PC USB Logic Analyzer - Who Has The Best Software?

[Mechatrommer]

- Can you combine arbitrary signals into a bus and display the bus waveform? I know that there is a simple parallel analyzer, but it's not what I expect

hantek sw (and of course logicport) is specialized at that, but just that. other than that, hantek sw tried to solve "chess problem - how to defeat kasparov" behind since apparently it struggling on displaying data (sim) and eventually got hanged when we press some button (i mean its only a simple thing it can do/show but processing it seems like hell in it). otoh, saleae gave more effort on GUI, but the functionality still can be improved. o well, cant get everything in one package, maybe i'll give more preference to something that open their software to SDK and add-on plugins. so future development is possible by 3rd party or diy.

IMHO there is no LA available which can do this in realtime. Only the dumb USB LAs could do this in theory

interesting point. are you saying that device like saleae logic (usb bluk transfer mode, no internal ram) in theory can do any kind of triggering possibilities in PC software? USB always the bottleneck i think. they have to implement excellent compression in HW before sending it to USB to improve sampling rate, am i thinking right? otoh saleae logic 16 has 10bils samples capability, so we can dumb capture in HW first and later download and analyze in PC in offline mode (as opposed to "virtually realtime but not so" original logic version). btw, i cant comment much on hardware as i'm no expert. but having experience guys talking here is great as it will be valueable for future reference later. keep it coming.

btw, i just realized i broke downloaded the zeroplus sw (138MB) last night, i have to restart download again just now, so still haven't play with it and the assessment still = ?
and i'm sure there are (should be) alot more USB LA i havent found yet. i only based on whats mostly mentioned here and there.

[Mechatrommer]

Its only the original salae but for $50 i couldn't say no
http://iteadstudio.com/produce/seleae-logic-usbee-ax-usb-blaster/

i have USB blaster, but slightly different layout (instead of 2 another chips, mine only got one), but the same processor CY7C68013. ok its almost the same layout as the ITDB01 board here http://iteadstudio.com/store/index.php?main_page=product_info&cPath=1&products_id=51 maybe i just have to find the FW for both "cloned saleae logic" and my original usb blaster FW? so i can switch functionality between LA and fpga programmer? if i buy it (even cheap at $50 not including shipping cost which could be the same or doubled price) maybe i'll just get another usb blaster HW? sound uneconomical for me. thanks for providing the link anyway.

[alm]

Well, the 10k€ Agilent desktop LA on my bench at work can't do this, neither can the 14k€ LeCroy Waverunner Scope with the ~1k€ SPI plugin. Well, the latter can trigger on SPI patterns, but of course only one the sample memory. So it actually already triggered and just filters the memory as described above.

I'm surprised, I would have expected them to have implemented real-time protocol decoding by now. Although I guess it gets slightly harder as the sample rate goes up.

Would the OLS come close if they ever finish the software side of the demon core?

No, it would just come close to the edge/time trigger capabilities of an old HP desktop LA (16500/16550):

I thought it might be possible to implement an SPI decoder in the state machine, although that would only work for very simple protocols.

Well, the (non realtime) operating system is always a bit of a problem. Sometimes windows decides to use 100% CPU load for whatever for dozens of seconds and it will be pretty hard to overcome this with a large buffer. Surely the reliability of USB analyzers could be improved a lot by using USB3.0 and larger buffers, but they will never reach the reliability of a real HW trigger.

But they always stream the data (semi) real-time, I believe they only have a few bytes of buffer in the hardware. Because of the large RAM, you can easily fit several seconds in the buffer at 24MS/s. The only limitation I see is that, on average, the trigger algorithm should be able to keep up with the data, but that shouldn't be too hard on a modern CPU. In the case of SPI you can ignore all samples except those on a rising/falling edge of the clock while SS is enabled, for example.

interesting point. are you saying that device like saleae logic (usb bluk transfer mode, no internal ram) in theory can do any kind of triggering possibilities in PC software? USB always the bottleneck i think. they have to implement excellent compression in HW before sending it to USB to improve sampling rate, am i thinking right? otoh saleae logic 16 has 10bils samples capability, so we can dumb capture in HW first and later download and analyze in PC in offline mode (as opposed to "virtually realtime but not so" original logic version). btw, i cant comment much on hardware as i'm no expert. but having experience guys talking here is great as it will be valueable for future reference later. keep it coming.

