Jack,
I have tried to look at the macro but it's written for the AP2700 too difficult to convert it to APWIN given I am not familiar with the functions it uses.
I will see if the interface can be set manually to plot the noise density.
Conversely, if Ido it the other way using the analog interface, given that the noise is average 10uV across board as shown by the noise plot, if I divide by the sqrt(H), means that at 1KHz I have 316nV/sqrt(Hz).
Since I used the equiripple, AP technote lists for each FFT window used and in this case it is 2.63, then the noise is 120.15nV/sqrt(Hz)
Since I didn't use bandpass I suppose I don't have to divide the result, anyway I don't know the factor used for bandpass I will look it up.
My quuestion is: The power supply input has its own noise floor, how do I factor the noise of the PSU alone? Should divide it by the noise rejection and subtract it from the noise at the output of the supply? Is the input noise taken into consideration with this?
Dukes suggestion from the AP 2722 is very good. Similar circuits are used in premium DVMs to protect against gross overload. And the effective resistance is less than a light bulb.
I believe the recommended D44Hxxx are 100MHz and have less interelectrode capacitances. They should be good up to 1A or so.
But I agree; what you have now is very good and hard to - meaningfully - improve. Go on to your next project
Jan
I didn't know that the macro was written exclusively for the AP2700. I've run macro's for their legacy systems on the 2722.
You don't have to do any translating to run the macro. You only have two choices. The white paper with the macro lays it all out.
After it runs, change the y-axis to "V" and log. If you want frequency represented log, put "10" in the start frequency and change the scale to log.
Jan,I believe the recommended D44Hxxx are 100MHz and have less interelectrode capacitances. They should be good up to 1A or so.
But I agree; what you have now is very good and hard to - meaningfully - improve. Go on to your next project
Jan
You are correct, I will move on onto the optimization of the shunt supply next, which is the regulator right after the Jung.
I have spent the last several hours on the Jung and made marginal improvement, by changing the pass device and switching to OPA182.
Below is the last result and I won't push it any further.
With the scope probe right at the output, I can see about 1.13mV pk-pk about 150KHz main oscillation frequency with higher harmonics
Play with grounds reduce or increases slightly the content of the waveform above.
Below the FFT of the noise out (not scaled to sqrt(Hz) yet)
Last edited:
I have read the paper it lays it how for the most part. I will try to set that measurement up.
Wat you have is very, very good. For the remainder, do the same test with the probe connected only to its ground lead but at the same location. Chances are that you measure basically the same grudge, meaning it's not the regulator.
Jan
Jan, yes you are correct. It’s a matter of grounding I got the plots to clear significantly. I have optimized to what I feel, as much as I could. I will post the final plots tomorrow as it is 1.30AM lol
Now I moved to the shunt section to which I found contradictory results where for example removing the cascode to the current source instead of decreasing performance, it actually increased it. Obviously the problem lays somewhere else.
I will post these “interesting” findings including the schematic tomorrow!
Now I moved to the shunt section to which I found contradictory results where for example removing the cascode to the current source instead of decreasing performance, it actually increased it. Obviously the problem lays somewhere else.
I will post these “interesting” findings including the schematic tomorrow!
Hmmmm, didn't know that.
I have a macro for noise calculation, in V/RtHz. Can PM it to you if you'd like. The Analog Analyzer is slightly lower noise compared with the DSP analyzer.
For power supply noise measurement with the Jung-Didden SR, LT3042, Silent Switcher, or Salas or Sjostrom etc it's really helpful to have a low noise amplifier with gain greater than 10. If you use the Pass P3 phono-preamp without the compensation network, you have a DIY device with noise floor on the order of 650pV/RtHz and gain >1000
Last edited:
Thank you so much jack. If you can send me the macro for noise I will greatly appreciated it. But don’t you have a micro for the AP2700 since you used the mighty 2722 now my dream machine? Or you have one for APWIN as well?
Regarding amplifier for measuring noise floor I do agree. I probably need to build or buy a dedicated one to make those measurements.
Regarding amplifier for measuring noise floor I do agree. I probably need to build or buy a dedicated one to make those measurements.
This is the shunt I am trying to optimize. interesting enough, I removed the cascoding M8 and performance instead of decreasing, improved. Maybe the cascoding needs more headroom to operate don't know yet.
Belo is the plot without the current cascoding also set to 0 ohm R155 and R156 and max the current on the current source I5, I5 by shorting the gate/source resistor and now probably deliver 1-2mA
I think the shunt should also provide similar performance to the Jung Regulator given the natural isolation it provides by it's current source, at least it should be better than this I assume lol
Belo is the plot without the current cascoding also set to 0 ohm R155 and R156 and max the current on the current source I5, I5 by shorting the gate/source resistor and now probably deliver 1-2mA
I think the shunt should also provide similar performance to the Jung Regulator given the natural isolation it provides by it's current source, at least it should be better than this I assume lol
There is a shunt regulator from Walt Jung employing the same principle as well known JD regulator. It should have excellent performance (judging by simulation results). Some members here use it for DACs but no one has made any significant measurement.
https://refsnregs.waltjung.org/UnivReg_122714.pdf
Walt Jung’s site is bandwidth limited so I’ve attached the document, just in case.
https://refsnregs.waltjung.org/UnivReg_122714.pdf
Walt Jung’s site is bandwidth limited so I’ve attached the document, just in case.
Thank you so much Tom. I had seen this topology just yesterday, this document toi attached is very useful in detailing everything.
Problem is that the board is designed with the shunt I have in the picture so I will have to fix what I have. Implementing this on the current board would be very challenging. Later on for my own purpose I will design a test board for this one and perform measurements maybe I will use it for future applications. Thank you again for the valuable info!
Problem is that the board is designed with the shunt I have in the picture so I will have to fix what I have. Implementing this on the current board would be very challenging. Later on for my own purpose I will design a test board for this one and perform measurements maybe I will use it for future applications. Thank you again for the valuable info!