VOLTCRAFT DOV series

[Martin72]

Hi,
What kind of signal are you feeding into the channel—a 10 MHz sine wave with a 10 mVpp amplitude?(int. AWG)

[Schenkenstein]

Hi,
What kind of signal are you feeding into the channel—a 10 MHz sine wave with a 10 mVpp amplitude?(int. AWG)

In the last picture, yes. The specific signal itself does not matter that much - I wanted to touch the horizontal line with a slow slope to not miss a bit (thus the sine on the top) but many times to cause many glitches (thus high frequency). "Unfortunately", that only gave me two, sometimes three glitches per trace. In the first three pictures, there is no signal applied: it's just the open BNC. I forgot to set the probe config to 1:1 though - sensitivity is 20mV not 200mV/div. It appears that it is only one specific value of the 2^(12) possible bits that is causing the glitch. Could be more but I have not found any yet. I'm going to investigate further tomorrow.

[Lutz21]

In the last picture, yes. The specific signal itself does not matter that much - I wanted to touch the horizontal line with a slow slope to not miss a bit (thus the sine on the top) but many times to cause many glitches (thus high frequency). "Unfortunately", that only gave me two, sometimes three glitches per trace. In the first three pictures, there is no signal applied: it's just the open BNC. I forgot to set the probe config to 1:1 though - sensitivity is 20mV not 200mV/div. It appears that it is only one specific value of the 2^(12) possible bits that is causing the glitch. Could be more but I have not found any yet. I'm going to investigate further tomorrow.

I can't reproduce the glitches on my DOV704 with open input.

[Aldo22]

@Schenkenstein
Which model do you have?
With 100M memory, it must be more than the 70/100 MHz version, right?

[Andi-61]

Hi, the FFT performance on my DOV 1002 is not very stable. I’ve attached two screenshots: one captured with a 1K memory depth and another with a 10K memory depth.

[Aldo22]

Hi, the FFT performance on my DOV 1002 is not very stable. I’ve attached two screenshots: one captured with a 1K memory depth and another with a 10K memory depth.

Thanks for the pictures! 
That doesn't look much better than mine.
I'm still wondering if it's worth sending the device in under warranty, going through all the hassle, and waiting—only to get it back with the same problem.

For me, it works OK at 1k, but not at 10k—except in Average mode. Anything above that is just flickering.

[Martin72]

1 kpt is also a bit low; you should adjust the bandwidth (sample rate) to match the signal—a 1 kHz signal followed by an FFT in the MHz range doesn't make much sense.
And you should pay attention to the window coefficients (Hann, Rectangle, etc.).
The Autoset function works well on the Owon/Voltcraft, as I've noticed once again.
(At 1k and 10k, only “half” of the FFT is displayed on the screen, which makes sense at just 25MSa/s, since the possible bandwidth then extends to about 12.5 MHz)

[Aldo22]

I’d be interested to know if the limited FFT issue is exclusive to 2-channel devices.
Andi-61 has a 2-channel scope, just like me.
His is a 100 MHz model, but I think there are hardly any differences in the hardware, if any at all.
So far, it looks a bit like my suspicion is correct, but I can’t say for sure.

If it’s confirmed, sending it in under warranty won’t help any more than it did with the 200 mV ripple problem.
It just costs money, takes time, and causes a hassle.
I think Conrad just has a fixed stock of these devices that they’re selling off, and that’s it.
There probably won’t be an improved version.

[Schenkenstein]

In the last picture, yes. The specific signal itself does not matter that much - I wanted to touch the horizontal line with a slow slope to not miss a bit (thus the sine on the top) but many times to cause many glitches (thus high frequency). "Unfortunately", that only gave me two, sometimes three glitches per trace. In the first three pictures, there is no signal applied: it's just the open BNC. I forgot to set the probe config to 1:1 though - sensitivity is 20mV not 200mV/div. It appears that it is only one specific value of the 2^(12) possible bits that is causing the glitch. Could be more but I have not found any yet. I'm going to investigate further tomorrow.

I can't reproduce the glitches on my DOV704 with open input.

Thanks for the insights! Yeah, it would have surprised me if I were the first one to notice such issue. However, it looks like some software issue rather than a hardware issue...

@Schenkenstein
Which model do you have?
With 100M memory, it must be more than the 70/100 MHz version, right?

Yes, I got a used DOV2504F LA for cheap.

I'm currently looking into histograms of samples (with Python) to see if some special value causes the glitches but nothing obvious so far even though I did not get the expected gaussian curve as well. Something I have to wrap my head around when I have some peace and quiet.

[Schenkenstein]

Hi, the FFT performance on my DOV 1002 is not very stable. I’ve attached two screenshots: one captured with a 1K memory depth and another with a 10K memory depth.

Hi, the FFT performance on my DOV 1002 is not very stable. I’ve attached two screenshots: one captured with a 1K memory depth and another with a 10K memory depth.

