NiklasFalk

iTrader: (1)

What's the penalty for creating and using a OBD-II spoofer with some simple inputs (TS, Revs, just to make the OBD-II values look live). The fine must be noticeable since there is no searchable info on it (or Chinese products), to my knowledge. They have started to check codes here too, sadly.

Call it "OBD-II for older vehicles", to enable the use of OBD-II for logging interfaces etc.

Bored with the ability it could be a nice reverse engineering project

Reverant

iTrader: (10)

In Sweden??? Wow.

Joe Perez

iTrader: (8)

Roman-style crucifixion.

Freaky Roadster

O.M.G this CA emissions thing sounds like the bloody Gestapo are in charge.

Wonder if the MS can be tuned to be even better at passing the tests then the OEM ECU?

bikersam717

iTrader: (5)

http://honda-tech.com/showthread.php?t=2013131&page=3
Not the exact thread I had in mind, but has a few different examples.

Joe Perez

iTrader: (8)

Ah, what a difference a day makes!

Jason, you'll be happy to hear that the FETs aren't oscillating and destroying themselves. It's working quite nicely, in fact. I've been driving this car on the stock ECU and hating the ever-loving hell out of it every morning for almost two years now. It's so refreshing to be able to start the car in the morning and just drive it, without all of the bucking and hesitation while trying to make it up that first little hill outside the driveway.

I have utterly no idea what makes the stock 1.6 ECUs so awful, but every one I have ever driven does that. I can now affirmatively judge that it's nothing to do with the stock sensors, fuel or ignition systems- there's just something wrong with those damn ECUs.

Seefo

iTrader: (5)

Oh god, I hate my stock ECU. It can't keep a regular idle to save its life. If I don't have idle droop, then I have idle increase. and now it randomly likes to run realllly rich when cruising. I am wondering if that rich running is due to my narrowband simulation signal from the MTX-L.

Braineack

iTrader: (62)

I mean you cant eat food or play football in CA, you think you can mod a car?

JasonC SBB

Good stuff. I didn't know the 1.6 ECU was that bad. I had a Euro 1.6 (96-ish) with sequential injection and it wasn't bad like that.

BTW Joe, here's discussion why a 2nd-order filter works so well for TPS and MAP:
http://www.msextra.com/forums/viewto...2233&start=100
See my posts pages 6 and 7.

The extraction of rate of change information amplifies noise. If the filtering is only 1st order, the noise goes back to about the size in the original signal. If it's 2nd order, the noise will go back to that of the size with first order filtering.

skidude

iTrader: (3)

I wish I could get my MS to act better than a stock ECU, nice job!

Joe Perez

iTrader: (8)

Not sure why you can't eat food (we have the best Baja-style Mexican down here) but while the smog stuff can be a pain, the roads more than make up for it.


Stipulated.

And here's a datalog screenshot which visually illustrates why first-order filters are entirely adequate for this particular application:



MAP looks smooth enough to me.


It's not difficult when the stock ECU in question is a piece of garbage to begin with.

I'd honestly like to heard from other owners of US-spec '90-'93 cars here. I've driven three of them, and all three exhibited the same annoying habit. When the engine is "cold" (anything less than full operating temperature) they have a massive flat spot from about 2000-4000 RPM at anything more than about 10% throttle, which means that if you need to climb any sort of incline at all, you have to do it in first gear and sound like a jerk who doesn't know how to drive.

With the red car, I proved that the VAF sensor was not at fault by removing it and substituting a synthetic airflow signal generated by an EMU. Now, with the blue car, I have demonstrated that changing only the ECU, and leaving all other stock sensors and hardware in-place and operational, cures the problem entirely.

I should have installed the wideband before I pulled the stock ECU just to see what the mixture was doing when this happened. I may put the stock ECU back in and do a test some morning.

Joe Perez

iTrader: (8)

There's been a lot of talk in the past about the difficulty of dealing with aircon, specifically in dealing with stalling at idle. I doodled a couple of circuits to deal with A/C in a very rudimentary way a few years ago, but never had a chance to test any of it. Since this is my first Miata with working aircon, I figured I might take this opportunity to try a fresh approach.

Here's what I have built:



Looks a bit more complex than the usual AC driver? Well, there's a good reason for this. I will explain by describing what the various flags connect to.

Starting at the top, "AC RELAY" is the pin in the factory harness which the ECU pulls low in order to engage the AC compressor and fans. This is an output from the ECU. In order for this pin to go low, two separate conditions have to be met, as follows:

• JS2 is one of the four general-purpose outputs of the CPU, specifically "Out 1". I have configured this output to be active when RPM is greater than 800, and turn off below that. 100 RPM of hysteresis is implemented in software. This pin, along with Q130, acts as the "first switch" for the AC relay. When RPM falls below 800, the transistor will turn off, disengaging the AC and (hopefully) preventing a stall.
  
