99+ MS-II install issues
 

Ok, looks like I have most of my MS-II built. There's a input that I never finished, so it's tied to +5, shouldn't kill anything, and there are two diodes which aren't quite right (have to get parts) - one in place of the zener diode on the knock line (so the CPU sees gnd all the time on the knock line) and a 30v diode verses a 70v diode on the boost solenoid line.

In other words, nothing that should mess with anything.

So I have two issues, otherwise things behave just swell with my jimstim:
1) The O2 just comes out way wacky. At the voltage increases is shows things being richer, then jumping leaner, and all around not working. I've set it for WB and NB, and I've put the jimstim's jumper on and off a few times for wb/nb, and no combination gives me in megatune what I'd expect. The voltage all throughout the circuit is fine, even up to the tinsy pin on the cpu itself, there's a 0-5v signal, varies nice and linear.

2) The baro correction shows way low. Like, 68 kpa. Again, the circuit is right out of the book, the two sensors share pins for ground/power, and they are within 1% of their output voltage, checked at the pins in the 68000 slot (cpu slot). But they read WAY different. The baro moves if I blow or suck on it, like it should, just like the other. But the reported value on the screen is way off.

Maybe this has to do with beta MS-II code. Not sure, but its annoying since everything else seems perfect.

[edit] I belive I'm running 2007-11-22 build, title bar on MT says: MT 2.25 p1 - MS/x pre2.0 beta 20071122 1040est (c) kc/jsm **ms2

1) You need to tell MegaTune you have a WB in 3 places. Once in MT config, once in EGO settings and once in tools->ARF sensors. Did you do all 3?

Jesus. No. I did all one. :-( Only in EGO settings, I couldn't find any others. Thanks!! I'll check it out.

The idle valve acted really funny till I power cycled the MS.

Oh, and why can't I find a way to mount the PWM idle valve diode in the case??

Anyone know what I'm doing here?
http://abefm.smugmug.com/photos/233650096-L.jpg



Anyway, perhaps if I could get raw voltages out of the MS, I could convince myself the hardware is fine. Not "12.7 afr", but 3.2V

It would appear that you're attempting to determine the compressive strength of your second MAP sensor. :bigtu:

Close. I was trying to figure out what was on the back of the board by the it made when it crushed. My only definitive answer is nothing that matters. :-)


Now I just have to figure out how to set up the MS. My thinking is if I dig enough into the settings, I'll get it.

Then it's harness making time.

Oh, and shockingly, radioshack had almost all the parts I wanted. Sort of. Their 4.7 zener for protecting the CPU was a 5.1v. Risky?

______________
Getting there! The KnockSenseMS came in (anyone want my old knocksense? Maybe I'll give it to my dad), so now I'm only waiting on my EGT (the guy never even got back to me?!) and the wideband. WOT's been helpful on this end at least.

Dude. We live in San Diego. There are two Fry's within a 20 minute drive of Qualcomm, and another five between here and Burbank. And you're shopping at Radio Shack?

Seriously though, what's with the Zener on the knock line?

Edit: nevermind, I see the document you're getting that from: http://www.megamanual.com/ms2/knock.htm

Still, the KnockSenseMS output is already 5v to begin with. I'd think the Zener unnecessary?

Yeah - I was kinda thinking that, which is why I was ok with the 5.1V. Nothing with any compliance should get higher than that anyway.

What I don't know if if the knocksense puts out an analog or digital signal. Analog seems a lot more useful, but I guess it's just "knock events" afterall. Only the threshold is different for different RPM.

That's a pretty slick way of hooking up your second map sensor. I thought about doing it that way, glad to see someone else doing it ahead of me.
I'm lucky to have a radioshack close to me. There are no Frys to be found anywhere around me though. They have things like TIP120s, Mica install kits, diodes, and a few transistors i need once in a while.
Take a picture of the top of your board, i'm curious to see what the MS2 board looks like with the daughterboard and all setup for a miata.
Did you decide to go with sequential injection right off the bat using all your outputs?

