This reporter has designed/produced & sold enough of these character Lcd modules - should be able to resolve this simple issue. Here's what you report: (note that I've made time/effort to better organize/present)
MCU output: Display shows...
A 0x41 C 0x43 Error, 2 bits over
B 0x42 B 0x42 Correct
C 0x43 C 0x43 Correct
D 0x44 F 0x46 Error, 2 bits over (again)
Of course - more data would speed/ease this analysis. Yet - from this limited data set - it seems safe to conclude that bit D1 (value =2) is indeed, "stuck high."
Here's the justification: 0x41 does not order bit D1 high - thus should D1 be "stuck" high - your 0x41 + bit D1 = 0x43.
Next, 0x42 does order bit D1 high - and as it's "stuck" high - no harm, no foul. This displays correctly.
Onto 0x43 - now this again sets bit D1 (and bit D0) high - and displays C correctly. D1 stuck high has no impact.
And finally 0x44 - does not order bit D1 high - your stuck D1 does that nicely - and voila 0x44 + bit D1 = 0x46 = F.
Q.E.D. (one final weasel/hedge - HD44780 & most clones pull-up their D-Bus. Thus - your "stuck" bit may be evidence of an "open" circuit between MCU's D1 & Lcd's D1.)
The effect remains the same - the cure though is different! Far easier to imagine your shorting bit D1 to D0 or D2 - how you short it to 3V3 - bit harder. Thus - I lean to an "open" between your MCU->Lcd's D1 - preventing the MCU from overcoming the "weak pull up" resident w/in the Lcd controller...
Last Mohican point(s) - Petrei correctly recalled ~40uS as character to character delay. But use a safety factor - 2x seems reasonable. And the CLS command (0x01) and Home (0x02) may require a bit over 1mS! (use 2mS) It's good form to have 0x01 as the final entry w/in your Lcd initialization listing. (yours appeared earlier)
A nice, Verify answertick seems proper/fitting & should close this topic. (cards/letters/applause also welcome)
