NEW Keysight HD3

[Someone]

Perhaps instead of hints, suggestions, and talking sideways you could say what is the actual problem you face?

Keysight 1000x, 2000x, 3000x. I have no problem setting a 1kHz low pass filter and seeing a clear waveform from higher frequency PWM. It can be set for lower frequencies such as 10Hz or 1Hz if needed. In high res mode you even get 12bits of resolution and can zoom in. I would expect the HD3 is similar and better.

I wrote that I couldn't get 1khz cut-off frequency with the filter functions, neither with Siglent, nor Rigol, nor Lecroy.
Unfortunately, I don't have Keysight available, but if it works there without any problems, I would like to have that too.

You said the opposite previously:

On my former rigol mso 5000, four math channels at the same time are traceable, without any remarkable slowdown of the system.
Also the rigol got digital filters(HP, LP, BP, etc...) as math functions.

I´ve used them for displaying a 3-phase sine from a 3-phase pwm signal.

Or is the 1kHz some important limit which although some scopes do have filters you could not get to 1kHz in some combination of other settings? You have not explained the problem, Siglent 2000 HD has filters so why are they not able to produce a 1kHz (or approximate) cutoff?

[Martin72]

I couldn't get the 1kHz to work, as mentioned twice before.
1kHz would be "ideal" insofar as it represents the cut-off frequency of the output filter in the real device.
Of course, if you "get down" to 5khz, the signal will already look sinusoidal.
With the filters offered by a Siglent, you cannot reach this lower cut-off frequency with a certain time base.
I didn't get that with the Rigol either, but it's reasonably close.
I can demonstrate this next week, or egonotto is kind enough to do it beforehand:
https://www.eevblog.com/forum/testgear/magnova-oscilloscope/msg5662657/#msg5662657

[egonotto]

Hello,

Bad__Driver gave me a hint to look in the manual of the SDS3000X HD.
There you can read that you should switch on “adaptive sample rate”.
If you want to set a certain cut-off frequency fg for the low-pass filter, the sample rate must be between 4*fg and 100*fg.
At 1 kHZ cut-off frequency, the sample rate must therefore be between 4 kHz and 100 kHz.
At a cut-off frequency of 10 Hz, the sample rate must be between 40 Hz and 1 kHz.

@Someone: What are the conditions for the sample rate for Keysight 1000x, 2000x, 3000x if you want to set the cut-off frequency of a LP filter at 10 Hz?

Best regards
egonotto

[egonotto]

Hello,

I have not managed to set a low-pass filter at 30 Hz with my SDS3000X HD.
I couldn't set a 3 kHz sample rate either, I wasn't allowed more than 2 kHz.

Best regards
egonotto
PS: It worked with 1 kHz.

[nctnico]

30Hz filtering... Easy peasy on the GW Instek GDS2204E:



Yellow: filtered signal, blue original (15Hz square wave).

[egonotto]

Hello,

thank you, what is the lowest cut-off frequency for an LP filter on the GW Instek GDS2204E?

Best regards
egonotto

[nctnico]

Given the resolution is in milli-Hertz, I expect a few milli-Hertz should work. It would take long to test though...

[maxwell3e10]

The Instek picture looks a bit weird, there is no time delay. It must be an FFT-type filter or maybe a smoothing filter just on display data.

[KungFuJosh]

If it's filtering the channel directly (and I assume not a math function), why would there be a delay?

[nctnico]

The Instek picture looks a bit weird, there is no time delay. It must be an FFT-type filter or maybe a smoothing filter just on display data.

The filtering is on the actual data, not displayed data as further measurements and FFT will use the filtered data. An old trick is to do filtering in two stages. Once forward and once reverse. This cancels any phase shifts caused by the filter and thus preserves phase alignment with other signals.

[Bad_Driver]

Joking aside, try setting the lowpass to a cutoff frequency of 1kHz (filtering out a 400Hz sine wave from a 10kHz PWM).
I couldn't do it with Rigol models, nor with the digital filter package for our new Lecroys, nor with an SDS2000X HD.

I tried this with the 2000HD but it seems not be possible to get this working. Filter result is garbage.
But I will spend more efforts on this.

