Lecroy WavePro 950 sampling problems

[NPKNIIDAR]

Hello everyone, I want to ask experienced people for advice in solving my problem.

I have a WP950 1GHz 16Gs/s oscilloscope, the problem is that during the rise and fall of the signal, inadequate amplitude readings appear, the oscilloscope draws a straight line between them, and sharp "saws" are obtained that spoil the reading signal. That is, the oscilloscope is not showing the sample correctly.
For testing, I assembled a simple high-frequency oscillator at about ~600 MHz.

RIS helps a little, but it doesn't solve the problem, it only reduces it. It is not always possible to use RIS, more often I need a single shot signal. The "saw" is also visible in RIS.
I also want to note that the "saw" goes through if you turn on the second channel, the third, or all at once. From this I conclude that the sampling frequency is to blame. At 16Gs/s the signal is bad, and at 8Gs/s and 4Gs/s it's already more or less normal. I would like to use the full capabilities of the oscilloscope. I suggest you watch my video where I showed how it all looks. -



I also have a question about the fact that the oscilloscope sometimes thinks for a very long time. You turn it on, work, work, and sometimes it freezes, and sometimes you have to wait up to a minute. Automatic recalibration on reboot turns on by itself, although I turn it off. At the same time, the time is right. This means there are no real-time battery problems.
I'm even thinking of buying a new device, I really want a DDA-3000 or WavePro-7300A, but it's very difficult to find in Russia and neighboring countries for adequate money

Thanks for the help.
-Tim

[DaJMasta]

On those WavePro scopes you have to turn on Sin(x)/x interpolation in the math menu, but if those datapoints are real and not an artifact of the sampling system, you may have a LOT of higher frequency noise on your signal (and if you're not filtering your 600MHz once it's generated, this is probably it).

If at the same sample rate you can visualize the test output and an unconnected DC signal and not see that trace noise, it is probably real.  You could try displaying just the measurement points instead of the lines and turning on a color graded persistence mode, which should give you a heatmap of the noise on the signal.

Also worth mentioning that measuring that fast means you need appropriate probes and probing.  Ideally, an active probe, but good passive probes may do the job, and definitely on the 50 ohm input termination (I believe the 1M inputs are only really rated for like 350MHz or so), and you want the minimum loop size between the probe tip and the grounding point.  If you're using alligator leads in any part of your probing, it is not going to be a good representation of any signal at 600MHz or above.

[NPKNIIDAR]

Also worth mentioning that measuring that fast means you need appropriate probes and probing.  Ideally, an active probe, but good passive probes may do the job, and definitely on the 50 ohm input termination (I believe the 1M inputs are only really rated for like 350MHz or so), and you want the minimum loop size between the probe tip and the grounding point.  If you're using alligator leads in any part of your probing, it is not going to be a good representation of any signal at 600MHz or above.

I've experimented quite a bit with this and it's definitely a software bug or accurate calibration.
In my generator, there can be no such "noise" with a frequency of 16 GHz, it produces a pure sine of 588 MHz.
I also want to note that noise (interference) with a frequency of 16 GHz simply cannot pass through the probe, and even more so through the input path of the device.

In confirmation of my words, I connected the BNC-BNC cable and applied a calibration signal to the oscilloscope input. If the calibration signal is applied to other inputs, the situation is the same.

[Kosmic]

I tried reproducing exactly the same setup with the calibrator output on my WavePro 960. I don't see any noise unfortunately.





This is most probably a hardware problem and not a bug in software.

[Kosmic]

I also have a question about the fact that the oscilloscope sometimes thinks for a very long time. You turn it on, work, work, and sometimes it freezes, and sometimes you have to wait up to a minute. Automatic recalibration on reboot turns on by itself, although I turn it off. At the same time, the time is right. This means there are no real-time battery problems.

If time/div and num recorded samples are too high, the scope will slow down and take more time to capture everything. Normally when it's happening, I hit Stop and reconfigure the num of recorded samples to get acceptable responsiveness.

