rusefi.com

I need to re-use the OEM injectors because the engine must also continue to run with original ECU. They are current RENAULT injectors : Some pictures of the tools I use :
Power supply Scale Original fuel supply RUSEFI console gives MRE the necessary command to dispense a timed pulse with number of repetitions you want.
I try to dispense always the same amount of fuel (say 5000ms) with various number of pulses.
First shot 1 pulse of 5000ms
Second shot 5 pulses of 1000ms
Then 10 of 500ms and so on until the pulse is too short.
Each shot is weighted with the scale and registered in a spreadsheet.

Some results :
14v supply
5000ms gives 11.92 grams of fuel in 1 pulse
The same amount of fuel is obtained with 200 pulses of 25.800ms
500 pulses of 10.800ms
So at 14v the dead time is 0.800ms.

12v supply
500 pulses of 10.800ms gives only 11.50 grams
To go back to 11.90 grams you need 11.100ms
So the dead time at 12v is 1.100 ms

9v supply
500 pulses of 10.800ms gives only 10.38 grams to go back to 11.91 you need 12.050ms
So the dead time at 9volts is 2.050 ms
This voltage is the lower limit, the fuel pump struggles to stay alive, but the injectors deliver precise fueling if you correct with the right dead time.

Great information. Method sounds very repeatable. I would not be surprised to see packages show up in your mailbox with various injectors in them asking you to repeat your findings for different vehicles...

Here is a graph showing injector linearity (or lack of linearity).

I never thought it needs a 8 ms pulse to become linear !

How have these data been mesured :

The weight of a consant theoretical open time has been measured (here 10000ms).
IE : 10000 shots of 1.8 ms (0.8 ms lag time)
then 5000 shots of 2.8 ms
and so on until 500 shots of 20.8 ms

This graph is not destinated to measure injector lag.
It's the result you will obtain with a constant lag time compensation with these injectors. So you can see what will be the gain of a non linear compensation method. It shows also where the linear zone begins and how wide the non linear zone is.

EDIT : THIS METHOD IS NOT CORRECT
Constant wheight is good for scale accuracy but induces a lot of error due to high pulses number multiplying individual lag error.

In this article from this link https://www.enginelabs.com/news/clearing-up-the-confusion-in-modern-fuel-injector-technology/

I found the same data I measure.

I put these data in the same format I use :
This injector has the same behaviour as mine. So my data and measurements are perhaps correct.
EDIT : EVEN WITH FACTORY DATA THIS METHOD DOES NOT WORK

New tool to search an explanation to strange results above
Matthew and James helped me a lot on slack to analyse the data and made me understand that the constant wheight system amplifies small pulses measurements. To keep fuel wheight constant I use more and more pulses so the effect of injector lag is multiplied. This system is necessary with my electronic scale because I know it has linearity problems when you use it over full range. It is very accurate over a small range and fast to use.
The mechanical scale has no linearity problem over full range because it uses reference weights. But it's slow because you have to wheight twice (empty and full).

Here is a video of how fast the fuel evaporates in the bottle and in the scale, almost 1.5 mg in 1 min.

Some results : Looks pretty linear not ?

To summarize the result of these tests :

First you can measure static flow by flowing 1 pulse of 5000ms.
Calculate maximum pulse width for your application exemple 20ms
Weight fuel flowed by a constant number of pulses (use a high pulse count to have good statistical results - I use 1000) for the complete range of your injector use (say 1ms to 20ms)
Calculate and plot a graph of g/pulse versus pulse time
Extrapolate pulse time for 0 g/pulse - the resulting time will be your lag at this pressure and this voltage.




Special thanks to Matthew and James who corrected my errors in the first method I tried.

Next to do is installing an adjustable pressure regulator an explore pressure effect on lag.