I have version 2.9.1 (I believe the latest, I downloaded it not long ago) of the MD_MAX72xx lib and yes it is in hardware SPI mode. I have a 10k pulldown (or pullup, either works to keep CS/LOAD stable) on the CS/LOAD pin 10. I tried with or without (started without initially), but the issue remained. Will try with another 10K as this one is less than 1/4w and it's legs are so thin they may not make always reliable contact with the protoboard pinch contacts. Maybe this can cause noise and generate unpredictable results. However this doesn't explain why without touching anything and simply changing from 6-5 or 5-6 modules in the code the failure is consistent and repeatable every time - I doubt it's a bounce or contact pulldown/pullup issue. Will try again with another resistor anyhow.
Cheers
- Update: just checked with another 10K resistor in another position, exactly the same behavior occurs as in the video when compiling with 5 and 6 modules.