Author Topic: Problem with my scope? Reading positive versus negative average DC voltages.  (Read 1544 times)

0 Members and 1 Guest are viewing this topic.

Offline wobblyTopic starter

  • Regular Contributor
  • *
  • Posts: 177
  • Country: gb
I have noticed a funny problem with my two-channel digital scope.  It looks like the two probes are not measuring voltages very well when one probe is measuring a voltage that is the opposite polarity to the other one!

The scope is a Keysight EDUX1052G, about 3 years old or so and has had an easy life without ever seeing voltages larger than +/- 30V DC.  Hasn't had many hours on it either, nor much of a temperature swing (in the UK).

To measure the average voltages, I'm not relying on my failing eyesight to read the waveform, I'm using the built-in "Measurement: Average Full Screen" or the "Analyse: DMM" features.  Both of which are pretty good when I'm using unipolar voltages.

Investigating further...

Lets say both probes are measuring the average voltage between GND and POSITIVE 10V; the scope reports the same results with a small discrepancy of perhaps +/-0.05V which I'm quite OK with.

Same thing goes for both probes are measuring the average voltage between GND and NEGATIVE 10V +/- 0.05V.

However, when I have one probe on +10V and one on -10V, there is suddenly a roughly 0.5V discrepancy between the measurements!  So when I'm expecting to see +10V and -10V respectively, I'm actually seeing +9.8V and -10.3V.  The same discrepancy exists when I swap the probes and measure the other way.

So I checked the obvious stuff...

  • Made sure both probes were set to "X10" mode on the physical probe switches.
  • Checked that the menu was also set to X10 for each probe.
  • Ran the self probe check on both probes (they passed).  But this is using the positive-only DC square wave on the front of the scope.  Hmm.
  • I'm using both GND clips on the probes connected to the same GND point.
  • Three separate multimeters (Fluke 117, Aneng 8008, Uni-T Clamp meter) all agree that the scope is at fault!

Is my scope faulty?  Or is this something to do with the way scope front-ends work?  0.5V is close to a diode drop.  What's goin' on?
« Last Edit: September 24, 2024, 08:25:11 pm by wobbly »
 

Offline wobblyTopic starter

  • Regular Contributor
  • *
  • Posts: 177
  • Country: gb
Captain's Log Supplemental:

I have only noticed this recently because I have been building a bipolar PSU (simple stuff: https://www.eevblog.com/forum/beginners/ua741-rail-splitter-with-complementary-darlington-emitter-follower/).

I was trying to use each scope channel to measure one voltage rail while I tuned the virtual GND to be exactly 0V, or half-way between the two rails.  I'd set the bench PSU to 20V for example and hook up CH1 to the nominal +10V rail, and CH2 to the -10V rail.

So after getting the scope's two readings to be exactly +9.97V and -9.98V (ish, given tolerances and losses) I was pretty happy it was set up right.  Only to verify with a multimeter to find it was still out by half a volt!

I don't have the experience to doubt a Keysight instrument's veracity, so I'm sure I'm doing something stupid.
« Last Edit: September 24, 2024, 08:03:25 pm by wobbly »
 

Offline Phil1977

  • Frequent Contributor
  • **
  • Posts: 737
  • Country: de
Did you check the voltages simultaneously with DMM and scope?

It seems that something shifts the GND-potential of the scope by 0.3V. Can you connect a DMM between the scope GND (e.g. at the signal generator output) and your GND point of the voltage source?

Is your +/- 10V voltage source floating or somehow connected to PE? I would nearly bet something small that there is a ground loop.
 
The following users thanked this post: wobbly

Offline wobblyTopic starter

  • Regular Contributor
  • *
  • Posts: 177
  • Country: gb
Did you check the voltages simultaneously with DMM and scope?

It seems that something shifts the GND-potential of the scope by 0.3V. Can you connect a DMM between the scope GND (e.g. at the signal generator output) and your GND point of the voltage source?

Is your +/- 10V voltage source floating or somehow connected to PE? I would nearly bet something small that there is a ground loop.

Good idea.  My Fluke 117 reports 10.1V (stable) measured with Fluke's GND probe on the benchtop PSU's GND and the positive probe on the Scope's sig-gen GND BNC outer metalwork.  This checks out, since the Scope probes GND clips are sharing the virtual GND of the bipolar supply.