Yeah, ok, hang on (goes on a journey across the web)

Back side of board
http://abefm.smugmug.com/photos/233649912-L.jpg

Front side, mostly done
http://abefm.smugmug.com/photos/233649946-L.jpg

For those who care:
Close up of wiring boost controller into Q16 high current ignition driver circuit:
http://abefm.smugmug.com/photos/233649973-L.jpg
The leads are all very heavy. Yellow jumper to ground shown, and note I put in the 1k resistor in the... I forget, I drew it all up here:
http://abefm.smugmug.com/photos/233649768-L.jpg
Ah! R57
But the idea is not to waste space in the protosection, and to give you a nice place to mount everything. The only non pcb mounted part was a resistor on the back, I soldered to js0. You can see in the board top pic where I will put the diode, I mounted the temporary one high in the air so you can see where it goes - there's +12V at that other smaller diode. You know, the "red kind" for you technical folks out there like Joe.

Note on the backside I jumpered the knock sensor circuit to ground since I didn't have my zener diode yet, that's getting replaced tonight.

I started wiring up an extra input on Js7, not sure what to use it for yet, maybe a clutch switch.

Other thoughts: Radio Shack sucks, but they are better than nothing. I drove right by Fry's without thinking. Sigh, I remembered, too, when I was in the RS parkinglot.

If you're thinking of doing the MAP that way, I'd probably just double the cap feeding the one sensor instead of soldering it to the leads. Really I just did it that way cause it looks cool. :-)

No picks of the MS-II yet, clearing the board with the can-style 2n2222A's I had was a concern, so I wanted to show how I routed everything below board level.

Also, don't use braided wire. It's hard to get in the holes and it runs funny, you can totally tell the one red wire on the top of the board that sucks was a braided wire.


Lastly: No, no sequential injection to start. Too many other uses for those inputs and outputs. I'd still like to do it though, maybe down the road - if I read up on CANbus, I can move all the other stuff over.

Oh, yeah - how do I wire the tach pulse on a 99+?

The MSx manual gives you two choices for an input, High and Low, and they say low can fry your CPU. Now, I would think just adding a zener to ground would work (what am I missing here?) but....

They say:
http://abefm.smugmug.com/photos/233863135-L-0.jpg
OR
http://abefm.smugmug.com/photos/233863131-L.jpg

But couldn't I do something more like:
http://abefm.smugmug.com/photos/233863138-L-0.jpg

I don't want to have to rewire every switch in my car to +5 volts, you know...

C'mon, someone here must know how a transistor works. Don't make me dig out horrowitz-n-hill.... :-)

Of the two schematics you drew, the first one will work, the second one will not.

Jim

Heh. Thanks. Your reward: Eye candy!

http://abefm.smugmug.com/photos/234090912-L.jpg

http://abefm.smugmug.com/photos/234090908-L.jpg

Doesn't that look nice and fast? Faster than the shipping I hope. The only thing slower is the wideband vaporware I ordered.... But it's a nice piece, no? $55 shipped from team rip engineering.

I got four exhaust probes from Motorsport Innovations at $70 a piece IIRC. It's almost a shame to put them on the header.

Jim

http://www.technospirit.gr/images/goodies/DSC06040.jpg

I'm curious to see the business end of them. They are all packaged up, hard to tell what I'm seeing....

My camera was shot a few weeks ago so I can't take a quality pic of them. Maybe tomorrow at work.

Jim

So, bits and pieces. The AEM forums have a lot of good info, Emilio does take stuff seriously. :-) Here's some stuff from miata.net, which I know has been seen but I want to put all the notes together.
http://forum.miata.net/vb/archive/in.../t-221938.html

I too initially thought that the sensor output was sometimes inverted, or inverted when cranking; or inverted randomly. By carefully examining the crank signal the moment the engine fires and RPM climbs from ~100 to ~500, it was obvious the berserk falling edges "de-berserk", and the signal doesn't look inverted enymore.
--------------
But I am currently using the falling edges and it just works fine. Engine starts fine and runs fine. It's been working like that for a while. No timing or syncing errors recorded by the EMS either ...
--------------
Then I entered the data in my crank cam signal generator program. Look at the attached picture - it is a scope of the '99 crank and cam signals generated by my program (not real '99 signals). The scope is of the program set to simulate the engine turning at 180 rpm. Looks the same as the real NB thing.

