Scope not showing what I expected.

[mathsquid]

Last week I was working on looking at a signal on my oscilloscope.  When I had it DC coupled, the signal showed spikes. I wanted to zoom in on them, but when I switched to AC coupling, they disappeared.  I attached two pics of the screen, and the only difference between the two is the coupling.  I did some other experimenting (trying the other channel, changing the triggering) but I never could get it to show the spikes on AC coupling.  Any idea what's going on?

[ataradov]

Those spikes are too slow for AC. You can still kind of see small bumps there, but they are not making it though the AC filter.

[DaJMasta]

Though if you zoom in, you probably can still see a very small signal at the same speed - it looks like there are single pixel blips on your picture at about the same intervals.

AC coupling puts a capacitor between your probe input and the front end of the scope, so as the signal gets closer to DC, the waveform starts to roll off.  The AC coupling on my scope seems pretty good, with a 1Vpp sine wave it has no problem seeing it at full amplitude at about 3.3Hz, but below that, the signal seen by the scope is smaller because it's going too slowly for the size and type of capacitor used on the front end, and while having a giant cap will give you better low frequency performance, they're also expensive, are bad for high frequencies, and have other design issues to address, so generally AC coupling on scopes can really only be counted on down to maybe to 10Hz or so.

A changing (AC) signal will still show until you hit the noise floor though, even if it reads a different value.  While I see 1Vpp at 3.3Hz, I see about 650mV at 1Hz, and about 95mV at 0.1Hz.  Since a spike like that is decently sized and has sharp edges, it should definitely show up as something when AC coupled, just at a much smaller amplitude.

[mathsquid]

Thanks for the explanations. I appreciate them (though part of me was looking for an excuse to get a new scope ).

[abraxa]

I wanted to zoom in on them, but when I switched to AC coupling, they disappeared.

Why do you switch to AC coupling when you want to zoom in?

[mathsquid]

I wanted to zoom in on them, but when I switched to AC coupling, they disappeared.

Why do you switch to AC coupling when you want to zoom in?

The baseline of the signal was at about 1.5v, and looked jagged. I wanted to zoom in on the jaggedness by going to a very low volts per division. Doing that on DC coupling would move the signal off-screen. I thought that by switching to AC coupling, the 1.5v DC component would be ignored and I wouldn't have to move the vertical down every time I changed the volts per division .  I thought that this was the right way to do it, but I'm open to advice from those who know more than I.

[tautech]

I wanted to zoom in on them, but when I switched to AC coupling, they disappeared.

Why do you switch to AC coupling when you want to zoom in?

The baseline of the signal was at about 1.5v, and looked jagged. I wanted to zoom in on the jaggedness by going to a very low volts per division. Doing that on DC coupling would move the signal off-screen. I thought that by switching to AC coupling, the 1.5v DC component would be ignored and I wouldn't have to move the vertical down every time I changed the volts per division .  I thought that this was the right way to do it, but I'm open to advice from those who know more than I.

That's about your only option, to use the channel offset but there are limits as to how far your scope will allow offsets and at low V/div ranges offset is limited. Offset limits should be listed in your scopes manual of datasheet.
Be sure to set scope and probe to 1x. 

[David Hess]

You could place an AC coupling capacitor between your probe tip and the circuit while leaving the oscilloscope on DC coupling.  If you are using a x10 probe, then the external coupling capacitor will be working into 10 megohms making it easy to extend the low frequency response.

[newbrain]

Sorry, but I don't understand the answers that have been given.
Even looking at the front-end schematics of some scopes in the same class (well, my Rigol DS1054Z and the DS1052E), I can't see any justification for statements as:

Those spikes are too slow for AC.

and

it should definitely show up as something when AC coupled, just at a much smaller amplitude.

Agree, but why at a much smaller amplitude?

How can a large capacitor in series -that is to say "AC mode"- stop narrow pulses?

The repetition  rate is , in fact, low: 500ms as can be seen, but this kind of pulses have lots of energy content at high frequency (as I'm sure mostly everyone here is aware).
They look to be more or less 1-2 pixel wide, at 50pixel and 200ms per division, they are shorter than 8ms, possibly much shorter.

So, what should happen, as in the expectation of the OP (and mine) is that the DC component is filtered out, but pulses go through.

If the pulses are very wide, you might also see sloping roof and floors, but the high dV/dt should go through without problems.

Just to make sure I was not deluding myself, I quickly built a demo with a Nucleo board and two resistors.
(Thanks to the CubeMX taking care of PWM configuration, I only had to add a single line to start it).

The results are below, and match my expectations.
The sample rate is in my case 5MS/s, with 12MS memory.

A very narrow pulse (220ns, the lower limit for this sample rate to catch something...) go through flawlessly, DC and AC.




As does a longer pulse (1ms), AC and DC.




A 10ms pulse shows the long RC time constant of the 220nF cap in the Rigol with all the rest of the input network.


So, maybe the OP really needs a better scope?

[ataradov]

That depends on what the scope designers consider AC.

It would also be helpful to see a close up of the rise time of that spike. I'm not sure that counting pixels is the right approach here.

[RoGeorge]

Looks like a defective scope to me.

Another explanation might be that the parasitic capacitance of the oscilloscope influences the device under test. Maybe the spikes appear only when the scope is DC coupled. To test this hypothesis, put the probe of the second channel in parallel, and leave one channel DC and the other AC coupled.

