Hi everyone.

Is there any reason why an (almost new) oscilloscope in AC coupling mode shows a small offset (a couple of milli volts)?
Example (using 1x probe):

Is this normal for relatively cheap new oscillosopes?
I didn't realize this behaviour with anolog scopes.
The offset spoils the (AC) RMS measurements, as it takes into account a mean value, and it is annoying.

Thank you!

 

Check the operating manual. There is usually a way to adjust the zero offset even in the most inexpensive scopes..
Regards,
Keith

 

I dont know, I've not ussed a scope to measure RMS,
but
I'd have thought the RMS needs to know the DC offset of the signal,
so can you take an RMS measurement if the scope is AC coupled ?

 

It's because the frequency of the signal is too low to fit on the screen.
I'm guessing that you have about 500us per division so the screen is about 7ms wide, and that the signal you are measuring is perhaps 50Hz, so that the negative-going half is off the screen to the right.

 

AC coupling mode shows a small offset (a couple of milli volts)? ...
Is this normal for relatively cheap new oscillosopes?

What is the vertical sensitivity? DSO's are infamous for their fuzzy traces.

I didn't realize this behaviour with anolog scopes.

Analog scopes can also have an offset. Unlike DSO's, you don't need to go through a menu to remove the offset; you just turn the knob.

The offset spoils the (AC) RMS measurements, as it takes into account a mean value, and it is annoying.

Just measure the peak value and divide by 1.414 to get RMS.

 

Just measure the peak value and divide by 1.414 to get RMS.

. . . if you want the wrong answer - it doesn't look remotely like a sinewave to me.

 

. . . if you want the wrong answer - it doesn't look remotely like a sinewave to me.

Then an RMS value wouldn't be applicable would it?

 

Perfectly. Root-mean-square is appropriate to any shape of waveform. It simply gives a measure of the average power it would deliver into a resistive load.
Its value is the square root of the integral over time of the square of the voltage, which just happens to be peak/root2 for a sinewave; or root3 for a triangle wave.
see https://www.allaboutcircuits.com/textbook/alternating-current/chpt-1/measurements-ac-magnitude/

 

Hi everyone.

Is there any reason why an (almost new) oscilloscope in AC coupling mode shows a small offset (a couple of milli volts)?
Example (using 1x probe):
View attachment 215759
Is this normal for relatively cheap new oscillosopes?
I didn't realize this behaviour with anolog scopes.
The offset spoils the (AC) RMS measurements, as it takes into account a mean value, and it is annoying.

Thank you!

Check the manufacturers websites for new firmware versions and install them as necessary.
After which and the scope has been running for some while so good and warm, run the Self Cal utility which should address AC coupling channel offset.
Better measurement accuracy can sometimes be got using measurement Statistics if your scope has such.

 

Just turn the timebase down to a slower speed, and the offset will be gone when you can see the rest of the waveform

 

Check the operating manual. There is usually a way to adjust the zero offset even in the most inexpensive scopes..

Unfortunatly no. The only reference in the whole manual is: "AC: block all the DC components and attenuate signals lower than 8 Hz. Use AC coupling to get a stable edge trigger when your waveform has a large DC offset."
Of course I can move the trace up and down (it is called offet too), but that doesn't fix the issue. The trace mean value is over the offset level.

I'd have thought the RMS needs to know the DC offset of the signal,
so can you take an RMS measurement if the scope is AC coupled ?

RMS can be measured for AC only, or for AC+DC. In my case, it takes into account DC (and a wrong DC!) in the AC mode, which is not as expected.

It's because the frequency of the signal is too low to fit on the screen.

No. I can set a 1 s/div and take a long capture, and the trace mean value is always above.

 

If you can move the trace up and down in AC mode then I don't understand your problem. Does the mean value shown change with the position of the trace?

 

Is there still an offset with the input not connected?

 

And is there still an offset with the input shorted?

 

If you can move the trace up and down in AC mode then I don't understand your problem. Does the mean value shown change with the position of the trace?

The mean value doesn't change if I move the trace up and down.

Is there still an offset with the input not connected?

Yes, there is.

And is there still an offset with the input shorted?

Yes, (approx) the same as with no input.

 

Is there an offset if you set the input to GROUND?

 

Is there an offset if you set the input to GROUND?

No offset whatsoever. Ground is right on 0 V.

 

The signal is AC coupled. Even if its AC coupled there will be a DC offset if the area under the positive peaks does not equal the area under the negative peaks (its called integration).
I believe what you are seeing is correct
Prove it yourself by coupling in a clsoe to perfect sine wave and see if the offset still exists.

 

The signal is AC coupled. Even if its AC coupled there will be a DC offset if the area under the positive peaks does not equal the area under the negative peaks (its called integration).
I believe what you are seeing is correct
Prove it yourself by coupling in a clsoe to perfect sine wave and see if the offset still exists.

Yes, there is. If I disconnect the probe (nothing connected to the BNC input), there's an offset too. If I set the time base to something like 10 or 20 ms (that's already pretty slow), there's still a non-zero mean value.

I updated firmware and did a self callibration successfully. Still, the offset is there.
Could it be a fault? Sometimes is larger, sometimes is lower (same setup).
Today, while working on a real circuit, I got zero offset for a couple of minutes. Then, the offset came back (same circuit, same probe, same attenuation, etc.)
Any idea?

 

Could it be temperature related? If so, it could be relatively easy to find the offending component.
Regards,
Keith