USB logic analyizers - current thoughts on 'the best'?

[three_jeeps]

I am looking for input regarding the 'best' options for a USB logic analyzer in the $200-$500 range 32 (34?) channels and having deep memory and at least 200 MS/s @ 32 channels.
Ive searched the forum and most of the comments/opinions seem rather sparse after 2021.  Some units I am considering are:
1 DreamSourceLab DSLogic U3Pro32
2 Intronix LA1034 LogicPort
3 Digilent Digital Discovery

The Intronix is a bit long in the tooth and doesn't have very deep memory. Reviews for the other two basically complain about software clunkyness or lack of features.  I am looking for feedback regarding these 3 or if there are any other 'newer' options out there.
Thanks
J

[tverbeure]

I have the DSLogic U3Pro16, which is identical to the U3Pro32 except for the number of channels. I haven't used the Intronix or the Digital Discovery, but at some point I have looked at them.

The Intronix and the DSLogic support external clock input, but the DSLogic one can only go to 50MHz, whereas it's 200MHz for the Intronix. The Intronix seems to have much better support for state tables, features that have been present in big iron logic analyzers from HP and Tektronix since the eighties. It's not very important for me: if I really need it, I'll use my HP 1670G. The triggering options of the DSLogic are not terrible, with the option do multi-state triggering with counters etc.

BTW, the DSLogic has an external trigger input pin, with associated probe wire, but that feature is currently not implemented, and I assume it never will.

The Intronix and the Digital Discovery have a very limited set of protocol decoders. The DSLogic has a truckload, and you can use PulseView if you want even more. This is a major plus for me.

The Digital Discovery only lists input voltages up to 3.3V. The DSLogic can go much higher, including 5V which is important for me because I have old 5V based test equipment that sometimes needs debugging.

The probe wires that come with the DSLogic are excellent: they're coaxial cables with a ground pin per probe wire. I expect the DSLogic to do best in terms of probing high speed signals. The Digital Discovery uses standard 0.1" Dupont wires but for $50 you can be a high-speed probe set consists of an adapter and twisted pair data/ground cables. Not as good as coaxial, but probably sufficient.

My biggest gripe about my DSLogic is that it doesn't do analog sampling, something the Saleae Logic Pro 16 can do. But neither can the Intronix or the Digital Discovery, and Saleae priced themselves out of the hobbyist market years ago.

I've been working for a while on a DSLogic teardown and review, but it's still incomplete. But if you're interested, here's the latest draft: https://github.com/tomverbeure/tomverbeure.github.io/blob/dslogic/_drafts/2023-03-20-DSLogic-U3Pro16-Teardown.md.

[tggzzz]

There's an aspect of logic analysers that is often overlooked: the desirability and ability to not capture data.

If you use the filtering and triggering capabilities of "decent" logic analysers, you can avoid seeing lots of irrelevant crap thus leading you to concentrate on the small proportion of interesting information. That has the benefit of making long capture buffers less important.

Examples:

• don't capture input except on the rising edge of an external clock. That allows you to concentrate on the digital state data (not the digitised analogue waveforms)
• don't capture input except when one input is active, e.g. a chip select. That allows you to concentrate on the data being sent to one specific chip
• don't capture input unless it occurs more than or less than a specified time after the previous data. That allows you to check for rare timing violations
• don't capture input unless it occurs after this and that and the other

Note that all those need to be executing at full input rates; postprocessing buffers inevitably means the "interesting" information will be missed because it occurred while the buffer was being processed.

[metebalci]

External clock input (state-mode) is quite useful in some cases, and also sampling at both the rising and falling edges of the external clock might be useful.

I also have a Digital Discovery, but AFAIK, it does not have an external clock input.

[hli]

For the Digital Discovery, the reference manual states:
Setting the voltage to 3.3V, 5V logic inputs are tolerated but the input threshold is 1.42V.
(It's hidden in a foot-note, though)

[ebastler]

Examples:

• don't capture input except on the rising edge of an external clock. That allows you to concentrate on the digital state data (not the digitised analogue waveforms)
• don't capture input except when one input is active, e.g. a chip select. That allows you to concentrate on the data being sent to one specific chip
• don't capture input unless it occurs more than or less than a specified time after the previous data. That allows you to check for rare timing violations
• don't capture input unless it occurs after this and that and the other

Thanks. I see your point, but am missing the connection to the OP's proposed shortlist of logic analysers (and potentially other products). What to look for in spec sheets? Are these essentially the "state table" features which tverbeure had mentioned as being strong in the Intronix but lacking in the others?

[tggzzz]

Examples:

• don't capture input except on the rising edge of an external clock. That allows you to concentrate on the digital state data (not the digitised analogue waveforms)
• don't capture input except when one input is active, e.g. a chip select. That allows you to concentrate on the data being sent to one specific chip
• don't capture input unless it occurs more than or less than a specified time after the previous data. That allows you to check for rare timing violations
• don't capture input unless it occurs after this and that and the other

Thanks. I see your point, but am missing the connection to the OP's proposed shortlist of logic analysers (and potentially other products). What to look for in spec sheets? Are these essentially the "state table" features which tverbeure had mentioned as being strong in the Intronix but lacking in the others?

