EEVblog #1146 - New Low Cost Rigol MSO5000 Oscilloscope

[Martin72]

By the way, the rating for the MSO5000 on batronix was written by me...
Did I see it right, the upgrade from 2ch to 4ch cost about 400€ ?!
Hmmmm….the difference between the 5072 and 5074 cost 100 bucks, I´ll buy the 5072 and upgrade it later for fourtimes the price difference...sure, sure...

[Noy]

Same price difference with bandwith Updates...

[Electro Fan]

Slightly off topic but can anyone with a Siglent 1104X-E confirm that they can decode data that isn't visible on the screen? With the 1104X-E can you scroll back and forth between data on the screen and data off the screen and get accurately decoded data and hold the timing relationships?  Or better yet, can you search and navigate decoded waveforms without just scrolling side to side?   In other words, do the X-E scopes perform better than the Rigol scopes in these respects?

First Time measuring at work, spi-decoding….
BIG disappointement…



It decodes from the screen, what you see is what you get, if not all data are displayed, then you couldn´t see all the data in the event table…..

Turn one channel off, all data are lost - Lecroy WS3024 and EVEN the smaller, much cheaper siglent 1104 can do it, they decode it from, the memory….and this one do it like the 1054Z 
Hope they correct it, so much memory inside and then decoding from the screen....
And:
The event table...You have to turn it on in the menu, good.
But:
If you press the "Back" button or if you turn on another channel, it disappears....
You have to go back in the menu and turn it on again....

Rigol, please fix it - Event table on/off only when you press the specific button.
And pretty please rigol....decode the data from the memory......PLEASE.

[tautech]

Slightly off topic but can anyone with a Siglent 1104X-E confirm that they can decode data that isn't visible on the screen? With the 1104X-E can you scroll back and forth between data on the screen and data off the screen and get accurately decoded data and hold the timing relationships?  Or better yet, can you search and navigate decoded waveforms without just scrolling side to side?   In other words, do the X-E scopes perform better than the Rigol scopes in these respects?

Is 1000 screen worths enough ?
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg2096347/#msg2096347

[Fungus]

Slightly off topic but can anyone with a Siglent 1104X-E confirm that they can decode data that isn't visible on the screen? With the 1104X-E can you scroll back and forth between data on the screen and data off the screen and get accurately decoded data and hold the timing relationships?  Or better yet, can you search and navigate decoded waveforms without just scrolling side to side?   In other words, do the X-E scopes perform better than the Rigol scopes in these respects?

The Rigol only decodes what's on screen but you can zoom out a long way and it still decodes correctly in the event table (ie. you don't have to be able to see the individual pulses on screen like on a DS1054Z).

Is using the 'scopes zoom functions to frame the interesting data better than simply decoding the whole of memory then navigating by scrolling up/down a huge event table? Some might think so, especially with 400Mb memory.

[TK]

In my opinion serial decoding on oscilloscopes is good for quick verification that the signals are correct by seeing some meaningful serial protocol information, but if you need serious protocol decoding, then you switch to a logic analyzer.  Even a cheap $10 saleae clone can do better protocol decoding than the most expensive oscilloscopes with serial decoding function.

[mrpackethead]

In my opinion serial decoding on oscilloscopes is good for quick verification that the signals are correct by seeing some meaningful serial protocol information, but if you need serious protocol decoding, then you switch to a logic analyzer.  Even a cheap $10 saleae clone can do better protocol decoding than the most expensive oscilloscopes with serial decoding function.

yes, i agree, until I want to use an analog source to trigger the start of a decode.  That is sometimes useful. 

[TK]

In my opinion serial decoding on oscilloscopes is good for quick verification that the signals are correct by seeing some meaningful serial protocol information, but if you need serious protocol decoding, then you switch to a logic analyzer.  Even a cheap $10 saleae clone can do better protocol decoding than the most expensive oscilloscopes with serial decoding function.

yes, i agree, until I want to use an analog source to trigger the start of a decode.  That is sometimes useful.