  
• "AC IN" (bottom left) is the button on the dashboard marked "A/C". When the driver pushes this button, U130 (a dual opto-relay) does two different things:
  1. It closes the "second switch" and completes the path to ground enabling the AC Relay to turn on, and
  2. It provides a closure to ground on pin JS4, which is a general-purpose input to the CPU.

Now, the JS4 connection may not immediately make sense, especially since the only documented purpose of JS4 is to read an EGT sensor. Well, a few years ago, JustinHoMi did a special build of code (HR10gm) which looks at this pin and, when it goes low, forces the idle speed setpoint to the FastIdle value.

I haven't gotten far enough to actually test this feature yet (still running HR10g for the moment), however I'm hopeful that the combination of these two things should make the engine more or less stall-proof. I should probably revise my whole "Great AC FAQ" thread with this new design. I suspect that there may still be an issue with idle droop due to the recovery time of the idle circuit (being reactive in nature) so I have an idea for a further modification to the code to implement a dashpot event when JS4 goes low. I have never actually modified the MS code, however, so I'm gonna have to do some learnin' up on how to properly hijack a spare bit somewhere to use as a first-entry flag.

Braineack

iTrader: (62)

It is in reference to the pre-pchooler who, this week, had her bag lunch siezed from her that was packed with a turkey and cheese sandwhich, apple juice, banana, and potato chips. The CA health official personal proceeded the deem it unacceptable and replaced it with three chicken nuggets.

/offtopic.

while you are at it, add a line or two about buying a MS2 CPU and running the a/c idle up code without any fancy circuitry other than an input and output relay

Code so nice, that a supercharged, intercooled, blow-through TB miata runs without any issues.

any chance you could built something that could add a 400-500ms delay betwen when JS4 is grounded and when the relay is?

y8s

iTrader: (8)

CArolina. You on the wrong coast, boss.

Joe Perez

iTrader: (8)

MS2? Bah. Who needs 32 bits? Retro is in, dude.



I do want to start doing MS3s, but I just can't bring myself to touch one so long as the schematics are being hoarded.


Probably, though it would raise the parts count because you need to isolate that delay from the part of the circuit that tells the MS that you're pressed the button.

Once Justin's code is in, it might be possible to simply raise the setpoint of the S2 output to a value slightly lower than the nominal fast idle, and increase the hysteresis to maybe 50% of the difference between slow and fast. I'd have to play with it. Still, this may be a problem that doesn't exist, as I haven't tried the idle-up code yet. Just verified that the hardware works.

JasonC SBB

Joe,

What's the time constant of your MAP filter, and did you try testing the fastest rise time it sees?

Also, are you using the 1.6 TPS switch and not using an auto TPS?

Joe Perez

iTrader: (8)

Post #70 of this thread lists all of my time constants. I went with 25ms on MAP and O2, and 10ms on TPS.

Nope.

Funny that you mention this, actually. My MAP signal is tapped off of the brake booster line on the driver's side (it's a very easy place to penetrate the firewall), and run to the ECU with relatively large-bore hose.

On the first drive, I had a serious problem with what I assume to be individual-cylinder pulsations making their way into the ECU while at steady-state high load conditions (eg: climbing a hill at low RPM), which subsequently caused the MAP accel to be triggering constantly, and setting the car into some pretty violent oscillation.

I solved this by constructing a tiny orifice out of a JB-weld-filled hose mender drilled out with the smallest drill I could find, and placed in series with the MAP hose about 3 feet upstream of the ECU. That damped things out nicely.



At the moment, I still have the 1.6 MT TPS, which is disconnected. I have an analog TPS from a BMW 735i automatic sitting on the bench which I plan to install this weekend.

JasonC SBB

2 pole MAP filter circuit to get rid of IM pulsations:

https://www.miataturbo.net/showthrea...filter+circuit

How do TPS and TPSdot look?

Joe Perez

iTrader: (8)

Yeah, I debated whether to filter the signal electrically or pneumatically. The mechanical solution won because I had all the necessary parts already lying around and it only took a few minutes to implement without needing to unbolt anything from the car. Just cut the hose and slipped the restrictor inline a few feet from the MAP sensor, using the volume of the hose itself as the "C" in the RC circuit.


At the moment, the TPS signal looks like a flat line at 0v, mostly because the analog TPS is sitting on my workbench and the TPS signal line at the ECU is grounded externally.

JasonC SBB

How does MAPdot look?