Then this morning I connected the signal to my AEM EMS, set in the AEM EMS to monitor raising edges only for both crank and cam. Then I adjusted the crank/cam options and tables the way I had predicted they should be few days ago (look the posts above).

Then I tried it and it works!

It synced instantly. Tried turning the EMS off and on about 10 times while the generator was running. Every time it synced the same way.

As a result:

The 0 tooth edge is the one 10 BTDCC cylinder 2. So, both ignition and injection sync options should be decreased by 1.0 (the NA way was counter zeroing at 10 BTDCC cylinder 4, so we should move the ignition and injection back by 1 cylinder = 180 crank degrees = 1.0 teeth )

The significant edges are the 10 BTDC ones
The alternative edges are the 80 BTDC ones

The Alt Fire mode is enabled. This means that during cranking, the dwell starts on the alternate edges (the 80 BTDC ones) and the spark happens on the significant edges (the 10 BTDC ones) 70 crank degrees later. Looked fine on the scope to me.

As soon as the rpm raise and the EMS goes out of cranking mode it starts using the timing to fire sparkplugs and injection, and it looked fine to me on the scope too.


I want to knock this down, and make a full pin-out chart for the 99-00 miata harness. I'll post whatever I come up with. Pretty it might not be.

Huzzah!
 

Sweet! My little test of Arga/Joe's input circuits seemed to work great. I took the two timing inpulses from the JimStim off, and ran those jumpers to snipped leads to aligator clips to paperclips I sanded down with a cut-off wheel, which I stabbed into the backs of the connectors on the car, trying not to tear the rubber. Everything else I let the stim run.

At the last second, I remembered to switch it to "Miata 99-00". Burn. Then.. It worked! I got ~300 rpm, but it kept bouncing, sometimes 0, sometimes 1000, sometimes 50, but mostly 350.
http://abefm.smugmug.com/photos/234520924-L.jpg
http://abefm.smugmug.com/photos/234520924-O.jpg <-Full Res Version
Then, after tearing it all down, I decided to reboot the MS (duh!) and after that I did it again, 235 RPM almost rock stable. Maybe 323-236.

So, I have some logs. I don't know what to do with them. Eveyrone's graphs are shockingly similar, so I assume there's some standard analysis program? I guess I'll attach it. Someone tell me if the RPMs look... I dunno, believable?

I'm still a little worried about the weird rev limiter thing, but I'm starting to think I could run this. Oh, one more thing: The signal to the coils should normally be "off", right? I.E. You ground while firing? And I want "inverted" output?

So I've finished building up my basic MS-II for a '00 miata. I've added a few goodies like boost control. I want to run VICS (variable intake runner - a solenoid you close at a given RPM) and my cooling fans (on at a certain temp).

I see I've still got D15, along with JS 11,8,6,3,2,1 all open. Doesn't that mean I could run 7 more devices? Which ones of these are ins, which are outs, which are analog? How much power can they drive (for instance, the stepper drivers must be able to deliver SOMETHING. Or I could do a trick like you do for boost control, putting a heavier duty chip where Q20 goes like I did for Q4.

Anyway, all input is appreciated. I keep seeing tables of only three or four options at once, but it looks like there are still 7 options open to me.

Let me know. I'd be happy moving stuff around if it opens up more I/O.

This will probably do the trick for the VICS.

http://www.diyautotune.com/catalog/p...ans-p-126.html

-Mike

They just use a 222a, er, I have a lot of those. Huge econo-pack.

But, can they handle the current of runing something like a fan solenoid or two? I have no idea what the draw on those is!!
-Abe.
<edit>Answer: Totally. The assortment pack I got from radioshack (what's the difference between them I don't know) says 800 mA on the back - both fan relays together draw 260 mA. So I'm in the clear on fams

P.S. Can't I use the LED15 output, too? Whichever that middle LED is. I'd love to drive my fans with it, if I could. I just have one wire going to both solenoids.