Then you can get a more expensive logic analyzer for around $150 like the Zeroplus LAP-C 16032 that can be triggered by an external signal.

[2N3055]

In my opinion serial decoding on oscilloscopes is good for quick verification that the signals are correct by seeing some meaningful serial protocol information, but if you need serious protocol decoding, then you switch to a logic analyzer.  Even a cheap $10 saleae clone can do better protocol decoding than the most expensive oscilloscopes with serial decoding function.

yes, i agree, until I want to use an analog source to trigger the start of a decode.  That is sometimes useful.

Then you can get a more expensive logic analyzer for around $150 like the Zeroplus LAP-C 16032 that can be triggered by an external signal.

Not the same thing. It is called MIXED signal scope for a reason. If you want to follow many thing simultaneously, and want to hop from analog to digital domain at will you need mixed signal scope for that.
Most of the time worst part is connecting DUT to instruments...
Many time, name of the game is not decoding 1000 I2C packets across 100 msec, it is 10 packets in 10 seconds interspersed with analog input/outputs...
During one session, I will go back and forth, triggering from serial protocol (start of packet, address, error) or analog trigger from other channel.

Also, it is point of view. From point of view when you are designing something from the scratch, you can test this and that, design special test firmware to test step by step..
You can then simply design test to use what you have, in which case you try to pay less, not more.. Smart.

If you have a system that you have to redesign/analyze/reverse engineer you need as many inputs at the same time, and have that time correlated. Enters mixed signal scope with decodes..
Tool for the job.

If you have something that is pretty much designed and works well from hardware standpoint, and you are now debugging software sending this and that message, you switch to logic/protocol analyzer.
It's a much better tool for that...Like you said.  I love my LAP-C 16128 for that.
And for simple serial protocols ScanaQuad 100. You drop it into laptop case and carry it all time.
If I need long time, I use Picoscope. It has huuuge memory and segments. For long sessions..

Regards,

[2N3055]

Slightly off topic but can anyone with a Siglent 1104X-E confirm that they can decode data that isn't visible on the screen? With the 1104X-E can you scroll back and forth between data on the screen and data off the screen and get accurately decoded data and hold the timing relationships?  Or better yet, can you search and navigate decoded waveforms without just scrolling side to side?   In other words, do the X-E scopes perform better than the Rigol scopes in these respects?

Is 1000 screen worths enough ?
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg2096347/#msg2096347

Respectfully in that post it shows that Siglent is also decoding only what is on the screen.
Explanation: Siglent captured 1000 segments, but on how it is shown on those RF's pics, it is decoding every segment separately. It is not one big table with 2000 packets. It is 1000 segments with 2 packets each, and you have to go from segment to segment to decode that one.
Both Keysight 3000T and Picoscope can show all packets from all segments as one big capture. I don't know if MSO5000 or Siglent have that. Would like to know.

Later he shows long snapshot with zoom technique that shows all the packets the same time (in table) and visual decode in zoom window. This is same as Rigol.

This nonsense with phrase "decode from screen" is annoying. Only DS1000Z does that.
All other Rigols decode full buffer.

Difference is that Rigol's "full buffer size" captures only specified memory depth and no more. That is very logical to me.
Some other scopes capture memory "before and after" the screen. And they can use it. It can be useful. But not necessary. I just deliberately capture wider time segment and then move around with zoom and search.

Sad part is that Siglent 1000X-E has always running "history buffers" that are VERY useful and very understated function. It is basically always on segmented memory. That is not mentioned enough.

[Electro Fan]

2N3055, Thanks for your posts.  I think they are insightful and helpful. 

Can you please elaborate on / distinguish between these two comments? Thx

"Respectfully in that post it shows that Siglent is also decoding only what is on the screen.
Explanation: Siglent captured 1000 segments, but on how it is shown on those RF's pics, it is decoding every segment separately. It is not one big table with 2000 packets. It is 1000 segments with 2 packets each, and you have to go from segment to segment to decode that one."

