MSO1074Z (DS1074Z Plus) Only decoding the data on the screen?

[DMKitsch]

Hey,

I noticed somebody posting about their DS series scope being able to decode everything that is in memory, but I'm having issues with accomplishing the same goal on my scope. Running an MSO1074Z, when I try to decode a single shot capture I'm only getting the data on screen decoded and this includes in the event table. Looking through the manual, I can't seem to find anything about changing this setting or any warnings that is the default behaviour.

Being a scope with MSO capabilities this seems like a hugely limiting factor and so I'd appreciate any advice. I have recorded a quick video to show what I'm getting.



Thanks for all the help guys,
Dave

[Fungus]

That's the way these Rigols work. You can use the 'scope to check data integrity, make sure your bus/baud rates are OK, but not much more. Even if it did a bit more, it still wouldn't be much use for long streams of data with that user interface.

You can get USB logic analyzers for $6 on eBay. With one of those you can fill up all your PC memory and decode to your heart's content.

[RoGeorge]

AFAIK the decoding is made with what's on the screen only, and the rest of the signals are disregarded, even they were sampled and present in the memory.  For example if there is a serial RS232 byte that is entering in  the screen starting from somewhere in the middle of an entire byte, it will be decoded wrong.

That is why, unless there is a need to look at mixed analog and digital signals in the same time, then any cheap logic analyzer would do a better job at decoding digital only signals.

[Karel]

You can download the complete waveform buffer from the scope to your pc and do the decoding there
but the process of downloading the data from the scope is very slow (LAN or USB).
For example, DSRemote can download and decode the whole waveformbuffer of 24Mpts.

https://www.teuniz.net/DSRemote/

Best option is to buy a separate logic analyzer.

[nctnico]

Best option is to buy a separate logic analyzer.

No, buy an oscilloscope which decodes full memory!

[Gandalf_Sr]

Didn't we just have a many-page thread on this very topic?

I recall that, setting the Rigol scope's memory depth to maximum, the capture mode to 'Normal', and then doing a single shot capture filled the available memory with data that was decoded as you scrolled left and right.

The 'issue' is that, by default, the memory buffer is auto-sized to only capture what's on the screen.

The OP should try playing with the settings I just mentioned.  I believe it's a feature, not a bug.

[Karel]

Didn't we just have a many-page thread on this very topic?

I recall that, setting the Rigol scope's memory depth to maximum, the capture mode to 'Normal', and then doing a single shot capture filled the available memory with data that was decoded as you scrolled left and right.

The 'issue' is that, by default, the memory buffer is auto-sized to only capture what's on the screen.

The OP should try playing with the settings I just mentioned.  I believe it's a feature, not a bug.

If I remember well, the issue is that when you do as you described, you still have the problem that if the visible trace
starts halfway a byte/character in the SPI/RS-232 stream, you'll see garbage in the output of the decoder.

[Gandalf_Sr]

If I remember well, the issue is that when you do as you described, you still have the problem that if the visible trace
starts halfway a byte/character in the SPI/RS-232 stream, you'll see garbage in the output of the decoder.

Yes but, in any data stream, if you start halfway through a data element, you'll get garbage initially but, at some point, the decoder will latch onto good data. If your trigger is the recognition of specific data  e.g. an I2C address, then the filling of memory is apparently some data before and after the trigger point. In the case of a fully maxed out Rigol MSO5074, that is up to 200 Mega points of data!  The point being, what do you care if the early decodes are corrupted? I have tested my MSO5074 and, by default, it only captures what's on screen but setting the memory depth to a large number produces the behavior I have described in this thread.

More practically though, scopes are not the ideal platforms to capture long sequences of data; it's much better to capture data via a PC using something like a Salea Logic adapter.

[Gandalf_Sr]

I found the post I was referring to on this subject, it's here.

It was carved out of another thread by Dave Jones because (I think) it was a) off topic but b) very interesting on its own right; that's why it appears to start in the middle of something.

