New Rigol DS1052E: ~20mV difference between CH1 and CH2 (at 5.15V offset)

[bao8]

Hi,

I've just started launching into electronics (after dropping out of an engineering degree 10 years ago), and one of the first things I grabbed was a Rigol DS1052E.
I was doing some testing of USB adapters this eve and discovered that on my 1052, there was about a 20mV difference between CH1 and CH2 (using the same probe).

How I noticed it: I had two USB cables modified to have the ground and +5V exposed, connected in a chain, and had one probe hooked up to each (one on CH1, one on CH2; both set to 10x). I set the offset to 5.15V and scale to 20mV/div on both (via VISA from computer). There was a small difference in V (i.e. 5.14V vs 5.16V) between the two, which seemed a bit odd. I pulled everything apart, plugged in one cable directly to the adapter, hooked a probe (still on 10x) up to GND and +5V, and connected it to CH1. Took a reading. Swapped the connector to CH2; different by ~20mV.

My response: I disconnected the probes and performed self-calibration (after leaving the 1052 running for 30 mins) and tried again. Same result.

The question: Is this normal? Should I be expecting an error of 20mV (when using probes on 10x) between CH1 and CH2 on a ~5.15V signal at 20mV/div?

Note: the 1052 came with 00.04.01. It was modded to an DS1102E for about 1/2 an hour using the file linked on the 1052e to 1102e thread, and then switched back using the same file with the patched by returned to the '50MHz' value (i.e. from 01 to 09, based on this post). There were no apparent problems with the update, nor the return to 50MHz. I just changed back because I figured I would maintain it in 'factory' state until necessary.

If there is important information I've not included, please let me know and I'll add it.

[theatrus]

Looking at the spec sheet, the oscilloscope has a typical gain error of 3% in this range. I'm not familiar with how this scope does offsets (and offset accuracy is not specified), however if we assume its using a digital display offset, a 5V signal would have a ~10mV gain error.  Your scope could still be in "typical" (who knows what max!) spec with one channel 10mV high and one channel 10mV low.

I'd make some more measurements to check if this is an offset or a gain error.

[larry42]

I'm sure that the probes also have an accuracy spec associated with them....

Sent from my One S using Tapatalk

[jlmoon]

Looking at the spec sheet, the oscilloscope has a typical gain error of 3% in this range. I'm not familiar with how this scope does offsets (and offset accuracy is not specified), however if we assume its using a digital display offset, a 5V signal would have a ~10mV gain error.  Your scope could still be in "typical" (who knows what max!) spec with one channel 10mV high and one channel 10mV low.

Correct me if I'm wrong.. but I think you could actually have accuracy errors up to 150mV (assuming >= (-75mV) to (<=75mV)  at 5 volts based on a  3% tolerance , am I correct? ... and that would be why they make precision digital volt meters for these test.

JLM

[theatrus]

Looking at the spec sheet, the oscilloscope has a typical gain error of 3% in this range. I'm not familiar with how this scope does offsets (and offset accuracy is not specified), however if we assume its using a digital display offset, a 5V signal would have a ~10mV gain error.  Your scope could still be in "typical" (who knows what max!) spec with one channel 10mV high and one channel 10mV low.

Correct me if I'm wrong.. but I think you could actually have accuracy errors up to 150mV (assuming >= (-75mV) to (<=75mV)  at 5 volts based on a  3% tolerance , am I correct? ... and that would be why they make precision digital volt meters for these test.

JLM

You are correct!

I need to go to calculator remediation school! 

[jlmoon]

nah.. just have to watch out for that sneaky decimal point.. happens to me all the time!

[bao8]

Thanks everyone for the really helpful responses.
I had looked at that accuracy info in the User Guide (and should have included it in the OP, really) but didn't quite get my head around it. Discussion here has helped. Basically, from what you have said it seems that a 20mV difference between channels at 5V is entirely within spec (which sounds reasonable, since that's actually only an error of <0.5%).