"Sad part is that Siglent 1000X-E has always running "history buffers" that are VERY useful and very understated function. It is basically always on segmented memory. That is not mentioned enough."

----

- This next part is not directed to 2N3055 but to the forum members overall.

FWIW, There appears to be a lack of consensus among the user community and the scope manufacturers regarding how decoding should work in a mixed-use use case.  This might be because there are multiple use cases.  It also might be because until recently the price of a useful MSO has been beyond the price range of many users, and sometimes users don't fully understand the architecture of the various entry level and intermediate level scopes we discuss around here.

As a result there is some ambiguity in the terminology we use to describe the technology and operations of MSOs, again, partly because of technology understanding or terminology and partly because different users have different needs and expectations for their various use cases.

I'll go first at making a strawman (and get riddled with criticisms, no doubt).  I think there is the capture process (how much of what type of data/protocol needs to be captured), there is the review and navigation process (how is the captured information examined - as waveforms with decodes and errors and timings, or as data in tables with decodes and timings and other annotations), and there is the search process (how do you find specific items of interest - by data strings, wildcards, etc. and/or by waveform triggers).  Admittedly there are different philosophies on all this including those who say I just want a scope to tell me about analog waveform integrity and maybe some high level digital data decode integrity and then I'll use a Logic Analyzer, and there are those who say I want to use the scope largely in place of the LA - in part because I place a value on seeing the timing relationships between analog and digital information.  For those who prefer the use of a LA, that's cool.  But what I think what would be helpful to give feedback to potential new users/buyers (and perhaps to manufacturers too) would be to improve our ability to understand and communicate what is the preferred use case or preferred most common use cases among those who want to use a scope primarily (if not entirely) rather than a LA for mixed-use work.  Personally, I'd like a MSO that would let you capture, review and navigate, and search while retaining the fidelity of the analog and digital information including timing relationships, in a reasonably user friendly manner, subject to the amount of memory available.  Just some thoughts.

And now, back to Rigol vs. Siglent, et al.....

Personally, I'd like to understand this a little better:

"This nonsense with phrase "decode from screen" is annoying. Only DS1000Z does that.
All other Rigols decode full buffer."

Does this imply that the DS1000Z decodes different than say, a Rigol MSO2072A?  If so, in what respect?  (I haven't seen any, or at least any that matters.  On the MSO2072A you can decode what is off the screen, but only to the extent you can adjust it back onto the screen with the time base.  Unless I'm mistaken about this, after that you are out of luck.) 

Net, net regarding Rigol vs Siglent X-E models:  how, in detail, do either the DS1000Z or the MSO2072A decode differently or similarly vs. the Siglent 1104X-E?  (I think this is a question that is on the table for some new buyers trying to decide between the Rigol 1054Z and the new Siglent X-E scopes.)
 

Slightly off topic but can anyone with a Siglent 1104X-E confirm that they can decode data that isn't visible on the screen? With the 1104X-E can you scroll back and forth between data on the screen and data off the screen and get accurately decoded data and hold the timing relationships?  Or better yet, can you search and navigate decoded waveforms without just scrolling side to side?   In other words, do the X-E scopes perform better than the Rigol scopes in these respects?

Is 1000 screen worths enough ?
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg2096347/#msg2096347

Respectfully in that post it shows that Siglent is also decoding only what is on the screen.
Explanation: Siglent captured 1000 segments, but on how it is shown on those RF's pics, it is decoding every segment separately. It is not one big table with 2000 packets. It is 1000 segments with 2 packets each, and you have to go from segment to segment to decode that one.
Both Keysight 3000T and Picoscope can show all packets from all segments as one big capture. I don't know if MSO5000 or Siglent have that. Would like to know.

Later he shows long snapshot with zoom technique that shows all the packets the same time (in table) and visual decode in zoom window. This is same as Rigol.

This nonsense with phrase "decode from screen" is annoying. Only DS1000Z does that.
All other Rigols decode full buffer.