<edit> I still want to know now on this. I guess I could take out the LED... Then I'd know the current. But I kinda wanted the LED as a fan-on-indicator.

Wow! The 2N3904BU's that run the lights on the front of the MS are only rated at 200 mA! That's really low. And the data sheet says 100mA is their max "useful" switching current.

I have piles of these 2N2222A's are rated at 800 mA. Why radioshack packaed transistos good to 100mA in with ones good to 800 and didn't label them as such bugs me.

Anyway, I'm pulling my three LED transistors out. I guess there's a chance these 2222A's are slower, but... No, even the rise and fall times look similar. That's it, out with the old, in with the new. :-)

Abe, looking good. Yes, two days into my vacation and I'm breaking the internet embargo.
I can't speak to the NB, but on the 1.6 the IGN lines are active high. The The coils begin to charge on the rising edge, and then fire on the falling edge.

Yellow trace is trigger voltage, blue trace is coil primary current.

Edit: That's odd, there was supposed to be an embedded image here. Oh well, here it is as an attachment. I took this one both to figure out polarity and to double-check the dwell times that everyone is using. I found that the EMU was giving me about 5.5 msec dwell at idle, but under 4msec at all positions off idle (this trace is at 2,000 RPM) As you can clearly see, the coil is not fully saturating here, so WTF!? Just another reason to go MS over EMU.

Glad its relaxed enough you have time to break it - just don't waste all day online. Everytime I go home, I'm so swamped I couldn't get to a computer if I wanted, then I need a vacation when I get back.

All excited to try the car out. Here's my complete list of shared pins:

Both B/Y A 3 Chasis GND
Both B/Y B 3 Chasis GND
Both B/L C 3 Chasis GND
Both B/R F 3 Sensor Ground
Both B/Y A 3 Chasis GND
Both B/Y B 3 Chasis GND
Both B/L C 3 Chasis GND

And complete list of ECU only pins
OEM BR/Y L 1 Data Link Connector
OEM G/R H 1 Const. Power (30A Fuel Inj Fuse, feeds Main Relay)
OEM W/B D 2 Instrument Cluster
OEM I/R A 1 Const. Power (80A Main, 20A BTN1 fuse and 10A Room Fuse)
OEM G/W D 1 Data Link Connector
OEM W/L E 1 Instrument Cluster
OEM G F 1 Brake Switch
OEM LG/B L 2 Mass Air Flow Sensor (B, Signal)
OEM GY/R O 1 Generator
OEM GY T 1 Generator
OEM G/O K 2 Data Link Con, Trans Ctrl Mod, Inst. Clust
OEM BR/R Q 1 Insrument Cluser
OEM BR/B K 3 Insrument Cluster
OEM P/L B 2 Air Intake Temperature Sensor Signal
OEM/none? BR/W I 3 Cruise Control Module + Clutch Switch


Basically, the computer only "shares" power and ground, and only sees :
brake switch, alternator, AIT, clutch/cruise switch, brake switch, MAS and the OBD-II port and guage cluster. I imagine I can lose the brake, clutch, and MAS (that's only on there so I can spy with OBD-II readers to see what it looks like) meaning I'll be down to guages and alternator stuff.

Once I put an alternator box from a mopar in, I'll be down to guages. It seems I'll QIUCKLY be rid of the OEM ECU altogether.

But tonight, try to start it. I have a drooping idle, but havent tracked it down. Maybe BOV or intake manifold leak or...?

Before you go chucking the alternator, figure out how that control line works. The two wires- one is a voltage sense and the other is field coil control (one's an in, one's an out.)

If you can document the relationship between them (and CLT) then it just might be possible to make the MS drive the alternator. Wouldn't that be cool!

Oh, totally cool. Only, I'm running out of ins and outs, and it's not even baking me cookies yet. What the hell?

