Recommendations and Suggestions Please!

[tweakedx]

Reposting my inquiry here as I believe this is the most appropriate forum.

Hello Experts,

I need to build an application that sends 2M digital patterns (~1.6V at VIH) using 8 signal lines at 18MHz to a DUT.  The tool should then read the same 8 signal lines plus an additional signal (R). R would then be compared to an expected response (E). If R != E, a flag (b010) would be raised on a GPIO line and a notification saying at which pattern (out of the 2M) didn't match. This step then needs to be repeated at higher frequencies up to 22MHz. It would be preferable that the tool also has a variable power-supply from 0.8V to 1.9V that can provide around 20mA to power the DUT).

The loading of the pattern could either be in csv or txt. I can match the required format by Waveforms. Scripting language preference to send the pattern, check the data, then trigger the flag would be either C or Python.

Is this possible using the Digital Discovery? If not, what tool would you recommend?

Thanks!

[JColvin]

Hi @tweakedx,

The Digital Discovery can do some of what you are asking, though some of it is not possible (I'll start with the latter in case that is a deal breaker).

The variable digital power supplies on the Digital Discovery (which correspondingly dictate the logic voltage levels used by the I/O pins) only operate between 1.2 V and 3.3 V, so it cannot provide the 0.8 V that you mentioned (though the supplies can provide up to 100 mA of current: https://digilent.com/reference/test-and-measurement/digital-discovery/specifications)

Sending and measuring the 2 million digital patterns will not be an issue for the Digital Discovery with its Play Mode on the Logic Analyzer instrument as that uses the DDR memory present on the device and can operate at up to 100 MHz with up to 100 million samples.

The tricky bit will be the implementation of comparison of the generated pattern to a known pattern. I have a few questions regarding your pattern / planned setup that would affect the feasibility.

- Is your pattern of a consistent style (a set of 4 words that get generated with known start conditions / lengths) or is the pattern much more arbitrary where the difference could be anywhere in the sequence?
- Would there be more than one difference in the pattern (so the possibility of generating a trigger condition once a difference is detected is a possibility)?
- Does the difference need to be detected live (as the pattern is generating)? Or could it be detected upon completion of the data? Or after the fact entirely?

Thanks,
JColvin

[tweakedx]

Dear JColvin,

Appreciate your response! The pattern could happen anywhere in the sequence. We are talking about just single bits here. The difference would most likely be 3-5 bits across the 2M. It is preferred to have both options to trigger during runtime (live) and trigger after date completion.