USB is the bottleneck, but this is independent of the triggering. USB is the reason why they can only sample at 24MS/s. With the Logic16 they employ some tricks so they can trade channels for sample rate, and I believe they added more buffer space. The reason why they can do billions of samples is because they store it all in your computer, the hardware alone is incapable of anywhere near that number. This is why you see a sharp divide between the USB bulk transfer ones that can do almost unlimited number of samples at modest sample rates (Saleae, Usbee), or the FPGA-based ones that can sample much faster but are limited to the RAM in the hardware (Intronix, OLS, Zeroplus).

[0xdeadbeef]

I'm surprised, I would have expected them to have implemented real-time protocol decoding by now. Although I guess it gets slightly harder as the sample rate goes up.

From an historical point of view, logic analyzers were mainly used to watch data/address buses. Our older HP LAs could even show disassembled opcodes. I guess therefore they are still aimed at complex edge/time triggers, not on realtime protocol decoding. Anyway, if you take into account that Agilent and Lecroy want around 1k€ for one simple (offline) protocol decoder in a scope, I'd guess they'd let you pay >10k€ for a few realtime decoders alone on top of a 10k€ LA.

I thought it might be possible to implement an SPI decoder in the state machine, although that would only work for very simple protocols.

The FPGA in the OLS is a pretty small Xilinx Spartan-3E (XC3S250E) with only 250k gates. A lot of this is used for the block RAM (216K). So there's actually not much space left for the LA.

[alm]

From an historical point of view, logic analyzers were mainly used to watch data/address buses. Our older HP LAs could even show disassembled opcodes. I guess therefore they are still aimed at complex edge/time triggers, not on realtime protocol decoding.

But how many people are still doing design with external address buses? Even the high-end ARM processors can have RAM and CPU in the same package. Not that many other parallel buses left either, although I guess real-time decoding of PCIe is quite tricky without dedicated hardware.

I thought it might be possible to implement an SPI decoder in the state machine, although that would only work for very simple protocols.

The FPGA in the OLS is a pretty small Xilinx Spartan-3E (XC3S250E) with only 250k gates. A lot of this is used for the block RAM (216K). So there's actually not much space left for the LA.

I was thinking about the HP 16500A-like state machine in the Demon core, but 16 states may not be enough to match all 8 bits.

[0xdeadbeef]

I was thinking about the HP 16500A-like state machine in the Demon core, but 16 states may not be enough to match all 8 bits.

And actually a lot of ASICs use 16bit per word and multiple words per message and then you can have even daisy chains of 2 or 3 devices etc.

[Balaur]

But how many people are still doing design with external address buses?

Heh, heh, you'll be surprised!

One of the major departments at my company is specialized on very specific reliability testing of components. Basically, we interface the Device Under Test to an internally designed tester that has a large FPGA (both in size but also on I/Os - hundreds of them). There is a DUT-specific board that ensures the electrical set-up for the DUT.
The FPGA implements an interface for the DUT to be happy and to work correctly.
The tester also have a standardized interface (up to 160 signals) to the LA to map the signals we need.

The DUT can be anything: SRAMs, DRAMs, FPGAs, TCAMs, CPUs, quite advanced devices really and sometimes even dedicated test vehicles or engineering samples. The tester copies the most interesting signals of the DUT to the LA to give us some visibility on what's happening there. That includes address/data, various buses, etc. Very useful.

We work with the component suppliers very early in the production (before release). The datasheet may not exist yet or not quite really fits the reality. Thus, in our case, a competent LA is a must.

Now, since we had tested hundreds of fairly complex devices until now, it looks like there is quite a lot of designing involving devices with a lot of interconnections. I wish those engineers all the best, since I know how things could get tough very easily.

