Injector control

[Chris Conlon]

Greg wrote:

> Anywayz--in the article, they mentioned that the FP unit has an injector
> pulse width RESOLUTION of 8 MICRO seconds--and bragged on this being far
> better than a resolution of 64 mi. s. I kinda assume this to mean that the

This is one issue I've thought a little bit about.

When you talk about the shortest repeatable "on" time, you're
looking mostly at the *jitter* in the injector open and close
time. Or to look at it the other way, if the injector took
500uSec +/- 1uSec to open and close, you could still get very
repeatable open times of much less than 1 msec just by starting
to close it before it's fully opened. Your fuel delivery would
be nonlinear wrt pulse width, but this is true of all short
"on" pulses where the partially-opened and partially-closed state
makes up a significant part of the "on" time.

As for what the jitter *is*, I have little to no idea, though if
a 1 msec pulse can be repeated precisely, that does set an upper
limit on what the jitter could be.

The only other thing I can think of offhand that'd mess you up
would be the lack of good power supply regulation, and good
current control in the injector drive circuit. (This is probably
made trickier by the injector temperature varying over such a
wide range, the inductive nature of the load, and possibly the
driver circuit used.)

> But the
> light bulb that came on is over the fact that an 8 mi sec adjustment to a
> pulse width of 1 ms. would make a 0.8% change in amount of fuel
> delivered--a reasonably fine adjustment increment when in a search for a
> good idle. Whereas a 64 mi sec adjustment to a 1 ms pulse width would only
> give a 6.4% change in amount of fuel delivered--way to coarse a

Well at this point you're certainly in that grey area where the
transition times are a meaningful percentage of the total on time.
It'd be interesting to see exactly how things work under those
conditions. If I had an injector flowbench I might even set it
up with a hi-res pulse generator to try and measure this behavior.
This would let you measure the nonlinearity of injector flow at
very short pulse widths under various (electrical) conditions.


FWIW I'm building a system that will eventually control injectors.
I had not decided on a final value for injector resolution, but if
it's practical, I'll try to get it down to 1 uSec. Ain't 16 bit
counters a pain? ;)

   Chris C.