I have to start over - Write Entire is SCARY!

[eficalibrator]

The errors become increasingly more important because as pulsewidth decreases the non-linear region comprises a greater proportion of total fuel delivery.

I'm not offended at all. I'm glad you seem the same.

Yes. On a practical level for the example you show above, the injector remains mostly linear down to about 7mg/shot. Since this is well below our minimum steady state fuel mass at idle or cruise, errors in the even smaller pulse widths become less relevant.

On the other hand, I have tested injectors from other sellers who provided "data" on them that varied by as much as 40% at the idle fuel mass! That's not to say everyone else's measurements are bad. When I measured Injector Dynamics units, I got within 2% of the rating across the board.

[Mr. Black]

Checking in to see how this turns out.

Wow jumbo 5 pages?? lol.

[JumboShrimp]

Hey Mr B!
Not complaining - Actually LOVE attempting to understand this all. I find it fascinating.

[SiriusC1024]

On the other hand, I have tested injectors from other sellers who provided "data" on them that varied by as much as 40% at the idle fuel mass! That's not to say everyone else's measurements are bad. When I measured Injector Dynamics units, I got within 2% of the rating across the board.

Oh that doesn't surprise me one bit. Thing is thousands of people have also tuned with that data. I guess what I'm getting at is you claim improvements over other's work and even your own previous work. Then I see something like the SPA that doesn't make sense. I just want perfect data and not good enough data. It's been good enough data. That's all.

Tuning large injectors can be a pain. Low resolution on the custom OS VE table coupled with low pulsewidth injector events makes things difficult.

[SiriusC1024]

Mr. Banish, I'd like to be clear with you and everyone reading. I'm not trying to shit on your work. Actually, I think you do stellar work. The fact that you do injector characterization for FIConnection is the reason I recommend them to people. Thanks for stopping in, and good luck with the tune.

[eficalibrator]

Oh that doesn't surprise me one bit. Thing is thousands of people have also tuned with that data. I guess what I'm getting at is you claim improvements over other's work and even your own previous work. Then I see something like the SPA that doesn't make sense. I just want perfect data and not good enough data. It's been good enough data. That's all.

Tuning large injectors can be a pain. Low resolution on the custom OS VE table coupled with low pulsewidth injector events makes things difficult.

I'm not offended. If my data didn't follow the science, I'd kick my own ass.

Here's an example of what I measure for short pulses. This was on a set of BIG injectors (~1600cc/min @ 4bar).
SPA capture.JPG

Notice how they actually behave pretty good down to about 9mg/shot. Below that, they diverge as the only measurable point I could grab was at 1.0ms and less than 2mg/shot. Even that point was a little noisy. We wouldn't really want to operate there if we can avoid it, but with the right short pulse adders lining up down to that point, we could theoretically be OK in the ~6mg/shot range. But even that is well below the typical idle fuel mass requirements, so it's kind of a moot point.

Arguing about short pulse adders for masses below 2mg/shot on this injector is ridiculous. We wouldn't really be there. So it doesn't matter if I use the same 0.225ms adder or a zero below that point, we're still not going to get any consistent fuel mass from the injector.

Here's the same graph from a stock LS3/LS7 Injector:
LSA SPA.JPG
Again, see how they're pretty linear all the way below 5mg/shot(1.2ms)? The injector basically didn't open at all at 1.0ms. We use the short pulse adder table to interpolate between these points (and make the TINY corrections between 1.2-4.0ms), but the practical application says that it's not that critical in this particular case.

Yes, there are other injectors out there with more non-linear behavior than these two examples. By running the same tests, we expose that and calculate the necessary SPA numbers. Yes, there can be a variation between the "average" data for a part number and one particular set of matched injectors. If that variation is of concern, the solution is to have your exact set tested and get the data for them rather than using the population average. It can make a bigger difference in some cases than others. Since I own the bench, of course I prefer to use the EXACT data when tuning my car or something for another client here in person.

YMMV


...meanwhile, Mike's car is making solid progress on remote tuning. Not my favorite way to collect data, but we are making big strides and will probably have it running well enough for him to drive it on road trips soon.

[Mr. Black]

...meanwhile, Mike's car is making solid progress on remote tuning. Not my favorite way to collect data, but we are making big strides and will probably have it running well enough for him to drive it on road trips soon.

IMO if there's no feedback on what you are actually doing there's almost no point in posting.

No disrespect intended.

[SiriusC1024]

I'm not offended. If my data didn't follow the science, I'd kick my own ass.

What you're saying makes complete sense now. Know what I was doing wrong? When I looked at the SPA graph I always thought integral. No, it's literally a point lol. The Y-axis isn't mass/t...about that ass kicking, huh?

That's why I thought every bit of the SPA had to be mapped. Thanks for helping me out with this