Even though it’s highly unlikely that this has anything to do with it, I’ll connect a dedicated 12V battery to the scope over the next few days to power it.
The ADS800 and ADS900 share the same main board, same firmware and same external power supply. So I was assuming Owon may had made some change or applied a fix between different batches that solved the observed oscillation.
I also have access to a ADS800 scope to test and I had already updated the firmware and swapped the power supply between the two models and nothing changed. So it's probably a hardware problem.
Today I had the chance to open up the ADS914A to take some pictures and sure enough, the first thing I noticed is an electrolytic capacitor bodged to the board. That is the only difference that I could spot so far*. Not sure if it's the real solution but I think it worth a try.
*edit: apart from the AFG circuit on the top left corner of the board, of course. I just noticed that the memory chips are different too, so it might not be so simple to "upgrade" the ADS800 to ADS900.
Today I had the chance to open up the ADS914A to take some pictures and sure enough, the first thing I noticed is an electrolytic capacitor bodged to the board. That is the only difference that I could spot so far*. Not sure if it's the real solution but I think it worth a try.
Thanks!
Are you referring to the 470μF/10V capacitor soldered to C1376?
Btw. I received an initial response from Conrad today:
“We have forwarded your inquiry to our quality assurance department. As soon as we receive a response, we will get back to you immediately.”
I don't have high hopes, but we'll see...
Hi,
Yes, that electrolytic capacitor is the only visible difference right now.
And it probably has something to do with the logic analyzer part, because:
I've already soldered it onto my scope, with no improvement...
Hi,
Yes, that electrolytic capacitor is the only visible difference right now.
And it probably has something to do with the logic analyzer part, because:
I've already soldered it onto my scope, with no improvement...
Thanks for the testing!
I'm not opening mine yet until I hear back from Conrad.
Just a note:
The ADS914A is the 125 MHz model, not the 70 MHz model we have.
Are there any other hardware differences to expect, or is everything handled in the firmware?
No, I think it's all handled through the firmware.
Keep in mind that the 70 MHz bands actually have a much wider bandwidth.
Thanks for the testing!
I think I'll test it again—I just noticed that one of the solder joints had come loose again.
I realized that when I took the scope apart further to access the calibration generator signal, which also has a ripple.
In the case of the generator signal, it becomes all too obvious why the square wave has a ripple....
I realized that when I took the scope apart further to access the calibration generator signal, which also has a ripple.
In the case of the generator signal, it becomes all too obvious why the square wave has a ripple....
Do you mean the void at C25?
Yes.
I've now installed a 10µF electrolytic capacitor, and the results are remarkable.
You could also connect a smaller, “quicker” capacitor in parallel to get rid of the “HF noise” or make C25 bigger, but this works just fine too.
Speaking of electrolytic capacitors:
No, the large 470µ electrolytic capacitor actually seems to be important only for the LA section; here, it doesn't change anything regarding the 200mV/div problem.
But I'll figure out what's causing it someday.
The calibration signal has a very simple structure.
A transistor (or FET) acts as a switching amplifier; the actual signal comes from the controller.
I sketched this out.
But I'll figure out what's causing it someday.
Do the calibration signal and the 200 mV/div ripple actually have the same frequency?
I'm seeing something around 40 kHz?
By the way, I'm seeing the same ripple at 10 V/div x1. Don't you?
It's really a shame, because there's a lot to like about this scope (OK, and a few things to hate, but we'll let that slide given the price).
By the way, I'm seeing the same ripple at 10 V/div x1. Don't you?
Actually, now that you mention it.
I'd never “touched” the 10V before...That was a helpful tip!
I'll show you why in a moment.
So, I’ve put together a small table—using Channel 1 as an example, but it applies to all channels.
Specifically, it shows the switching states of the two relays per channel in the front end.
What stands out in this overview is that there are only two situations where the ripple appears: at 200 mV and 10 V—and in both cases, RL30 is switched on.
That’s a lead I’ll be pursuing further.
Martin
As part of my research today, I took a look at the key components and listed them.
