Icedawg

I had the MS3PnP starting the car just fine at 40 F and above, but now that winter arrived it is very difficult to tune the startup procedure. So I am hoping someone who tunes for frigid starts can offer some advice.
DeastchWerks injectors around 600 cc/min, stock 62 psi
FM Garret 2560 turbo & intercooler kit
NB2 2001 Miata

At 32 F or below, as low as -17 F, it is either very hard starting, starts & fails to catch, starts & runs a while then dies, and if lucky eventually starts catches and keeps running.
I think I had it too rich at first, and have been leaning out the cold crank pulses and reducing the PWM on the IAC Valve during cranking, while enriching the priming pulse. It seems to be the right direction.
But it still catches, runs, then dies. When it does there is a characteristic sudden drop in the PWM, associated with a drop in the rpm. When it does not die and stays running the PWM goes through a smooth downward ramp, along with the injector PW and the idle. It is like the controller functions are confused, but I do not know what to adjust to reduce the confusion.

Below is an image for the most recent start that matches the current tune.
The IAT sensor is a GM and seems accurate, the coolant sensor is the stock Mazda one and always reads high at these sub freezing T's. Car has sat overnight, so T's should be the same.
The cranking rpm is set high, but it seems to need that or it simply thinks it has started and dies and I can never get it to start.




Here is another older example at lower temp, with an older, richer startup setup on cold crank pulses and so on.



I have many more files, at various temps as low as -17 F, but these get the issue across I think.

DNMakinson

iTrader: (3)

1) Closed Loop Idle Settings: PID Delay(s) from 2 to 3
2) I would add some prime delay, unless you have a really good fuel pump with really good check valve. Else, you are not getting a valid priming. (Cranking / Startup Settings). I'm using 2.1 seconds and do not move key from ON to START until fuel pump stops (at the end of the priming pulse). That may be excessive, but I would use 1 second at least, then pump will stop in 2 seconds, as usual. (mine stops in 5 seconds).
3) I use WAY more Priming, Cranking and ASE. With your issue, I'd start with more Cranking and ASE. When you get it to start and run in about 2 seconds, add Priming to get a faster start.
4) At 30F, PP=65 (with 640cc injectors) (the maximum): CP = 400: ASE = 113%
5) I would use ASE taper in 0.1 seconds rather than cycles. Minor issue.
6) I would use MAT, not CLT for Table Lookup in C-L Idle settings. Set RPM with coolant then use that RPM target with MAT to fine tune.
7) I would have 20* idle RPM to more like 1450.
8) Your VE table at around 700RPM and 90 kPa seems low. This could attribute to lack of fuel just after start-up.

9) Do you have any logs from starts at 70F?

10) One general help on start-up tuning I used way back... a set of diodes and a 12V power supply to allow the O2 sensor to run during the entire procedure. Then you can get an idea when you are too lean, and when to fatten. A big pain, but gave results.

Sometimes, below 35F, I do need (2) starts. I just added some CP down there, so we will see how that works.

DNM

Icedawg

Thanks for this DN. I will look into trying a number of these, but have a few quick questions.

2) Certainly the fuel pump is normal, not strong. And on my SM 1999 I usually cycle it on and off three times before starting it, which does get it to start on the first try (at summer temps). I always thought that was because the non return fuel line takes longer to build up pressure, so this makes sense and looks like a useful way to get around that.

4) Acronyms got me. Is PP the priming pulse? Is CP cranking pulse?

6) I am confused on this point, and need to think it through. Usually a car's warm idle rpm is independent of outside air. I have closed loop set to turn on at 160 F CLT, although based on the logs it seems to turn on much sooner. And I am not sure how there is a fine tuning? Is this a blended second table you are referring to?

8) Very good point. I can't actually drive to VE tune in this 700 rpm, 95 kPa range, so have left it pretty much alone down there.

DNMakinson

iTrader: (3)

4. Yes
5. The MAT is a trim, but it is handy. No. If you set the setting in Closed-Loop Idle Setting to MAT, then the table uses MAT vs RPM. You already set RPM vs CLT, so the table now utilizes both. (CLT -> RPM)[curve] Then MAT + RPM -> PWM Idle % [table]. So yeah, CLT controls RPM target. MAT only comes into play in setting the initial PWM Idle %. Same table. The Closed Loop Idle Initial Values table. Just the Y axis becomes MAT if set for that. I don't see were you can set a temp to switch from open to closed loop on idle valve control.

8. My 100kPa at 700 is about 81, while at 4300 it is 87 (for reference). With braking, one can hit those cells. It feels weird, though. Probably better to do it at 1500RPM, and just copy left.

Adding 9. You are going to have AFR issues at idle if you do not allow EGO. For my idle area, I allow +/- 17%. Over summer / winter, I need this much to keep 14.7. Not a starting issue. To keep some richness during warm-up, I don't allow EGO until 120*F.

Note: I did

curly

iTrader: (25)

You might look at ignition settings too. You're at 30 degrees when it's running, it drops to 14 by the time it dies, I might try and keep it up there with cold advance or idle correction and see if helps. You're also only adding 7 degrees by the time you're 300rpm under target, I add 10 by the time I'm 100 rpm under. So you could probably add some and see if it helps.

I'd probably agree that fueling is more of an issue, but thought I'd chime in.

Icedawg

Thanks all.
I tried some of what was suggested and it seemed to get worse.
So, will have a rethink, and a retest. Another 0 F morning predicted, so good for another test.

Overall fuel enrichment is around 180 % or more when it first runs, with combined ASE and WUE modes, and AFR is around 10.5 before stranger things happen. Fuel may be wrong, but it does not seem way too low.
So a timing check is a good idea.
Looking closely at thetraces, the weird changes in PWM seem to mostly appear immediately after the sudden drop in rpm, not just before.
So I will look at keeping timing high longer. Although when it does catch and keep, timing is around 20 or more.