Speeduino is now on Github Sponsors (Rather than Patreon): https://github.com/sponsors/noisymime
Help with building your Speeduino, installing it, getting it to run etc.
First, great thread.

Good work on your project and a good call to ask for help.
I can only support what you have been told.
Check polarity
check calibration of all sensors.
Rising or falling for crank settings.
Nullify iat, map baro and on the new firmware cold start advance and anything that can change your ign timing of fuel amount like iat and map baro until you get a good solid tune done.
The less things that make changes in the beginning the less to confuse the tuner.
There is differing opinions with ego correction so make your own choice.

What I found fascinating is the comment on MAP sensor filtering/sampling.

I am at the point where I am seeing my tune change as driving style changes.
It is much as if my lambda delay is off but it is not. I have noticed lambda delay changes as my tune gets better. Probably port velocity is changing as the tune gets better.
But there is still something that is acting as if lambda delay is off.

Could it be the map sensor filtering/sampling that is causing the tune delay response?
I will be trying different settings to find out for my self but I sure find this very interesting and if any one has found out if this is sorta true please post it.

It is very true if you can hold a steady load and rpm it tunes accurately but normal driving has many accelerator speed changes so if your load sensor is delayed from filtering/sampling could this cause tuning issues??????

I hope I am not hijacking or changing the subject here.

Very nice info.

Just a side note.

I too have a 85 subaru rx turbo. I went with a 2.2 swap and bell housing adapter.
Mine is propane turbo. 10 to 1 comp and 6 lbs boost dyno at 180 hp. Could go to 12 lb boost but it is my wifes car and it would not go over well if we broke her car from over power.

Very cool posting. Good info.

Thanks to all here.

Black Knight
By JoostXT
@LPG@CV; Previously I was unable to get anything closed-loop to work, I understand wholly how a PID should work and it just didn't. Looking back this was probably due to (amongst others) the MAP sampling rate and the large/slow/old injectors. I might give it another go, but the EGO usually just makes the ride jerky and irregular, I feel like I'm better of without (for now).

The injector timing had me confused as well! But I've sweeped both channels to all 360 degrees to find a happy spot and this is it, it really needs to be like this to run right. I've swapped the injector harnesses about as well but it just changed which channel had to be 180 and which 360. Could still be something wrong though! I'll give it a double check.

The first .msq posted was actually an attempt by me to see what lowering ignition timing did to the driveability (not much if other settings are wrong!). I'll attach the current .msq and a recent log. Both of them are a lot more promising.

@Black Knight I will just swap over the polarity and see if the spark latency changes. I found that setting the EA at a low tps/dot but with very low to little fuel added the VEAL triggered it's EA filter and left the cells unchanged, allowing you to only tune the steady state cells. Right now I've dialed in some EA adder because it really improved driveability. Most of the added functionality is already turned off (like idle advance) because I didn't think the old ECU had this and it ran fine (until ofcourse it didn't anymore!).

If this engine gives me anymore grief it's getting replaced by a EJ18 with a turbo strapped to it, the EA82T is a really bad engine design, very un-subaru-like so I totally understand you going with a EJ22!
(58.2 KiB) Downloaded 22 times
(2.85 MiB) Downloaded 18 times
By JoostXT
I just went out and checked the polarity of the VR sensor, or actually, since I don't know how it should be I used tried different combinations.

I tried 3 different combinations, polarity fipped or 'correct' and triggering in rising or falling edge. See attached teeth loggers. Seems to me that polarity correct and falling trigger is the right one, although polarity flipped and rising trigger is almost the same. This is however against the suggested settings for the DIY-EFI VR conditioner.
polarity flipped, falling edge.jpeg
polarity flipped, falling edge.jpeg (255.54 KiB) Viewed 501 times
polarity flipped, rising edge.jpeg
polarity flipped, rising edge.jpeg (176.94 KiB) Viewed 501 times
polarity correct, falling edge.jpeg
polarity correct, falling edge.jpeg (202.58 KiB) Viewed 501 times
By jonbill
I think falling is correct for current diy-efi dsc - current one is the psig dsc. much of the docs assume the max 9926 based one which is inverting so needs to read rising edge.

