Siglent SDS2000X Plus - Bugs / Missing Features / Feature Requests

[nctnico]

You are not understanding the use case here. My GW Instek can low-pass filter to less than 20kHz with 1Gs/s input data just fine...

Is that filtering pre or post-trigger?  IOW, can it prevent triggering on HF spikes?

It is post-trigger (post acquisition filtering; you can change the settings without needing to re-acquire on the GW Instek). Triggering typically already has various high-pass / low pass filter settings so having extra filters is not really necessary. With enough memory you can usually just take a random snapshot and have plenty of data to work with; precise triggering becomes less important if you have deep memory.

[Performa01]

You are not understanding the use case here. My GW Instek can low-pass filter to less than 20kHz with 1Gs/s input data just fine...

Oh, interesting. Are you sure that the effective sample rate used for the filter trace is actually 1 GSa/s for such low frequency limits?
Have you ever tried that with the RTM3000?

For a project I did a couple of years ago I had a lot of noise coming from a circut at around 150kHz (nothing to do about that due to the nature of the circuit) but I needed to see signals in the several kHz range that got buried in the noise (edit: well, not noise perse but non interesting content in a different frequency band). In such cases there is no other way than to use filtering and it is extremely handy if the DSO can do that digitally (adjustable).

Yes, no need to repeat the obvious over and over again. I never disputed that and already said that we might get something in the future.

In the end Eres and bandwidth limiting only get you so far because the unwanted content can alias back into your signal with lower samplerates; the minimum samplerate has to be high enough to meet nyquist for the signal content you want to get rid of otherwise it will alias. And there may also be cases where a highpass or bandpass filter come in very handy. Think about protocols that are modulated onto mains.

Huh?

Òn which scope will the bandwidht limiter affect the sample rate?
And in what aspect do you think ERES is different from a LP filter math function, so that it might affect the sample rate? Yes, ERES frequency range is closely related to sample rate - like any digital filter.
And yes, (re-iterating it once again) high- and bandpass filters can be useful as well. Does the GW have a bandpass? And what about the R&S?

On the other hand, when Siglent provides an offer in the future, it will be a complete filter-package.

[nctnico]

My GW Instek, R&S and Micsig, Yokogawa and Lecroy scopes have low-pass, band-pass and high pass filtering (although R&S uses decimated data). So it is not some kind of outlandish feature. And they all do that today (well, if I had the extended math option on the Yokogawa; but it is there), not at some undefined point in the future. On the GW Instek I can set it to 1 Hz if I want with a 1Gs/s samplerate; don't care how they do it -it works-.

Regarding Eres: you suggested to lower the samplerate in order to bring the cutoff-frequency of the filtering Eres down, but that comes at the risk of aliasing. So in the end that is a severely limited filtering function and not much of a replacement for having dedicated filter functions. That is the point I wanted to make.

[2N3055]

My GW Instek, R&S and Micsig, Yokogawa and Lecroy scopes have low-pass, band-pass and high pass filtering (although R&S uses decimated data). So it is not some kind of outlandish feature. And they all do that today (well, if I had the extended math option on the Yokogawa; but it is there), not at some undefined point in the future. On the GW Instek I can set it to 1 Hz if I want with a 1Gs/s samplerate; don't care how they do it -it works-.

Regarding Eres: you suggested to lower the samplerate in order to bring the cutoff-frequency of the filtering Eres down, but that comes at the risk of aliasing. So in the end that is a severely limited filtering function and not much of a replacement for having dedicated filter functions. That is the point I wanted to make.

We spoke about this already few time. ERES is lowpass filtering on full sample rate, NOT decimation. High frequency content gets filtered out, it cannot alias....
It is actually an digital antialias filter...

[mawyatt]

My GW Instek, R&S and Micsig, Yokogawa and Lecroy scopes have low-pass, band-pass and high pass filtering (although R&S uses decimated data). So it is not some kind of outlandish feature. And they all do that today (well, if I had the extended math option on the Yokogawa; but it is there), not at some undefined point in the future. On the GW Instek I can set it to 1 Hz if I want with a 1Gs/s samplerate; don't care how they do it -it works-.

