-
bug found? Hantek DSO5102P: different measures in AC vs DC coupling
Posted by fmnardini on 09 Apr, 2021 15:20 -
Hello all!
I recently bought a Hantek DSO5102P. Today, I measured the main voltage (220Vrms, 50Hz) in my office (using an insulation transformer, obviously) and what I see on the screen changes significantly in case I use AC or DC coupling.
In details, when using:
DC coupling: I read 592Vpp, -296Vmin, 296Vmax
AC coupling: I read 664Vpp, -332Vmin, 332Vmax
the testing condition is the same, i.e., probe connected to the output of the insulation transformer. the only thing that changes is coupling, AC vs DC.
I know that AC coupling exploits HPF and so, 50Hz could be attenuated. However, I am experiencing exactly the opposite. When using AC coupling, I see increased voltages.
What's going on here guys?
I am attaching two photos showing the oscilloscope and the measure.
Thanks a lot in advance for the help!
best,
Franco Maria (from Tuscany, Italy)
-
There's a post on Keysight blog about that:
https://blogs.keysight.com/blogs/tech/bench.entry.html/2018/10/18/when_to_use_ac_coupl-X0K7.html -
thanks, unfortunately it does not help!
I did some other experiments and I believe I found a bug of the oscilloscope.
I did some experiments by adding in parallel to the probe two TRMS recently calibrated testers (Brymen 869s and Fluke 117). They both return the same Vrms value returned by the oscilloscope when in AC coupling. In DC coupling, Vrms, Vpp, Vmin and Vmax returned are WRONG! (-10% more or less than the correct value).
Moreover:
1) I tested with four different probes (Hantek branded and others, both 1x and 10x): problem always present
2) I flashed the latest firmware available on your website (dated 2021-02-25) and after I run self-calibration. Nothing changed.
3) I tested several different voltages (220Vrms, 30Vrms). Error in DC is always more or less 10% of the AC readings.
4) I tested AC vs DC with the square wave from the internal generator: problem NOT present
5) I then tested a square wave sweeping from 10Hz to 1KHz. For frequencies greater than 200Hz the problem seems no more present.
6) I also tested a sin wave sweep from 10Hz to 1KHz. Again, from 200Hz on, the problem seems no more present.
In addition, a friend of mine owns a DSO4202C. I asked him to run the same test on the main voltage (220Vrms, 50Hz). The result is the same. The error rate is different (less than 10%). However, DC coupling readings are far less than the correct ones provided again by AC coupling.
I wrote to the Hantek service. Waiting for their reply.
ciao from Italy!
Franco Maria -
4) I tested AC vs DC with the square wave from the internal generator: problem NOT present
5) I then tested a square wave sweeping from 10Hz to 1KHz. For frequencies greater than 200Hz the problem seems no more present.
6) I also tested a sin wave sweep from 10Hz to 1KHz. Again, from 200Hz on, the problem seems no more present.
I think that answers the question.
Old oscilloscopes used an AC coupling capacitor in series with the input which was switched in and out of circuit using a relay. More modern designs tend to use a dual path input amplifier with the feedback implementing the AC coupling function because this is much easier to switch electronically, however this requires the time constants of the DC and AC paths to be matched for a flat frequency response above the transition region.
To verify this, look to see how the low frequency transient response changes between AC and DC coupling using a low frequency square wave. When AC coupling is used, a single pole (exponential) high frequency response should be present. A clean response for comparison purposes can be made by placing an approximately 0.19 microfarad capacitor in series with the x1 or x10 probe tip with DC coupling; 0.18 or 0.22 will work fine for testing purposes. In your examples, you can see the problem where the power line signal has a different shape when AC coupling is used.
Another way to test this is to compare the phase and amplitude of low frequency sine waves with one channel set to DC coupling and the other channel set to AC coupling. The difference should be a clean single pole response, but it won't be.
I consider any oscilloscope with this behavior to be defective by design; there is no excuse for it. I would return it for a refund if possible.
-
thanks a lot David!
that was my guess too and you confirmed. Just few hours ago I returned it. I tried to contact the service. I sent an email to service@hantek.com, info@hantek.eu.I also posted an issue on their forum. No answer after 5 days. Complete refund is the way!
thanks David!
best,
Franco Maria -
Good that you returned it, Franco!
When I looked at the images I immediately noticed the very distorted waves. Not even close to ideal sinusoids. You should’ve run an FFT analysis on those signals (DC coupled), maybe they had like 2-3% THD(!). I hope this is not the type of power you are being fed in Italy.
