@fabio
Sorry, I didn't explain clearly beforehand, so apparently there is some miscommunication here.
For any power supply there are three major sources which will contribute noises at its output. (A) from its input (B) self-generated noise (C) from its output load.
Usually PSRR is used to specify the input influence. I haven't done any simulation yet, but from your previous testing results, it is not too bad --- at least not the major concern right now. Yes, some rectifier switching noise (the odd harmonics of 50Hz) is obviously present in all your test result.
For self-generated noise: in your measurement it is about 2.4uV @48k bandwidth. Let's check spec. of OPA2177, its voltage noise is 8nV @ >30Hz. The 48k bandwidth total noise is roughly 8nV*sqrt(48k) = 1.753uV. This is for one opamp, OPA2277 has two opamps, so total noise will become 1.753uV*sqrt(2) = 2.479uV. It match your measurement very well, so pretty much all the noise comes from OPA2217. This also means the fastest way to reduce self-generated noise is modifying the circuit such that only one opamp (the error opamp) is used.
The third noise source usually is the dominate one, as in your case. From your single tone measurements, it is 40~50dB above the self-generated noise. Playing music at -12dB maybe a little too loud, at a more realistic -30dB level, the noise still will be 20~30dB above self-generated noise.
Will this -90 ~ -100dB level noise influence the output of Katana? (audible?) I strongly double it. For a good audio opamp, its PSRR usually > 100dB, e.g. LM4562 has PSRR of 120dB. Katana use a customized discrete one, so its PSRR is unknown, but I assume it should be decent.
Just for curiosity, can this load introduced noise be reduced? Yes, maybe. That's why I asked you to take measurement at close to R115 rail. This is where the circuit sense the output voltage and try to eliminate difference between it & reference voltage. From your measurements, there is about 6dB difference @10kHz from Katana input pin, not as much as I hope. Maybe it is because Katana only draw 70mA or because OPA2277 is simply too slow, in which case taking another measurement using TL072 can clarify it. Anyway, to reduce load introduced noise, the most effective way from power supply side is using remote sensing, as I described in previous post. In this configuration, the circuit will try to regulate at the load.
So no, all these measurements are not related to calculating output impedance. For which the simplest setup involve one fixed load and one dynamic load, both with precisely known current draw, the dynamic load will be controlled via a mosfet or similar switch and a signal generator which generate periodic square wave. And the measurement will be taken in time domain, and switching overshoot and undershoot will be used to estimate output impedance.
Sorry, I didn't explain clearly beforehand, so apparently there is some miscommunication here.
For any power supply there are three major sources which will contribute noises at its output. (A) from its input (B) self-generated noise (C) from its output load.
Usually PSRR is used to specify the input influence. I haven't done any simulation yet, but from your previous testing results, it is not too bad --- at least not the major concern right now. Yes, some rectifier switching noise (the odd harmonics of 50Hz) is obviously present in all your test result.
For self-generated noise: in your measurement it is about 2.4uV @48k bandwidth. Let's check spec. of OPA2177, its voltage noise is 8nV @ >30Hz. The 48k bandwidth total noise is roughly 8nV*sqrt(48k) = 1.753uV. This is for one opamp, OPA2277 has two opamps, so total noise will become 1.753uV*sqrt(2) = 2.479uV. It match your measurement very well, so pretty much all the noise comes from OPA2217. This also means the fastest way to reduce self-generated noise is modifying the circuit such that only one opamp (the error opamp) is used.
The third noise source usually is the dominate one, as in your case. From your single tone measurements, it is 40~50dB above the self-generated noise. Playing music at -12dB maybe a little too loud, at a more realistic -30dB level, the noise still will be 20~30dB above self-generated noise.
Will this -90 ~ -100dB level noise influence the output of Katana? (audible?) I strongly double it. For a good audio opamp, its PSRR usually > 100dB, e.g. LM4562 has PSRR of 120dB. Katana use a customized discrete one, so its PSRR is unknown, but I assume it should be decent.
