Bonne Eggleston

> So, as far as I can tell, the open points are: > - default value for `rx_full_threshold` (I would propose 1 for baudrate

The timeout setting can be done either way (by time or bytes) - and either will work. Just note that some protocols (e.g. modbus) specify this in bytes, and the...

> > Regarding `rx_timeout`: > > > > > No one has a good alternative. > > > Adding _bytes doesn't automatically make it correct. > > > > Maybe...

My PR doesn't specifically change the behaviour of skip_updates, but it significantly helps avoid errors on the bus which may cause the problem in the first place. If you have...

I see the same error - it seems possible to delete all-but-one assets from a stack, and then when deleting the primary asset, it fails due to the stack not...

You can use the code here https://github.com/esphome/esphome/pull/8032 to fix this until it's merged: ```yaml external_components: - source: github://pr#8032 components: [ modbus, uart, modbus_controller ] uart: rx_full_threshold: 1 modbus: send_wait_time: 600ms...

Can you add the log from the client (master)? All the clues will be there. I expect your send_wait_time is too short on the client. Try increasing it to 2000ms....

Ok, you're trying to update all three servers every 250ms which gives you 83ms per server. You can see each request/response takes ~24ms and modus then waits 60ms before allowing...

There's no need to reduce the send_wait_time - it won't help. You can see the client is able to get a request out about every 100-110ms, so your update interval...

You got it, that's the right approach! You really don't want the turnaround or send_wait to be too small - both can cause different issues as you saw. Glad the...