[rtl] TWD idle when NoACK during reading #1158
Conversation
|
Explanation: this prevents the MSB to pull the SDA line down and thus preventing the stop command to be recognized. |
|
Ah I see. So we should also check for a STOP condition in the ACK time slot? Shouldn't we do this for read and write accesses? |
|
I guess checking for the stop anytime makes sense, but this problem might occur anyway since in my example the stop signal comes afterwards: |
| if (ctrl.enable = '0') then -- disabled? | ||
| if ( | ||
| (ctrl.enable = '0') or -- disabled? | ||
| ((engine.cmd = '1') and (smp.scl_rise = '1') and (smp.sda = '1') ) -- or no ACK while read? |
There was a problem hiding this comment.
Basically, cou are checking for a STOP condition here, right?
Can't we use the smp.stop signal?
((engine.cmd = '1') and (smp.stop = '1')) -- STOP while read?There was a problem hiding this comment.
No, as the stop signal is a SDA rise with SCL being one and not the other way around. This is more like reading the 9th bit: is the ack set or not? If not => I'm done (technically waiting for the stop), if yes => I send the next byte.
|
Something isn't right with the TWD... 🙈 |
|
I think there are several design flaws in the TWD 🙈 |
|
Works with #1161, thanks! |
|
Great to hear and thanks for the feedback! |
Fixes #1157 and to my understanding should be TWI compliant.