4

Suppose I need to view voltage between two arbitrary points of a circuit with an oscilloscope. For that I can use two probes and A-B math, but should I connect the ground lead to the circuit? How to maximize CMRR?

  • Yes, connect the ground lead to get best results providing it's safe to do so. – Andy aka Apr 27 '16 at 16:39
  • Several years ago one man say me just tie ground leads of two probes and *don't* connect it to the circuit under test. But he couldn't explain me why. Maybe it's wrong. –  Apr 27 '16 at 16:45
  • @Andyaka At least one ground probe must be connected to the circuit for this to work, no? – KnightsValour Apr 27 '16 at 16:47
  • @KnightsValour for "what to work"? – Andy aka Apr 27 '16 at 16:48
  • @Andyaka For the oscilloscope measurement. Why would it be unsafe to connect a ground lead? – KnightsValour Apr 27 '16 at 16:51
  • I repeat "for what to work" i.e. what are you talking about? – Andy aka Apr 27 '16 at 16:59
  • @Andyaka Corvus wants to measure a voltage difference in a circuit. Presumably, these two voltages are across a device or in some location where neither of them are ground on the circuit. Therefore, he/she is using 2 standard probes and the difference mode on the oscilloscope where probe B is subtracted from probe A. When measuring in this configuration, doesn't the ground lead from at least one probe need to be connected to the ground reference on the circuit for the measurement to be valid? And as a follow up, why wouldn't this be safe to do? – KnightsValour Apr 27 '16 at 17:14
  • @KnightsValour why don't you either answer this question or raise a new one. Or maybe wait to see what the OP says. – Andy aka Apr 27 '16 at 17:21
  • @KnightsValour: Yes, you're correct. Even when making a differential measurement, you need to have a ground connected in order to control the common-mode voltage on the scope inputs. However, if that common-mode voltage is being supplied by a low-impedance source, it might be unsafe in the sense that excessive current will flow in the ground leads. – Dave Tweed Apr 27 '16 at 17:22
  • @DaveTweed by differential measurement you mean using two channels right? And not measuring using differential probes. – seetharaman Apr 28 '16 at 04:02
  • @seetharaman: Yes, that's what this question is all about. – Dave Tweed Apr 28 '16 at 11:04

1 Answers1

3

Yes, you need to still connect it to ground of the device under test because you are doing two single-ended measurements, not a true differential measurement.

A true differential probe has no ground and uses the ground of the test equipment.

Without a ground reference, there is no guarantee that ground reference of the two single-ended probes will match. So then you might say, OK, I can connect the two probes grounds together. But then you might hit the max voltage limit of your single-ended measurement. For example, what if the scope probes ground reference floats up to 8V from external charges and your scope clips at 10 V (at 1V per division setting), now you can only measure up to 2V before the signal clips.

So basically, because you are actually doing two single-ended measurements, you still need to ground your device to your scope or else you'll pick up all sorts of noise and weirdness (AC line, etc).

Vince Patron
  • 3,566
  • 8
  • 17
  • As you say the common mode voltage range is one thing and then the non-linear common mode gain characteristics that will cause unpredictable (usually minor on a modern scope but high in an old valve scope I had) distortion if the commons are left floating. If the inputs are truly isolated from each other including the grounds then the grounds HAVE to be connected or you are measuring static charges and bias currents etc that may be much larger than your differential voltage. – KalleMP Apr 27 '16 at 18:22