How do you guys get a bead on injector dead time (latency, opening time) without a test bench? Sure, it would be swell to do it on a test bench with a variable power supply, but I don't have one, and I sure as hell am not pulling all the fueling parts, Megasquirt, and wiring out of the car to build one.
I thought about trying to dangle the still-attached rail plus injectors over a catch basin inside the engine bay, and then do some testing at different pulse widths, decant into a graduated cylinder or weigh the gas on a postal scale or something, but I have visions of struggling with the cramped quarters, getting explosive fuel everywhere, and doubting if I am getting meaningful numbers. 
then I stumbled across this technique:
http://www.miataturbo.net/ecus-tuning-54/how-find-your-injector-dead-time-56061/
In a nutshell, these guys idle while logging, and vary injector dead time in steps until the AFRs vary from 16:1 down to 11:1. Then they plot MAP divided by AFR vs injector pulse width and find the best line, and plot the intercept.
I can understand how logging while varying dead time could be useful, but am struggling to wrap my head around how MAP divided by AFR is what you want to compare to pulse width. Does anyone understand the theory better than me?
Alternately, do any of you have success stories of measuring your injector dead time on the car? What technique did you use?
If you assume that the VE curve is flat (which is a fairly reasonable assumption over the narrow range of RPMs and MAPs in the test), then MAP tells you the mass of the air in cylinder, and AFR tells the ratio of air to fuel, so by dividing them you wind up with the mass of the fuel. Graphing that against injector pulse width will tell you the mass of fuel per per injector on-time, and the zero-intercept is the dead time.
I get the car to a smooth stable idle and then go between 1 and 2 squirts/cycle. If the dead time is correct, measured AFR will not change. If the dead time is too high, it will be richer on 2 squirts/cycle. If the dead time is too low, it will be leaner.
Thanks, I knew this place would point me the right direction! Gotta love the GRM forums. 
codrus, that makes total sense! I was so hung up on the fact that the measurement with MAP is pressure, that I lost sight of the fact that for our purposes it represents (maybe slightly imperfectly) the mass of air. I was sitting here struggling to understand why dividing pressure in the manifold by fuel ratio would net something useful! Now I feel a little silly.
Knurled, I had read some about the doubling/halving the squirts method, which seems like it would get you right in the ballpark. I was planning to use this method as a test to confirm what I measure using the AFR method from the turbo Miata forum.
I got a log tonight while idling warm, with fixed advance. The injectors are RC Engineering high impedence 270cc/min. ECU is MS3Pro. The 11:1 AFR ballpark was about 1.6ms dead time, and the 16:1 AFR ballpark was acheived with 0.2ms dead time. I went by 0.1ms increments for the meat of the test, with 0.2ms increments at the outer edges. Voltage was a rock-solid 13.9V for the entire log. A short doubling the squirts test points towards the 0.7-0.8ms range as being about my dead time. I plan to crunch some numbers on the log once I finish wrestling it into a spreadsheet and see what the log suggests.
My TunerStudio and MegaLogViewer are on my Mac laptop, and Numbers really doesn't like MegaLogViewer log files, so I am currently moving the log files to a PC, opening them in Excel, then saving them as an Excel file to transfer back to the Mac to open in Numbers for manipulation...there's gotta be an easier way!
Knurled wrote:
I get the car to a smooth stable idle and then go between 1 and 2 squirts/cycle. If the dead time is correct, measured AFR will not change. If the dead time is too high, it will be richer on 2 squirts/cycle. If the dead time is too low, it will be leaner.
This, and I would recommend testing at 2000 RPM in neutral rather than idle to avoid effects from reversion.