[slateraptor]

There is a salae sdk and clone hardware can be had for $50. My clone works as both a salae and a usbee.

It's sad to see all the hard work that Joe and Mark of Saleae have put into their product being ripped off by yet another Chinese hacker lacking principle. It's even more sad to see the hobbyist community supporting the adulteration of a fair product sold at a fair price.

A disappointment to see hacking without principle for the sake of marginalizing that which is already within reach.

[Mechatrommer]

There is a salae sdk and clone hardware can be had for $50. My clone works as both a salae and a usbee.

It's sad to see all the hard work that Joe and Mark of Saleae have put into their product being ripped off by yet another Chinese hacker lacking principle.

open source fanatics? can we make it as a new term? i do agree with you at some points. but realizing the HW is just the same as another device, just difference in FW/SW and the bundled SW cannot give all you want is sometime mind disturbing. if they want to make a "good for money" device i think they need more hardwork, ie better SW and functionality etc, instead of just try make make a balance in the force (cheap but not complete SW). but i dont know about others and you.

[Rasz]

yes i'm currently looking for "bang for buck" USB LA. but since this is dependent on PC and USB, so software is equally important part just as the hardware i think. most USB LA can provide decent hobbiest level feature, but the kicker usually the "crappy" software.

http://www.aliexpress.com/product-fm/446147460-CY7C68013A-USB-Logic-Analyzer-Core-Board-Schematics-wholesalers.html

its Salea clone for $15, you can get clones with a jumper that can work as both Salea and USBee for $20-25. Thats as "bang for buck" as they come.

[Mechatrommer]

thanks mate and welcome to this forum. its been proposed by Chet T16 here https://www.eevblog.com/forum/index.php?topic=5361.msg70133#msg70133 i think its the same hardware, but your link is far more cheaper
edit: with one chip less missing? (RAM?) and the funny thing, the ad showing Saleae software picture with the original saleae logic16 HW in it... as usual... from China with Love!

[alm]

Since the software represents the far majority of the development effort for the dumb USB bulk logic analyzers (Usbee, Saleae), cloning them is more harmful than cloning something hardware focussed. If companies clone a scope front-end (eg. the whole Hantek/Rigol/Atten/Instek mess), they at least have to spend some R&D effort into getting the layout and parts right, and they have to actually build the hardware. This is just a Cypress SoC plus EEPROM on a PCB, which probably costs much less than $15 to make without case and probes. Usbee at al. are basically being punished for not annoying their users with protection schemes like online activation.

On the more practical side, the screen shot appears to show the old software, which is no longer maintained or available. I would verify that it also works with the current version before buying. I believe there have been some troubles with clones and the new software.

[Mechatrommer]

going back at this. know what i found? i really dissapointed still not found the "ideal" software that i think suitable. the most featured and user friendly imo so far is scanalogic software, but still i want it to be able to show analog waveform and the HW is a bit on the low end side (4ch 20MSps) though it provides FFT, what the heck FFT is for in logic analyzing? . saleae SW looks cute (ipod scroll, iron man website), but cannot show analog, cannot find data value, logic16 price near ds1052 cost, what did they do?, sigh.

there's new guy coming...
http://www.ebay.com.my/itm/Ideofy-LA-08-USB-Logic-Analyzer-I2C-Master-and-I2C-SPI-UART-decoders-/270852968239?pt=BI_Analyzers&hash=item3f10184f2f
http://www.ideofy.com/la-08_en

his HW looks promising, but the SW still in very early stage. maybe i should encourage this guy, i like the way he did the sampling rate and channel selection. i think i'm going PM him if he can add saleae+scanalogic feature in his SW. if he can respond positive then good for him.

[alm]

How about just concentrating on finding the best logic analyzer software, as opposed to a jack of all trades, master of none? Most logic analyzers suck at analog, and most DSO's suck at working many logic channels. Unless you spend $$$ on a real MSO (i.e. not Rigol), I think you're better of getting a good DSO and a good logic analyzer.