[Karel]

If I remember well, the issue is that when you do as you described, you still have the problem that if the visible trace
starts halfway a byte/character in the SPI/RS-232 stream, you'll see garbage in the output of the decoder.

Yes but, in any data stream, if you start halfway through a data element, you'll get garbage initially but, at some point, the decoder will latch onto good data.

It's useless imho. Most of the time it's all garbage except when the visible trace starts exactly in between. At least this is the case with the rigol.

[Gandalf_Sr]

It's useless imho. Most of the time it's all garbage except when the visible trace starts exactly in between. At least this is the case with the rigol.

Hmmm, I'd be interested in seeing your claim backed up with some data or a screen capture.

I won't be home until Sunday so I can't do any tests until then.

[Karel]

For a test, generate a stream of consecutive bytes.
Because there are no gaps between the bytes, the scope can't use a timeout to re-synchronize.

[Fungus]

Didn't we just have a many-page thread on this very topic?

Yes, there have been several.

I recall that, setting the Rigol scope's memory depth to maximum, the capture mode to 'Normal', and then doing a single shot capture filled the available memory with data that was decoded as you scrolled left and right.

You can fill the memory and zoom/pan but it only decodes what's visible on screen at any given moment.

It's easy to get bad decodes if the left edge of the screen doesn't align nicely with the data.

[stafil]

Best option is to buy a separate logic analyzer.

No, buy an oscilloscope which decodes full memory!

Does the Instek 2000 decode the full memory? (Genuinely asking)

[nctnico]

Best option is to buy a separate logic analyzer.

No, buy an oscilloscope which decodes full memory!

Does the Instek 2000 decode the full memory? (Genuinely asking)

Yes. One of the reasons I kept it after buying a Siglent which didn't do full memory decoding.

[tautech]

Best option is to buy a separate logic analyzer.

No, buy an oscilloscope which decodes full memory!

Does the Instek 2000 decode the full memory? (Genuinely asking)

Yes. One of the reasons I kept it after buying a Siglent which didn't do full memory decoding.

Rubbish !
Decoding is only limited by the # of total packets, not the amount of memory.

[nctnico]

If it decodes only what is on screen then what is the use of having so much memory? It doesn't make sense at all to decode only what is on screen. But let's not start that discussion again. The question from the OP makes it clear that the screen centric way some oscilloscopes (like the ones from Siglent) work is not a good fit for doing embedded software/hardware debugging. There are oscilloscopes available which do decode full memory. On the particular Rigol it is likely a limitation of the speed of the platform.

[Fungus]

If it decodes only what is on screen then what is the use of having so much memory?

For doing analog stuff.

It doesn't make sense at all to decode only what is on screen.

I think it's a hardware limitation. I believe the main CPU doesn't have direct access to the sample memory, only to the downsampled version of it that's produced by the FPGA for display purposes. That's why all the measurements, the decoding, the FFT, etc. only work on screen data.

Yes, there was a firmware update that made the FFT use a bit more memory, but not a lot more.

There's probably a way to process the sample memory in little batches but I'm guessing it would be slow and everybody would whine about how slow the decoding is. They can't win.

[Fungus]

There are oscilloscopes available which do decode full memory.

But they cost more.

[nctnico]

There are oscilloscopes available which do decode full memory.

But they cost more.

Not necessarily.
 

[Martin72]

I noticed somebody posting about their DS series scope being able to decode everything that is in memory, but I'm having issues with accomplishing the same goal on my scope. Running an MSO1074Z, when I try to decode a single shot capture I'm only getting the data on screen decoded and this includes in the event table. Looking through the manual, I can't seem to find anything about changing this setting or any warnings that is the default behaviour.

Yepp, this scope can only doing decode from whats´s on the screen.
I´ve done a comparison between this ( DS1000Z, it´s the same but without MSO) and the rigol mso5000 series.
The rigol 5000 does decoding from the memory, will look for the link.