Difference is that Rigol's "full buffer size" captures only specified memory depth and no more. That is very logical to me.
Some other scopes capture memory "before and after" the screen. And they can use it. It can be useful. But not necessary. I just deliberately capture wider time segment and then move around with zoom and search.

Sad part is that Siglent 1000X-E has always running "history buffers" that are VERY useful and very understated function. It is basically always on segmented memory. That is not mentioned enough.

[nctnico]

Explanation: Siglent captured 1000 segments, but on how it is shown on those RF's pics, it is decoding every segment separately. It is not one big table with 2000 packets. It is 1000 segments with 2 packets each, and you have to go from segment to segment to decode that one.
Both Keysight 3000T and Picoscope can show all packets from all segments as one big capture. I don't know if MSO5000 or Siglent have that. Would like to know.

AFAIK Besides Keysight and (according to you Picoscope) I don't know any oscilloscope manufacturer which shows the decoded data from all segments in one bus table. To me this is an obvious thing to do but the only oscilloscope I had in my hands which did that was an (older) Agilent DSO7104A. OTOH the Keysight units don't have much memory compared to the competition so you are much more likely to use segments on a Keysight scope where you might be able to get away with a single acquisition on a scope with more memory.

BTW there are quite a few scopes which only do 'on screen decoding'. AFAIK most (if not all) Siglent scopes and some tests Martin72 did with his Rigol MSO5000 seem to indicate that the MSO5000 also decodes only what is on screen. One of the 'problems' associated with decoding is that it needs a parallel process and memory which runs in parallel with acquisition OR it is a post processing operation. The first needs extra hardware, the latter needs a lot of processing power especially when the oscilloscope has a lot of memory (like the MSO5000). Given the high samplerate of the MSO5000 it may not even be able to do decoding in the hardware and thus making decoding only possible as a post-processing step. If the latter is true then Rigol made a serious mistake in the hardware design.

[2N3055]

Gladly. let's go one by one.

2N3055, Thanks for your posts.  I think they are insightful and helpful. 

Can you please elaborate on / distinguish between these two comments? Thx

"Respectfully in that post it shows that Siglent is also decoding only what is on the screen.
Explanation: Siglent captured 1000 segments, but on how it is shown on those RF's pics, it is decoding every segment separately. It is not one big table with 2000 packets. It is 1000 segments with 2 packets each, and you have to go from segment to segment to decode that one."

Older Rigols (2000, 4000 series) will not decode from screen pixels. 1000Z  does that, it sucks. If you have on screen 10M points, they will decode from 10 Mpoints with full resolution. To the point that you can have packets that are visually on screen all smeared together in one big block, but in a decode table you will see them all nicely decoded.
Hence usual advice to capture big block of data and then zoom in.

Both 1000E-X  and older Rigols (2000, 4000 series) can also capture in segments. You capture 1000 separate shorter buffers, and when you go through them it will decode them.
If somebody can check do the new ones (5000 and 7000) also can decode from segments. It is called recording in Rigol parlance.

But both Rigol and Siglent will decode and show only current segment buffer (basically what's on screen, in terms time interval.) At least that is how it was shown on RF's images for Siglent.
That means that if you use segments, in order to decode third burst of data (3 history segment) you go to 3rd segment manually, and it will decode it for you.

Keysight (and Picoscope) decodes in a table, where you can choose to see only current buffer (like Rigols and it seems Siglent 1000E-X), but you can tell it to scan through all buffers, and decode them to single unified decode table. So you can see decode and move around seamlessly. If 1000E-X cannot do that, that would be GREAT upgrade.

And that connects sort of with this:

"Sad part is that Siglent 1000X-E has always running "history buffers" that are VERY useful and very understated function. It is basically always on segmented memory. That is not mentioned enough."
[/i]

Picoscope has same always running segmented memory. Meaning that at any time, you can stop and go several hundred trigger events back.  R&S 2000/3000/4000 have something similar.
Picoscope also have fast segmented mode (also Siglent), that doesn't update screen to minimize blind time. LeCroy has similar thing.
On all other scopes (Keysight Rigol) you have to manually enter segmented mode, an it is only fast mode.
1000X-E has it running all the time with a bit larger blind time but very useful.