I am also assuming you were not using an ultra-low sampling rate (e.g. less than 100 Sa/s) on the scope.
I checked a sinusoidal 4Vp-p wave on my Siglent 2104X Plus with DC coupling. Switching to AC coupling I achieved - 3dB at 4.7Hz, which is within the published specs, which indicate 5 Hz (typical). So, it’s highly likely that the Hantek scope that you had was either defective or, worse, its design was seriously flawed.
Ciao from Canada! Lots of Italians here, too...
-
Old oscilloscopes used an AC coupling capacitor in series with the input which was switched in and out of circuit using a relay. More modern designs tend to use a dual path input amplifier with the feedback implementing the AC coupling function because this is much easier to switch electronically, however this requires the time constants of the DC and AC paths to be matched for a flat frequency response above the transition region.
To verify this, look to see how the low frequency transient response changes between AC and DC coupling using a low frequency square wave. When AC coupling is used, a single pole (exponential) high frequency response should be present. A clean response for comparison purposes can be made by placing an approximately 0.19 microfarad capacitor in series with the x1 or x10 probe tip with DC coupling; 0.18 or 0.22 will work fine for testing purposes. In your examples, you can see the problem where the power line signal has a different shape when AC coupling is used.
Another way to test this is to compare the phase and amplitude of low frequency sine waves with one channel set to DC coupling and the other channel set to AC coupling. The difference should be a clean single pole response, but it won't be.
I consider any oscilloscope with this behavior to be defective by design; there is no excuse for it. I would return it for a refund if possible.
Here is sine wave and square wave 100 Hz from a signal generator connected through a T-BNC and two coax cables (the same length) into CH1 and CH2. The channel CH2 is always DC coupling. The channel CH2 taken with AC and DC coupling.
Does it means that my unit is defective by design?
-
Here’s mine on DC and AC coupling, for a square wave at 100 Hz. Indeed, as David mentioned previously, it seems that that input capacitor on AC coupling, or dual-path-input amplifier really, does not like much the sharp-transitioning, transient signal up to about 300Hz. Absolutely no difference in shape on a sine wave, as expected. (Note the I’ve got a 250mV DC offset in the signal I input.)
Here is sine wave and square wave 100 Hz from a signal generator connected through a T-BNC and two coax cables (the same length) into CH1 and CH2. The channel CH2 is always DC coupling. The channel CH2 taken with AC and DC coupling.
Does it means that my unit is defective by design?
-
Hurray! Our units are all defective by design!!!
Where’s the queue for returning them? 
I consider any oscilloscope with this behavior to be defective by design; there is no excuse for it. I would return it for a refund if possible.
Here is sine wave and square wave 100 Hz from a signal generator connected through a T-BNC and two coax cables (the same length) into CH1 and CH2. The channel CH2 is always DC coupling. The channel CH2 taken with AC and DC coupling.
Does it means that my unit is defective by design?
-
I don't see an RMS measurement on the scope--how are you comparing that to the meters?
-
In your examples, you can see the problem where the power line signal has a different shape when AC coupling is used.
Another way to test this is to compare the phase and amplitude of low frequency sine waves with one channel set to DC coupling and the other channel set to AC coupling. The difference should be a clean single pole response, but it won't be.
I consider any oscilloscope with this behavior to be defective by design; there is no excuse for it. I would return it for a refund if possible.
I agree as to the defectiveness (pending an answer to my RMS question) but this comparison should be done on a clean sine wave, not mains power. One factor here may be that the transition frequency of the AC coupling is ridiculously high, so I'd measure that first. -
Thanks a lot Michael!
that was my guess: poor design for all the units and this affects also other families of Hantek oscilloscopes (I have experience on three of them, same results).
I returned it and I will go for a Siglent!
thanks! ciao!
FM
ps: yes, I know... lot of italians in Canada, although italians are everywhere!
-
Thanks bdunham7,
I did a lot of tests. The photos above do not show Vrms, you are right. However, I tested that both on a clean sin wave and on the main power.
Thanks!
ciao from Italy!
FM -
OK, assuming everything was set up right, if the calculated RMS displayed on the scope was 10% low compared to the DMMs at 50Hz with DC coupling, then the scope is garbage.
As Keysight Daniel Bogdanoff explained, you would typically use DC coupling looking at any low frequency AC signal. If you use AC coupling, you have to be aware of the AC coupling response and how it might naturally affect your signal. As David Hess explained, this should be a single-pole response such as what you would expect from a simple RC filter. Anything other than that will cause some pretty strange problems.
