I've been looking at this on the scope and *may* have found the issue.

There is no separate chip-select line to the AD8400 dimmer chip, as we ran out of pins on the PIC. So, the CS is driven by an unused monostable from the clock line of the dimmer - when there is a clock, CS goes low and when the clock is absent for a while it goes high again.

There appears to be an issue that the R/C timing circuit for this is a bit close to the edge - ie, if the PIC is running slightly slow there will be a phantom rising edge on CS halfway through the dimmer setting byte - this is only ~20ns wide, but the dimmer chip is very fast and catches it.

The solution appears to be either to change the R/C circuit (hard, as it's under the display) or to add capacitance to the CS line to hide the glitch. On *most* mk2a players, the single chip on the back of the display board is a DIP 8400 (we couldn't source SMT ones for love nor money - they were used in some mobile phones, allegedly).

Anyway, if someone with this problem, a 10nF capacitor (SMT is ideal, very neat) and electronics knowledge/a soldering iron can try fitting it between pins 2 & 3 (ground and CS) of the AD8400 - ie, the middle 2 pins of the chip closest to the top of the display board, and see if this helps them. The one unit we've had returned which would repeatedly show this problem was fixed by this capacitor addition.

We're awaiting some more confirmation that this does address the issue before we do any warranty work.

Hugo