, 0dBFS is approximately 10 VRMS and there is heavy averaging to be able to see any particular harmonics etc.
On the bench and with some C-L-C (10mF-2.2mH/2.5ohm-1200uF) filtering prior to the regulators and measuring the regulated output between +36 and -36 Volts best case I get this:
There are some interesting differences depending on if I swap the mains phase and if earth the trafo or not which is interesting, worst case looks like this:
These all look reasonable or should I say even pretty good given the setup but when I implement the same supply in the amp and measure the same way I get this:
This is simply way too good and I don't understand how this happens, anyone have any ideas and/or should I not trust this result?
To measure I am using a Scarlett G3 with a differential, floating ground probe built like instructed on this forum, 0dBFS is approximately 10 VRMS and there is heavy averaging to be able to see any particular harmonics etc.
On the bench and with some C-L-C (10mF-2.2mH/2.5ohm-1200uF) filtering prior to the regulators and measuring the regulated output between +36 and -36 Volts best case I get this:
There are some interesting differences depending on if I swap the mains phase and if earth the trafo or not which is interesting, worst case looks like this:
These all look reasonable or should I say even pretty good given the setup but when I implement the same supply in the amp and measure the same way I get this:
This is simply way too good and I don't understand how this happens, anyone have any ideas and/or should I not trust this result?
Could be radiated mains but for measurements like this I move things around to make sure they are not affected by their immediate environment. The application/amp environment is similarly open air lab'ish style but on another bench.
There are indeed additional caps on the board which I also simulated on the bench albeit not as much or the same type but it did not make a noticable difference.
I have a hard time trusting the measurement as it's almost perfect (the rise from 50 khz is the AD converter in scarlett USB interface).
Is this what one can expect from a well implemented 317/337 supply?
BTW, the measurements above was with Analog Devices 317/337 and I swapped them for another pair that came with the kit and it got me around 18dB lower under 40 Hz which was very unexpected.
There are indeed additional caps on the board which I also simulated on the bench albeit not as much or the same type but it did not make a noticable difference.
I have a hard time trusting the measurement as it's almost perfect (the rise from 50 khz is the AD converter in scarlett USB interface).
Is this what one can expect from a well implemented 317/337 supply?
BTW, the measurements above was with Analog Devices 317/337 and I swapped them for another pair that came with the kit and it got me around 18dB lower under 40 Hz which was very unexpected.
I am using the scarlett G3 USB interface and REW's RTA feature measuring the rails +to - directly the same way I would measure THD and/or noise at the output of an amp or preamp. As I capture the results it moves within REW to a tab called SPL and phase which later gets added to the jpeg, don't know why but its still the same as the RTA plot.
This is how it looks before the capture (it's not the same measurement, it's the amps output at 1khz without averaging)
Rayma, what do you mean by a loopback measurement in this context?
This is how it looks before the capture (it's not the same measurement, it's the amps output at 1khz without averaging)
Rayma, what do you mean by a loopback measurement in this context?
Interesting and you are right (again
) not that I understand how it happens but this is how it looks with both probes on ground (it looks the same with the probes on + or - 36V;How can I minimize this, would it help shielding the probes or would I need to skip the floating ground and run the measurement computer on batteries?
This is with the probes shorted to themselves:
Edit last RTA done at the same computer as the first.
Last edited:
Ok, will do but later tomorrow as I just cleared the test bench for other things tomorrow.
On the PS schematics I don't have it documented but took a few pics that I think will give you a good understanding. They are different between the "test bench" and "application bench" in that the test bench has a pair of bridge rectifiers and shares a common ground at the smoothing caps that are 24mF per rail and a more expensive/supposedly better Audio trafo at 150VA.
The application bench only has one bridge rectifier center tapped grounded and "only" 10mF per rail, 160VA std transformer.
Between this and the 317/337 board there are inductors at 2.2mH/2.5 ohms with another 3,400 uF per rail plus 1uF MKT caps just before the regulators. Current draw is a steady/non dynamic 125mA and I will be using smaller trafos for the final build once I get there.
Application bench supply:
Test bench supply:
And the regulator schematic:
On the PS schematics I don't have it documented but took a few pics that I think will give you a good understanding. They are different between the "test bench" and "application bench" in that the test bench has a pair of bridge rectifiers and shares a common ground at the smoothing caps that are 24mF per rail and a more expensive/supposedly better Audio trafo at 150VA.
The application bench only has one bridge rectifier center tapped grounded and "only" 10mF per rail, 160VA std transformer.
Between this and the 317/337 board there are inductors at 2.2mH/2.5 ohms with another 3,400 uF per rail plus 1uF MKT caps just before the regulators. Current draw is a steady/non dynamic 125mA and I will be using smaller trafos for the final build once I get there.
Application bench supply:
Test bench supply:
And the regulator schematic:
Twisting the probe cable hard (it was loosely twisted before) did not make a big difference in this setup albeit still a difference, maybe 3-6dB less harmonics.
I played around with a few things and eventually got one of my old laptops that still had some battery life in it and set up REW with the USB interface and the mains harmonics were gone! Plugged in the charger and they were back in a similar manner so I must have an issue with the mains supply in this section of the house.
Either way I am now trusting the measurement and will use it in the final build even though since 20 years or so my perception has been that the 317/337 are not good enough for this type of application.
I do have a 250mohm resistor between the regulated output and the VAS in the amp which subjectively dampened the 317/337 sonic signature I had experienced before (it does not change the measurements).
I will now experiment with a few low ohm inductors to see if I can improve things further.
I played around with a few things and eventually got one of my old laptops that still had some battery life in it and set up REW with the USB interface and the mains harmonics were gone! Plugged in the charger and they were back in a similar manner so I must have an issue with the mains supply in this section of the house.
Either way I am now trusting the measurement and will use it in the final build even though since 20 years or so my perception has been that the 317/337 are not good enough for this type of application.
I do have a 250mohm resistor between the regulated output and the VAS in the amp which subjectively dampened the 317/337 sonic signature I had experienced before (it does not change the measurements).
I will now experiment with a few low ohm inductors to see if I can improve things further.