[Mechatrommer]

Most logic analyzers suck at analog, and most DSO's suck at working many logic channels.

err i mean, combining say 8 or 16 digital io and show as byte/word values in one graph. kind of debugging DAC on the digital side. thing like Hantek LA5034 software does.

[alm]

Is this useful for anything but old DAC/ADCs with an old-style parallel interface? Most modern parts, especially the ones likely to be used with a low-end MCU, are using some sort of serial bus. You'd have to decode the serial protocol first before being able to graph the values.

[Mechatrommer]

well at least parallel decoding could be handy. btw, its not like we have to add a resistor in the HW? its just in the SW, only few lines of code to add will do, isnt that hard isnt it? esp if you are making money out of it.

[alm]

I'm not saying it's hard or that I'm against software incorporating this feature, but I wouldn't consider it one of the must-haves. The ability no navigate large traces smoothly would probably among the most important ones. This includes both search and smooth zoom/pan. Also good support for decoding the popular serial protocols. I should be able to zoom out (to some degree) while still seeing the decoded data. Of course YMMV.

[Mechatrommer]

de-ghosted my PC and now i managed to test previously crashed SWs, some other apps was messing with the dotnetfx config/file i think. USBee tested, about similar interface with older saleae (minus the ipod scroll). but this Lecroy LogicStudio, so far its the best software, list in Reply #8 updated, i gave it HW 9, SW 9 its the best so far compared to others at several order magnitude. but $999? gulp! i hope the next update they will incorporate Saleae's scroll and centered over mouse zoom style, we cant get everything right but top navi bar of Lecroy's is adequate enough to navigate all data easily. now they got rectangle lens/magnifier too! very very nice! thats how a LA software should look like from my eyes.

[roli_bark]

Bumped into it while looking for a good USB based LA. Sorry for reviving this thread.

The Intronix LogicPort has only a Sample buffer of 34 x 2048 samples ? Or is it a Typo ?
If so, why did you grade it as HW8 [ relative quite high grading] ?
Sorry, but in my book a 2048 samples per channel LA does not deserve a 2nd look ....

[amspire]

Bumped into it while looking for a good USB based LA. Sorry for reviving this thread.

The Intronix LogicPort has only a Sample buffer of 34 x 2048 samples ? Or is it a Typo ?
If so, why did you grade it as HW8 [ relative quite high grading] ?
Sorry, but in my book a 2048 samples per channel LA does not deserve a 2nd look ....

It is an excellent logic analyzer - one of the best in the sub $500 price bracket.  It uses compression to store samples so it can often store millions of clock cycles in that 2048 samples. It also has good triggering, so you hopefully do not need to capture as much data. Because the samples are stored in memory on the Asic, it can capture at a genuine 500 MHz for a 34 bit wide data with data compression - much faster then the analyzers that stream the data through a USB port to be saved on the PC. The streaming analyzers have a slower capture rate but can store huge samples.

So having both types of analyzer can be a good tactic. The right tool for the right job.

Yes there are cases when a big data buffer is needed to say capture the whole conversation of a communications port for later analysis, but if you always have to capture MBytes when debugging, then you are probably not working efficiently.

If you understand how to work within the limitations of the Logicport, it is a fabulous fast logic analyzer. It most definitely deserves a look if you need a high performance analyzer.

[Mechatrommer]

If so, why did you grade it as HW8 [ relative quite high grading] ?

the guy talked out of his ass

[Bored@Work]

If so, why did you grade it as HW8 [ relative quite high grading] ?

the guy talked out of his ass

No, the guy managed to explain his reasons. You, on the other hand, seem to inhale some very bad batch of glue.

[Mechatrommer]

If so, why did you grade it as HW8 [ relative quite high grading] ?

the guy talked out of his ass

No, the guy managed to explain his reasons. You, on the other hand, seem to inhale some very bad batch of glue.

if you have trouble understanding, under alcoholic influence or in a very bad day... here i was refering to this post https://www.eevblog.com/forum/projects/pc-usb-logic-analyzer-who-has-the-best-software/msg70033/#msg70033 that he (roly_bark) pointed out.