RomRaider

I expected a little more understanding from you, but here goes:

You have that back to front. Injectors flow about 15-20% higher with gas compared to testing fluid.


If MAF can be +/-10% out from factory, and then you add the extra 15-20% from the incorrect injector size, you are immediately past the point where the ECU can adjust fuel trims to meet your targets in CL and your OL will be dangerously out.


As you know, load is an axis in almost all tables. By using incorrect injector sizes and correcting it with MAF scaling you throw load out across all those tables.

Leslie

10% high or 10% low, either way it's 10% off.

If anyone installs these with a proven MAF scaling, I'd be in favor of leaving the MAF scaling as-is, and setting the injector scaling to whatever it takes to get the desired AFRs across the board (with minimal AF Learning).

However I don't think that's any better or worse than installing these with the advertised scaling and latency, and tuning the MAF scaling to get the desired AFRs across the board (with minimal AF Learning).

Not sure yet which approach I'm going to use, but as long as the AFRs are on target, and AF Learning A-C are within 5%, and AF Learning D is near zero, I'll be satisfied.

I just don't think it's a big deal if the injector scaling is +/- 10% from what Subaru would have used, or 50% of what Subaru would have used (as people with 16-bit ECUs and MAF over 300 g/s routinely do).

Since Subaru has not been calibrating injectors with a test fluid that isn't gasoline, the values in the load axes are not accurate now, in any stock tunes. If you truly cared about the number of grams per revolution, you would have to multiply your load axis values by 1.1 (or is it 0.9) to get it. Just like 16-bitters have to multiply them to 2 to get the true number of grams per revolution after halving the MAF scaling and injector scaling. But the fact is that we don't much care - the actual numbers are useful for two things:

1) The injector scaling and MAF scaling have to be balanced so we can get the AFRs we want. This is absolutely critical.

2) It's nice, but not necessary, for everyone to be fudging the numbers by the same amount, so that numbers from one tune can be compared to numbers from another tune. That is, if someone says they're pulling 300 g/s and 3.0 g/rev, you can make better sense of those numbers if you know that what their ECU thinks is one gram of air is equal to what your ECU thinks is one gram of air. Whether those ECUs are in fact accurate isn't all that important as long as we're all off by the same amount. And at this point, I'm convinced that we've all been off by 10% this whole time. Not that it matters.

However I have just convinced myself that I should use 850 (or 900 or whatever it takes) for my injector scaling. That is, even though I believe that my injectors flow gasoline 1000 cc/min, I am going to tell my ECU that they flow 900 cc/min. Since my load axes will continue to be off by 10%, I won't have to re-do my timing and fueling tables to align with the rate at which my injectors actually flow gasoline. And, as a side-benefit of continuing to lie to my ECU about how much my injectors flow, I won't have to constantly translate between what my ECU thinks is 1 gram of air and what every other Subaru ECU thinks is 1 gram of air.

Litle note on testing standards:
I was told yesterday, that flow difference between gas and n-heptane (which is industrial standard) only ~3%, which makes sense to me. What's the reason to test gasoline injectors with something that flows absolutely different? Testing wrx '420cc' (as in ROM) injectors I.D. way and getting 447cc makes sense also.

I tend to end up with these at about 900cc @ 45 psi, with latencies of 6.5v = 3.5, 9v = 2.15, 11.5v = 1.4, 14v = 1, 16.5v = .78

M@

It's weird, but once new tank of gas my A/F trims which were in +/-2% rapidly raised to +7..8%. Flow scale set to 975 now, and latencies back to sheet values - that gives me ~ +2% a/f trims, makes me believe that accurate number for me must be in 950-ish range.

Hard hot cranking issue appears to be vapor lock issue. It never ocurs when cold, just after car stays under the sun for more than 10mins

How is the tip-in response with these values? What is your process like to verify proper tip-in?

I generally test tip-in in 1st or 2nd gear since it's easier for my butt-dyno to detect abnormalities. I also make sure I try to stay in closed loop for the duration of the tip-in event. OL masks a lot of the jerkiness. I do slow roll-ons (almost always smooth), quick jab then hold, quicker jab then hold, varying levels of jab then hold etc.. I log delta TPS% and A/F Sensor # 1 Direct. I have spent many miles tuning the tip-in with these modded stock injectors and I've made decent progress. The slow logging rate is causing all the issues. My numbers are quite lower than yours even though I have smaller injectors than these. And in my next iteration, I plan on reducing tip-in even more. Would I be going in the wrong direction? Though, I do run my latencies at the 1.20ms range for 14v... So I guess I can just add .2ms to my values to get to yours?


I added the difference of our latencies to my tip-in and plotted this. Pretty close but my tip-in is still lower than yours. And I want to reduce it even more. Am I going in the wrong direction? It's very difficult to tell.

That's weird, but my logs call for even more tipin in 3-7% tps delta range.
Honestly, I didn't go through many iterations tuning tipin, just made sure that car drives fine and there are no too lean / too rich spots.

How did you verify that the 'car drives fine'? Just punch it from low throttle and make sure it doesn't fall on its face?

No Did alot of logging jabbing the pedal to different angles.
Occasional hard warm starts is the thing that worries me.

update
I'd say hot start issues gone away. All I did simply multipl 'Initial Start' tables (in my rom there are 8 in 'Fueling - Warmup Enrichment' tab) by 0.7. Not sure why, but this worked

Cranking tables values were already reduced almost same way. My best suggestion on cranking pulsewidth conversion formula:

new_cranking_PW = (old_inj_size/new_inj_size)*old_cranking_PW + new_inj_latency - old_inj_latency

where latencies are at typical cranking voltage. Let me know what you guys think.

actual rom for refernce

Leslie,

Thanks you VERY MUCH for posting this...

Unfortunately, I found your post 1/2 why through the ID1000 installation... otherwise, I would have considered some other options.

The bottom line: The ID1000's flow a little bit more than my old DW850's...

In order to get the daily driving performance (like stock) I had to set my injector scale to 900cc injector sizing (908 actually). I am using the latency values from my 850DW's and it is working fine for me.

I am a bit disappointed in the spec that Tony has given for the ID1000's but considering that when I installed them I went with high pressure lines (aeromotive braided, AN fittings etc) I know I can over come the under spec ID1000's with more fuel pressure.

For reference with the ID1000's... with 43.5 pressure, I am at 100~108 % IDC at 7500RPM with 24psi of boost. My engine: Axis/RAW built/ported heads, cams and a FP red turbo... tuned by me with help...

Thanks again Leslie.. you saved me a lot of grief tuning these in...

Jeff B

Its nice to see a few other people having the same results as me.

Matt

I tried searching, but didn't find anything.

Anyone here use the ID1600s yet? I doubt they will idle better then stock still, but can you keep the same idle rpm, and comments?

I also know that according to Subaru's scaling system, the ID1000s seem to flow about 880 cc/min or so. (43.5 psi) Any results on the ID1600s?

Nobody?