Regarding Eres: you suggested to lower the samplerate in order to bring the cutoff-frequency of the filtering Eres down, but that comes at the risk of aliasing. So in the end that is a severely limited filtering function and not much of a replacement for having dedicated filter functions. That is the point I wanted to make.

We spoke about this already few time. ERES is lowpass filtering on full sample rate, NOT decimation. High frequency content gets filtered out, it cannot alias....
It is actually an digital antialias filter...

Think you pointed to this LeCroy document awhile back.

https://teledynelecroy.com/doc/differences-between-eres-and-hires

Best,

[nctnico]

My GW Instek, R&S and Micsig, Yokogawa and Lecroy scopes have low-pass, band-pass and high pass filtering (although R&S uses decimated data). So it is not some kind of outlandish feature. And they all do that today (well, if I had the extended math option on the Yokogawa; but it is there), not at some undefined point in the future. On the GW Instek I can set it to 1 Hz if I want with a 1Gs/s samplerate; don't care how they do it -it works-.

Regarding Eres: you suggested to lower the samplerate in order to bring the cutoff-frequency of the filtering Eres down, but that comes at the risk of aliasing. So in the end that is a severely limited filtering function and not much of a replacement for having dedicated filter functions. That is the point I wanted to make.

We spoke about this already few time. ERES is lowpass filtering on full sample rate, NOT decimation. High frequency content gets filtered out, it cannot alias....
It is actually an digital antialias filter...

Think you pointed to this LeCroy document awhile back.

https://teledynelecroy.com/doc/differences-between-eres-and-hires

I stand corrected but still Eres is a far cry from having filters which have samplerate independant adjustable cut-off frequencies and band pass / high pass filtering in addition to low pass filtering.

Edit: reading the website again, I don't see that the filter length for Eres is adjusted for a lower samplerate. And given that Eres preserves the original data (being a post processing math operation), I don't see how Eres would prevent aliasing when the samplerate is pushed down in order to make the cut-off frequency lower. Perhaps someone can eloborate on that.

[2N3055]

My GW Instek, R&S and Micsig, Yokogawa and Lecroy scopes have low-pass, band-pass and high pass filtering (although R&S uses decimated data). So it is not some kind of outlandish feature. And they all do that today (well, if I had the extended math option on the Yokogawa; but it is there), not at some undefined point in the future. On the GW Instek I can set it to 1 Hz if I want with a 1Gs/s samplerate; don't care how they do it -it works-.

Regarding Eres: you suggested to lower the samplerate in order to bring the cutoff-frequency of the filtering Eres down, but that comes at the risk of aliasing. So in the end that is a severely limited filtering function and not much of a replacement for having dedicated filter functions. That is the point I wanted to make.

We spoke about this already few time. ERES is lowpass filtering on full sample rate, NOT decimation. High frequency content gets filtered out, it cannot alias....
It is actually an digital antialias filter...

Think you pointed to this LeCroy document awhile back.

https://teledynelecroy.com/doc/differences-between-eres-and-hires

I stand corrected but still Eres is a far cry from having filters which have samplerate independant adjustable cut-off frequencies and band pass / high pass filtering in addition to low pass filtering.

As Performa said, nobody disputes that fully featured filters are very useful.

[bdunham7]

Edit: reading the website again, I don't see that the filter length for Eres is adjusted for a lower samplerate. And given that Eres preserves the original data (being a post processing math operation), I don't see how Eres would prevent aliasing when the samplerate is pushed down in order to make the cut-off frequency lower. Perhaps someone can eloborate on that.

I think it works for the same reason a sampling RF voltmeter 'works'--there can be aliasing in the acquired data, but after 3-bit ERES is applied, it all gets filtered out.  You might be thinking that perhaps a signal just over Nyquist would alias all the way down to the frequency you want to see and I can't think of a good reason offhand why that wouldn't happen.  I suppose an experiment is in order.

[bdunham7]

OK, is there some reason the MATH traces don't show up in the zoom window???     I guess I'm in the correct thread to complain about that...

Anyway, as far as I can tell using 5ms/div, 20 kpts which results in 400kSa/s and 201kHz, the ERES operation removes any aliased signal.  It is completely flat with a 201kHz signal input and shows normal filter response as you go from 1 to 5 kHz and so on.

