My instruments are lying. Or maybe not.

[fc3000]

Hy all.
I am measuring the output of a bench PSU.
I am using a Rigol DS1054 SCOPE, a Rigol DM3068 Bench Multimeter and a Fluke83 V Multimeter.

So I set the PSU at 1V and Get:
Scope: 1.02V DC
Bench Meter: 0.942 V
Fluke: 0.942V.

I do get it that the scope has a higher impedance. But does that make a difference if the PSU can supply a 3A?
I would like to know what to trust.

I have tested the output up to 30V and the difference is throughout the range.
I also have done self cal on the scope.

Any thoughts.
Regards.

[TimFox]

What is the manufacturer's specification for DC accuracy on the 'scope?

[bdunham7]

How are you connecting to the scope and what range are you using?

[dobsonr741]

You are complaining about a 2% error. It’s well within the scope’s specification.

In general, a scope with an 8bit ADC will not make up to the precision of a 3.5 digit multimeter. Use the scope what’s it’s made for - waveform analysis.

[bdunham7]

You are complaining about a 2% error. It’s well within the scope’s specification.

Assuming the two meters are accurate (a pretty good bet), it is actually an 8% error.

[wraper]

To measure DC more or less accurately, you need to run scope self calibration first. And scope must be warmed up for at least 30 minutes (before calibration and when measuring) as DC offset tends to drift a lot. Also selected vertical scale must be the lowest possible as you have just 8 bits of full range resolution.

[tunk]

If you e.g. use 1V/div, then full scale is 8V and 3% accuracy is 0.24V.
At 0.5V/div it's 0.12V. Both within your measurement.

[coppice]

Using a general purpose scope as a precision measurement tool is a strange usage. They don't control the DC offset very well, and an 8 bit ADC reading changes in steps that are about 0.5% of full scale. A few scopes have a precision measurement system that is separate from the waveform capture system. General purpose scopes are great for studying waveform details, but only suitable for getting a rough estimate for most measurements. If a multimeter says 0.942V, and a scope shows somewhere in the ballpark of 1V, that's considered agreement.

[fc3000]

Thank you to all the replies.

I was not complaining at all. I was trying to establish which tools were more credible.
Your comments are very useful.  So my trusty old Fluke is still OK.

Thank you very much.
Regards

[u666sa]

What kind of resistance you getting with your probes crossed? 

[calzap]

Suggest you buy a voltage standard.  Search online to find several inexpensive ones.  Can get a 5 V standard for less than US$50.   I use DMM Check which costs more but has voltage, resistance and current standardization.  Company has been in business for years.  Standard can be sent back to the manufacturer for recalibration when desired.

Mike in California

[BeBuLamar]

While it shouldn't make a different but I believe the scope has lower impedance than the 2 meters.

[TimFox]

Normal DMMs have an input resistance close to 10 megohms, while normal 'scopes are 1 megohm (at DC).
Some DMMs have very high input resistance on low-voltage full-scale settings (specified only as a minimum).
To see 3% difference between a 1 megohm load and a 10 megohm load, the output resistance of your source would need to be about 30 k\$\Omega\$, which is unlikely.

[coppercone2]

the voltage you are testing at is 1 volt. the scope is consuming 1 microamper of current. the hand held dmm are consuming 100 nano ampers of current.

a 1kohm input impedance would consume 1 miliamper of current. (low z multimeter)

your supply losses accuracy when it supplies more current sometimes.

The only thing I can think of is that your supply might suck when its unloaded. Put a 100k resistor on it as a load and measure it with all 3 meters?

the meter should all read the same regardless of input impedance if you load the supply. or even 10k load is probobly fine, it might drift a tiny bit with a bad resistor


the other thing you do is connect all the meters together at the same time. this impedance issue would only show up if you individually measure it. Or if you happen to be measuring voltage through a salty pasta noodle

[MarkT]

Mmmmm, salty pasta noodles!

I suspect a cheap x10 probe might not have very accurate 9M resistor in it, so that's another source of possible error with the 'scope.

[vk6zgo]

You are complaining about a 2% error. It’s well within the scope’s specification.

In general, a scope with an 8bit ADC will not make up to the precision of a 3.5 digit multimeter. Use the scope what’s it’s made for - waveform analysis.

I would not totally dissuade the OP from using a 'scope to look at DC, as it can be convenient when troubleshooting to just probe power rails to see if they are "round about" the right voltage.

I don't have a DSO, but have used an analog 'scope for this for years.
If I want to find the actual DC voltage, I use a Fluke.

