It's for this PSU and will go between the PSU & the SSHV2.
http://www.diyaudio.com/forums/attachments/power-supplies/317423d1355438587-simplistic-mosfet-hv-shunt-regs-salas-sshv2-problem.jpg
http://www.diyaudio.com/forums/attachments/power-supplies/317423d1355438587-simplistic-mosfet-hv-shunt-regs-salas-sshv2-problem.jpg
So you're making the Kmultiplier into a common mode filter as well as a differential mode filter. That is an interesting idea. This should work for single rail supplies. However the problem with using the Kmultiplier on ground is that ground needs to accept positive and negative current, and the Kmultiplier can only accept one direction. You are looking for a virtual ground circuit.
I'm guessing that this noise is generated between switcher output and switcher input ground. In this case Kmultiplier ground needs to be grounded at the switcher input ground with the shortest wire possible.
It needs to be an isolated output SMPS.
I'm guessing that this noise is generated between switcher output and switcher input ground. In this case Kmultiplier ground needs to be grounded at the switcher input ground with the shortest wire possible.
It needs to be an isolated output SMPS.
Attachments
Hi,
I have managed to clean up smps positive rails in the past but the ground stays noisy, so I am wondering how to deal with that, without raising circuit impedance, or using big caps / coils.
A virtual gnd is perhaps not what I'm aiming for - perhaps a clean active ground is ?
Can a cfp be modified to allow sinking and sourcing ? Or should I look at a filtered long-tailed pair as a push-pull ? The supply for the ltp could be taken from the Kmultiplier, and the inverting/non-inverting inputs could be used to cancel the ground noise ? Perhaps the differential output of the ltp can then be used as the supply to the device ?
To be honest, I'm trying to re-learn something that I never fully understood even when it was taught many years ago....
but it is good for the grey cells
I have managed to clean up smps positive rails in the past but the ground stays noisy, so I am wondering how to deal with that, without raising circuit impedance, or using big caps / coils.
A virtual gnd is perhaps not what I'm aiming for - perhaps a clean active ground is ?
Can a cfp be modified to allow sinking and sourcing ? Or should I look at a filtered long-tailed pair as a push-pull ? The supply for the ltp could be taken from the Kmultiplier, and the inverting/non-inverting inputs could be used to cancel the ground noise ? Perhaps the differential output of the ltp can then be used as the supply to the device ?
To be honest, I'm trying to re-learn something that I never fully understood even when it was taught many years ago....
but it is good for the grey cells
Last edited:
from I have been told, thats what smps does
and why it needs proper earth to get rid of it
would balanced drive help to 'isolate' ground
btw, got my 1930/5171 today
all the way to northern EU from Grand Prairie USA and after a tour to Paris .... even more special now SMPS cleaner
Yes, I think a little diff amp is needed.
SMPS does have many advantages over linear, but noise and EMI make it unsuitable for audio, so I use trafo_rectifiers, and lots of isolation trafos and mains filters too for my hifi because the power in my home is polluted from the fridge, strip lights, TV, etc. SMPS is everywhere...
So if we can fix the noise with your kmulti and just half a dozen more parts for the diff amp... well, I definitely think this is worth exploring ?
My thought for the ltp is to feed the non-inverting with 1.2V less than supply, and the inverting with 1.2V more than ground, using IR leds. It's another 2.4V dropout but should provide a clean supply from the output of the diff amp ?
The diff amp is two transistors and one resistor or ccd, and the sources are a resistor and led each. Perhaps add some caps and rc filters ... the whole thing including the kmulti could be very small and light, just like the smps supply. No big trafo, smoothing caps, etc...
What do you think ?
Yes, I think a little diff amp is needed.
SMPS does have many advantages over linear, but noise and EMI make it unsuitable for audio, so I use trafo_rectifiers, and lots of isolation trafos and mains filters too for my hifi because the power in my home is polluted from the fridge, strip lights, TV, etc. SMPS is everywhere...
So if we can fix the noise with your kmulti and just half a dozen more parts for the diff amp... well, I definitely think this is worth exploring ?
My thought for the ltp is to feed the non-inverting with 1.2V less than supply, and the inverting with 1.2V more than ground, using IR leds. It's another 2.4V dropout but should provide a clean supply from the output of the diff amp ?
The diff amp is two transistors and one resistor or ccd, and the sources are a resistor and led each. Perhaps add some caps and rc filters ... the whole thing including the kmulti could be very small and light, just like the smps supply. No big trafo, smoothing caps, etc...
What do you think ?
Here's what I came up with. It will be difficult to use fewer parts. Of course I haven't tested it in real life.
This scheme does not really need a dual rail SMPS input, though it may help at DC. It should not be built, it is probably unstable.
Instead of an LTP I used a buffered singleton which will perform better here.
This scheme does not really need a dual rail SMPS input, though it may help at DC. It should not be built, it is probably unstable.
Instead of an LTP I used a buffered singleton which will perform better here.
Attachments
This wouldn't work ? R3-5 could be ccd and that would improve performance....
An externally hosted image should be here but it was not working when we last tested it.
from I have been told, thats what smps does
and why it needs proper earth to get rid of it
would balanced drive help to 'isolate' ground
btw, got my 1930/5171 today
A proper earth - actually this is a bit of an understatement - the earth in my home is heavily polluted - I need a big copper pole deep into the ground and a thick, screened, cable from that to my hifi ;-)
Nice to get special parts eh ? Do update us when you get your build done. I'm still waiting for my parts...
Thanks... and the question is... how well ? Your kmulti should really help - most smps supplies have about 5mV of broadband noise/ripple under load, so 60db off that and we have a few uV. Too good to be true ? I have no way to measure. But I'll put it on my projects list and try it when i get my DAC finished.
One thing you could try is isolate the SMPS from line and ground with shielded transformer and drive the shield with a buffer fed by the SMPS, with earth as it's input. This way there is not nF of capacitance between SMPS and earth for ground noise to leak through. This is a the rejection approach as opposed to the filtering approach, and I suspect it will work much better.