RIGOL DHO900: Logic analyser can only do fast

[D-Jan]

Hi,
just noticed that you cannot use the 16chan. logic analyser (LA) in the DHO924 at slow timebase speeds:
Whe the unit comes up it always has this annoying "Auto Roll mode" enabled.
Default timebase is 2µs. When I enable the LA and then reduce the timebase to 50ms the LA traces disappears. Pushing the LA button sais "Operation disabled"
When I disable the "Auto Roll mode" to have classic scope behavior the same happens at 200ms timebase setting.
     What is the problem of recoding digital data at slow speeds?
I don't get it.

Hey Rigol guys please read this and fix it.

[NE666]

It speaks to not having a variable sample rate for the LA section, I think.

I think this also speaks to selecting the right tool for the right job. IMHO the LA section of a MSO is principally there for tasks which require correlation of analogue and digital domains and the characteristics you see largely fit with that. For example, no-one has a major need for debugging signal integrity issues with rising edges in the order of 100's of ms. Other principally digital tasks such as long captures, long running protocol exchange traces and in your case slow recording speeds, are the domain of 'real' LAs. The good news is that 'real' LAs can be pretty inexpensive for low speed operation.

I coughed for the PLA2216 for my MSO5000 but I sill have and use a 'cheap' PC-based USB LA for what you describe. They are, again IMHO, different instruments with different use-cases. The LA section of the Rigol's isn't really a LA, it just provides a subset of that functionality.

[D-Jan]

Hi NE666,

I cannot believe that there is no variable sample clk available in the DHO. As you mention it also features 4 analog channels. When I go to 1sec/Div these ADCs  also need a slower sampling clock.
My old Agilent 54622D has no issues with that whatsoever. With slow timebase you can see analog and digital signals develop at the same time from left to right. For the DHO924 the LA updates the LA in one shot when the analog channels have reached the right border of the screen.
I still think an MSO is the correct tool for the job. My project features a MCU that generates PWM and controls various other circuits. With the MSO I can monitor the Buck output voltage in analog (up to 60V). For a single shot I have all my control signals at a view, plus I can zoom in to even measure PWM duty rate since modern MSO have huge memory depth.
I have a standalone LA (ikalogic), but I just like the benchtop better. Also a LA will never show you analog waveform matching with your digital data.
Having the MSO function in my 54622 I never wanted a standard oscilloscope.
At work I have a RIGOL MSO8204A and here the LA section is behaving as expected.

I have no idea if someone at Rigol has ever tested the LA in the DHO924: You can disable some channels for better readability, but every time you press the LA button to hide /show the LA traces all channels are automatically enabled again. It is a pain ...
It is nice bit of hardware, but the firmware of this thing is clearly still in beta stage.
I would expect Rigol to have some developers to fix the bugs in the firmware and provide SW-updates on a regular basis.

[NE666]

My old Agilent 54622D has no issues with that whatsoever. With slow timebase you can see analog and digital signals develop at the same time from left to right. For the DHO924 the LA updates the LA in one shot when the analog channels have reached the right border of the screen.

Hi.

That's what leads me to believe that the digital sampling system has a separate sample clock which is either fixed rate, or has a minimum rate for some architectural reason. It's not that the LA just doesn't work at low speed, it is being actively prevented from doing so, and there must be a design reason for that. If I had to bet, I would say it's because it's fixed sampling (or fixed within a narrow range) and doesn't have the memory depth to collect enough data to meaningfully decimate and recombine in post-processing with a slow analogue capture. Your other scopes use different hardware and architecture and therefore could well be different in how they handle this. However, I'm an enthusiastic amateur, not a qualified engineer, so I'll be the first to admit that's a hunch based on what you observe.

I have no idea if someone at Rigol has ever tested the LA in the DHO924:

If it's anything like the three Rigol products I have purchased, not much testing was done period. And probably no usability testing or customer feedback whatsoever. That might just sit ok with me, at this price point, if your other expectations of Rigol actually being interested in fixing defects and committing to a consistent, sensible FW release process and cadence. Sadly that's not been my experience either and thus why I'd be hard pressed to buy another Rigol product right now.