Just for curiosity, can this load introduced noise be reduced? Yes, maybe. That's why I asked you to take measurement at close to R115 rail. This is where the circuit sense the output voltage and try to eliminate difference between it & reference voltage. From your measurements, there is about 6dB difference @10kHz from Katana input pin, not as much as I hope. Maybe it is because Katana only draw 70mA or because OPA2277 is simply too slow, in which case taking another measurement using TL072 can clarify it. Anyway, to reduce load introduced noise, the most effective way from power supply side is using remote sensing, as I described in previous post. In this configuration, the circuit will try to regulate at the load.
So no, all these measurements are not related to calculating output impedance. For which the simplest setup involve one fixed load and one dynamic load, both with precisely known current draw, the dynamic load will be controlled via a mosfet or similar switch and a signal generator which generate periodic square wave. And the measurement will be taken in time domain, and switching overshoot and undershoot will be used to estimate output impedance.
He probably mean DC (output voltage) this board ain't work with 6VAC I've tested with 6 VAC (I ordered transformer with two outputs 5 and 6AC ) and then I had to connect transformer in series to get MIN 9VAC (in my case was 12VAC that let me pushed the board to steady 6.4 VDC, though I soldiered resistors L/M with lowest values marked on the board)
thanks to bButcher who was my guide😁
Couple of questions....
Hi all - great thread, and lots of genuinely useful input - especially from bButcher - thanks!
I've just got a couple of the eBay dual supply kits, and will be using them as the starting (maybe permanent....) power for IanCanadas latest group buy stack (RPi3b, FifoPi, dual es9038 board etc). Before I complete the build, any input on the following two questions would be greatly appreciated:
1 Anyone found an opamp that performs at least as well as the TL072 in this circuit that would lower the opamp supply rail needed to ~5.5V so I can use a 6v or 7V transformer? Like bButcher, less heat across pass transistor @ high current when driving 5V load @ 1A to 2A per rail would be useful. (Currently planning 9v trafo and bigger heat sinks).
2. Any downside (stabilty, longevity, performance) to the bButcher mod for reducing R102 from 10k to 5k at the Zener Diode?
Regards Adrian
Hi all - great thread, and lots of genuinely useful input - especially from bButcher - thanks!
I've just got a couple of the eBay dual supply kits, and will be using them as the starting (maybe permanent....) power for IanCanadas latest group buy stack (RPi3b, FifoPi, dual es9038 board etc). Before I complete the build, any input on the following two questions would be greatly appreciated:
1 Anyone found an opamp that performs at least as well as the TL072 in this circuit that would lower the opamp supply rail needed to ~5.5V so I can use a 6v or 7V transformer? Like bButcher, less heat across pass transistor @ high current when driving 5V load @ 1A to 2A per rail would be useful. (Currently planning 9v trafo and bigger heat sinks).
2. Any downside (stabilty, longevity, performance) to the bButcher mod for reducing R102 from 10k to 5k at the Zener Diode?
Regards Adrian
@adrian
Fabio have tried OPA2277 & OPA2134. Both have lower noise ... at least spec-wise, than TL072 and rated to work at 5V, although OPA2134 need a little mod. So they are good candidates. But dropping a 5V opamp doesn't automatically mean you can lower the AC input to below 9VAC, which will also need modification of 10V opamp power supply circuit.
Since Zener 5V1 in the kit has unknown spec. and most Zener diodes need at least 1mA to push over its knee voltage. If you use 9VAC input, the voltage across R102 will be around 6V~7V, and 10k will only generate 0.6~0.7mA current. In the original design, voltage across R102 is about 45V which will generate 4.5mA current.
Fabio have tried OPA2277 & OPA2134. Both have lower noise ... at least spec-wise, than TL072 and rated to work at 5V, although OPA2134 need a little mod. So they are good candidates. But dropping a 5V opamp doesn't automatically mean you can lower the AC input to below 9VAC, which will also need modification of 10V opamp power supply circuit.
Since Zener 5V1 in the kit has unknown spec. and most Zener diodes need at least 1mA to push over its knee voltage. If you use 9VAC input, the voltage across R102 will be around 6V~7V, and 10k will only generate 0.6~0.7mA current. In the original design, voltage across R102 is about 45V which will generate 4.5mA current.
@bButcher
Thanks for the clarification. With regard to the ref voltage - any thoughts on using a blue led? Should give around 2.6v, although noise would be a bit higher. Also, if using opa2134 - any other mods around the opamp supply voltage recommended to enable maybe a 6v trafo?