Edit:  It seems to be the case that MATH doesn't appear in a zoom window on at least the SDS1000X-E and SDS2000X+ series.  I never noticed before, but this seems egregious on a scope with memory management that essentially requires you to use zoom for things that other scopes don't and particularly egregious on the SDS200X+ where ERES is implemented as a math channel instead of an acquisition mode.  Am I missing something?  What do other scopes do?

[gf]

If your ROI is 0..5kHz then the potential alias regions are N*fs-5...N*fs+5 kHz, so rather try to inject an unwanted signal of (say) 399 or 401 kHz, if the sample rate is 400kSa/s. In the absence of an apropriate analog anti-aliasing filter in front of the ADC, this frequecy will appear at 1kHz then, and can't be removed any more with a digital filter w/o disturbing the wanted signal in the ROI as well.

[Performa01]

OK, is there some reason the MATH traces don't show up in the zoom window???     I guess I'm in the correct thread to complain about that...

Even though we have deep measurements and math (in contrast to just using screen data), at one point we need to start decimating the original sample data in order to be able to handle them with reasonable performance. For a Keysight MSOX, the size of the decimated data is some 64 kpts, whereas for the Siglents it is at least 12.5 Mpts.

In certain situations, the record length might be longer than 12.5 Mpts and then all math and measurements are using the secondary buffer that contains the decimated data. In this case, a zoomed version of the original math function would not be quite what we want.

Solution: We have the zoom channels Z1...Z4 as additional data sources. With these, we get full resolution math and measurements, no matter how long the original record is and what zoom factor we use.

[Performa01]

My GW Instek, R&S and Micsig, Yokogawa and Lecroy scopes have low-pass, band-pass and high pass filtering (although R&S uses decimated data). So it is not some kind of outlandish feature. And they all do that today (well, if I had the extended math option on the Yokogawa; but it is there), not at some undefined point in the future. On the GW Instek I can set it to 1 Hz if I want with a 1Gs/s samplerate; don't care how they do it -it works-.

I never claimed it was an outstanding feature.

You did not answer my question about low filter frequencies on the R&S at full sample rate - or yes, maybe you did, as you admitted that it uses decimated data. And you are not afraid this might cause aliasing problems?

And as cool as it might sound "I don't care how it's done", it should be quite obvious, that 1 Hz corner frequency just isn't feasible at e.g. 1 GSa/s - and no serious instrument will have this feature - all the more so without any mention how it works, because the latter would be absolutely required in order to warn the user - for the exact reason that you brought up: sample rate reduction might cause problems with aliasing. Because the only way to implement this will of course be internal data decimation, hence reduction of the effective sample rate. All the more so on a not so high performance embedded platform.

I already hinted on the FFT, which shares the same basic problem: one might want to keep the effective sample rate high because of some unwanted signals in the HF region, but still closely inspect the audio frequency range at the same time. It just doesn't happen. At 1 GSa/s, even 1 Mpts FFT has a the frequency step of ~2 kHz, resolution bandwidth would be some 7 kHz with the FlatTop window. In order to get a more useful RBW of less than 100 Hz, we'd need a 128 Mpts FFT. and for 1 Hz frequency step (not RBW) we'd need 2 Gpts.

[nctnico]

My GW Instek, R&S and Micsig, Yokogawa and Lecroy scopes have low-pass, band-pass and high pass filtering (although R&S uses decimated data). So it is not some kind of outlandish feature. And they all do that today (well, if I had the extended math option on the Yokogawa; but it is there), not at some undefined point in the future. On the GW Instek I can set it to 1 Hz if I want with a 1Gs/s samplerate; don't care how they do it -it works-.

I never claimed it was an outstanding feature.

You did not answer my question about low filter frequencies on the R&S at full sample rate - or yes, maybe you did, as you admitted that it uses decimated data. And you are not afraid this might cause aliasing problems?

I checked my old notes: The decimation used by the R&S scopes uses some form of subsampling (likely to retain the shape/outline of the signal) which causes aliasing problems with filtering; the R&S wouldn't be my tool of choice when I need filtering (which I also wrote in the review I did on the RTM3004 a couple of years ago).