(note, I'm only pretending I know what I'm talking about)
All I can say is good work to all of you.

Good info.

I have one more idea to help with your low end injector problem.

I started to experiment with paired injector set up as I am using a 2 channel ecu.

With paired it is possible to get the inj deg closer than batch. Pair up intake and exhaust for inj timing may stop the puddling. I did this on my subie running engine stand months ago and saw some gains. You just reminded me of this and will go back and try it on a running car.

I found that above 1500 rpm none of this makes a big difference. Just my opinion.

If you try this let me know what pairs and deg for inj firing you like and I will do the same.
Subie firing order (right bank left bank) makes for interesting channel location.

Also the EJ motors come with well known sensors and crank wheels so it is easier to set up the ecu.
But you will be smarter than us because you are learning more from what you are doing.

You are doing some good changes. Keep going.

Black Knight
As to map sensor location or as to location of the vac line.

The location on subaru and others, if you are in one runner you will get pulse back giving map fluctuations. Too close to the throttle plate you end up in a temporary dead zone when you open the throttle quickly from inrush of air.
If you have pulse you can filter it out but as we just learned this may cause a map signal delay.

I found the sweat spot about 1.5 inches from the throttle plate. Clean signal with little pulsing and good fast reacting for map changes.

Hope this helps.

Black Knight
By JoostXT
So, this has been a while, but I have been plugging away at it and managed some results;

Setting the MAP sample rate has proven to be quite experimental, I couldn't find any guidelines on which setting to use and have settled on instantaneous, which is usually not recommended, but it seems to work the best? The vacuum point is actually almost exactly 1,5'' from the throttle plate and feeds directly into a large symmetrical plenum, so this must work pretty well.

After this I was also able to set the injector timing back to 355 for both channels and it runs fine. Still getting the strange hiccups but in general the driveability has increased 10 fold! Fuel consumption is down (still not enough though), VE table is stabilising, power is up (I can actually accelerate without going into boost). This brings up an interesting issue though; whilst cruising at 3000rpm I'm at 90kPa, so I want it to be quite lean to gain some mileage. If however I am accelerating I also pass this point and so I want it to be quite rich. But the TPSdot or MAPdot is high enough to reliably trigger an AE!

Also, this engine seems to like a lot of timing, I'm up to almost 40 degrees high rpm cruising which seems a bit high to be honest. Perhaps there is still some spark latency? Haven't checked this after all these changes..
User avatar
JoostXT wrote:
Mon Aug 03, 2020 3:06 pm
… The vacuum point is actually almost exactly 1,5'' from the throttle plate and feeds directly into a large symmetrical plenum, so this must work pretty well.
You may have to test for the most stable MAP location in your manifold, typically away from the inlet and runners, but dynamics can be weird.
JoostXT wrote:
Mon Aug 03, 2020 3:06 pm
This brings up an interesting issue though; whilst cruising at 3000rpm I'm at 90kPa, so I want it to be quite lean to gain some mileage. …
Also, this engine seems to like a lot of timing, I'm up to almost 40 degrees high rpm cruising which seems a bit high to be honest.
90kPa is oddly high for cruise, but could be your MAP source point or the MAP mode. Generally, from best-torque fuel Lambda/AFR at stable cruise, lean to the point that you feel lean surge (slow push-pull feel), and enrich just out of it. Check MAP. Don't be surprised if it's higher, as lean is actually less efficient (assuming it was originally fueling for max torque).

In order to recover efficiency at the new Lambda; at stable cruise again, begin slowly testing timing advance until timing is at minimum value for MAP just at minimum value, indicating peak timing efficiency. Advance may be quite high relative to power Lambda timing, in order to compensate for the change in cylinder pressure rise rate. The table cell may change during this exercise. See where that gets you. Don't forget you are tuning for best target performance (and resulting Lambda) at any point, not a Lambda number, which is only a reference.

Fuel Injected Two Stroke

. The AFR lower Yellow line stays consistent for t[…]

Gamma Enrichment?

So, either take a log of it happening and post it,[…]

Hi All For everyone that is interested: I designe[…]

STM32 development

Standard dec to STM32 pins in variant.h is: // Ri[…]

Still can't find what you're looking for?