Does not look like there is any ground loop causing a voltage shift.  The bench PSU's GND output (Aim TTI PL303-P, which is also about 3 years old and I trust completely) is definitely floating.
« Last Edit: September 24, 2024, 08:47:55 pm by wobbly »
 

Offline Phil1977

  • Frequent Contributor
  • **
  • Posts: 737
  • Country: de
You definitely have very trustworthy equipment!  8)


Good idea.  My Fluke 117 reports 10.1V (stable) measured with Fluke's GND probe on the benchtop PSU's GND and the positive probe on the Scope's sig-gen GND BNC outer metalwork.  This checks out, since the Scope probes GND clips are sharing the virtual GND of the bipolar supply.

I hope it reports around 0V between both GND... But I think you measured the right things. Then I´ve got no idea about the reason, I agree that 3% deviation would be more than expected for a reputable scope.

Do you have such banana-->BNC adapters? You could measure with the DMM through the scope probe.
 
The following users thanked this post: wobbly

Online tautech

  • Super Contributor
  • ***
  • Posts: 29812
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
......
I hope it reports around 0V between both GND... But I think you measured the right things. Then I´ve got no idea about the reason, I agree that 3% deviation would be more than expected for a reputable scope.
Cough.

+ 3% vertical accuracy has been industry standard for decades but now improved near 10x with the advent of cheap 12bit DSO's.

Spec from the datasheet for the DSO OP uses is:
DC vertical gain accuracy:
+3% full scale (≥ 10 mV/div)
+4% full scale (< 10 mV/div)
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline wobblyTopic starter

  • Regular Contributor
  • *
  • Posts: 177
  • Country: gb
Do you have such banana-->BNC adapters? You could measure with the DMM through the scope probe.

Might be able to scrounge something together at work.  Can you go into more detail about what you mean? :)
 

Offline Phil1977

  • Frequent Contributor
  • **
  • Posts: 737
  • Country: de

Cough.
A good rule of thumb for industrial machinery is if you specify 3% tolerance then your standard deviation should be a tenth of it - 0.3%.

It may be in spec, but are the scopes really that bad? I must admit I rarely ever used a scope for something the DMM should measure.
 

Offline Phil1977

  • Frequent Contributor
  • **
  • Posts: 737
  • Country: de
Do you have such banana-->BNC adapters? You could measure with the DMM through the scope probe.

Might be able to scrounge something together at work.  Can you go into more detail about what you mean? :)

Just an adapter so that your DMM can measure what comes out of the scope probe.
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29812
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.

Cough.
A good rule of thumb for industrial machinery is if you specify 3% tolerance then your standard deviation should be a tenth of it - 0.3%.

It may be in spec, but are the scopes really that bad? I must admit I rarely ever used a scope for something the DMM should measure.
8bit, yes but it has never stopped anyone from doing valuable work while they are aware of the limitations of their instrument.
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Online bdunham7

  • Super Contributor
  • ***
  • Posts: 8178
  • Country: us
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #10 on: September 25, 2024, 12:15:32 am »
Good idea.  My Fluke 117 reports 10.1V (stable) measured with Fluke's GND probe on the benchtop PSU's GND and the positive probe on the Scope's sig-gen GND BNC outer metalwork.  This checks out, since the Scope probes GND clips are sharing the virtual GND of the bipolar supply.

This is not adding up for me.  Are the ground clips of both scope probes connected to the PSU GND connection?  If not, can you quickly draw out a schematic of exactly how you have everything connected?
A 3.5 digit 4.5 digit 5 digit 5.5 digit 6.5 digit 7.5 digit DMM is good enough for most people.
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 17526
  • Country: 00
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #11 on: September 25, 2024, 12:24:31 am »
It may be in spec, but are the scopes really that bad?

Yes. Scopes are accurate horizontally, not vertically.

I expect them to be linear but vertical numbers on 'scopes are a ballpark figure.
 
The following users thanked this post: wobbly

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 17526
  • Country: 00
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #12 on: September 25, 2024, 12:36:00 am »
I agree that 3% deviation would be more than expected for a reputable scope.
Cough.

+ 3% vertical accuracy has been industry standard for decades but now improved near 10x with the advent of cheap 12bit DSO's.

Spec from the datasheet for the DSO OP uses is:
DC vertical gain accuracy:
+3% full scale (≥ 10 mV/div)
+4% full scale (< 10 mV/div)

What's the spec for crosstalk between channels?
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29812
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #13 on: September 25, 2024, 12:40:11 am »
I agree that 3% deviation would be more than expected for a reputable scope.
Cough.

+ 3% vertical accuracy has been industry standard for decades but now improved near 10x with the advent of cheap 12bit DSO's.