Are those spikes still visible on the DC coupled channel when both probes are in parallel, but not visible on the AC coupled channel?
If so, something is wrong with the oscilloscope's AC coupling.

[newbrain]

That depends on what the scope designers consider AC.

Please elaborate. I don't really see what you mean here.
If you intend that the high-pass filter (AC mode) starts from a frequency above some tens of Hertz (a bizarre choice, IMHO), that would not change the result at all.
Moreover, the SDS6062 is specified as having -3dB at 5Hz, exactly as the DS1054Z.

I tried with a 6.8pF cap in series with the probe (TC ~= 50us) and could still see the pulses.
But, of course, they have sharp edges (see below).

It would also be helpful to see a close up of the rise time of that spike. I'm not sure that counting pixels is the right approach here.

Agreed, that would make it easier to understand what's going on.
Counting pixels gives an upper limit to the pulse width, and hence a lower limit of dV/dt for the pulse to go to 1.5V and come back.
So, I "slowed" the edges with an RC and still could see the pulses in DC and AC.
But, yes, with slow edges on a long pulse (less HF content), and 6.8pf in series, I was not able to see the pulse, swamped in the noise.

Looks like a defective scope to me.

I agree, by fault or by design.

[ataradov]

If you intend that the high-pass filter (AC mode) starts from a frequency above some tens of Hertz

This. I did not try do do even basic calculations. Intuitively, I agree, it looks like those things should have passed DC filter.

So the experiment with observing the same signal as DC on one channel and AC on the other channel, is still useful.

[DaJMasta]

Now that I'm thinking about again, I think you're right, such a pulse would have shown.  Given the really slow time base, that pulse could be the positive side of a 100Hz sine wave in a single shot.... but that still shouldn't be at the point where it starts to rolloff because of the cap, and while I still sort of see the little peaks in the signal, they also could be in the noise.


Rather than just a faulty design, I assume they'd at least get a cap that would pass 50Hz fine, it could just be a bad solder joint or part.  As far as we know, no other AC sources have been tested, so maybe the connection from the input to the front end just isn't being made in AC mode.  It sounds like replacement was probably the best course of action.

I think there's probably some case where you could see this result because of the increased capacitive loading and a very sensitive circuit under test.... but the likelihood that the thing being tested is such a special case is basically low enough to be ignored in favor of a more likely option.

[RoGeorge]

An edge will always pass through a High Pass filter. The cutting frequency doesn't matter, because the spectrum of a step signal is infinite, so as long as there is at least a sharp edge, that edge will be seen at the output of a High Pass Filter.

Even more, that spike is very thin, so it's more like a Dirac pulse, which is even more rich in high frequencies than a single step (edge) signal. There is no reason for the oscilloscope's AC filter to cut that spike.

The distance between spikes doesn't matter here. Even if the spikes are an hour apart, they will still appear when the scope is in AC mode.

[DaJMasta]

The distance between spikes does matter, because it means you know less about the actual characteristic of the spike.  While again, I agree that it's probably a pretty square pulse and it has a lot of high frequency content, but from this zoom level it's impossible to tell a 100ns pulse from a 10ms pulse, and if that content in the 10ms is a gradual rise and fall instead of a square wave, the spectral content would be much lower and close (though still above) to the point where the input capacitor could be questioned.  Because of how little detail we can see from the rise and fall of the pulse, there's no guarantee it's a sharp edge, though it looks it at this time scale.

Again, though, I don't think that's a particularly likely scenario, but with the information on that fairly low resolution display at 200ms per division... a sharp looking spike could be almost anything.

[newbrain]

After my experiments, let's make some back-of-the-envelope signal analysis.

The distance between spikes does matter, because it means you know less about the actual characteristic of the spike.

If we look at it in the frequency domain (Fourier transform) making a signal repetitive (i.e. going from an energy signal to a power one)  will "just" change the spectrum to a series of discrete pulses (delta distributions) spaced and scaled as 1/T rather than a continuous function, but the relative frequency content will not change so I do not expect any major effect on what we see.

[--8<--]
and if that content in the 10ms is a gradual rise and fall instead of a square wave, the spectral content would be much lower and close (though still above) to the point where the input capacitor could be questioned.

Yes, exactly, that was indeed the purpose of my pixel counting.
If we imagine the worst case for the dV/dt, a triangular pulse with an 8ms base (I know I should use a gaussian, but it's not much different), the Fourier transform should be A*sinc²(f/250) - can't remember the amplitude offhand, but it it's immaterial for our topic.
This means that the first zero is at 250Hz, leaving a lot of content in a band that should easily go through any self respecting AC mode (and the specs say -3dB at 5Hz).

In fact, a square pulse in the frequency domain would be a sinc(f/250); losing the ^2, the fall off is gentler and much more energy is contained in the higher portion of the spectrum, but the global picture is not radically different.

[mathsquid]

I really appreciate everyone's interest in this problem. I'd post a zoomed-in view of my original signal, but I can't seem to replicate it. I got it while probing the serial connection on a casio calculator, but I had some sort of problem with my setup. I changed the setup and now I'm reading the serial just fine, but I cannot seem to replicate what was going on before when I got this signal.

If I have time later I'll try a few more things.  Thanks again.