DIY Fuel injection thread.

[0R10N]

Cheers for that. Apparently it has knock control too but as you say it's fairly rudimentary compared to modern ECUs.

 

Speaking of which, does anyone have a DIY knock detection "kit" in Auckland that I could possibly borrow for a couple of days? May have to scour the internerds/Aliexpress and source some componentry to build one.

[Roman]

Yeah easy enough to make, just need a Bosch donut style knock sensor, shielded cable and then wire it as an aux in to your laptop.

 

Then listen with headphones.

Just need to turn off software processing or whatever so you dont get any latency.

Handy that you can record it as well though.

I had it setup this way but ended up wiring knock sensor directly into ECU.

[kpr]

yeah i wouldn't even bother with the knock control on those old links.    just listen  when tuning and keep timing  a good few degrees away from knock

 

as roman said  all you need is knock sensor  head phones  and some kind of amplifier to get the level up to power headphones.

 

some engines it can be a pain to hear over the valve train noise. (probably a rattly old 3s aye dave) so need to know what you are listening for.    also when you are just on the verge of knock, you may only  hear one "tick" throughout the rev range under high load. which is where you want to catch it.  needs to be done at highest load possible.   as may not get any knock in lower gears, but will show up in 4-5th 

[Roman]

some engines it can be a pain to hear over the valve train noise. (probably a rattly old 3s aye dave)

I'll get back to you after I've tested now that the screw driver is gone

[Stu]

Be careful Dave, that screw driver was poking the bearings back into place each time they threatened to fall out.

[Roman]

Oh man this test rig thingy rules:

 

 

Making something similar currently but just to test 1 injector at a time.

Keen to get some good injector deadtime values and see how spray pattern and CC rating changes at varying pressure differentials.

[Ned]

keen!

 

What you using to control the injector?

 

Also, is there a set pulse width and frequency you use for that sort of testing? like, is there an industry standard? guessing there might be a range where injectors are more consistent and reliable for repeat tests etc, and that at the extremes they probably arent as great?

[Roman]

I'm going to control it with my ECU - an extension cord on Injector plug #1. So I've got full control over it with laptop, and I'll have datalogging of what's going on.

 

There's an ECU test function where I can input how many ms I want the injector triggered for, and I can also make it fire a set amount of times.

 

So for finding deadtimes, you could reduce the MS until no fuel comes out.

If I want to find the CC rating, I can run it at say 50% duty cycle at 100hz for say 1000 events and then see how much it fills the tube in that time period.

Once the deadtimes are figured out, you can test to see how linear the injector is at low opening times.
There's a compensation table for "short pulsewidth adder" to address this.
If the ECU would want to spray in say 15ms of fuel, but that's an area where the injector is non linear.
Then it actually sprays 13ms or 17ms or whatever to compensate, if you've got that table filled in.

So these things just make a really little difference, but accurate information makes all of the compensation tables work way better for changing temps etc.
Makes quite a difference where I'm trying to blend together two sets of injectors though, as currently I'm not confident that my deadtimes are at all accurate.

 

At this stage, because I dont have spare fuel pump or FPR I'm going to be using the stuff from my car.

But I'd like to have a voltage regulator so that I can set it somewhere between 8 volts and 15 volts for finding deadtimes - Any ideas on an easy way to go about this Ned? 

[Ned]

if you wanna tune it for on the car, then best thing is to use it as close to in situ as possible and as real life as possible. Especially when it comes to voltage corrections etc. Batteries behave differently than power supplies, and different (size/type) batteries behave different, so you should do things like check dwell and injector stuff on the car as it would be used IRL really. That way you use the ECU to measure the battery voltage and control the things. If you did it off the car, and you get to an area thats weird and your ECU measures the voltage a little higher or lower than your test setup does, or the resistance in the leads is more or whatever, then it's 'shit' data already.

 

When we did dwell on cams car, we just charged the battery right up, measured dwell, turned on the headlights and measured every 0.5V all the way down to like 10V i think? took all afternoon...

[Roman]

Oh yep cool.

 

The only problem in my case is that because I'm using parts from my car to set this up (fuel pump, FPR, ECU, injectors) I cant run the car while the test is going.

 

So can only go to a max of 12 volts when I really need 14+

But regardless, seeing how the deadtimes change with fuel pressure will be interesting in itself even at 12v.

