SPI overlap
Posted: Tue May 24, 2022 9:37 amI have been playing with SPI overlap lately. It's not that I really need it, but it can spare some pins when really in need and I can't stand it I couldn't get it working earlier.
I tried it about two years ago and then, whatever I did, the esp8266 would not boot. In hindsight I think that was because the device I connected (Eastrising display) does not seem to release the MISO line. Which is resolved now as I implemented an MISO-less-mode (wait for a certain period instead of polling status).
I am now testing with ilitek based displays, largely comparable but they're not using MISO at all, anyway.
With some sample code from Espressif I now had it sort of working, but some thinks remain strange and I can't explain.
- the signals are not moved to other pins but duplicated. I had the display still connected to the "regular" HSPI pins and it just remained working. So apparently you actually need to reassign the original HSPI pins to something like output or input.
- chip-select works (CS2 -> GPIO 0), MOSI works, well DCX works because it's outside from SPI anyway, but I can't get the SCK (clock) working. MISO may or may not work, didn't test. I attached my simple protocol analyzer and it can't make head or tail from the signal. This usually means it's very high frequency, possible 80 Mhz, it can only use 10 Mhz and below.
So that means I'd still need to take SCK from the original HSPI pins (and then it works). Little sense in that then, it would only spare one pin.
Anyone familiar with this matter?
I tried it about two years ago and then, whatever I did, the esp8266 would not boot. In hindsight I think that was because the device I connected (Eastrising display) does not seem to release the MISO line. Which is resolved now as I implemented an MISO-less-mode (wait for a certain period instead of polling status).
I am now testing with ilitek based displays, largely comparable but they're not using MISO at all, anyway.
With some sample code from Espressif I now had it sort of working, but some thinks remain strange and I can't explain.
- the signals are not moved to other pins but duplicated. I had the display still connected to the "regular" HSPI pins and it just remained working. So apparently you actually need to reassign the original HSPI pins to something like output or input.
- chip-select works (CS2 -> GPIO 0), MOSI works, well DCX works because it's outside from SPI anyway, but I can't get the SCK (clock) working. MISO may or may not work, didn't test. I attached my simple protocol analyzer and it can't make head or tail from the signal. This usually means it's very high frequency, possible 80 Mhz, it can only use 10 Mhz and below.
So that means I'd still need to take SCK from the original HSPI pins (and then it works). Little sense in that then, it would only spare one pin.
Anyone familiar with this matter?