I was hoping to be able to free up 40 or 50 of the 14 wasted ground pins on the DB-37, where I'm sure 4 or 5 would do - and recycle them. But since I've gotten no information about whether they are connected on an inner plane, I'm just assuming they are, so it's looking like anothe connector is the answer.

Anyway, right now I'm thinking of buying the $12 NAPA regulator which will do it. Otherwise, it may involve something more sophisticated than a silicon relay. I'd like to incorperate it on the board, but if I have to make an analog measurement and have an analog output, I think I'm going to waste functionality I could employ elsewhere.

All down the road. Right now it's making a cable as described above and plugging it in to see what gives, on the stock 265 cc injectors. If they work, in go the 650's. If it'll start and idle there, I might take it out of the garage.
-Abe.

This is really just for notes. Maybe it'll help someone someday.

Joe, you'd mentioned the amount of stuff in the car there to fail you on emissions. Well, I found another wire I didn't know what to do with:
CDCV. It's a solenoid that works in concert with another and with the vacuum sensor in the tank to both keep you from polluting (good) and make things more complicated (bad). I'm just hoping not hooking this stuff up doesn't do something dumb like have me always sucking fresh air in through the charcoal canister. I'll try leaving it off, initially. And maybe later I'll set up a circuit to make it purge that ONCE when my temps cross 130* F or something. Wouldn't be hard to do. I can lose the pressure sensor and the self check, but something to clean out the charcoal as described below (for an MPV) might be a sane thing to do.

Not that i'm into pulluting or that I would do this myself. You can remove all of that crap and replace it with a small filter venting to atmosphere if you don't want to keep it anymore. Pretty sure rmcelwee did that.

Yeah - I was thinking about it - then again, if I ever have to put my car back to stock, that's a big PITA. I thought it might be really easy (just one output) to get that to work. Maybe I could even run it in parallel with my fans - when the fans turn on, it vents the carbon canister. I'm not a fan of check engine lights or hurting performance, but if I can have the same amount of fun and pollute less I'm into it.

Lots of Progress, Major Sucsess, Medium Disaster
 

Sigh. Well, the car didn't want to start, swapped the coil packs back and forth a few times - it kept losing the trigger pulses while trying to start.
I remember someone's comment about things working much better looking at falling edges on the triggers and that solved that issue. It's doing the thing where you have to crank two or three full revs before it knows where it is, but it starts reliably now.

A couple buddies dropped by and got my idling right quick. I was very impressed. Shortening the dwell time substantially picked me up all kinds of smoothness, and instead of barely staying alive at 1800 it idles rock solid at 7 or 800. Still running on the stock injectors.

Fuel seemed to work fine. I had three issues:
1) My battery charger wasn't working and I didn't know, so running my stereo for two days of soldering left my battery down at 10 volts, and was cooking my buddy's alternator while he tried to keep my car working. Ooops.

2) The "test" part of megatune was screwed up. First it ran the coils (which were acting funny) but the time was way wrong - I couldn't get under 5 ms dwell time no matter what I did! Changed some settings, and it started running the injectors *instead* of running the coils. The weird part is it wouldn't go back to running right even after rebooting MT and the MS.

3) Here's the bad news: I cooked a coil! Shit. They are like $600 a set, I'm pretty annoyed about it. Thankfully, I had one coil left over from my original set that went bad, so I've swapped it in. But I really don't want to do this anymore. I didn't flash anything. It got hot, too hot to touch, just from being on. The longer I ran it, the hotter it got. Then I was running on two cylenders.


So, yeah, I'd like a little help on number three! I don't want to keep cooking coils. I'm thinking that I had the "inverted" or active high setting on the coils, and maybe that was wrong? When I was doing the testing (which was buggy so I don't trust it) the pulses were definately down-going. Now I think this was a bad thing. The car was running like that, save for a couple issues. I had to set the trigger offset to 18 degrees to get the spark right on 0, the setting under basic/timing.


