This article describes briefly how we do injector testing. I can’t tell you all the secrets, because some of them are not mine to share I’m sorry to say.
Firstly I want to give a big thank you to Paul Yaw from Injector Dynamics who has helped me understand a lot of the issues in testing injectors. Thank you!
The setup to test the injectors is mechanically fairly simple. Test fluid is stored in a conventional fuel cell. This goes through a 100 micron filter, and then into a high pressure EFI pump.
Fuel tank (on right), fuel filter, then fuel pump. Power supply at the bottom, and modified ECU at the top left. At the back you can see a tank, which is filled with pressurized air.
The liquid then goes through a 50 micron filter, before going to an FPR2000 pressure regulator. From there, the fuel is filtered again using a 10 micron filter, and after that it goes into the injector test rail, after which is the final pressure sensor. After the fuel goes through the injector, it drains back into the tank. Of course there is some measurement in there somewhere but that’s the secret stuff I can’t tell you about.
Side-feed Toyota test rail with Dash 6 fittings at each end
Side-feed Toyota test rail with Dash 6 fittings at each end. It is a bit makeshift because there are many rails for different manufacturers and building a solid test platform with dedicated geometry for each injector to be tested wasn’t viable for the volume we do.
14mm top-feed injector test rail with Dash 6 fittings at each end
We have several test rails for testing different types of injectors, for example side-feed Toyota, Nissan and Mazda injectors, 14mm top-feed injectors and 11mm top-feed injectors.
The pressure of the fuel is controlled by adjusting the air pressure fed to the pressure regulator in the reference port. An air tank (shown in the first picture) contains pressurised air, and two LPG injectors are used to control its pressure; one to admit pressure from an air compressor, and the other to allow air to vent to the atmosphere.
Electrically, there is a 15V power supply which powers the pump and the air control solenoids. There is a separate 19V supply which provides power to the ECU via a switching regulator, whose feedback voltage can be modified by the ECU. The ECU can control the supply voltage over a range from 7 to 16V.
The ECU also controls the fuel pump on/off, the air pressure control solenoids, and the injector under test. It also gets feedback from multiple pressure sensors at different parts of the hydraulic circuit to detect filter failures, it measures the level in the tank as a run-dry protection function, and it measures the amount of fuel delivered by the injector.
The way the test progresses is that the ECU first starts by setting the target fuel pressure and supply voltage. The fuel pump is enabled, and once pressure is built-up, the ECU attempts to regulate the fuel pressure using the air control solenoids. Once this is achieved and everything is stabilised, the ECU begins pulsing the fuel injector.
The ECU pulses the injector for over 50 different pulse widths so that an accurate curve of fuel delivered vs on-time can be plotted for the given battery voltage and fuel pressure. Once that curve has been plotted, the ECU automatically advances to the next battery voltage and restarts the test at the same pressure.
Once that pressure has been finished, the overall flow rate can be determined, and then the dead time at each voltage can also be calculated. The ECU then moves on to the next pressure, and the software plots the curves as they’re measured as well as an ongoing flow rate vs pressure curve so that you can check for any outliers, that might indicate that the experiment needs to be repeated at that pressure.
Overall it takes usually between 1-3 days to fully characterise an injector at the 10 different voltages and 10 different pressures that we measure at; the time variation depends on how “well behaved” the injector is. For example if it creates a lot of water hammer, then often this excites resonances in the pressure reg so instead we need to run it at a fixed frequency, which is much slower.
The end result is for each pressure, we have flow rate and 10 different dead times, and heaps of data points on each which can give us confidence in the result.
Thank you and happy learning!