And as cool as it might sound "I don't care how it's done", it should be quite obvious, that 1 Hz corner frequency just isn't feasible at e.g. 1 GSa/s - and no serious instrument will have this feature - all the more so without any mention how it works, because the latter would be absolutely required in order to warn the user - for the exact reason that you brought up: sample rate reduction might cause problems with aliasing. Because the only way to implement this will of course be internal data decimation, hence reduction of the effective sample rate. All the more so on a not so high performance embedded platform.

The likely way it is implemented on the GW Instek is by cascading filters and decimate the signal a couple of times to get to a lowpass or bandpass filter. But that doesn't mean that there is aliasing! And the decimation filters don't have to be very tidy; an elliptic filter doesn't need much resources. Or even averaring a number of samples to form a lowpass filter could be enough. In my experience cascading filters in order to decimate the samplerate is a good way to start from a high sampling rate if the aim is to filter a relatively low frequency. Using one filter is usually problematic due to finite resolution and stability issues.

I already hinted on the FFT, which shares the same basic problem: one might want to keep the effective sample rate high because of some unwanted signals in the HF region, but still closely inspect the audio frequency range at the same time. It just doesn't happen. At 1 GSa/s, even 1 Mpts FFT has a the frequency step of ~2 kHz, resolution bandwidth would be some 7 kHz with the FlatTop window. In order to get a more useful RBW of less than 100 Hz, we'd need a 128 Mpts FFT. and for 1 Hz frequency step (not RBW) we'd need 2 Gpts.

There is a limit to everything.

[rf-loop]

Eres.
Least with these example oscilloscopes, SDS1204X-E and some unknown oscilloscope  ERES do not prevent aliasing.
Here decimated samplerate in both cases is 500kSa/s
In both cases ADC true hardware samplerate is 1GSa/s.
Both oscilloscopes input is roughly equal signal, 50501 kHz level roughly 5.6dBm (naturally same works with 501, 1001, 1501, 2001 kHz and so on. )
Both oscilloscopes set for ERES ench.bits 3.0
Both show 1kHz without even attanuation. Both 50ohm inputs 200mV/div.

Same 1kHz is of course available using what ever signal what is sampling frequency or decimated sampling frequency harmonic (as I have used 101th)  + 1kHz or -1kHz (if want 1kHz alias)
Of course also some other frequencies, just then alias is different.

This is also why fixed  ADCs can use for frequency converter. And, in digital side there is not any single method for regognize what is alias or not alias (if we do not get some more information for detect this.)  (naturally there is methods for hide aliasing on oscilloscope screen while at the same time accepting the nasty things it causes.

Why I use so high freq, this result can see alsousing just 501 or 499kHz. (for get 1kHz as IF if just think ADC is like mixer where LO is 500kHz and 501 or 499 is freq in.)

But why I use so high freq... ERES is 3.0   and ADC true HW sampling frequency is 1GHz. With 3 bit enchange -3dB is perhaps around 0.008 x fSampling. (0.016 x fNyquist) in LeCroy paper. But I do not know Siglent filtering. (looks like -3dB is bit lower)  But least it is somehere under 10MHz at this 1GSa/s.
But because decimated samplerate 500kSa/s it can easy see that example using 50.504kHz this alias drops more than -3dB (same happen if use 504kHz signal.)
If look LeCroy paper -3dB is 250kHz * 0.016  it is 4kHz but when I look it with example SDS1204X-E it is somewhere around 3.5kHz.

But I can not see  attenuation in alias if example alias is 1kHz and input freq is this bit over 50MHz.
(If I look this same bit over 50MHz signal with 1GSa/s samplerate and use same ERES 3.0 of course then this 50MHz is highly attenuated, just as expected)

Naturally this 1kHz alias do not trig so its position jumps around.

With ERES user need understand what is alias and how to avoid traps caused by it, specially when use low decimated samplerate and wide band front end.

[2N3055]

This conversation just went wrong direction.

The original statement that I responded to (as far as I understood it)  was that if you apply ERES on a signal sampled at 5 GS/s  you will get a lowpass filter effect. If you go for 3 Bit ERES you get effective lowpass filter with less than 100 MHz of bandwidth.
And then was said that if you apply a high frequency, say 200Mhz, it will alias.