Regards Adrian
Thanks for the clarification. With regard to the ref voltage - any thoughts on using a blue led? Should give around 2.6v, although noise would be a bit higher. Also, if using opa2134 - any other mods around the opamp supply voltage recommended to enable maybe a 6v trafo?
Regards Adrian
@adrian
Yes, blue led is noisier than red, but the low pass RC filter's cutoff freq. is set at ~0.34Hz, so most of its high freq noise won't reach to error opamp's input, it should be O.K. To be honest, LM4040 or LM336 isn't that expensive, and much more stable under temperature and line voltage variation.
If you plan to use 6/7VAC input, than change R104 to 3.3k will give you ~6.5V op rail voltage. And also change R102 to 1K to maintain the proper zener current. But the problem is at high current (>1A), MJE15034's hfe will drop significantly so OPA2134 need more than 6.5V to supply the needed driving current. The solution is change it to some higher rating power transistor like D44H11.
6V transformer is kind of risky, you must consider rated lowest line voltage is 94%*230V.
Yes, blue led is noisier than red, but the low pass RC filter's cutoff freq. is set at ~0.34Hz, so most of its high freq noise won't reach to error opamp's input, it should be O.K. To be honest, LM4040 or LM336 isn't that expensive, and much more stable under temperature and line voltage variation.
If you plan to use 6/7VAC input, than change R104 to 3.3k will give you ~6.5V op rail voltage. And also change R102 to 1K to maintain the proper zener current. But the problem is at high current (>1A), MJE15034's hfe will drop significantly so OPA2134 need more than 6.5V to supply the needed driving current. The solution is change it to some higher rating power transistor like D44H11.
6V transformer is kind of risky, you must consider rated lowest line voltage is 94%*230V.
@bButcher
Thanks again. It looks a lot more straightforward sticking with the 9v solution for the 5v feeds, and using more heatsinking. I'll go with one dual board for the +/- 12-15v supply for the DAC output (12v trafo), and one dual board for the separate 5v feeds for the RPi and the isolator board. I'll do a few of the mods you suggest (R107 to 5k and the cleaner feed for Vref) for these, but I'll probably buy another board to try the other bigger changes later.
Regards Adrian
Thanks again. It looks a lot more straightforward sticking with the 9v solution for the 5v feeds, and using more heatsinking. I'll go with one dual board for the +/- 12-15v supply for the DAC output (12v trafo), and one dual board for the separate 5v feeds for the RPi and the isolator board. I'll do a few of the mods you suggest (R107 to 5k and the cleaner feed for Vref) for these, but I'll probably buy another board to try the other bigger changes later.
Regards Adrian
Thanks to everyone who's contributed so far - Ive learnt a lot and have have also ordered a STUDER for a RPI3B+ Digione Signature build.
Really interested to see what comes from the opamp experiments and maybe trying my hand at the remote sensing mod.
How would one float the OP power supply from the output voltage? I have a spare tap on my smallest 9V-3.5Va transformer.
Really interested to see what comes from the opamp experiments and maybe trying my hand at the remote sensing mod.
How would one float the OP power supply from the output voltage? I have a spare tap on my smallest 9V-3.5Va transformer.
...it would be useful to know what is the current consumption of the device you are intending to use your 'Studer' PSU with...
Here's a general rule from bButcher:
For moderate load current (<=500mA), use 9V AC for 9V DC output or below, use xV AC for xV DC output for x >= 9.
You need at least 9V AC bc opamps need 10V dc to operate properly.
For best result, keep Vce of pass transistor > 2V under worst condition, please refer to post #97 part (2) for more detail calculation
For moderate load current (<=500mA), use 9V AC for 9V DC output or below, use xV AC for xV DC output for x >= 9.
You need at least 9V AC bc opamps need 10V dc to operate properly.
For best result, keep Vce of pass transistor > 2V under worst condition, please refer to post #97 part (2) for more detail calculation
It will be used in a phono preamp wich, according to the seller, has a max current consumption of 100mA.
I think 2.5A at 12V is probably asking a bit too much of this reg. However there are others who have posted here who may be able to give you a more technically accurate answer, if they are around.