As for as decoding what is not on the screen, that is a funny topic.  In my mind (analog scope) you have ONLY what you have on screen. If I want to see more I expand timebase.  It seems some scopes capture samples you didn't tell the to capture, so you have pre-screen and post-screen data. And some people got used to it using it that way. I actually prefer that scope takes exactly amount of data I told it to take. And if I need more I will set scope to capture longer set. Long capture +zoom is exactly for that. In which decode table will be for whole capture, and zoomed portion will have visual decode under it. It will not break up because it knows there is more data. Rigol subscribes to this PHILOSOPHY of thinking.

Some other manufacturers are doing it differently. If you are used to this, it feels more right to you. That is fine with me. I have many different tools and they all have their own idiosyncrasies. I learn how they work and adapt to them. Such is life.

For all these reasons I got Picoscope for decoding. Despite their own problems, their way was most logical to me, and let's me use it without problems.

Regards,

[2N3055]

Explanation: Siglent captured 1000 segments, but on how it is shown on those RF's pics, it is decoding every segment separately. It is not one big table with 2000 packets. It is 1000 segments with 2 packets each, and you have to go from segment to segment to decode that one.
Both Keysight 3000T and Picoscope can show all packets from all segments as one big capture. I don't know if MSO5000 or Siglent have that. Would like to know.

AFAIK Besides Keysight and (according to you Picoscope) I don't know any oscilloscope manufacturer which shows the decoded data from all segments in one bus table. To me this is an obvious thing to do but the only oscilloscope I had in my hands which did that was an (older) Agilent DSO7104A. OTOH the Keysight units don't have much memory compared to the competition so you are much more likely to use segments on a Keysight scope where you might be able to get away with a single acquisition on a scope with more memory.

BTW there are quite a few scopes which only do 'on screen decoding'. AFAIK most (if not all) Siglent scopes and some tests Martin72 did with his Rigol MSO5000 seem to indicate that the MSO5000 also decodes only what is on screen. One of the 'problems' associated with decoding is that it needs a parallel process and memory which runs in parallel with acquisition OR it is a post processing operation. The first needs extra hardware, the latter needs a lot of processing power especially when the oscilloscope has a lot of memory (like the MSO5000). Given the high samplerate of the MSO5000 it may not even be able to do decoding in the hardware and thus making decoding only possible as a post-processing step. If the latter is true then Rigol made a serious mistake in the hardware design.

Hi Nico!

One of the reasons I bought Pico was the fact that it can decode current buffer or all buffers in one big table. It works well. It is also obvious to me and was wondering why nobody does it as standard.

New Rigol series decodes in FPGA.  And as I said, it decodes "screen length of data". Not from screen pixels. So you just grab really long capture. It has 100s of MS of memory. It will capture it with high sample rate and will decode thousands of packets. Then you search and zoom. In that regard it is actually similar to Tektronix mso3000/4000 series without segments. With 10x more memory.
I wouldn't say it is useless. Just not as good as it could be.
Regards,

[Martin72]

BTW there are quite a few scopes which only do 'on screen decoding'. AFAIK most (if not all) Siglent scopes and some tests Martin72 did with his Rigol MSO5000 seem to indicate that the MSO5000 also decodes only what is on screen.

On the siglent 1000X-E series, you could turn the channels "off" and it still decodes( showing in the event table).

The 5000 rigol decode what´s on the screen.
But there is a difference to e.g. the rigol ds1000.
Where the ds1000 sucks with decoding when you set the timebase (in my case) from 20 to 50µs, the 5000 decodes further untill 20ms.

[tautech]

Yeah well if you know your tools, you'll know how to get the best from them.
How many packets do you want to place on the display, your choice, a few or many 100's ?
A couple of simple examples from another thread:





Both screenshots are with the X-E in RUN mode using just edge triggering and the highlighted line in the Decode table shows the packet immediately after the trigger, so whatever # it is in the table, it's halfway through the Decode list and time stamped.

