LSswap

I built an automated fuel injector tester and characterization workbench. Unlike other testers I've seen, it can do a broad set of tests and send the results to a laptop, without having to read graduated cylinders. The accuracy of the measurements is around 100 times more accurate than graduated cylinders.

In addition, it can chart startup current and dead time as well as flow at various pulse widths and duty cycles. Each injector can be pulsed separately to get more accurate results.

A touch screen is used to select preprogrammed tests for a single injector or for all injectors, as well as to assist with filling the and emptying the fluid.

In the video above, the left injector is a siemans deka 80 and the rest are generic 1000cc. Even though it looks like all the injectors are pulsing at the same time, they are being turned on and off exactly as they would be in your engine. Before and after each test, the weight of the fluid is measured extremely accurately, to determine how much each injector delivered for that pulsewidth and number of revolutions.


Below is a dead time chart of the starting current in one of the 1000cc Injectors I tested. The Excel pivot chart shows when the pintle began opening ad when it settled fully open. Left side is the current through the injector and the bottom of the chart is the time since the injector was fired. It's in microseconds for accuracy.


These 1000cc injectors would barely fire at 1.5 millisecond pulse-width. They appeared to flow some by 1.7ms. The chart shows why. One would think that the split second that the injector is turned on, current would be on. But in the real world, the injector is an inductor. So current doesn't flow right away. There is an interesting dip in the current around 1.45 milliseconds. I believe the ball hasn't moved off the seat until around 1.27 ms. Once the pintle began to move, the current dropped to a low around 1.45ms. Then the pintle was rising until around 2.356ms when it reached maximum throw. There it looks like it bounced around 2.892ms and then settled full open around 3.24ms.


Here is another type of chart characterizing 3 different injectors varying pulsewidth from .7 to 3 milliseconds.
Injector 1 is siemans dekka 80 lb
2 is turbo dodge 2.5L stock low impedence
3 and 4 are generic 1000cc injectors.

y axis is milliliters


You can see that the low impedance injector started delivering fuel at around .7ms. The Siemans Deka 80 at around 1.1ms and the generic 1000cc at around 1.3ms.

I used this to match my 1000cc injectors in my C3 Vette. One of my friends was running 2000cc injectors and was having idle issues. I ran his injectors to help him solve the idle.

The computer controlling the tests and touch screen is and Arduino Mega, same computer I use for controlling my 4L80E transmission.

Comments and questions are welcome.

jonathans130

Hey, came across this while looking at various diy fuel injector test setups. I'd love to know more about the specifics of the setup (particularly the control circuit & the datalogging aspect) if you're willing to share. Would've sent a PM but not able to due to a fresh account haha. Thanks!

LSswap

I have no problem sharing code and schematics.

In general, there are two types of logging.
There is a .1 ohm resistor in series with the current supplied to the injectors. The voltage drop across that resistor is measured by the arduino every 25 microseconds to give you the current characteristics.
Each of the collection bottles sits on a 1 kilogram load cell with an HX711 A/D converter for 24 bit resolution. They are unbelievably sensitive and ridiculously cheap to buy. The weight of the collection bottle is measured before and after each test to provide the volume delivered by the injector.

The tester can be run in standalone, just with the arduino but results are not logged. Logging is done to the PC USB port if it is connected. The data is comma delimited so it's easy to create charts with excel.

The majority of the code is related to the touch screen.

Also, I've added a the ability to completely flush all the fluid out of the injectors and lines with compressed air to make it easy to change injectors. There is a small air pump that flushes the fluid back to the reservoir.

Let me know what parts you're interested in and I can elaborate.

LSswap

Here are some more charts.

This one is in 1 ms increments to 93% duty cycle for the deka80 turbo dodge low impedance, and the two 1000cc injectors. Notice how the two 1000cc injectors are not matched.


This one is for the same 4 injectors and shows the current every 25 micro seconds. The low impedance one was scaled down by a factor of just so you can see the timing of it relative to the high impedance ones.

SS2win

iTrader: (11)

This is cool stuff and I'm surprised that there's not been more interest. I would love to have a tool like this

LSswap

SSTwin Yeah, I thought there would be more interest. I really think the "knee" or dip in the response of the injector is interesting especially when paired with how much or how little fuel is delivered at those short duty cycles.

If anybody want to build one, I can share the software and draw up a diagram. Hell, if there alot of interest in building these, I might even build a PC board

If anybody wants to get their injectors checked out or identified, PM me. No charge but you pay for shipping. I'll post the results here.

SS2win

iTrader: (11)

send a parts list and I'll considering building one

LSswap

More info including schematic are here: https://ls1tech.com/forums/tools-fab...arduino-2.html

turbolx

Nice effort. By now, you are hopefully starting to see some correlation with known injector data on known samples. It took a while to go from getting raw measurements to something that consistently lined up with known GM/Bosch data. It's easy to get a false sense of security with precise measurements that are not accurate due to a number of noise factors in testing. Sorting all of that out can be very time consuming. Like you, I ended up building my own bench from scratch after realizing that other "clean and flow" benches were simply not capable of generating what we needed without completely reinventing their hardware and controls.

LSswap

I have not cross checked against the manufacturers specs yet. That's a good idea. Is there some database that has specs for various injectors?

I have some stock LS injectors I'd like to test against as well as a set of dekka 80s. There will definitely be a correlation, but a slight calibration might be in order to compensate for differences is supply voltages, fuel pressure regulators, fluids used, fluid temperature, and so on. If I was trying to get the absolute best characterization, I could increase the number of injection cycles for each test as well as increase the number if weight samples for the fluid delivered. At some point there is a diminishing return on the time it takes to run a set of tests.

I agree that the "clean and flow" benches are not capable of generating any meaningful data. The accuracy of my setup is orders of magnitude more accurate than visually looking at a graduated cylinder, especially at short duration pulses.

72rssecgen

I have a set of Bosch green giants original Bosch cannot get correct data on them if you are interested in characterizing I currently have them installed in my second GENLS one conversion if you’re interested let me know and I’ll pull them out. Send me your shipping address and if you’d like attach a phone number and I’ll speak to you personally thank you Al

LSswap

I wouldn't mind running them through the tester. Assuming they are standard type O ring sizes and EV1 or EV6? I don't think you have enough posts for me to PM you.