Siglent CML+ Firmware ?

[StillTrying]

What firmware should be on the CML+?  I don't seem to be able to find any new.

Currently 6.101.12 or something,
I think the "12" is possibly: 'The target number of freeze ups per hour'.

[tautech]

What firmware should be on the CML+?  I don't seem to be able to find any new.

Currently 6.101.12 or something, I think the "12" is possibly: 'The target number of freeze ups per hour'.

There is a later version, I am trying to get Siglent to post it on their websites, but.....what we are a little concerned over is it might be used in error and installed in the non-plus models. It's quite large...7Mb IIRC.
Be patient or PM me with your email and I'll shoot it over to you.

[StillTrying]

I moved it and left it just displaying it's own 1khz for over a hour, seemed to keep going until I tried turning on Averaging, turned 16 down to 4 and then it froze, and had to use Math on power up, so I'll stay away from there!

I like to use dot mode and 1 sec persistence, in RT or ETS because I think it helps to make dso waveforms look more like a proper  scope(even if I'm the only one who does), so I notice the strange 'features' of wrong colour for ch2's persistence dots, and no persistence at all on 100nS/div - just that speed!, although there's still enough thin dots anyway.

I don't update anything unless I have to, is there a list of the changes?

[tautech]

Sorry I haven't seen a change log but I'd hope there's one when its posted on Siglent sites.
I was sent it just after the plus models were released and I install it in each plus unit I sell.
Just yell and I'll get it to you.

[tautech]

Finally got some time to follow this up for you from my bench.
The version I have is SDS1000A_V100R006B01D01P16.ADS and the file name bears no resemblence to any of the previous models. Also unlike older models, once installed the DSO self-boots whereas you previously had to manually power cycle the DSO.

I've asked Siglent for a link to this FW and the latest is; there is even newer P18 FW on its way soon.
When I get the link I'm promised you can be sure I'll post it here.

[StillTrying]

Any links to try?

[tautech]

Any links to try?

Not yet.
Pop me your email by PM and I'll send the P16 version I have.
There is a P18 version coming but no word as yet when.

[tautech]

Any links to try?

Did the P16 FW I sent you help some?

[StillTrying]

Yes, got the 10MB attachment no problem.

Seems mostly improvements in .16

Definitely needed a reboot after update because of trigger anomalies.
X and Y position controls faster, but still fine enough.
Variable/fine Y amplitude works without the waveform moving away from 0V and falling off the screen.
No freezes at all really, but perhaps not on long enough.
Audio frequency, low level, low noise, still the same, still very good!

But the trigger level indication no longer lines up exactly with the displayed waveform. It can be up to 1/4 div. off the waveform and still trigger, or 1/4 div. inside the waveform and not trigger, and with changes of Y amplitude control the trigger doesn't always follow the waveform accurately, occasionally failing outside of the waveform's amplitude completely, minor but it didn't do that with .12.

Still no persistence on 100ns/div and 50ns/div.
Still some jumping waveforms if trying to stop or save them.

[tautech]

Thanks for documenting those observations, please do add edits if you find anything else.
I'll grab a unit and see if I can replicate some......

Let's see what Siglent can improve on with this expected P18 FW update.
Like I said I'll provide a link to it when I get one.

[StillTrying]

Thanks for documenting those observations, please do add edits if you find anything else.

I've got loads, I think it's difficult when not everyone, all 1 of us! would agree whether it's a bug or a feature,
- and you know what happens to scope threads. I've never found a CML+ thread anywhere(the "+" makes searching difficult), this could be the only one!
Of course a change log would be very useful but as it's still on it's 1st and 2nd FW, I can imagine why they don't want to publish one.

[tautech]

Thanks for documenting those observations, please do add edits if you find anything else.

I've got loads, I think it's difficult when not everyone, all 1 of us! would agree whether it's a bug or a feature,
- and you know what happens to scope threads. I've never found a CML+ thread anywhere(the "+" makes searching difficult), this could be the only one!
Of course a change log would be very useful but as it's still on it's 1st and 2nd FW, I can imagine why they don't want to publish one.

