'747 and Proportional Stuff

[rr]

There is an interrupt that drives a counter. Ram loc L0070 is the counter. It
is inc'd at 160Hz. Once it reaches 160 (L0070 that is), it is reset to zero, and

the 1-sec counter is inc'd.

Fuel and ignition split the 160Hz, each doing 80Hz. There are times that the
counter is tested to see if it's time to do another code section. After the 80Hz

fuel is done, there are two code blocks (more fuel), that each run on every
other 80Hz fuel tick. The PID is one of them.

Other blocks of code run at 10Hz, and I think 20Hz, depending upon
the requirements. IE: coolant is done on one of the slow loops.

BobR.

Garfield Willis wrote:

> On Thu, 13 Jul 2000 22:40:36 -0400, rr <RRauscher at nni.com> wrote:
>
> >(been busy)
>
> That's OK, it was worth the wait. :)
>
> >The PID loop is executed 40 times a second (every 25 msec).  The prop
> >duration is a lookup as described. Note that the '747 tables for duration
> >is in seconds based on: TablVal = sec's * 40.
>
> Ooops, well I was only off by a binary order of magnitude! :)  Yup,
> gotcha; yes I WAS asking about the PID loop execution period. In these
> ECM's, is that time period interrupt driven, or is there a timed major
> loop with a counter to maintain periodicity? Just wondering if the
> periods show up plainly somewhere in the code, either as a counter value
> that's read or as a timed interrupt handler that's set up. Not a biggie,
> just curious. Been awhile since I played with real-time code.
>
> Main thing is I have the period of computation now, tank you tank you;
> would sure like to get some time to put some flesh on dem bones and
> build up a model of the PID control loop to play with. So many
> eXperiments, so little time. Ya know there are books written on how to
> tune PID loops, if you can create an accurate system model, but I'm such
> a neophyte at this, I couldn't really tell ya at this point if such a
> venture would pay off with anything useful. Sure like to sniff around at
> it to see, tho. :)
>
> Gar
>
>

----------------------------------------------------------------------------
To unsubscribe from gmecm, send "unsubscribe gmecm" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org