Potentially. EDIT: clarification: some LAs allow the user to specify traditional FSMs, which may or may not use state tables[1].

I am not familiar with the devices mentioned, and do not know the OP's "use cases". It is up to him to match capabilities to his use cases.

For all we know he might be better off with a protocol analyser, or something like the BusPirate5.

[1] example from a datasheet

[alm]

I also have a Digital Discovery, but AFAIK, it does not have an external clock input.

I'm pretty sure it supports state mode in addition to timing. Look for sync capture mode. It just does not have a dedicated input like some other logic analyzers.

[tszaboo]

BTW, the DSLogic has an external trigger input pin, with associated probe wire, but that feature is currently not implemented, and I assume it never will.

Does the trigger output work? Have you tried connecting it to an oscilloscope external trigger input?

[tggzzz]

I also have a Digital Discovery, but AFAIK, it does not have an external clock input.

I'm pretty sure it supports state mode in addition to timing. Look for sync capture mode. It just does not have a dedicated input like some other logic analyzers.

Some logic analysers have such a state mode without having dedicated clock inputs. They interpret a 0->1 transition on a general input as a clock rising edge, which implies the state mode's capture rate must be 4 times slower than the asynchronous mode's capture rate.

[metebalci]

I also have a Digital Discovery, but AFAIK, it does not have an external clock input.

I'm pretty sure it supports state mode in addition to timing. Look for sync capture mode. It just does not have a dedicated input like some other logic analyzers.

I think you are right, I think I remember something like the trigger and ext clock cannot be used at the same time.

[metebalci]

I also have a Digital Discovery, but AFAIK, it does not have an external clock input.

I'm pretty sure it supports state mode in addition to timing. Look for sync capture mode. It just does not have a dedicated input like some other logic analyzers.

Some logic analysers have such a state mode without having dedicated clock inputs. They interpret a 0->1 transition on a general input as a clock rising edge, which implies the state mode's capture rate must be 4 times slower than the asynchronous mode's capture rate.

Maximum for external clock is I think around 100MHz and half for DDR. This part of digital discovery is not document well to my opinion.

[three_jeeps]

Thank you for the very helpful information!!! Much appreciated!

[three_jeeps]

There's an aspect of logic analysers that is often overlooked: the desirability and ability to not capture data.

If you use the filtering and triggering capabilities of "decent" logic analysers, you can avoid seeing lots of irrelevant crap thus leading you to concentrate on the small proportion of interesting information. That has the benefit of making long capture buffers less important.

Thank you. Appreciate the point.  I used LA's extensively through the 1980s and well aware of being able to configure/specify discriminators.  The Gould/Biomation units were the absolute best at doing that.  Later the HP 1671G more or less caught up to it .  I used various Tek and HP LA's and was part of an eval team for HP to put a couple of their LA's against some prototype CPUs we were developing for NASA.  I've been away from the hardware edge of things for a number of years and am curious at to what is out there now.  I do have one HP and one Tek in my basement lab but they are beasts to lug around compared to the smaller USB ones.   I do like the logic pro 32 but the prices is a bit out of my budget....

[tggzzz]

There's an aspect of logic analysers that is often overlooked: the desirability and ability to not capture data.

If you use the filtering and triggering capabilities of "decent" logic analysers, you can avoid seeing lots of irrelevant crap thus leading you to concentrate on the small proportion of interesting information. That has the benefit of making long capture buffers less important.

Thank you. Appreciate the point.  I used LA's extensively through the 1980s and well aware of being able to configure/specify discriminators.  The Gould/Biomation units were the absolute best at doing that.  Later the HP 1671G more or less caught up to it .  I used various Tek and HP LA's and was part of an eval team for HP to put a couple of their LA's against some prototype CPUs we were developing for NASA.  I've been away from the hardware edge of things for a number of years and am curious at to what is out there now.  I do have one HP and one Tek in my basement lab but they are beasts to lug around compared to the smaller USB ones.   I do like the logic pro 32 but the prices is a bit out of my budget....

Similar experience here. When I managed to return to embedded systems, I was delighted and appalled at how little had changed in 30 years: still C on 8bit processors. The only major differences are smaller/faster/cheaper, adc/dac performance, and energy harvesting.

Yes, I know that's an exaggeration, but there is more than a grain of truth in it. There are also signs of light, especially w.r.t. highly parallel software and hard real-time software.

Still, at least I can afford things like an Agilent 1682 LA, an HP8562 spectrum analyser, far too many scopes, an Agilent 53310 modulation domain analyser, 8 digit DMMs, and other weirder test equipment. All you have to do is watch and be patient

And I'm waiting for the new BusPirate 5, to use as a protocol analyser/stimulus.

Now I just have to make time to get back to playing with a circuit I first used in the.late 70s: an n-path filter and it's modern variant, the Tayloe mixer. Oh, and a scanning tunnelling microscope.

[tverbeure]

Does the trigger output work? Have you tried connecting it to an oscilloscope external trigger input?

It does! It creates a 1.8V pulse that's 500ms long.