AFAIK you're the first to bring any CML+ matters to our attention, despite that there have been many sold in the few months they've been out.
The FW I've shot to you will be a beta of the yet to be released P18 so until there is an official FW release there won't be any public changelogs I'd imagine, but despite this if there's any functionality you think isn't right (don't know how much DSO experience you have), by all means either post it here or shoot it through to my email and I can forward it to the factory.

Whether this turns into a scope thread or not doesn't matter, if something's not right it need be fixed. Period.
Basic functionality should be identical to any other CML, the Plus only signifies the addition of a higher res display and LAN to the previous models.

[tautech]

Thanks for documenting those observations, please do add edits if you find anything else.
I'll grab a unit and see if I can replicate some......

But the trigger level indication no longer lines up exactly with the displayed waveform. It can be up to 1/4 div. off the waveform and still trigger, or 1/4 div. inside the waveform and not trigger, and with changes of Y amplitude control the trigger doesn't always follow the waveform accurately, occasionally failing outside of the waveform's amplitude completely, minor but it didn't do that with .12.

New SDS1102CML+ with P16 FW.
I did notice some evidence of the above but ONLY after using Auto setup.
But once you take control back from the scope with some X and Y user initiated adjustments, all returns to normal. Some small trigger inaccuracies/inconsistencies remain evident when in Peak Detect mode, unsurprisingly. 

Still no persistence on 100ns/div and 50ns/div.

I tried both these frequencies at 1, 2, 5s and Infinite persistence settings in both Sin/X and dot display modes and observed no waveform persistence problems. 

Still some jumping waveforms if trying to stop or save them.

How much?
Point 1 or 2 of a div I saw but never more.
This I expect to be a result of processing delays from signal detection/processing and then to the display.
Does it show in captured screenshots?
I wonder if this too is associated with Auto setup and a reboot with user settings negates any jumping saved waveform.
Can you find a simple way to replicate this?


Feel free to throw any further things forward for checking or clarification.

[StillTrying]

Thanks for your efforts.

Don't think I'm not quite happy with it because overall I am, I'd have got, or swapped it for an X with free (but never to be used decoding) or something else if I wanted. I think the low noise on 10mV/5mV/2mV is excellent.
I'll have to post a couple of full size screen captures.

HW B-64-5.1
I've put the trigger level right on the border of triggering/not triggering as shown by the trigger counter showing 530Hz, a pixel or 2 up or down would change it to <10Hz or 1.000kHz, and obviously I can see when the waveform is alive or dead.
I'd like to say a self-calibration fixed it but it didn't.

Persistence captures didn't save, I'll try again. - Done.

If just a single channel is ON, then persistence is missing from 50ns as well. 

[rf-loop]

Thanks for your efforts.

Don't think I'm not quite happy with it because overall I am, I'd have got, or swapped it for an X with free (but never to be used decoding) or something else if I wanted. I think the low noise on 10mV/5mV/2mV is excellent.
I'll have to post a couple of full size screen captures.

HW B-64-5.1
I've put the trigger level right on the border of triggering/not triggering as shown by the trigger counter showing 530Hz, a pixel or 2 up or down would change it to <10Hz or 1.000kHz, and obviously I can see when the waveform is alive or dead.
I'd like to say a self-calibration fixed it but it didn't.

Persistence captures didn't save, I'll try again. - Done.

If just a single channel is ON, then persistence is missing from 50ns as well. 

Persistence:
Missing persistence (50ns) feels like bug in FW what you use.
If I remember right (long time ago when last used CML model) when scope is stopped, persistence disappear but if save screen image "on  fly" it also save persistence.

Trigger:
It looks normal (with this your single example and explanation).

It need understand and remember that this model do not have full digital  trigger system.  It is very different what can find in example SDS1000X and 2000X and many other modern technology scopes what have full digital trigger system and what make them  totally different animals.

