EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: saber63 on May 19, 2016, 10:51:58 pm
-
So, trying to wrap up an embedded project and running into some bugs and require some tools... but which one? We have been using a Saleae logic 16 Pro and logic 16 for day to day dev work. But they aren't quite up to this task.
We have roughly 36 IO we need to probe. Mainly SPI, I2C and UART, But we also have a Parallel Flash RAM, and a few GPIO. Nothing faster than ~20Mhz all 1.8 or 2.5V signal levels.
The current batch of issues require that we sample all the lines for roughly 20 days straight.
We only need to trigger on interesting stuff, Like missing ACK/NAK, Incorrect/ missing SPI data. Or 0x00/0xFF pages being written to flash, etc.... Being able to trigger and save data more than once without user intervention would be perfect. The Saleae's aren't stable enough to leave alone for more than an hour. let alone issues with triggering, data size etc...
No budget for a new Tek/Keysite etc... piece of kit.
But a used piece should be doable! I don't mind over gearing a bit to make it usable on future projects. But it's been 7 years sense i've used a real Logic Analyzer, and they were already 5-10 year old units that the university had and i dont recall them being able to do serial...
In this case UI and software is most likely more important than hardware specs.
Some of the kit i've been looking at... the TLA 715-7xxx series kit. the keysight etc... 1690x and newer.
Would love some first hand insight!
-Levi
-
Hi
Simple answer:
Get a demo board from any of a billion places and program it up for the triggers. Spit the captured data out via ethernet to a server. Hardware cost < $50. If a single board does not have the bandwidth (36 x 20 MHz X 2 samples ... could be a problem) put an $75 FPGA board in front of the micro. Total cost is still way less than you paid for the logic analyzer.
Bob
-
The issue with that sort of a solution is time.
IMO the time it would take to validate the software running on said dev board to ensure that it is indeed sampling fast enough, the trigger logic is all working correctly, Creating the serial and Parallel decoding logic to allow the trigger etc... AND than setup the data out, and setup a server etc.... far far outweighs the cost of a used Logic analyzer. Saleae has how many people working on their stuff for how long? And still cant get it to do this stuff? Yeah....
I doubt that I can get ALL of that working perfectly in 40hours. At which point its cheaper to buy a used TLA 714 or up.
Time is money and all that.
-
The issue with that sort of a solution is time.
IMO the time it would take to validate the software running on said dev board to ensure that it is indeed sampling fast enough, the trigger logic is all working correctly, Creating the serial and Parallel decoding logic to allow the trigger etc... AND than setup the data out, and setup a server etc.... far far outweighs the cost of a used Logic analyzer. Saleae has how many people working on their stuff for how long? And still cant get it to do this stuff? Yeah....
I doubt that I can get ALL of that working perfectly in 40hours. At which point its cheaper to buy a used TLA 714 or up.
Time is money and all that.
Hi
Well, to capture a months worth of data, you are going to need a pretty fancy setup. There will be validation and the like no matter what you do. Relying on newly purchased used gear for that sort of thing has a lot of risks.
Bob
-
I think you need a hybrid solution. First of all the firmware needs to be able to detect something went wrong (which is always good so it can try and recover or die gracefully) and toggle a GPIO to trigger a logic analyzer or MSO. A Tektronix TLA715 has the ability to do repetitive acquisitions and save each acquisition to a different file or sample block consisting of 63 samples (the latter is a bit like segmented recording).
Another option is to use an MSO with data logging (like the GW Instek GDS2000A series). That way you can also look at things like the power supply in the analog domain. If you get one I'd get it on loan first and see if it suits your needs / works as advertised.
For the problem at hand however I'd try to figure out if there is a common cause like the power supply. I'd also add a stack check to the firmware, triple check variables which are shared with interrupts and use a tool to check the code for potential buffer overflows. Even better if you can compile/run the core routines on a PC and use a realtime checking tool like valgrind.
-
Rental?