[OpenTRV-dev] Residual Receive Strangeness

[Gary Gladman]

That is not so simple.

I have looked into the problem a bit further and I had a version of the 
code that had ZERO !RXerr F2 but plenty of !RXerr F5.

As I understand the F2 issue at the moment it is a tension between RX 
FIFO almost full mark and the rate at which PollIO is called. As I 
increase the poll rate by various means the F2's disappear i.e. the RX 
terminates sooner and the FIFO overfills less and thus no longer loops 
around and corrupts the front of the FIFO (!?). Clearly this approach 
has power implications.

The F5 issue seems to be partially related to the RX FIFO almost full 
mark which if I understand correctly does not strictly account for JSON 
status receptions. As the poll rate increases less of the JSON TX is 
received and thus its tail is increasingly truncated resulting in 
increasing F5's. Simply increasing the RX FIFO almost full mark seems to 
recreate the F2 issues which requires an increase in the poll rate.

However, as the TX get larger there comes a point at which some other 
strangeness kicks in and the transmissions seem incomplete for another 
reason as yet unidentified.

There is no simple answer - you might consider increasing the high water 
mark, accounting for and reducing the max size of JSON stats and 
increase the poll rate to prevent RX overflow. However, as the high 
water mark increases less headroom is left leaving less time available 
to terminate the RX. This is likely to involve a lot of trial and error.

Ultimately, the RX FIFO cannot be allowed to overflow by staying in 
receive too long once a preamble and sync has been seen as the FIFO 
(over) fills (and corrupts) until RX mode is exited it seems.

RX Polling may simply be the wrong approach.

Regards Gary

On 12/07/2015 12:07, Damon Hart-Davis wrote:
>> On 12 Jul 2015, at 00:04, gary.gladman at talktalk.net wrote:
>>
>> Hi,
>>
>> I am now confident that I have evidence that receive overrun is the primary cause of !RXerr F2.
>>
>> Regards Gary
>>
> Very interesting; what do you think could/should be done to reduce or work round the problem?
>
> Rgds
>
> Damon
>
> _______________________________________________
> OpenTRV-dev mailing list
> OpenTRV-dev at lists.opentrv.org.uk
> http://lists.opentrv.org.uk/listinfo/opentrv-dev
>

-- 
Gary Gladman B.Sc(Hons), CEng, MBCS, CITP
LinkedIn: http://uk.linkedin.com/pub/gary-gladman/32/309/81b
tel: +44 (0)7941 107 280    (mobile)
tel: +44 (0)1245 259 746    (home)
mailto: gary*gladman.name   <NEW * -> @>
Please don't print this email unless you really need to.