Something I have done many times is to measure a supply rail with a Fluke, then whilst probing the same voltage, setting my CRO to "vert uncalibrated", & adjusting the display to a convenient scale more closely resembling the measured voltage.
Analog 'scopes have poor calibrated accuracy, but their "reset accuracy" is good so I can easily determine smaller variations than those visible in the calibrated position.

Of course, most people with DSOs hate "counting squares" & will simply take the readout as gospel.

[radiolistener]

So I set the PSU at 1V and Get:
Scope: 1.02V DC
Bench Meter: 0.942 V
Fluke: 0.942V.

This is pretty expected result. Usual oscilloscope precision is ±5%, so if it shows 1V as a value within 0.95...1.05 V this is a good result. 

[CaptDon]

Oscilloscopes make very poor frequency counters and generally poor DVM's unless specifically equipped for that purpose. Even when you use the cursors to try to measure it is a problem because of the thickness of the trace. If you need accurate frequency measurements buy a frequency / timer / counter (freq counter) with an OCXO timebase reference. If you need accurate voltage measurements buy a Keithly or some other upscale 4 1/2 digit DMM. And if you do RF / IF alignment work always have an analog meter handy for 'peaking' and 'zero beat' use.

[coppice]

Oscilloscopes make very poor frequency counters and generally poor DVM's unless specifically equipped for that purpose. Even when you use the cursors to try to measure it is a problem because of the thickness of the trace. If you need accurate frequency measurements buy a frequency / timer / counter (freq counter) with an OCXO timebase reference. If you need accurate voltage measurements buy a Keithly or some other upscale 4 1/2 digit DMM. And if you do RF / IF alignment work always have an analog meter handy for 'peaking' and 'zero beat' use.

Oscilloscopes are actually very valuable tools in their frequency counting role. You can get very confused by the results of a simple frequency counter, when the waveform has any complexity to it. An oscilloscope lets you see what is really going on, and the timing you are truly measuring. If you are looking for something like 1 in 10^9 accuracy in the timing measurement a scope is not tool for you, but they do have an important role to play.

[bdunham7]

Oscilloscopes are actually very valuable tools in their frequency counting role. You can get very confused by the results of a simple frequency counter, when the waveform has any complexity to it. An oscilloscope lets you see what is really going on, and the timing you are truly measuring. If you are looking for something like 1 in 10^9 accuracy in the timing measurement a scope is not tool for you, but they do have an important role to play.

The same applies to voltage measurements, AC and DC.  A DSO might struggle to acheive 1% accuracy, but it's ability to deal with such issues as crest factor (with peaks within bounds, of course) and bandwidth make it useful where most meters fail.

[fourfathom]

I agree that the scope measurement accuracy likely accounts for the difference.  But if you are concerned about loading effects then just connect all three (meters, scope) to the voltage source.  Then they will be truly seeing the same input.  And then disconnect the scope and see if the meter readings change (I consider this unlikely.)

[DavidAlfa]

At which V/div setting? People seem to keep forgetting scopes usually have a 8-bit ADC, which means only 256 positions for the entire vertical span.
So for 1V, you should move the trace all the way to the bottom and set it to 200mV/div, which will give an ADC resolution (200mv * 8-10 divisions depending on the scope / 256) of about 6-8mV.
Also 10x probes might not be very precise, try measuring 1V in 1x mode.

[vk6zgo]

Oscilloscopes make very poor frequency counters and generally poor DVM's unless specifically equipped for that purpose. Even when you use the cursors to try to measure it is a problem because of the thickness of the trace. If you need accurate frequency measurements buy a frequency / timer / counter (freq counter) with an OCXO timebase reference. If you need accurate voltage measurements buy a Keithly or some other upscale 4 1/2 digit DMM. And if you do RF / IF alignment work always have an analog meter handy for 'peaking' and 'zero beat' use.

I don't know how effective modern DSOs are for adjusting for a voltage peak, or a "Zero beat", as the early ones were useless, especially for the latter, as they kept changing sample rate as the beat note got lower in frequency, giving a very confusing display.

Analog 'scopes" are very useful for just such jobs, & if you have one which presents a sample of the vertical signal on the rear panel, can be used as a sensitive preamplifier for a frequency counter, giving "the best of both worlds".

[tggzzz]

I was not complaining at all. I was trying to establish which tools were more credible.
Your comments are very useful.  So my trusty old Fluke is still OK.

All tools lie, but some are useful.

To determine the boundary, read and understand each tools' specifications and manual.