The deadtimes for my two set of injectors are WAY different based on current info.

Outer injector deadtime is roughly double.

[anglia4]

Could you have another car running next to you with jumper leads to boost the voltage to running conditions?

[Roman]

Putting a battery charger on while it's doing it's thing might work!

 

If not...

 

I've looked at known data from a shitload of injectors to see what kind of curve there is for the deadtime changing with Voltage, and see if it's predictable.

 

95% of the ones I've looked at, if you add an "exponential" trend line it matches the curve pretty much bang on.

 

So even if I could only go to 12v I'm confident that I could extrapolate to 14v-15v reasonably accurately.
showing only 3 here so it doesnt look cluttered, but they mostly all look the same.

Interestingly my aliexpress injectors which supplied a deadtime data sheet, the deadtimes are exactly the same as RC Engineering 750cc injectors.

 

So probably rubbish data because they just copy pasted from somewhere else!
 

[Roman]

HERES WHY I LIKE GRAPHS.

 

I was looking up the deadtime info on the Injector Dynamics page (As you do)

 

When you put it on a graph it's pretty obvious that there's a typo in one of their figures.

 

Should be 1140ms instead of 1040ms at 12v and 65psi pressure differential.

But it's hard to tell when it's just lost in a sea of numbers listed on the page.

[Stu]

DaveScience - I have fuel rails and regs and pumps here if you want to use them?

[0R10N]

All right, experts. Thinking of setting myself a long-term goal (once squid is on the road) of decommissioning the G1 V5 in the TX3 and replacing it with a G4 Atom setup because we all know the G1 is stone-age garbage now.

 

I've had a look at the Atom specs and on paper the viability seems perfectly fine for my needs, however I seem to perhaps inaccurately recall that the Atom had some drawbacks/limitations which meant they weren't used all that much on turbo cars, and people were encouraged to buy Storms instead.

 

Was this even a thing? Or am I getting my wires crossed between the Atom and some other Link product?

 

(Heck, even I got upsold into buying a Storm for the Skyline.)

[kpr]

atom  doesn't come with a built in map sensor, so need to  buy an external one.    probably the only reason  someone told you to buy the storm for turbo

[Celica RA45]

dave what are your out board injectors ,and could you start the car on them and turn the inboards off and do a full tune on the out boards from idle to wot and then redo the inboards and then start the out boards from say 3500 to blend in to the inboards 

[Celica RA45]

also when i was doing a std beams with 2 stage, we used the std  altezza injectors 365cc and on the out boards we used 550 cc with same ohms of 13.8  ,but we also pushed the fuel pressure up to 75lbs to help 

[CamB]

I've had a look at the Atom specs and on paper the viability seems perfectly fine for my needs, however I seem to perhaps inaccurately recall that the Atom had some drawbacks/limitations which meant they weren't used all that much on turbo cars, and people were encouraged to buy Storms instead.

 

 

Might depend on what you want as inputs - that seems to be the Atom's weakness on a 4 cylinder. It has 3 analog inputs which can typically get used up by MAP sensor, TPS and wideband. I've got a Storm that will eventually see use and I plan to use 2 more analog inputs for (1) switching between boost settings and (2) 0-5V oil pressure. This is a compromise for me, as if I had more analog inputs I would do more. 

 

It also only has 2x digital inputs, but that might be enough (eg high boost / low boost (if you have a boost control solenoid) and speed sensor).

 

Also doesn't have "motorsport features" if that's your thing. Which is good, coz with limited inputs its hard to use them.

[0R10N]

Cool, thanks guys.

 

I want to keep it simple on the TX3, being such old tech and only a street car, so MAP/TPS/O2 seem to be the most logical ones to have. What about knock control? Is that even a thing still? I didn't even recall seeing it anywhere in the spec sheet.

 

My existing G1 has no knock control, no closed loop O2, and no TPS, it purely relies on MAP.

 

Boost control is handled externally of the ECU and I don't have existing speed/RPM switches so I guess the digital inputs remain free for logging etc.

 

Crafty that it doesn't have a MAP sensor though. Almost expect a scenario where I buy one and straightaway a big pop-up window appears with "CONGRATULATIONS! YOU BOUGHT AN ATOM! NOW ADD ON A MAP SENSOR FOR $499.95! ORDER IN THE NEXT 10 MINUTES AND GET A FREE LINKECU STICKER!"