Interestingly, instead of an SD card like in Rigol’s 12-bit oscilloscopes, the Voltcraft/Owon has an eMMC on board.
Hmm...
Otherwise, it’s the usual stuff: a Spartan-7 FPGA and a Samsung SoC with a total of 4 cores and so on.
Those 2 ADCs are rated to 1GSa. I wonder how well at 1.25 GSa they will perform.
I guess they've tested them. Although I had no idea that ADCs could be “overclocked.”
Especially given the maximum bandwidth, that wouldn't have been necessary.
But 1.25 GSa/s sounds better than 1 GSa/s.
Especially given the maximum bandwidth, that wouldn't have been necessary.
But 1.25 GSa/s sounds better than 1 GSa/s.
From what I’ve skimmed through about this ADC, it doesn’t seem possible to increase the sampling rate at all.
Did the 1.25 GSa/s figure come from the marketing department?
Or is there actually a way to do it?
From what I’ve skimmed through about this ADC, it doesn’t seem possible to increase the sampling rate at all.
Did the 1.25 GSa/s figure come from the marketing department?
Or is there actually a way to do it?
Sampling rate is 1.25 GSa/s (800ps).
Hi Lutz,
How did you capture that?
How did you capture that?
I think I get it.
Just save the waveform on the scope to a CSV file (1k here), clean the headers and then import it into Pulseview, for example.
The "dot spacing" is then 800ps (1.25 GHz).
I don't know enough about this to say whether that allows us to draw any direct conclusions about the ADC.
P.S. In two-channel mode (on the DOV702), the "dot spacing" is 1600ps, as expected.
How did you capture that?
I think I get it.
Just save the waveform on the scope to a CSV file (1k here), clean the headers and then import it into Pulseview, for example.
The "dot spacing" is then 800ps (1.25 GHz).
I don't know enough about this to say whether that allows us to draw any direct conclusions about the ADC.
P.S. In two-channel mode (on the DOV702), the "dot spacing" is 1600ps, as expected.
Yes It is just a nice visualisation of the CSV file, the file contains time values and the differnce between each value is 800ps.
Model,DOV704
Firmware Version,V1.0.1.0.14
Horizontal Units,s
TimeBase,4.00E-8
Horizontal,0.0
Sample Interval,8.0E-10
Record Length,1000000
Channel,CH1
Probe attenuation,10.0X
Vertical Offset,-3.00div
vertical Scale,1.000V
Label,?
Frequency,80.00kHz
Period,12.50μs
PK-PK,5.250V
Average,1.501V
Vertical pos,-150
Per ADC value,100.000000
TIME(Units:s),CH1(Units:V),
-0.0004000000,4.935
-0.0003999992,4.935
-0.0003999984,4.930
For comparison the same signal captured with the DHO804:
You can also measure it directly in the scope. If you switch interpolation to "x", you can make out the points.
Unfortunately, the DOV doesn't have a dedicated dots/points mode like the Hantek DSO2000. I mean, the switch is there, but the lines are still connected horizontally.
Unfortunately, the DOV doesn't have a dedicated dots/points mode
You can switch between vectors and points in the “Display” menu—or are you referring to something else?
Unfortunately, the DOV doesn't have a dedicated dots/points mode
You can switch between vectors and points in the “Display” menu...
Yes, but that's already enabled in the screenshot above.
Still, I don't see individual dots like I do with the Hantek (in Stop mode).
With the Hantek, I can see right away that there's one dot per 1 ns.
Then that won't work with the DOV.
Rigol's DHO/MHO models don't even have the option to switch between dot and vector.
They might as well leave that out of the Voltcraft/Owon models too—one less bug...
With a Siglent, it looks as expected, though it also displays the correct number of visible points.
Then that won't work with the DOV.
Rigol's DHO/MHO models don't even have the option to switch between dot and vector.
They might as well leave that out of the Voltcraft/Owon models too—one less bug...
It does make a difference, but not in the way I expected.
It’s probably similar to the Hi-Res mode: Depending on what you’re expecting, you might not notice the effect.
In this case, though, I think the Hantek (or your Siglent) is better.
So the mode isn't dead after all...