Here are four screenshots that should be self-explanatory and show a Siglent SDS1104X-E with an approximately 1Hz cutoff with AC coupling, both with a direct 1Vrmsinput from a 50R impedance signal generator and with calibrated 10X probes.
-
I get a difference of -0.5% for AC coupling vs. DC coupling on the scope. Compared to my DMM which measures 1.421V TRMS, AC coupling on the scope is -0.63% different and DC coupling on the scope is -0.13%.
OK, assuming everything was set up right, if the calculated RMS displayed on the scope was 10% low compared to the DMMs at 50Hz with DC coupling, then the scope is garbage.
-
Thanks for your tests Michael and bdunham7. I returned mine. I will go for a siglent too in the near future.
Thanks!
FM -
Here are four screenshots that should be self-explanatory and show a Siglent SDS1104X-E
It looks that Siglent made some change in X-E model RF frontend. Do you hear relay switch when changing DC to AC coupling? -
It looks that Siglent made some change in X-E model RF frontend. Do you hear relay switch when changing DC to AC coupling?
No relays. I think you've posted about your scope before and perhaps we've discussed it. It doesn't look good to me.... -
I think you've posted about your scope before and perhaps we've discussed it. It doesn't look good to me....
yes, this is SDS1102X -
Here is sine wave and square wave 100 Hz from a signal generator connected through a T-BNC and two coax cables (the same length) into CH1 and CH2. The channel CH2 is always DC coupling. The channel CH2 taken with AC and DC coupling.
Does it means that my unit is defective by design?
That response looks bad but a lower test frequency is needed to see more of what is going on. At higher frequencies, the square wave will look "tilted". Old AC current probes have the same response and this specification is actually given as "tilt".
-
That response looks bad but a lower test frequency is needed to see more of what is going on.
I think the issue with his scope is the DC-coupled response, not the AC-coupled slope, which I would expect to a certain extent. The DC-coupled just looks wrong and there's no good explanation for it AFAIK. -
I get a difference of -0.5% for AC coupling vs. DC coupling on the scope. Compared to my DMM which measures 1.421V TRMS, AC coupling on the scope is -0.63% different and DC coupling on the scope is -0.13%.
Obviously there should be an amplitude difference as the input frequency approaches the cutoff frequency, but it should be small at 50 or 60 Hz since this it would be common to make AC coupled measurements down to this point.
Since on old oscilloscopes have the AC coupling capacitor in series with the input, the low frequency cutoff also depends on the probe's input resistance, so x10 probes extend the frequency response to 1 Hz instead of 10 Hz.
-
Here are four screenshots that should be self-explanatory and show a Siglent SDS1104X-E with an approximately 1Hz cutoff with AC coupling, both with a direct 1Vrmsinput from a 50R impedance signal generator and with calibrated 10X probes.
Just found that you're doing it for 1 Hz. My previous screenshots was for 50 Hz. Can you please test your scope for sine with 50 Hz?Here is sine wave and square wave 100 Hz from a signal generator connected through a T-BNC and two coax cables (the same length) into CH1 and CH2. The channel CH2 is always DC coupling. The channel CH2 taken with AC and DC coupling.
Does it means that my unit is defective by design?
That response looks bad but a lower test frequency is needed to see more of what is going on. At higher frequencies, the square wave will look "tilted". Old AC current probes have the same response and this specification is actually given as "tilt".
The same measurement 3 Vpp for 10 Hz and 1 Hz:
-
Just found that you're doing it for 1 Hz. My previous screenshots was for 50 Hz. Can you please test your scope for sine with 50 Hz?
I did. There are 4 shots, the first at 1Hz showing the approximately 3dB/45 degrees you expect from a first-order filter at the nominal cutoff frequency, the second at 50Hz showing nearly perfect coincidence between AC and DC coupling.
The next two are the same thing, except that I'm using freshly calibrated 10X probes instead of connecting directly to the SDG2042X-h with BNC cables, and I have reversed the channels vis a vis the DC/AC coupling. Same result. It appears to be working correctly with a 1Hz first-order AC coupling response that does not vary with probe impedance.
Your scope is weird. It appears to have a peak in the response at some very low frequency. Perhaps you could do a low frequency sine wave sweep from 0.1Hz to 10 Hz and see if there is a peak? -
Your scope is weird. It appears to have a peak in the response at some very low frequency. Perhaps you could do a low frequency sine wave sweep from 0.1Hz to 10 Hz and see if there is a peak?
Two sweeps, first from 5 Hz to 100 Hz and second from 5 Hz to 1000 Hz.
Below 5 Hz amplitude response is flat. Above 100-200 Hz amplitude response is also flat.