For CML and CML+ specifications tell (only) trigger sensitivity (what is perfect definition for this?)
Trigger Sensitivity:
1 Divisions: DC-10 MHz
1.5 Divisions: 10 MHz - Max BW

In this model (Whole SDS1000 series) trigger system is traditional old two analog pathway system.  Simplified: Analog signal is splitted for  trigger system after final buffer amplifier before  ADC. Then routed and level edjusted to analog trigger comparator. Comparator other port have this analog signal.  Then there is system generated (DAC + some adjustments) trigger level. When it cross this level comparator output change and this signal is binary trig signal. (this is also what counter counts, in principle).
Digitized analog signal what is displayed on the screen is not signal what trigger comparator get. There is many small differencies due to different  analog signal handling, trigger comparator hysteresis (what is very important) and this signal also include different signal noise, different bandwidth shape, inaccuracies  and noise in trigger level signal. Also trigger comparator is not school book ideal component.

As can see in image it do not generate trigger signal after every rising edge. Because your trigger level setting is just borderline if it can see this edge or not. It really need realize that on the screen is signal what human look with eyes but it is not what trigger comparator see. Also there need be hysteresis. Comparator need see that after every rising edge what crosses hysteresis upper level is then returned down and reliable below hysteresis lower level. Both of these must happen. But on the screen we can not see real this signal what trigger comparator see and what is see in its other port for compare.
Of course this kind of system is possible to do very accurate but after then we do not talk simply cheap machines. (analog pathway trigger can be, if really do "rocket science" very advanced...(because analog processing is fast, because analog processing do not drop to Nyquist-Shannon trap etc. But extremely good analog pathway trigger system in fast digital advanced oscilloscope is - expensive and complex. I'm not ssure but perhaps example LeCroy have done there some experimental things...)




Shortly: With this kind of trigger system this (what you explain with this one image) is just normal. It is how it works. Important is:  "know your equipment" limits.

Many times (but not always) and example in this case, just push trigger level button for autoset half level trig and you get quite reliable trigger and right counted trigger pulses for frequency counter) 


(Why I can say it is normal (if look this one example). 
I have tested "lot" of individual SDS1000CML and CNL models. (Not this new CML+)) Also I have tested many individual older Rigol DS1000E series (what have same trig principle) and also example lot of some  older Owon models and even in bottom some Hantek models. I have used "truck load" of some older Tektronix and HP digital scopes what also have same trigger principle but more expensive and mostly  better - if not look this one "famous" total junk Tek "school" model. (and of course even more both of these analog scopes - and still in use)

In full digital trigger system  all is different. (simplified principle) Trigger system "look" same digitized signal after ADC what we look on the screen. What is digitized is what trigger system see (but it also see only this)

Take this seriously. Remember that all you see on the scope screen is sum of all kind of errors mixed with undefined true signal. Never you can see there whole pure truth  and only this. Independent of if your scope price is $100 or $100000.

[StillTrying]

Can you find a simple way to replicate this?

I forgot about the jumping, but here goes.

I must have done 100s of STOPs at low X speeds between 100ms/div and 1ms/div and I don't remember seeing any jumps on STOP of even 1 pixel. But at higher X speeds jumps occur.

If I enable ETS at 100ns the waveform looks same as the above with the trigger position exactly right and a STOP is only about 1 pixel out which is fine. (I don't think it uses ETS until 50ns even when it's set in the menu.)

At 50ns the ETS waveform looks like ETS1.gif. With the extra sampling the live waveform looks very good, but the trigger position is off by 25ns.

If I STOP this waveform it jumps to do one last sweep through the correct trigger position. ETSjump.gif

Although it was a clean switch-on just since I saw your post about 2 hours ago, and it's only been looking at it own 1kHz, enabling ETS soon enough gave problems as in Strange.gif. 

Thanks rf and taut I read your detailed posts.

[rf-loop]

There is some kind of "de ja vu" when I look these images and when I try remember some very old cases with SDS1000CML with its some quite old FW version.

I think if there is now P16 and as tautech told there is possible soon coming P18 is is better to wait it.

Btw, using ETS with extremely low frequency, nearly DC (1kHz is nearly DC ) signal with very very slow rise time (for what purpose)....  it may take nearly infinite before it have collected all samples, specially if go faster t/div. (also best display mode for ETS is dots)

Also in "old times" it need be well stabilized temperature (example one hour) and then  self cal.  If scope is then fully thermal stabilized ETS works better, this I do not know why. I remember old time when ETS did not work nearly at all just after scope is turned on until it have thermal stabilized. But this was with some old CML and old FW. 