Thanks for the pictures! 
That doesn't look much better than mine.
I'm still wondering if it's worth sending the device in under warranty, going through all the hassle, and waiting—only to get it back with the same problem.

For me, it works OK at 1k, but not at 10k—except in Average mode. Anything above that is just flickering.

I tried it on my scope and the FFT with the same setup looks all good. Stable measurements and as expected. I used an SDG2042X as a source but the internal AWG worked fine as well.

[TomC]

This week I've contacted Owon that confirmed the small ripple in the 200mV range is a hardware issue and that they already fixed that in the later revisions.

Well, you've already gotten further than I have—I haven't heard back yet.
It was kind of obvious that Owon had “quietly” fixed the issue and wouldn't be recalling the versions that had already been shipped.
The additional capacitor at the OP's input and the removal of the circuitry around the “ST3SK” chip are likely the measures taken to reduce ripple.
But it would be better to know for sure rather than just assume that's the case.
It would be great if I could get a close-up photo of the area with the additional capacitor near the IC, close to the relay...@TomC

Some better photos of the areas that are different. Unfortunately, C2081 has no markings, I'll try to get some in circuit readings.

[TomC]

Predominant resistance can't get any meaningful capacitance reading. I don't expect this to be very helpful, but here it is just in case.

[Martin72]

Thank you very much, TomC!
I'll take a closer look at the pictures later and compare them.

[Aldo22]

I have a question.
Either I'm doing something wrong, expecting the wrong result, or this is a bug.

I want the cursor to automatically follow the Y value of a Math calculation.
I expect the cursor, in Auto mode, to measure the Math Y value at the intersection with the X cursor. Here, it basically doesn't matter where X is set (DC).
When I measure manually, I get the correct result, 88A (Screenshot 1).
When I measure in auto mode, the Y cursor jumps to a location that has nothing to do with the Math Y value (178A, Screenshot 2).
Is this happening to you too? Or what am I doing wrong?

Thanks.

[Martin72]

Hi,

As always, please specify the conditions—that is, what are you feeding into Channel 1, and which mathematical function are you using?

[Aldo22]

Hi,

As always, please specify the conditions—that is, what are you feeding into Channel 1, and which mathematical function are you using?

Hi,
As you can see, the image shows approx. 23 mV DC, which is labeled as mA.
The function is simple: 2/Ch1
Maybe later I'll use Ch2/Ch1, but I didn't want to complicate things right now.
Thanks.

[Martin72]

Hi,

No problems with Auto Mode...

[Martin72]

Some better photos of the areas that are different.

But they certainly didn't hold back over there when they manually “optimized” the front end...

Unfortunately, C2081 has no markings, I'll try to get some in circuit readings.

The only option would actually be to desolder it and then measure it, but I wouldn't really want to do that either.

[KungFuJosh]

They stacked caps?

It might be worth measuring it out of circuit just to do a better job replacing it with a better single cap.

[Martin72]

That's not all that uncommon when the optimal value falls between two standard values.
With Testec differential probes, you might come across something like that every now and then—but in those cases, it's soldered neatly and with care to avoid damaging the other components.

[KungFuJosh]

That's not all that uncommon when the optimal value falls between two standard values.
With Testec differential probes, you might come across something like that every now and then—but in those cases, it's soldered neatly and with care to avoid damaging the other components.

I'm aware, but it seems to be missing that element.

I'm also going to go ahead and guess it was more of a quick fix rather than a precise value issue.

[Martin72]

I'm aware, but it seems to be missing that element.

That was also meant to be a bit sarcastic toward the manufacturer—basically, the resistor needs to be replaced; it doesn't look too good anymore...

[Martin72]

In the early evening I looked up the individual window coefficients for the FFT function.
There are six variants in total, including some that are (to me) exotic, such as “Bartlett” and “Kaiser.”
It would be nice to see the RBW and the current bases displayed in the FFT—no, actually, that's a must.

[TomC]

Some better photos of the areas that are different.

But they certainly didn't hold back over there when they manually “optimized” the front end...

Unfortunately, C2081 has no markings, I'll try to get some in circuit readings.

The only option would actually be to desolder it and then measure it, but I wouldn't really want to do that either.

The component labels got me confused, not the first time! After looking more closely I think the capacitor I measured in-ckt is C2076, next to the IC and missing on the older PCB. I didn't look at the capacitor stack at all; I think that should be C2120. The capacitor I mentioned C2081, now I think is one of the missing components below the IC.

[Martin72]

While I’m still hoping that Microcheap gets a response from Owon—which will either confirm or refute my assumptions—I decided to try something else with the scope and actually took some measurements again...
Waveform update rate...
The result is fine—no complaints there.
I might run the same measurements again with multiple channels active, but basically this is enough, since it already shows the maximum achievable rate.
Martin