While setup like that, if you apply 200 MHz signal it won't alias. It will show 200MHz signal. It will have amplitude defined by ERES filter stopband. But it won't alias.

Then people started talking about different scenario: If you setup sample rate too low for the signal (10 MS/s for 30 MHz signal), and then apply ERES there will be  aliasing...  Of course there will be aliasing.
Because you sampled with sample rate too low in a first place, got aliased samples and then applied ERES to already subsampled signal. What was ERES supposed to do: magically reverse destroyed signal?

[Martin72]

The original question was this:
https://www.eevblog.com/forum/testgear/siglent-sds2000x-plus-bugs-missing-features-feature-requests/msg3988559/#msg3988559

Fits "perfect" to this thread from the meaning.
The answers to this question are drifting more and more away from it.

[nctnico]

Because you sampled with sample rate too low in a first place, got aliased samples and then applied ERES to already subsampled signal. What was ERES supposed to do: magically reverse destroyed signal?

No, it just shows that Eres is not a replacement for having real filtering. It is simple as that.

[rf-loop]

My previous answer, if it wasn’t clear enough, was largely a caricature example. I tried to stop all possible stories that ERES was like some kind of anti-alias filter.
I think I should have said it in a simple short sentence so that it also reaches potential readers with little knowledge on the subject.

ERES (which we are talking about in this context) does not in itself remove the alias regardless of the sample rate. Filtering (for prevent aliasing) signal max f   below the aliasing frequencies should occur before the AD converter. It cannot be done properly after the AD converter for fixed frequency sampling or decimated directly from it.

ERES does not prevent aliasing. Period.
It may prevent aliasing at some known spot frequencies, but the oscilloscope must work to analyze unknown signals to resolve the signal.

It is the user's responsibility to examine only signals that do not pass significant amounts above the fNyquist frequencies at the input of the AD converter.

And as observed, this is strongly on the sidelines for one of the questions raised previously.

But any misunderstandings that ERES would act as an anti-alias filter must also be stopped so that no one is left with false beliefs about it.
However, ERES is very good when used correctly. Some of its benefits are described in LeCroy’s white paper.

[2N3055]

Because you sampled with sample rate too low in a first place, got aliased samples and then applied ERES to already subsampled signal. What was ERES supposed to do: magically reverse destroyed signal?

No, it just shows that Eres is not a replacement for having real filtering. It is simple as that.

No. It shows GIGO principle. Garbage in garbage out.
As far as aliasing goes, ERES is working better than HiRes on Keysight. HiRes on Keysight, for instance, can be made to alias fairly easy.

[2N3055]

My previous answer, if it wasn’t clear enough, was largely a caricature example. I tried to stop all possible stories that ERES was like some kind of anti-alias filter.
I think I should have said it in a simple short sentence so that it also reaches potential readers with little knowledge on the subject.

ERES (which we are talking about in this context) does not in itself remove the alias regardless of the sample rate. Filtering (for prevent aliasing) signal max f   below the aliasing frequencies should occur before the AD converter. It cannot be done properly after the AD converter for fixed frequency sampling or decimated directly from it.

ERES does not prevent aliasing. Period.
It may prevent aliasing at some known spot frequencies, but the oscilloscope must work to analyze unknown signals to resolve the signal.

It is the user's responsibility to examine only signals that do not pass significant amounts above the fNyquist frequencies at the input of the AD converter.

And as observed, this is strongly on the sidelines for one of the questions raised previously.

But any misunderstandings that ERES would act as an anti-alias filter must also be stopped so that no one is left with false beliefs about it.
However, ERES is very good when used correctly. Some of its benefits are described in LeCroy’s white paper.

ERES is not an aliasing REMOVER. That is for sure.
But it won't cause aliasing, if sampling is proper for signal in question before ERES is applied.

As Martin pointed out, this all has nothing to do with bugs or missing features. ERES is one of the things that works well.

The crux here is SDS2000X+ at the current time has no filters. Will Siglent port their filter package to SDS2000X+ and in which form remains to be seen.