Unfortunately I don't have a power supply that can generate a 'known' voltage, and the error on my cheap multimeter is probably higher than the Rigol (and I have no way to check anyway). However, I do have a range of random power supplies around (up to about 18V) and if they generate a reasonably stable voltage I can check if the signal further diverges with increasing voltage which would suggest a gain error. Of course, there is no way to know where the gain error lies - CH1, CH2 or both.

@theatrus I haven't yet watched the various 'learn about oscilloscope' videos yet (sick baby = time comes in 10m snippets) which I will, but in the interim: just regarding digital offset - I don't think the DSO could be reading the value and then just doing a 'display' offset (rather than actually offsetting the incoming signal prior to ADCing it) because if it did then it would be using 8 bits to sample a >5V range. However, I'm able to see <0.01V variations (5V/256) so I think it must be adjusting the incoming signal prior to digitizing. The other thing that makes me think that is that when I do scroll, rapidly adjusting the vertical offest (i.e. too fast for the DSO to capture new signal) will sometimes get me to a clipped signal (i.e. I can see part of a waveform but a proportion of it looks clipped) before the DSO catches up. I suspect what I'm seeing there is the edge of the captured range.
As I said, though, I haven't watched the videos so I could be misunderstanding.

Anyway, thanks all for the help. I think the next step for me is to try to work out how to produce a set of accurately known voltages on the cheap (I'm on spending lockdown at the moment) to quantify the 'error', see if it is stable, and try to separate gain and measurement error. I have an uncle who is an ex-TV-repairman so I think that'll be my next avenue. In any case, I'll post findings here if/when I have them in case they are of use to someone else.

Thanks again.

[theatrus]

A cheap multimeter might still have an accuracy spec of <<1%, even 0.1% would be common. Top end multimeters get to 0.05-0.02% for DC volts.

[AG6QR]

The DC accuracy might be improved somewhat by switching the probes and scope to X1 instead of X10.  This removes any issues around imprecision in the voltage divider between the probe and the scope's input impedance.  But if the problem is elsewhere, this trick won't help.  Generally speaking, for AC signals, you want to use X10 any time that you can, in order to flatten the frequency response, but that's not an issue when measuring DC.

One point that is made in the "scopes for dopes" course video linked on this site is that a typical scope is a fairly poor voltmeter.  Its accuracy on the vertical axis isn't spectacular, and almost any multimeter can beat it there.  Compared to multimeters, scopes usually sacrifice some accuracy in favor of speed.

[bao8]

theatrus: Unfortunately I've long lost the manual for my multimeter and haven't been able to track down an electronic copy so I don't know the specs for it. That's good info though. (Edit: I have contacted the manufacturer now to try to get a copy of the manual and will hopefully get a response)

AG6QR: Thanks for pointing out the possible voltage-divider issue and AC/DC difference there; I'm planning to watch the videos but I was acutely aware that it will take me a while to get through them all and if the observed behaviour was something to worry about (which it seems it isn't, so that's all fine) then I wanted to jump on it immediately while the scope was only weeks old.

(My multimeter is a Dick Smith Electronics Q-1426, which is probably just a rebranded something else but hard to know)


The reason I noticed anything in the first place is simply because I was comparing three different USB voltage/current meters (Mini Voltage and Current Detector USB Charger Doctor Tester Meter Blue E0Xc, Double Function PC USB 2.0 Cable Device Voltage Current Meter Tester (which has been relisted due to 'an error in the listing'. hmm.) and USB Power Current Voltage Amp Tester Red LED Digital Charger Monitor Meter) and was checking they were not introducing any horrible noise in the voltage output. For that, the absolute value isn't so important so I can check the voltage using the multimeter and the noise using the scope.

[miguelvp]

My DS2072 has an offset of 15mV at power up on CH1. So I calibrated it after being turned on for 30 mins. Not sure how long it takes it to warm up but it comes down to nominal after a while.

Like others said. The scope is to measure times, for voltages use a DMM

Edit: the funny thing is that CH2 stays in place overtime no drift at all. Not a big deal anyways.