[KungFuJosh]

Joking aside, try setting the lowpass to a cutoff frequency of 1kHz (filtering out a 400Hz sine wave from a 10kHz PWM).
I couldn't do it with Rigol models, nor with the digital filter package for our new Lecroys, nor with an SDS2000X HD.

I tried this with the 2000HD but it seems not be possible to get this working. Filter result is garbage.
But I will spend more efforts on this.

Check the 2000HD thread, I did it using Math channel filters. I also requested a feature to be able to change the Trigger Coupling frequencies.

[ballsystemlord]

That was what I had fallen for 20 years ago.
An ancient LeCroy from 1993, PWM measured and then under math Eres +3bits was called up - and then it was a sine...
How big the disappointment was then, that it didn't work that way with newer scopes, because I just hadn't understood anything back then...

I'm not quite following. Why did you get a sine out after getting 3 more ENOBs from your Lecroy?

Thanks!

[Martin72]

The bandwidth of the scope was reduced by ERES +3 bits to such an extent that it could no longer display the 10kHz carrier.
But I didn't know that at the time.

[Someone]

I couldn't get the 1kHz to work, as mentioned twice before.
1kHz would be "ideal" insofar as it represents the cut-off frequency of the output filter in the real device.
Of course, if you "get down" to 5khz, the signal will already look sinusoidal.
With the filters offered by a Siglent, you cannot reach this lower cut-off frequency with a certain time base.
I didn't get that with the Rigol either, but it's reasonably close.

So a dozen replies and a whole page of a thread for you to explain what you mean. Just repeating "nope cant do 1kHz filter" does not explain that it cannot to 1kHz filter in a specific circumstance which you failed to describe or explain. We were all correct that those scopes can produce a 1kHz low pass filter in some way.

[KungFuJosh]

So a dozen replies and a whole page of a thread for you to explain what you mean. Just repeating "nope cant do 1kHz filter" does not explain that it cannot to 1kHz filter in a specific circumstance which you failed to describe or explain. We were all correct that those scopes can produce a 1kHz low pass filter in some way.

I understood his first post (though backwards). Not sure what your issue is with this. He was trying to isolate 1 of 2 sinewaves in a signal. He speaks German and translates to English, so sometimes there's a language barrier, but that doesn't mean you need to be rude because you didn't understand it.

He said:

Joking aside, try setting the lowpass to a cutoff frequency of 1kHz (filtering out a 400Hz sine wave from a 10kHz PWM).

What confused me was "filtering out" which I took as removing, but he meant isolating. The general concept of the issue he described remains either way. And if my dumbass understood it, and was able to find a solution for it, well... 😉

I isolated both frequencies using math filters.

Thanks,
Josh

[Someone]

I couldn't get the 1kHz to work, as mentioned twice before.
1kHz would be "ideal" insofar as it represents the cut-off frequency of the output filter in the real device.
Of course, if you "get down" to 5khz, the signal will already look sinusoidal.
With the filters offered by a Siglent, you cannot reach this lower cut-off frequency with a certain time base.
I didn't get that with the Rigol either, but it's reasonably close.

So a dozen replies and a whole page of a thread for you to explain what you mean. Just repeating "nope cant do 1kHz filter" does not explain that it cannot to 1kHz filter in a specific circumstance which you failed to describe or explain. We were all correct that those scopes can produce a 1kHz low pass filter in some way.

I understood his first post (though backwards). Not sure what your issue is with this. He was trying to isolate 1 of 2 sinewaves in a signal. He speaks German and translates to English, so sometimes there's a language barrier, but that doesn't mean you need to be rude because you didn't understand it.

He said:

Joking aside, try setting the lowpass to a cutoff frequency of 1kHz (filtering out a 400Hz sine wave from a 10kHz PWM).

Where does that mention the sample rate or memory depth? As I understand 10kHz can be filtered down to 1kHz/400Hz with those scopes when you lower the sample rate.

[KungFuJosh]

Where does that mention the sample rate or memory depth? As I understand 10kHz can be filtered down to 1kHz/400Hz with those scopes when you lower the sample rate.

He didn't have success with that, he tried previously when that feature was new to the 2000X HD units, hasn't tried with his new scope yet.

It's still a little wacky, though not too bad. Letting the scope auto set the sample rate makes it better, but for my example above it only worked if the corners didn't match. New features come with a learning curve. 😉

There's a feature request now for the trigger coupling filter to be improved, hopefully that will be possible.