For autocal it will happen more often when the instrument is cold. After 10min it shouldn't happen that much.

[shakalnokturn]

I'm not sure disabling auto calibration is helping, the front ends drift with temperature, if they don't all drift exactly the same way and you're sampling in 1 channel mode (scope interleaves channels during the sampling) expect that kind of problem.

Give it a try enabling auto cal. or forcing a recalibration once the scope is warm. (The auto cal. that annoys everyone using LeCroys is actually a good feature.)

If you are sampling in interleaved mode and the above didn't help. Try going to the highest single shot sample rate it will give with 4 channels enabled. Feed the 4 channels the same input with matched lengths of coax and compare (superpose) the 4 waveforms.

[Kosmic]

I guess abusive autocall could be a symptom of a hardware problem (drifting). The instrument would try to compensate and activate autocall.

[shakalnokturn]

I haven't RTFM for the Wavepro 950, but I don't think they actually test for the drift but rather do the auto cal. periodically and quite regularly while the instrument is still cold.
All scopes drift while warming, most of them are adjusted to meet their specified accuracy once warm, LeCroy try to make them accurate from the power-on.

[Kosmic]

I haven't RTFM for the Wavepro 950, but I don't think they actually test for the drift but rather do the auto cal. periodically and quite regularly while the instrument is still cold.
All scopes drift while warming, most of them are adjusted to meet their specified accuracy once warm, LeCroy try to make them accurate from the power-on.

I didn't measure mine but I didn't notice any regular pattern.  So I don't think it's a simple timer. I would say it's mostly irregular and the period increase over time (internal temp ?). Also, If you change the settings this will inevitably re-trigger an autocall.

[shakalnokturn]

From the user manual:

Automatic Calibration
The WavePro DSO’s automatic calibration ensures the overall vertical accuracy. Vertical gain and offset
calibration, and horizontal (time) resolution take place each time you change the volts per division setting.
Periodic and temperature dependent auto-calibration ensures long-term stability at the current setting.

So it's a mix...

Still I don't know how much adverse effect disabling auto cal. could have on 4 channel interleaved sampling, I'd definitely try with it enabled in OP's situation.

Edit:
One minute dead time is very odd though. Have you looked at the error log in maintenance mode?

[macboy]

Autocal is even more essential when real time sampling rate is higher than 4 GSps. This is because the instrument has four separate 4 GSps sampling channels, and when it is used with only one channel enabled at 16 GSps, it uses all four channels with a phase shift on each sampling clock to create the interleaved 16 GSps signal. If all channels are not calibrated exactly alike (both phase and amplitude) then significant artifacts appear very easily. This is one reason why autocal runs frequently especially as the scope is warming up. The autocal on my wavepro 960 takes just a few seconds to complete each time it runs.

[shakalnokturn]

That's what I was suggesting.

The long freezing isn't normal though, I don't know if a defective sampling clock PLL will cause the scope to freeze, I suppose that failure to lock correctly could cause that sort of waveform.
I'm pretty sure I saw a red error message at the top of the screen in one of the videos.
Maybe OP will find his way back here and give some error log information one day.

[Converter]

I'm not sure disabling auto calibration is helping, the front ends drift with temperature, if they don't all drift exactly the same way and you're sampling in 1 channel mode (scope interleaves channels during the sampling) expect that kind of problem.

Give it a try enabling auto cal. or forcing a recalibration once the scope is warm. (The auto cal. that annoys everyone using LeCroys is actually a good feature.)

If you are sampling in interleaved mode and the above didn't help. Try going to the highest single shot sample rate it will give with 4 channels enabled. Feed the 4 channels the same input with matched lengths of coax and compare (superpose) the 4 waveforms.

OK. This is likely. Well, let's say we connected the source to all inputs through lines of the same length and saw that the delays in the channels are different. How can this be calibrated? Do you know the methodology and the name of the required parameter, which we should edit? I looked in the service menu - there are hundreds of parameters. I am lost among them. Also, I don't know if I can fix them manually.