Spec from the datasheet for the DSO OP uses is:
DC vertical gain accuracy:
+3% full scale (≥ 10 mV/div)
+4% full scale (< 10 mV/div)

What's the spec for crosstalk between channels?
Dunno, a while ago closed the datasheet tab.
Easy to find though.....
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 17526
  • Country: 00
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #14 on: September 25, 2024, 01:49:10 am »
Easy to find though.....

Not THAT easy...

I found one and it isn't specified.

https://eu.mouser.com/datasheet/2/878/5992_3484-1759582.pdf

I suspect that's what this might be though.
« Last Edit: September 25, 2024, 01:53:59 am by Fungus »
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29812
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #15 on: September 25, 2024, 02:03:29 am »
Easy to find though.....

Not THAT easy...

I found one and it isn't specified.

https://eu.mouser.com/datasheet/2/878/5992_3484-1759582.pdf

I suspect that's what this might be though.
0.5V over 20V is far more than one would expect from channel crosstalk but within accuracy spec.
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 17526
  • Country: 00
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #16 on: September 25, 2024, 06:02:30 am »
0.5V over 20V is far more than one would expect from channel crosstalk but within accuracy spec.

Yeah, I guess.

The next question would be does the reading change on a channel between when they're both connected to +10V and you connect the other channel to ground, not -10V.
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29812
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #17 on: September 25, 2024, 07:11:02 am »
0.5V over 20V is far more than one would expect from channel crosstalk but within accuracy spec.

Yeah, I guess.

The next question would be does the reading change on a channel between when they're both connected to +10V and you connect the other channel to ground, not -10V.
Revisiting the OP, what one thing has the OP not done ?
 :popcorn:
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline Phil1977

  • Frequent Contributor
  • **
  • Posts: 737
  • Country: de
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #18 on: September 25, 2024, 07:11:33 am »
I´ve just checked the Rigol 1054 out of curiosity: (without probe, DC directly fed into BNC, normal sampling mode, AVG-measurement, all tests with 1V/div)

2382151-0

It´s not good, but it seems to follow the 1/10th-of-the-tolerance rule. I checked for DC-crosstalk with a negative voltage on Ch2 and there was nothing.

@wobbly: Is it possible your trigger is tricking you? Sometimes the 50/60Hz-hum mixes into the triggering and then it looks on the screen like a DC-offset for fast horizontal settings. May you try to set the timebase to something slower or use roll-mode?

PS: Is there a good way to copy tables from libreoffice (<=> excel) into the forum without using images?
 
The following users thanked this post: wobbly

Offline wobblyTopic starter

  • Regular Contributor
  • *
  • Posts: 177
  • Country: gb
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #19 on: September 25, 2024, 07:36:30 am »
I just tried this experiment at work (with completely different equipment) and got almost the exact same result!

The work scope is an old Tek 3034.

I guess I just wasn't expecting it to be such a large discrepancy.  If that's just the way I things are then so be it!

Shame it's not something that can be adjusted by the user.
« Last Edit: September 25, 2024, 07:38:46 am by wobbly »
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 17526
  • Country: 00
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #20 on: September 25, 2024, 07:43:52 am »
I just tried this experiment at work (with completely different equipment) and got almost the exact same result!

If that's just the way I things are then so be it!

It really is...  :)
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29812
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #21 on: September 25, 2024, 07:51:45 am »
0.5V over 20V is far more than one would expect from channel crosstalk but within accuracy spec.

Yeah, I guess.

The next question would be does the reading change on a channel between when they're both connected to +10V and you connect the other channel to ground, not -10V.
Revisiting the OP, what one thing has the OP not done ?
 :popcorn:
Not worked it out yet ?
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 17526
  • Country: 00
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #22 on: September 26, 2024, 01:51:52 am »
Self-cal?
 
The following users thanked this post: tautech

Offline Aldo22

  • Super Contributor
  • ***
  • Posts: 1275
  • Country: ch
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #23 on: September 26, 2024, 06:38:27 am »
With my cheap scopes I have found that the positive DC measurement is quite good (definitely within 3%, usually better) but negative DC polarities can be off by 5-6%.
Is this a general observation or is it a “special feature” of cheap scopes?

Yes, I know, scopes are not made for that...  ;)
 

Offline Phil1977

  • Frequent Contributor
  • **
  • Posts: 737
  • Country: de
Re: Problem with my scope? Reading positive versus negative average DC voltages.
« Reply #24 on: September 26, 2024, 06:52:32 am »
And please check if the voltage levels change when playing with the trigger settings. I´ve seen it for too often that Auto-mode was expected to be free-running but was in phase with something unexpected. This exactly looks like a DC-offset and would affect the "average full screen" in the same way.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf