Android tablet injecting digital noise to power lines

I have another hack idea , how to adjust down converters output voltage , but no implementation yet, without breaking galvanic isolation . Idea is to shift trim pin voltage by lets say +0,2V upper than output voltage. It can be a schottky diode and resistor to +12V simply, no high current needed. But ground must be common then , for diode to operate. This would extend trim range to lower level.
You can test that , by using 10k potentiometer ,any diode, and 47k resistor , but join input and output grounds together.
 
Problem with diode is not constant ,also temperature dependent ,forward voltage drop.With low current schottky diode may have even 0,1V , but with several amperes it may reach 0,5V maybe . While standard Si diode varies from 0,5V to 1,3V ... It may be tricky to select matching diode , but luckily we have trim pin .
 
Sounds realistic . Just need to find p-mosfet , which can fully open to few milliohms at less than 5Volts. Probably logical level pmos .Lm358 have its drawbacks , like zero drift , had to replace it to some better to make working solid. Also output voltage is supply voltage - 1,5v , so sensitive mosfet may be unable to fully close. At best , rail-to-rail ic with low drift to be used there.
 
Two ideas about home made LDO.
The #1 uses a charge pump to create twice the DC/DC voltage output to have large enough gate voltage over its source to operate NMOSFET as series pass.
The #2 uses a PMOSFET plus a CMOS OpAmp and a micropower reference. As PMOS needs gate voltage below its source, no need for charge pump.
Obviously each has their pros and cons.
 

Attachments

  • 16747343563852697919292738108606.jpg
    16747343563852697919292738108606.jpg
    345.7 KB · Views: 50
  • 1674734384207807742967724388056.jpg
    1674734384207807742967724388056.jpg
    323.5 KB · Views: 49
  • Like
Reactions: ximikas
Hy,
Thank you both for thinkering on LDO situation.
I had just wired the diode & 2200uF cap in series.

Voltage is 4.45v when tablet is completely off and 4.24 - 4.27v when on, 4.31v when in stand by mode (locked screen, no activity).

0 AC on power rails.
I would call it good, would you also (keep in mind my current as small as possible situation)?


I do still get a bit of noise on the audio path... slight hiss before the playback starts and after it stops. This hiss starts and ends with a click.
I needed to remove the ground loop isolator on audio path as there was loud buzz as grounds are not common anymore.
I won't bother you with this as amp in question is cheap class D amp that has common audio gnd and power gnd. This amp will be removed and I must still add the eq (eq has gnds seperated).

However I would like you ask something else... Do you have any knowledge about USB lines?

I forgot that using dc dc converter will also seperate usb hub gnd and tablet gnd. Usb hub is powered by another smps, I disconnected vcc+ and gnd on the USB, only Data+ and Data- are wired from tablet to usb hub.
Voltage between gnd and isolated gnd differs for 0.08v max and usb is working fine.
Should I worry about that?
If this is not ok, I will need to add usb isolation.

EDIT:
If I'm not mistaking, usb uses differential signal, no gnd needed?
 
Ok, here I have more freedom as I don't need to stay in current limits.
Ground loop isolators with 1:1 trafos seem to "colour the sound", is this also the case with phocoupler?

I would like to execute my plan first (eq, different amp) and go from there.
I believe that this chinesse class D amp is the big problem remaining...