Warning, this is gonna be long, but I want to get as much of the information out there that I think I have...

I've been working on getting my standalone MS-II running on my NA '93 for a couple months now and have made good progress. I'm at the point where the car idles and I can drive it around the neighborhood (albeit running very poorly).

I believe I have some sort of fueling issue. My initial VE table was derived by interpolating the MSPNP's table and scaling it to topping out at 100kpa. The issue is - I have to run REQ_FUEL at 25.0 to get anywhere near 14.7afr indicated by the MS.

I was convinced this was a problem between my JAW wideband O2 and the MS. I figured the JAW's signal wasn't getting read by the MS correctly. That assumption had me chasing my tail for quite a while, culminating with me eventually watching a multimeter on the JAW's outputs and confirming it's working correctly. I base this on being able to run the car on the stock ECU and getting voltages appropriate for around 14.7afr. With the MS, voltage spikes to 5 (extreme lean) unless I run the 25.0 REQ_FUEL. The MS is currently configured with the LC1 0-5v=10-20afr and I've downloaded the same preset to the MS via the tool helper.

At 13.4 REQ_FUEL my pulse-widths are in the 4's, which I read somewhere should be over 6. At 25, they're in the 10-12ish range. The car will drive around at 25, but it's clearly pig-rich as it's popping out the exhaust pretty regularly and has no pickup power. It'll cruise fine.

I've been through my settings a bunch of times, but I could just be snowblind by now. If someone can review it, I'd be grateful. I've also reviewed my hardware build, but wouldn't rule out some issue there either. Oh, finally, everything seems to work just fine on the Jimstim.

TIA. MJ.

 

I looked through it and it all looks ok to me. Take some datalogs and post them up. That would help a ton.

 

FWIW you have the acceleration enrichments mostly on map based, but the threshold limit is very high at 300. Try 80. That will help on the response for sure.

Best thing is to calculate what your req fuel should be and use that. Then make a datalog and start tuning the VE table.

 

I'll work on a datalog. I have the acceleration enrichments at 300 so they're not activated while I'm just trying to get a baseline going.

I calculate my REQ_FUEL as 13.0, but that makes it all the more lean. I'll datalog at both the 13 and 25 ranges. Not sure if it's drivable in the 13s.

 

Have you tried just adding 75% to all the VE values? The VE tables can be a little funky; they aren't the same between 029y and HR10, so I wouldn't assume they are the same for MS1 and MS2.

 

Here's my VE table I'm running. Try it with your 13 req fuel. Don't run a different req fuel than what you calculate as that causes other things to get messed up too.

EDIT: it won't let me attach a .VEX WTF? Send me your email and I'll attach it that way.

 

pat - try renaming the .vex as something else, anything else - .xev, .abc, .pat and Mike wil know to change it back when he downloads it.

Mike - did you try running the stock ECU from a narrow-band JAW output? It'll probably involve some un-fun wiring, but would help convince me that this isn't JAW related.

do you still see the AFR change at idle as the car warms up? that behavior makes me think that maybe the heater element of the WBO2 is coming into play somehow - you get one reading when its cold, another when its warm... i can imagine scenarios where this is either normal or a problem.

 

There's a countless number of people on the JAW forums who have megasquirt problems. In fact, I've yet to find a single person who said it works as expected. I'd get it working with a narrowband before attempting to use the JAW.

 

I have tried running the MS from both the JAW's NB simulated and from the Miata's NB. Both have the MS running lean with my current VE table, but having JAW providing NB simulated is closer.

When I've monitored the JAW's output, there are definitely different values until it gets warm. Usually about a minute into switching it on.

Also remember that I have these uber-lean values at idle. The MS shouldn't even be paying attention to the JAW until RPMs rise above idle. At idle, it should just be providing information.

I've attached 2 datalogs. The first is by far the largest and is having JAW provide simulated NB to MS. There is the corresponding .MSQ for this log as well. The 2nd is a much shorter one with the same VE table, but having JAW provide simulated LC-1 output. The semi-normal values seen towards the end were made my me revving then abruptly letting the throttle close, thus causing an after-fire.

 

I agree. I was hoping I was smarter then they are.

I'm getting closer to springing for the LC-1, but was really hoping for a true DIY solution.

 

For now, why not go ahead and run open-loop? Just set the EGO authority to zero in Megatune. This way, you can focus on getting your Req_Fuel and VE numbers into the ballpark, without worrying about gEGO thwarting your efforts with potentially bogus data. Leave the JAW installed with a meter on its output, and if you've still got your NB sensor, put a meter on that too so you can reality-check the JAW.

 

I'd still try a REAL narrowband... not just the JAW emulation.

Do you have an aftermarket exhaust manifold?

 

I like the open-loop idea. I'll give that a shot. I have both the stock NB and the JAW's WB sensor in the exhaust. I'll monitor them both.

I did pipe the stock NB into the MS and it's still running very lean.

 

Unfortunately, the car's covered in a shell of ice, so no testing, but I have been thinking...

What if this is a fueling problem? My rambling thoughts go something like this:

To get a "decent" idle at around 800 RPM using the MS, I set REQ_FUEL at 25 and the VE Table cells I'm in are set in the mid 60s. This produces a smooth-ish idle, but horrible stumbling upon throttle application.

When I reconnect the stock ECU I get a very smooth idle at a slightly lower RPM, but a completely different engine note. Higher in pitch and just healthier sounding.

So, I put the MS back on the Jimstim and have some interesting results. I first just hooked up a jumper to the INJ1 and INJ2 pins to make sure the MS is triggering some sort of signal to the injectors. It is. The really interesting result is the two pulse-widths. Both are right around 11 and pretty steady. This is with REQ_FUEL back down to 13.4. If you look at my first post in this thread, the same setting with the MS in the car resulted in a pulse width of 4. I thought it may have something to do with the O2 sensor, but there is no change regardless of the stim's O2 pot setting.

Any thoughts about this possibly being fuel related?

 

you can now.

 

 

Here ya go.

 

Pat et. al. -

I tried Pat's fuel map without much difference. Thanks for the effort. BUT...

I have made a huge discovery. Injectors on cyls. 3 and 4 (closest to firewall) are not firing at all! I used a home-brewed probe stethoscope and all four injectors click away happily on the stock ECU. Only 1 and 2 do on the MS. This explains everything. The leanness, lumpiness, etc.

I've troubleshooted the following:
1. Harness. My injectors are wired in the harness the exact same way as the DIY build doc in the Sticky threads. I have good continuity between the two ECU side injector ins and the four DB37 outs, spit of-course.

2. Using the Stim, I am able to get a good fuel injector led flash on both INJ1 and INJ2 pins. The problem with this is you can't discretely test all four DB37 outs. So I opened up the MS box and...

3. Measured voltage between Pins 32 through 35 and ground. I get a steady 3.2v at all four.

I'm now stumped.

 

This would only be possible if they weren't plugged in. If #1 and #2 are injecting, and assuming a good harness and connection at the injector, then #3 and #4 HAVE to be injecting as well.

This is because #2 and #4 are wired together and #1 and #3 as well.

 

I agree with you. But you'll have to take me at my word. 3&4 aren't tapping. It's bizarre.