3
\$\begingroup\$

I'm poking at the inside of a USB oscilloscope, with the hope of modifying it so that the scope probe commons can be referenced to different voltages.

When I use my USB oscilloscope to measure a 24V DC signal with a 10x probe, I get reasonable results. enter image description here

By my figuring, this 10x probe is a 9 MOhm resistor, in series with the scope's 1 MOhm internal resistance. Thus, a 10:1 divider.

schematic

simulate this circuit – Schematic created using CircuitLab

By adding 10 MOhm to the probe, I should turn it into a 20:1 divider. When I insert 10 MOhm into the positive signal path I do get a rough 20:1 effect, but I start seeing AC 60 Hz noise.

schematic

simulate this circuit

enter image description here

When I instead put the resistance in the common path, between the signal common and the scope probe common, I see vastly more AC noise.

schematic

simulate this circuit

enter image description here

My USB scope is isolated from my PC with a commercial USB isolator. Can anyone tell me what's going wrong, and how to fix it? I thought this might just be a consequence of having all this strewn about my desk instead of pleasantly soldered together with minimum signal paths. But I would have expected the noise to be much lesser magnitude, and much more responsive to physical rearrangement. Moving leads and parts has zero observable effect. And it still wouldn't explain the difference between inserting the resistance in the positive and negative paths.

Why is this happening, and how can I fix it?

\$\endgroup\$
  • \$\begingroup\$ Schematics please? Preferably of the probe's presumed resistances plus yours, and the ADC tap point. \$\endgroup\$ – AaronD Nov 24 '15 at 22:02
  • \$\begingroup\$ Is this just a consequence of having all this strewn about my desk - quite likely, yes \$\endgroup\$ – pjc50 Nov 24 '15 at 22:06
  • \$\begingroup\$ @pjc50 So why the asymmetry between the positive and negative legs? \$\endgroup\$ – Stephen Collings Nov 24 '15 at 22:08
  • \$\begingroup\$ Exactly what is producing the '24v DC signal', and which USB isolator are you using? \$\endgroup\$ – Bruce Abbott Nov 24 '15 at 23:04
  • \$\begingroup\$ @BruceAbbott the 24V signal is a Phoenix Uno Power supply, and the USB isolator is a B&B UH401. \$\endgroup\$ – Stephen Collings Nov 25 '15 at 20:45
2
\$\begingroup\$

Look at the case represented by your first schematics. Let's say there is some noise source that is coupling to the connection between the probe and the 24V supply. Just for example, the coupling is weak and is equivalent to many mega-ohms of impedance. Since the 24V is relatively low impedance, the noise coupling in would not register at all in the measurement.

Loot at the case represented by the second schematics. Let's say the same noise source is coupling to the connection between the probe and the 10Mohm resistor. The impedance of that point to the ground reference is around 5Mohm. So even weak coupling in many mega-ohms of impedance from the noise source will show up in the measurement.

My guess is that the situation can be improved by attaching the 10Mohm resistor to the probe as closely and directly as possible. Then extend the ground around the probe tip with metal foil to cover the connection and the resistor completely.

Look at the case represented by the third schematics. If there is ground loop current between the scope and the 24V supply, the addition of the 10Mohm would cause that to show up. Theoretically, 1uA of ground loop current would show up as 10V with the 10Mohm resistor.

It is asymmetrical to case 2 because any external current coupled to the whole 24V apparatus would show up across the 10Mohm resistor.

The fix could be to hunt down and eliminate any ground loop coupling. For example, it is not unusual for power supplies and equipments to have their functional ground (directly or indirectly) connected to chassis ground through a resistor of mega-ohms and/or a small capacitor in parallel.

But in this case, perhaps decide if this is indeed an issue, possibly aided by an estimate of the ground loop current (for example, take your scope measurement and divide that by 10Mohm). If decided as not a real issue, tolerate it when using the scope and do not use the set up as in schematic 3.

\$\endgroup\$
  • \$\begingroup\$ Unfortunately my whole goal is to be able to reference two scope probes to different commons, which schematic 3 would allow me to do. If it's not possible, I'll need a different means of accomplishing that. \$\endgroup\$ – Stephen Collings Nov 25 '15 at 20:47
  • \$\begingroup\$ Adding resistors to the probe grounds will not do what you want (even if the noise issue is much reduced). \$\endgroup\$ – rioraxe Nov 25 '15 at 22:27
  • \$\begingroup\$ Why not? It seems like it should. \$\endgroup\$ – Stephen Collings Nov 26 '15 at 19:51
  • 1
    \$\begingroup\$ Unless you have isolated probes (then you would not need the kludge in the first place), the grounds on your probes are one and the same. If you connect one ground through a 10M resistor to one voltage, and another through a 10M resistor to another voltage. Current will flow through the two 10M resistor in series and the probes ground would be in the middle. \$\endgroup\$ – rioraxe Nov 27 '15 at 0:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.