Re: LXI, IEEE1588 and TTP




Paul E. Bennett wrote:
> Does anyone on here have any experiences to relate about implementing any
> of the above protocols. How much effort did it take, how much processor and
> memory resource and how accurate a timing correlation can be achieved
> across nodes on the network. I would also be interested in knowing if some
> of the schemes used other timing synchronisation distribution schemes in
> addition to these protocols.
>
> I am looking at a situation with approximately 50,000 sensors, some time
> critical control requirements and and a need to ensure that all the data is
> on a closely relateable timebase.

For an accurate time synchronisation, the method I have used
on party line protocols was based on measuring the time with the
following procedure:

A sends a synchronisation frame to B and starts a timer (1) when
a specific character is sent
B reads the time at wich it receives the character and starts
a timer (2)
B sends a reply frame, reads the timer (2) when the equivalent
character is sent and writes its value in the frame as it goes.
A reads the timer (1) when it receives the reply, substracts the time
value (2) in the frame and divides the result by two in order
to compute the time it takes to send a frame.

The timers should be started in an ISR upon transmit interrupt

When you send a time synchronisation frame you add the time it takes
to send a frame to your local time. You read the time when you send
the specific character, in the ISR.

I could achieve 80us accuracy on a fieldbus with this method and 1ms
accuracy on long distances at 1200bds

.



Relevant Pages

  • Re: Turning off some messages - is it possible? Is it unnecessary?
    ... 10ms per frame. ... 20-25cps so at the most you can only get one character every 40ms or so. ... the first thing I did was create a timer to ... > O.K. DirectInput for getting controller input, ...
    (microsoft.public.vc.mfc)
  • Re: Scrolling text in a stationary label help needed
    ... single pixel scroll just once in each Timer event, ... and displaying the "scrolling text in a transparent picture box" if it ... exactly correspond to the video "frame rate" (75 frames per second in the ... rest of your code could use that 3.3 milliseconds to do something else. ...
    (microsoft.public.vb.general.discussion)
  • Re: How to inform NDIS the Max Frame Size?
    ... Pause and restart only stop data flow temporarily so they do not ... I guess it would be possible to require NDIS 6.0 protocols to re-query the ... > legacy protocols that would not honour any new dynamic frame size status ... But are there reasons for not adding such a status ...
    (microsoft.public.development.device.drivers)
  • Re: Turning off some messages - is it possible? Is it unnecessary?
    ... However, in the gaming domain, responsiveness (not efficiency) is the problem. ... >10ms per frame. ... the first thing I did was create a timer to ... MVP Tips: http://www.flounder.com/mvp_tips.htm ...
    (microsoft.public.vc.mfc)
  • Re: How to inform NDIS the Max Frame Size?
    ... I understand there are historical reasons. ... legacy protocols that would not honour any new dynamic frame size status ... Thus if such legacy protocols are ... message giving a hint to the protocol to reduce the frame size dynamically. ...
    (microsoft.public.development.device.drivers)