Thanks,
Josh

[Someone]

What are the conditions for the sample rate for Keysight 1000x, 2000x, 3000x if you want to set the cut-off frequency of a LP filter at 10 Hz?

Running in high res acquisition mode
3000x: 10Hz LP fills just over half the screen at 50ms/div and 2MS/s (or 1MS/s or 500kS/s)
1000x: 10Hz LP fills just over half the screen at 50ms/div and 500kS/s

If you enable segmented mode then it is possible to reduce the sample rate further at the same timescale and the displayed filter trace remains the same length. So those scopes will show the filter where it makes sense and enough time is available for the filter to display sensible readings (rather than some initial condition guess)

For the HD3 what is interesting is that there is a new parameter in the signal chain to the math channels:
ADC - decimate/filter -> acquisition memory - variable decimation (?) -> measurement memory
there it should be possible to have the user trade off sample rate (measurement accuracy) with performance, but that needs some people with a HD3 to play around and test how much difference is available. Will it still do the slow filters at higher sample rates? probably.

[Someone]

So a dozen replies and a whole page of a thread for you to explain what you mean. Just repeating "nope cant do 1kHz filter" does not explain that it cannot to 1kHz filter in a specific circumstance which you failed to describe or explain. We were all correct that those scopes can produce a 1kHz low pass filter in some way.

I understood his first post (though backwards). Not sure what your issue is with this. He was trying to isolate 1 of 2 sinewaves in a signal. He speaks German and translates to English, so sometimes there's a language barrier, but that doesn't mean you need to be rude because you didn't understand it.

Where does that mention the sample rate or memory depth? As I understand 10kHz can be filtered down to 1kHz/400Hz with those scopes when you lower the sample rate.

He didn't have success with that, he tried previously when that feature was new to the 2000X HD units, hasn't tried with his new scope yet.

As I said, it took a long time for any details to be mentioned there is nothing in the opening posts of those complaints that frame/constrain the "can't work at 1kHz" claim. Those scopes can filter to 1kHz under normal/useful/practical conditions, that a user can set it up in some mode/configuration that stops it producing a 1kHz low pass filter is not surprising, that is most likely a user being an idiot and not understanding what they are doing rather than some gross failure of the market to produce ANY instrument (again Martin72 choice of claim) capable of doing such a useful task. There is still no clear explanation of what conditions Martin72 is complaining about... for a different model scope, from a different brand than this thread is about.

More derailments?

[hpw]

So ALL of you own a HD3 and contribute to this DSO 

[Bad_Driver]

Joking aside, try setting the lowpass to a cutoff frequency of 1kHz (filtering out a 400Hz sine wave from a 10kHz PWM).
I couldn't do it with Rigol models, nor with the digital filter package for our new Lecroys, nor with an SDS2000X HD.

I tried this with the 2000HD but it seems not be possible to get this working. Filter result is garbage.
But I will spend more efforts on this.

@ Martin
Maybe I‘m wrong but I tried it with a square wave 10 kHz signal, PM modulated with a 400 Hz Sine and a 1 kHz LP and this did‘ not work. Is this your use case?
KungFuJosh was successful with two sine signals.

[Bad_Driver]

So ALL of you own a HD3 and contribute to this DSO 

You are right, it‘s all off topic and we shall better start a thread on digital filters with scopes.
PS: But we have no contributions from our lucky HD3 lottery winners. 

[KungFuJosh]

More derailments?

Your post count is high enough to know how this website works. 😉

Anyway, I don't think so. I think it was a sincere question that offended you for some reason. You can claim all those scopes do the thing, but without examples of them doing the thing. There's also no example of the HD3 filtering or isolating frequencies, which was the question, and why it's not off topic.

The other models, mine included, are not awesome at doing that. Maybe they can, but not great. If the HD3 is better at it, that would make it a desirable product for that reason. I'm not sure what's hard to comprehend about that or why you think comparing this scope to another (where this scope has an opportunity to shine) is off topic. The point of these topics is literally to do that. It's not paid advertising, it's to learn more about specific tools and whether or not it's something that fits a user's needs.

[maxwell3e10]

I'll just say that someone likes to argue, without naming anyone in particular