Memory timings tuning

[Lardman]

Any chance of a little guidance from the collective ?

I'm currently working a UFD that was DOA on the USB side, after working out the layout its a ec1aa8de with 2 parts. Recognised in pc3k Flash and read fully with the default settings. Device appears to be the same as http://flash-extractor.com/library/Mono ... a8_de__1x2

Part 1 reads well after ECC/RR around 200mb of invalid data out of 16GB.
Part 2 is total garbage. It can't even determine ECC to do RR map.

I tried my usual voltage, timing mods and temperature changes, but as I was getting nowhere fast I kicked it back to TS. TS has had a look and think that things can be improved with continued adjustment of the timings.

I've adjusted things up and down, left and right but I'm not seeing any improvement in data quality, but nor am I follow how the PC3K flash timing relate to what Im seeing out of my logic analyser which appear to be wildly different. Can anyone point me towards some actual information to help interpret my LA results into PC3K.

Or if anyone has other suggestions that I can try - I have a lot of time invested in this one and I'd really like to get something back to the client.

Attachments

[csava]

The reader of PC3K always provides the worst reading performance, which is caused by hardware defects. Changing the timing can usually help, but not much. I like using PC3K to analyze dumps, but I don’t use it for reading chips. Neither reading nor RR are handled well by PC3K. RR can be improved through software updates, but the defect in the reader can only be addressed by waiting for ACE to identify the problem and produce a new reader. However, it is clear that ACE is not currently focusing its R&D efforts on 3K flash. I designed and built my own reader, so I am well aware of its shortcomings, but I lack the motivation to improve it. FE and VNR are usually better options for reading chips, especially if you are familiar with chip configurations, and the work can easily be switched between the three.

[pepe]

i swapped the buffers for stronger ones on the wire adapter. However, before swapping my dump looked different, as in, the beginning of the pages were ok, the end was unstable due to clocking problems.

[Lardman]

The reader of PC3K always provides the worst reading performance.

These are things I couldn't establish until after purchase - unfortunately and Im kind of stuck with what I have. FE's have disappeared and Rusolut aren't sending me VNR FOC anytime soon. As a simple "user of the tool" I'm not able to tell if the problems I see are me, just another bad case or the tools themselves.

I can't think of another situation where I've had nothing at all read without there being an actual problem though. I thought initially that I must have identified the wrong line as CE1.

i swapped the buffers for stronger ones on the wire adapter.

I'm on the multi-board with monolith wiring adapters - but I have tried genuine Acelabs, multi-com uni adapter and also some cheap AE clone ones - everything is pretty much the same. Wire lengths are also as short as I can make them.

[Gregory]

This is a problem with the power supply, i.e. I mean a problem with the VSS/GND ground. Try soldering additional wires to ground on the monolith to GND, not only the one connected to the USB connector.

[Lardman]

Try soldering additional wires to ground on the monolith to GND

I'll give it a go, currently a 4hr wait to read with the latest timings.

[Lardman]

Adding an additional GND on the mesh didn't improve things. Any other suggestions ?

[csava]

If you only have a PC3K reader, try using a Spider adapter, which can provide a better reading experience.