[StillTrying]

Btw, using ETS with extremely low frequency, nearly DC (1kHz is nearly DC ) signal with very very slow rise time (for what purpose)....  it may take nearly infinite

Just using it's own stable 1khz so that others can very easily reproduce, or maybe not!
I'd say at 1kHz the ETS takes about 0.75sec to form.

I agree the rise time (125ns) of its 1khz may be too slow. I've had some supposedly 20ns rise/fall time CMOS 556s sitting here for a month or four, so a good excuse to test one!

Persistence stays put when STOPped, if you expand the stopped waveform the persistence doesn't expand - just a new version of the wave.

[tautech]

As yet I've not been able to replicate the 50ns Persistence issue with frequencies <25 MHz.
I've found another tiny issue where in Dots mode at fast timebase's there's some kind of waveform reconstruction happening. No effect on the waveform per se, just it not dots. 

Anyways, Siglent will be linked to this thread to examine findings and make any changes needed.

[rf-loop]

Just using it's own stable 1khz so that others can very easily reproduce, or maybe not!
I'd say at 1kHz the ETS takes about 0.75sec to form.

Ok, it is then faster than with some very old FW in CML model what I remember.
Specially if go to higher "equivalent sample speed" (more fast t/div)

If you zoom when scope is stopped it do not keep persitence. Naturally, because persistence iis not in sample buffer. It is in display memory. But when you zoom in (stop mode) it read new set from sample buffer. (It is NOT display memory zoom!)
In sample buffer there is just one single data stream and not any knowledge about previous "sweeps".

[StillTrying]

rise time.
The 15ns rise and fall times of the test CMOS 556 completely(near enough) solved the 25ns ETS trigger offset.    Still jumps on STOP but only very small.

At 84kHz the time to form the first ETS waveform still around 0.75sec.

Still missing persistence, and still some trigger level offset.

[StillTrying]

Also in "old times" it need be well stabilized temperature (example one hour) and then  self cal.  If scope is then fully thermal stabilized ETS works better, this I do not know why. I remember old time when ETS did not work nearly at all just after scope is turned on until it have thermal stabilized. But this was with some old CML and old FW.

I think this is probably it. If I do a self-cal immediately before using ETS, it works without problems. After about 1 to 1+1/2 hours the first effect I see is the double waveforms 25ns apart on STOP, which gives me a chance to do another self-cal and then the double waveforms disappear. I don't know how long the second self-cal lasts - not been on long enough.

[StillTrying]

Normal/Long Memory
Normal Mem No persistence DOTs on 100ns/div and 50ns/div, if I switch to Long Memory persistence DOTs then appear on 100ns and 50ns, but off on 25n/div and 2.5ns/div instead!

STOP works differently on Normal/Long Memory.
STOPping on Normal Memory simply captures the displayed waveform against any new triggers as you would expect.
On Long Memory STOP clears the grid screen and awaits one last trigger, (Even though it's in STOPped mode) which it then displays. FW .12 was the same.


No persistence dots in Zoom/Delay mode, all speeds, but persistence is on with interpolation.

[StillTrying]

As yet I've not been able to replicate the 50ns Persistence issue with frequencies <25 MHz.
I've found another tiny issue where in Dots mode at fast timebase's there's some kind of waveform reconstruction happening. No effect on the waveform per se, just it not dots. 

Running in DOTS mode at the fastest timebase speed it draws small horizontal lines, STOPped it just shows the (few) DOTS.

[StillTrying]

Am I the only CML+ user in the world.

Long Memory, Single trigger, Force trigger gives as pic, I often get the impression that it doesn't know where the trigger position is in long memory, or where the start and end position(s) of memory are, either way the (repeatable) forced trigger trace makes no sense to me.
The sine curved bit is part of a previous trace/capture.

Edit: More.

A Forced trigger, Long Mem, both probes shorted to gnd clip.
Usually the 'cliff edge' is not expandable even if it's right under the trigger, but this one was expandable.
LongMemEdge.gif  and  LongMemEdgeX20.gif"