[Performa01]

I checked my old notes: The decimation used by the R&S scopes uses some form of subsampling (likely to retain the shape/outline of the signal) which causes aliasing problems with filtering; the R&S wouldn't be my tool of choice when I need filtering (which I also wrote in the review I did on the RTM3004 a couple of years ago).

Isn't it funny, how a prohibitively expensive DSO from a reputable A-brand wouldn't be "the tool of choice"?

And what about other serious instruments, e.g. from Keysight or LeCroy?

Sorry, but I don't buy the "Goodwill can magically circumvent the principles of digital signal processing" narrative.

And as cool as it might sound "I don't care how it's done", it should be quite obvious, that 1 Hz corner frequency just isn't feasible at e.g. 1 GSa/s - and no serious instrument will have this feature - all the more so without any mention how it works, because the latter would be absolutely required in order to warn the user - for the exact reason that you brought up: sample rate reduction might cause problems with aliasing. Because the only way to implement this will of course be internal data decimation, hence reduction of the effective sample rate. All the more so on a not so high performance embedded platform.

The likely way it is implemented on the GW Instek is by cascading filters and decimate the signal a couple of times to get to a lowpass or bandpass filter. But that doesn't mean that there is aliasing! And the decimation filters don't have to be very tidy; an elliptic filter doesn't need much resources. Or even averaring a number of samples to form a lowpass filter could be enough. In my experience cascading filters in order to decimate the samplerate is a good way to start from a high sampling rate if the aim is to filter a relatively low frequency. Using one filter is usually problematic due to finite resolution and stability issues.

So, the truth is that you're just speculating (not the first time) and actually don't know. Therefore, my quoted statement is still valid.

Have you ever tried to find out the true frequency resolution for e.g. a 100 MHz filter (no matter if it's low-, high- or bandpass)? Do you really believe you can set the frequency in 1 Hz steps up there? Yes, that's hard to find out, but maybe the actual frequency steps are coarse enough so that it's not that difficult to do.

[nctnico]

I checked my old notes: The decimation used by the R&S scopes uses some form of subsampling (likely to retain the shape/outline of the signal) which causes aliasing problems with filtering; the R&S wouldn't be my tool of choice when I need filtering (which I also wrote in the review I did on the RTM3004 a couple of years ago).

Isn't it funny, how a prohibitively expensive DSO from a reputable A-brand wouldn't be "the tool of choice"?

And what about other serious instruments, e.g. from Keysight or LeCroy?

Simple: there ain't no such thing as the perfect scope. You may want to downplay the GW Instek but you might want to try the GDS-2000E series some time. You'll find these are very effective instruments within their capabilities, a real joy to use. Mine is currently sitting on top of a Lecroy Wavepro 7k series at my second work area. Only when things get though (like needing high time resolution or complicated measurements) the Lecroy gets powered on; otherwise it is the GW Instek.

And as cool as it might sound "I don't care how it's done", it should be quite obvious, that 1 Hz corner frequency just isn't feasible at e.g. 1 GSa/s - and no serious instrument will have this feature - all the more so without any mention how it works, because the latter would be absolutely required in order to warn the user - for the exact reason that you brought up: sample rate reduction might cause problems with aliasing. Because the only way to implement this will of course be internal data decimation, hence reduction of the effective sample rate. All the more so on a not so high performance embedded platform.

The likely way it is implemented on the GW Instek is by cascading filters and decimate the signal a couple of times to get to a lowpass or bandpass filter. But that doesn't mean that there is aliasing! And the decimation filters don't have to be very tidy; an elliptic filter doesn't need much resources. Or even averaring a number of samples to form a lowpass filter could be enough. In my experience cascading filters in order to decimate the samplerate is a good way to start from a high sampling rate if the aim is to filter a relatively low frequency. Using one filter is usually problematic due to finite resolution and stability issues.

So, the truth is that you're just speculating (not the first time) and actually don't know. Therefore, my quoted statement is still valid.

Have you ever tried to find out the true frequency resolution for e.g. a 100 MHz filter (no matter if it's low-, high- or bandpass)? Do you really believe you can set the frequency in 1 Hz steps up there? Yes, that's hard to find out, but maybe the actual frequency steps are coarse enough so that it's not that difficult to do.