[Fungus]

The analog part goes into roll mode at slow speeds, too.

I'm guessing it because it simply isn't useful to take ten seconds (or two minutes) between screen updates.

[D-Jan]

Hi Fungus,
the DHO924 can sample (one channel) at 5sec/div in roll mode. So you can watch the screen to fill up for 50sec. If you then press Stop on the timebase it is possible to examine the trace at any position down to 4µs/div timebase with good resolution (2µs/pt). So that is quite impressive. The timebase ranges from impressive 500s/div down to 2ns/div. So the hardware looks quite capable at a first glance.
Roll mode is ok when you want to observe a signal over a long period of time and inspect the signal by visual. The DHO924 goes into roll mode at 50ms/div due to a default setting that is loaded every time at bootup. So far I found no way to change this default, so it has to switched off EVERY TIME, similar to the probe gain settings that also come up at "1x" when the unit is switched on.
If you know what to look for, some rare thing happening, you would normally setup a single trigger and go for a coffee. Why should you not want to observe slow speed signals and have the scene frozen at a well defined trigger point? There is plenty of things that could go wrong overnight. With a single trigger you can stop sampling at that trigger point and then carefully examine the traces to understand what happened before something went wrong.

[NE666]

Probably not helpful if you already own the DHO scope but for reference lest it interest anyone else, and since I was using it today, here's the state of play with my MSO5000. Perhaps it might assist a purchase decision for someone.

The 5000 can indeed continue to operate its LA section all the way down to the lowest timebase setting, 1000sec/div. However, notice the effect on the sample rate and memory depth - down to 250S/s and 2.5Mpts respectively at that speed. Note also that the analogue and LA sections have different sample clocks and available memory depths.

Nonetheless, it can still capture and decode a train of slow (20kHz) I2C frames when set at a timebase of 1sec/div (at which the LA section sample rate is equivalent to about 10x the clock rate of the bus signal.). It's just a shame that it doesn't support segmented / compressed sampling as would a non-trivial standalone LA, which is where of course those devices come into their element for long captures.

[With apologies for the lens flare in the attached photos]

[D-Jan]

Hi NE666,

yep I actually bought an DHO924 for our company. I was attracted by the size of the unit and the proposed feature set.
And still I am thrilled about the feature set it can do. The serial decoder is really nice and throws in the decoded pattern while the unit is triggering on a signal. Awesome!

Working for a long time with a HP/agilent 54622D I really love to have both analog and digital in one unit.

So I was really shocked when I tried the logic analyser in the DHO924. Using the logic analyser even for slow timebase was something that I would take for granted, and there is also no comment in the products data sheet that the function of the LA is limited whatsoever.

Reading through the DHO900_UserGuide_EN I fould this however in sect 7.6 Roll Mode:

OFF: the Roll mode is disabled. The oscilloscope operates at slow sweep speeds
when the horizontal scale is 200 ms/div or slower. The slow sweep mode is not
available for digital channels.  

I have written to Rigol already. They have already confirmed a bug for switching LA on/off. Which will re-enable all channels. Even if you just want to use 4 signals.
So I am anxious to get a reply from them.

For my last debugging I borrowed a colleague's MSO8204A which is a bit oversized regarding speed. But it worked fine also in slow timebase mode. I could record for several secs and then analyse PWM ratio and the analogue output signal.

The software on the MSO8204A feels a bit crusty compared to the DHO. The latter is more responsive. So I am hoping that RIGOL is going to work on the firmware.
But it seems there are few updates. 

There is the electronica fair in Germany in November.
If I do not get a reply from Rigol I might take my unit and some glue and bond myself to the floor of their booth... 

[TomKatt]

I have written to Rigol already. They have already confirmed a bug for switching LA on/off. Which will re-enable all channels. Even if you just want to use 4 signals.
So I am anxious to get a reply from them.

You may get a reply, but my perception is that Rigol doesn't have the best history of prompt FW fixes so I wouldn't hold my breath.