Other than that, it idled great, I turned off the idle valve and it was still idling solid. The IAC works (it would hunt like mad when it was still not running right) - but it's LOUD. Turn the car on and you can hear it pulsing away. It can't be good for it. My guess is that the frequency is too high, I'd love to know how to set it right. I have a couple of scopes, but don't know what to look for. So tomorrow, if I can find a new battery, I'm going to try to get it running again - and check into the weird issue where I'm not sure my injectors I bought will fit my car.

Abe- this reminds me of a problem that some 1.8 users were having- frying their coils after installing an EMU.

I never actually witnessed it myself, but my understanding is that it turned out that while the documentation for the EMU said to set the ignition triggers to 5 volts (which was correct for a 1.6) a couple of 1.8 owners scoped their ignition outputs and determined that they were +12. Setting the EMUs ignition trigger output to +12 apparently solved the coil-eating problem on these cars.

I wonder if on an NB the ignition outputs pullups ought not to be from the +12 line?


Did you look at my attachment above? The coil primaries begin to conduct when the trigger goes high, then discharge when it goes low. I'm pretty sure this is the same for all Miatas.


FWIW, I scoped the IAC valve on a 1.6, and it operates at 160Hz. On the MSI anyway, that's 10000/x = 62.

Huh. Interesting. I assumed it was 5V. I set the settings opposite from the directions, and got to put the timing back where it belongs instead of 17 degrees off. :-) After playing with the scope, the best dwell ended up being 5.0ms, which is exactly what the DIYAT docs say. Maybe they said 4.5, but anyway, it's just where the coil current starts to taper off.


Yeah - I think I read some MS-I stuff, you guys use a different circuit maybe for driving them. My circuit inverts, so what the ECU thinks is active high is really active low. Anyway, it's great right now. My biggest issue: The f'ing alternator doesn't work, battery is dead. So much for keeping the OEM ecu. The only think it's supposed to be doing fails. I'm off to NAPA to get one of those quick-and-dirty regulators.

How do you know? You mean, you look at how often the OEM pulses it? My idle valve is the loudest thing in the car. It vibrated the whole intake, you can hear it out of the air cleaner. It works great, but seems wrong. I gotta read up on that, right now I'm just happy to be idling, watching the battery slowly tick down.

I know. Aren't I hot?

Oh! Now there's a picture. :-) My buddy was telling me (for reasons I cannot comprehend) that saturating the coils leads to weaker spark. Not sure I get it, but it does run very very well when you turn it off JUST before it gets there. It's where I set mine, adding in a half ms at a time. I'll revisit this when everything's running well.

Yes. Remember that current probe I showed you? I used that to measure one of the two wires going into my IAC and found that the current through it was being pulsed at 160Hz.

On MSI anyway, you enter the frequency as 10000/x for some reason. DIYAutotune puts 62 in that box (which equates to 161Hz) so I was basically confirming their observation. Should be easy enough to repeat this experiment on the MSII if you still have the stock ECU available to drive the IAC.

Yeah - I might have to hook the OEM back up for that. I dunno. I used my leftover 50 mOhm resistor for measuring current on the dwell time on the coils and it worked out great, the pins were easy to remove on the coilpacks. I could likely do the same there. I've tried just wrapping wire about a line before, it basically doesn't work too well. :-)

Still, it's an open question to anyone if you know the right frequency to use for an '00 IAC, or have a reasonable guess.


Now, my real reason for posting: I HAVE NO TACH! I have no idea where to get a tach signal from, nor what the dashboard is used to seeing, or where it gets it. Someone MUST be running a stand alone. Maybe all I need to do is feed the tach the same signal I'm feeding the MS, 4 pulses per rev. Anyway, I don't care what or how, but having the needle just sitting there sucks - and I have NO idea how to start on this, though I guess I could get out the scope and trace the line, I'm hoping someone is just running a MS-stand alone, be it MS-I,or MS-II. If someone knows how to get the MS-II to put out a signal, all the better. I have more unused output circuits.

Other than that, yeah, it's idling, running the IAC valve (got the PID set up ok, but trying to get it to idle without it. It's working, but I don't yet know how to smooth it out), charging, and just generally "working". Stumbles like mad from like 2k-3k, so I have to look at the tables, but I'm not there yet and I'm falling asleep at the wheel. :-)

Anyone know how the tacha and speedo work? They both go into the computer, then back out to the guages, and I found no docs on either. I don't even know which color wires go to which guage.