I can try some time but I have no reason to doubt it works. You may think that it is impossible to do but as I wrote before: by cascasing filters, getting 1Hz resolution even when starting from 1Gs/s is perfectly doable. But this likely takes using test signals in the single digit Hz range. With a bandwidth of 100MHz a 1Hz resolution will get lost in the noise so that is not very useful to test. From using the GW Instek for real work I have found that the frequency resolution of the filters is really fine. The coefficients are calculated on-the-fly; it is not a bunch of preset filters.

[Martin72]

Digital filters:
Tomorrow I guess, I´ll get a 30-days-trial license for our waverunner scope, the digital filter package.
Will post some pics here, maybe we got a clue what will await us, when siglent offers digital filters too.

[mgams]

Hi everybody,
I would like to report a bug regarding the SDS2104X Plus with SPL2016, Firmware v1.3.9R6.

Issue: Digital Bus Hex decode not in sync with digital lines

Situation:

1) Trigger on Strobe (D7) falling edge, Timebase 1ms/div (see pic A)

2) "Zooming in" by changing Timbase to 200ns/div (see pic B), the digital lines D0-7 show 08H, but the bus Hex decode shows FFH (incorrect)
   Note: if Timebase is changed from 200ns/div to 100ns/div (or 500ns/div) the bus decode shows the correct value for a short time and then "jumps" back to the incorrect value.

3) If triggered on Strobe (D7) falling edge, 200ns/div (see pic C), the Bus Hex decode shows the correct value 08H

This behavior does not occur on my SDS1104X-E with SLA1016.

Please confirm, thanks, Matthias

[RBBVNL9]

[Apologies if the below question was already answered before; I saw a couple of posts on it but no real solutions.]

I want to load a custom waveform – preferable a measured wave - in the SDS2000X+ function generator (current FW 1.3.9R6). According to the specifications and manual, there are three ways of using custom waveforms. Two ways do not seem to work at all for me, and one way only half.

The first way is to copy a trace to the AWG. The manual is pretty silent on this (except saying it is possible) but it appears that the AWG > ARB TYPE > STORED > CHANNEL is the way to do this. When I do so, however, I invariably get a ‘File does not exist!' error. I tried different ways to see whether I can store a waveform (saved a trace as 1.CSV, saved a trace as 1.BIN) but all at no avail.

The second way is to send a wave from the Siglent EasyWave Windows software, as mentioned in the manual. The software recognizes my SDS as such, and via COMMUNICATION | SEND WAVE I can send a wave to the SDS in what is called 'store location' “ARB1” to “ARB4”. (*) When done so, I hear some relays in the SDS click, had have the feeling something is really happening. But the problem is that after that, the transferred file cannot be found anywhere in the ARB selection menu of the SDS (and cannot be found in the File Manager of the device either…) So upload from EasyWave is unsuccessful on the SDS. The same procedure, however, DOES work on my Siglent SDS2042X function generator. 

(*) another strange thing: creating a new waveform only allows the user to choose between 20Vpp and 20mVpp, I had to choose the latter as the SDS does only support up to 6Vpp for ARB (as indicated by both on the SDS itself and the EasyWave software, that gives an error that 20Vpp is not allowed for this 6Vpp device).

The third way is to load an *.CSV, a *.DAT or an *.BIN waveform from a USB flash drive (unfortunately not from the internal memory, where such wave can be saved by the SDS itself). There is no description in the manual of how that CSV (or another file) should be formatted. After some experimentation, I found that the HaversineExampleFile.csv in the ZIP file at https://siglentna.com/wp-content/uploads/2017/12/EasyWaveCSV.zip can serve as a starting point (this file is meant for the EasyWave software but apparently works on the scope too). With trial and error, I get some waveforms to upload, but it seems that the parameters at the top of the file (amplitude, frequency) are not correctly read by the SDS.

So, long story. Questions to anyone with experience:
A.   did anyone get options 1 or 2 (copy trace, import from EasyWave) to work? How?
B.   Does anyone know the actual specifications of the CSV file the SDS ARB can read properly?

Thanks, Rudi