Unlike the previously linked 'decoding in Sequence mode' example by rf-loop the above screenshots are using the split screen in Zoom mode are typical of how to get the best from Siglent DSO's when decoding.....all Siglent models that offer decode can work this way.

If anyone wants to see more jump into this thread and have a hunt or ask for further examples:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/

[nctnico]

The downside of the 'Siglent way' is that you are cluttering the (already not so big screen) screen with a mandatory zoom window. Otherwise AFAIK it will only decode what is on screen.

@2N3055: I wouldn't want to start a semantic discussion by 'what is on screen' means  Let's just define it by the trace shown on the screen. There is a way too large variety of how oscilloscopes do decoding to make a good definition. Some scopes even decode properly when the samplerate is way too low so even though the trace looks like garbage the decoded data is still good.

Either way, one of the downsides of oscilloscopes which decode only what is on screen is that they miss the start of the message and can no longer decode when looking into a detail of the data. Also in many cases you'll want to look at a specific message. If the amount of data being decoded depends on the amount of trace displayed and you need to look at the 20th message at 10ms/div then which message do you need with the time base set at 1ms/div? With the bus decoding table changing you'd need to write down the timestamp.

[Fungus]

Given the high samplerate of the MSO5000 it may not even be able to do decoding in the hardware and thus making decoding only possible as a post-processing step. If the latter is true then Rigol made a serious mistake in the hardware design.

What should they have done to fix it? Lower the sample rate to a point where it's possible to do it?

[Fungus]

Yeah well if you know your tools, you'll know how to get the best from them.

That's true.

In the case of this new Rigol the information you need was posted a couple of pages back:

https://www.eevblog.com/forum/blog/new-rigol-scope/msg2034745/#msg2034745

Apparently you can zoom out until 1000+ SPI clock pulses are contained in a single pixel "on screen" and it still decodes correctly.

This, combined with the 400M memory of the Rigol makes it a clear winner vs. the Siglent IMHO.

(plus people are busy making little PCBs to break out the digital part of "MSO" cheaply)

https://www.eevblog.com/forum/testgear/rpl1116-active-logic-probe-pod-for-1000z-series-teardown/

[seronday]

   I recently had the use of a Rigol MSO5074 for a period of time and have been able to measure and record the frequency response with both the original bandwidth and the full bandwidth.
The signal source was connected via a "T" piece with a 50ohm termination on to the Ch1 input connector of the MSO.

Several comments have been made about the lack of brightness on the MSO5000 series display .
To my eyes the display appears to be similar to other Rigol DSO's that I have used in recent times ( DS2000 and DS4000 series ), and is certainly not dull.

[Martin72]

Several comments have been made about the lack of brightness on the MSO5000 series display .

It´s not the brightest on the market, for sure  - But it didn´t disturb me as much as others do, who hadn´t a 5000 rigol….

[TK]

Keysight announced the 1000X series 4-channel scopes and they start at USD 998 in the US for the base model, $1204 with AWG...

https://www.eevblog.com/forum/testgear/new-keysight-4-channel-scope-(1200-x-series)/msg2120590/#msg2120590

Would you still buy the Rigol MSO5074 or go with the new Keysight?

[Vtech]

Would you still buy the Rigol MSO5074 or go with the new Keysight?

Keysight 1000X: 2Gsps, 1M memory, 7 inch 800x480 TFT no touch, no logic analyzer option.
Closer to Rigol DS1000z than to MSO5000 series. I would definitely go with Rigol's MSO5000.

[supercilious]

Keysight's lower-end scopes are too low memory to be even considered for my purposes, 1M (or even 10M) is just not acceptable in 2019. Shame as they are nice scopes otherwise. No other UI feels as fast as the Keysight instruments in my experience.

[Fungus]

Would you still buy the Rigol MSO5074 or go with the new Keysight?

Rigol is hackable so you get all options/upgrades for $999