I can't imagine it's worth having the MS watch speed...

I guess there's a microprocessor IN the guage cluster, so I don't know where it gets speed from at all..... I have this nice tach circuit, but nowhere to feed it yet.

The microprocessor is in the small box behind the Odometer. The speed signal is derived from the speed sensor in the gearbox. IIRC the output signal from the speed sensor is 5V square-wave and the output frequency at 60mph is 133Hz. The tacho, I think 12V, square wave and output frequency at idle is in the 25Hz range (a complete pulse every 40ms on the scope).

Jim

Ah, well, as long as the speedo is right off the tranny, I can ignore it. And I like ignoring things. :-)

The tach is more confusing. "Idle" on my car is 800 rpm. I shipped off the numbers to NASA and their super computer came back with:

800 rpm / 60 sec/min = '13 and change'.

Which means you're seeing a two pulses per revolution? That's weird! I could see four (just passing the crank sensor) and I could see 1 (passing the ignition pulse for channel A).

Hmmmm, ok, I guess it's any-time-a-cyl-fires.

Ok, well, even if the MS can't put that out on it's own, it'd be trivial to have an OR gate feed a transistor and put a 12V square wave to the tach. Weird it's 12 volts though. Maybe I could just watch the power line feeding the coils? Seems risky.

On the NAs, tach is active-high 1 pulse per ignition event or two pulses per revolution.

Can't see why it would be any different on the NB. Forget an OR gate, just use a pair of diodes. (ok, technically a pair of diodes is an OR gate...)

She Runs!!

Oh yeah. It's my birthday, have a party. :-)

A little story telling, questions at the end.

I spent like hours sitting next to the car, trying to perfect the idle..... Realizing that I had such a wide range of values where it worked that I had no idea when I was making it better. Then Arga sent me some files, an older version of the code he said worked better. Somehow I thought it started faster - it starts well but takes a while to do it (syncing late?) though this morning, oddly, it didn't start so fast.

It was very cool taking off from a stop sign, deciding it felt underpowered at the stop-and-go type clutch feathering, and running into a parking lot to work on it, and five minutes later having the car better-than-stock at that one simple task.

The speedometer works, and so does the cruise control - but I still have the OEM ecu plugged in (it's back behind everything else). I'm going to pull it, and if it works, the link piggy, o2 clamp, o2 simulator, and the old knock sense come out (actually, I could run the two in parallel till I get the settings right.)

Now I'm going to make another thread for the weird tuning issues, try to keep this segmented, easier to follow.

Just an FYI - figuring it wasn't doing anything anyway, I unpluged the OEM ECU today. She ran fine. Speedo works, cruise control works. The only thing missing is EGR. Oh, and I think it took the CEL with it. :-)

Once I get the tach working, I'd say I'm back up to normal (you know, minus the 5k stumble and the no boost, etc. ha!).

Basically, buying the connectors for the OEM ecu was a total waste of good money. Anyone looking to do a late model car, buy the mopar altenator controler with the money you would have spend on connectors. I can turn my fans on and off as I like, same with VICS, and after a bit of messing with it, the idle even seems to be working nice.
-Abe.

Ok, after a whole day of messing around, soldering, unsoldering, etc... And getting traces just like yours first for the crank sensor (it would "stick" every 4th tooth) and then for the cam sensor (didn't happen till I fixed the crank) I got it all working *perfect*.

I get the same signal before and after my simple input circuit. Right now I have ~880 ohms to the sensor, and a 10k pull up on the back side of that to +5V - that intersection between the two resistors feeds my 1970's 2n2222A.

The issue was: Sensor position. The crank is speced ~0.025"-0.050" or so, and it was very nearly 50. I thought it was too close, but just for the heck of it, I moved it into ~25 mils. Bingo, it's been flawless since.

