The Sinclair QL Forum

I have been working on patching the Minerva I2C driver to work with the Q68.

This is my first working version (00.4). But I don't have any I2C devices to test it with other than the Q68's built in RTC chip.

Is anyone prepared to do some testing with external I2C devices to see if it is working OK.

At the moment the driver is only running at a clock speed of about 34KHz. The next job is to try to get it up to the standard 100KHz

This test program will set and read the Q68's Real Time Clock

There are some 8 bit parallel I2C breakout boards on ebay - https://www.ebay.co.uk/itm/273468863517 ... SwuzFgR-Be

That may be useful for testing - you can have several of these on the bus. Also - to protect the Q68 - you may want one of the isolated buffer boards that's available for I2C - I think they work upto 400KHz SCL clock - so should work for most boards.

Thank you Martin! I looked onto this a few years ago, but the task was too difficult for me. Does it runs only from normal memory or do you need to use some of the Q68 fast memory ?

François

Hi,

Looks great, excellent software, as usual from Martin.

I was going to assemble similar boards as Prof mentioned to emulate TF Services I2C interfaces.

There are a lot of good boards out there already - at reasonably cheap prices - I picked up a few good I2C boards when I ordered a jumbo experimenters pack for Arduino - came with all sorts of goodies, but did also include some interesting I2C devices - like temp / humidity sensor, I2C display, RTC, and some general I/O devices. Was a kit of over 100 parts in total.

FrancoisLanciault wrote:Does it runs only from normal memory or do you need to use some of the Q68 fast memory ?

Normal memory. I have not yet tried using fast memory. It may go a bit faster, but it uses the Q68 hardware timer for the serial clock. Which would limit it.

It uses a subroutine to introduce delays to time the serial clock and data reading. Initially just to get it working, I disregarded processor instruction times and set the delays to be about 4.5uS serial clock pulses. Which would be about 100KHz. But I found that the clock pulses came out about three times longer than wanted. Hence it 's running at about a third of the desired speed.

Yesterday I had a play with the delay routine, figuring out the processor instruction times overhead. I found that if I just had an empty delay subroutine, with just the return. Then the serial clock was pushing 200KHz. At this speed I could still read the RTC. But on my Q68 the rise time of the SDA line was starting to get in the way. The rise time of the SCL was about 0.2uS, but the rise time of the SDA line was about 2.2uS

I am going to try some trial and error tests of different delay counts to try to get the SCL to about 100KHz.

Hi Martin,

excellent job! The rise time on the SDA line comes from the 10 K Pullup on the Q68 which is typical for 100 kHz standard mode. It is possible to add smaller values like 2.2 K externally if 400 kHz fast mode is desired.

All the best
Peter

Version 0.06
After some speed optimization of the timing delay routine, The driver now runs at the I2C standard speed of 100KHz in normal RAM.
Interrupts are now disabled during the I2C_IO function.

Hi Martin, many thanks for the good work. One correction to the manual: Pullups of 2.2 K or larger to 5V are allowed.

Here's version 0.07

This version auto-configures itself for the Q68's normal or fast memory.

As there are no other changes, I have not bothered about a manual update.