Then the cam sensor was weird, and since it's not adjustable, I didn't know what to do. I moved it radially in and out from the sensor as far as I could, no effect. Then I loosened the bolt, pulled it away from the valve cover... and bingo, great signal. The "sticking" went away. So I put a washer under the bolt, an o-ring under the sensor, and now all my signals are perfect.

Basically, the thing you forgot to check is the input of the circuit. It's not your VR circuit (I don't think you want to use a VR anyway), it's the sensor itself. Why the sensor got weird, I don't know. Might be a sign of something bigger around the horizon.


Anyway. Now you can see that my signal inputs match exactly my signal outputs (to the MS-II ecu, one read on the 40pin DIP, one read at TSEL). The signals look awesome, too.

My issue now is: I still don't sync! And I'm out of ideas about why. I thought it was a problem with the coils, but no, they work perfect (recently switched to COP) when testing. And things don't improve if I unplug them.
Unless it has something to do with my pulling the fuel pump relay (and I doubt the current to that is being measured) I'm at a loss.
I tried it with b15 and arga's file on my old chip, and b17 and a cleanly generated file on the new chip. Neither one is working - I get "sync" for a moment once every few revs.

I wish there were a way to log the pulses the computer counts, when does it see a falling edge?

pics:
Sticking Crank Sensor:
http://abefm.smugmug.com/photos/240163520-M.jpg

Sticking Cam Sensor:
http://abefm.smugmug.com/photos/240163667-M.jpg

Fix for cam sensor:
http://abefm.smugmug.com/photos/240163890-M.jpg

Trouble with too small a pull up resistor on Cam:
http://abefm.smugmug.com/photos/240163864-M.jpg
Note how it can't quite sync all that current, and the tops are wavy. This was with 1700 ohms total between sensor and 5V. At 11,000 ohms, it's fine.


What the sensor outputs look like, with 11k of pullup to 5V:
http://abefm.smugmug.com/photos/240164108-M.jpg


Finnally, this is what my car is doing:

http://abefm.smugmug.com/photos/240164291-L.jpg
The smaller traces are the inputs to the ECU, the larger ones are the raw signals.

I would really like some help with why my car won't sync! Could it be because of noise? My guess is it has to do with the different spacing of the pulses. My cam pulses are NOT co-incident with the crank pulses
Higher res/additional pictures available at:
http://abefm.smugmug.com/gallery/4117931#240164291

Man. You know a $75 CAS shipped to your door would have given you so many more hours of drive time while you figure out these issues... :(
What is the world could be different between your car or MS build and Argas.
I can offer some encouragement though. That cam sensor fix is pretty darn clever. I had to replace my crank sensor, used a feeler gauge it space it within factory specs. The previous owner had broken it and epoxied it out of line. I think he was trying to adjust the timing that way, similar to what FM does with their bolt on timing wheel.

From the MS forums:

And here is the the data for the M2 (NB) Miata, 1999-2005.

The numbers are crank degrees relative to the #1 TDCC (negative being before and positive after)

Crank signal edges in one engine cycle ( engine cycle = 720 crank degrees), 16 edges total:

falling edge at -80 ( == 640 )
raising edge at -77 ( == 643 )
falling edge at -10 ( == 710)
raising edge at -7 ( == 713 )
falling edge at 100
raising edge at 103
falling edge at 170
raising edge at 173
falling edge at 280
raising edge at 283
falling edge at 350
raising edge at 353
falling edge at 460
raising edge at 463
falling edge at 530
raising edge at 533

Cam sync signal edges in one engine cycle, 6 edges total:

falling edge at 37
raising edge at 57
falling edge at 377
raising edge at 397
falling edge at 421
raising edge at 441

Jim
_________________

Abe, how are you taking control of the VICS system?

Er, just using a generic output, and grounding the line when I'm at 4k rpm. It's probably not the right number, but it's a start.

Thought so, just checking.

http://abefm.smugmug.com/photos/243230862-L-0.jpg

Injectors all installed. With some of the carbon canister and EGR stuff removed and the VICS stuff redone, it's a whole new engine bay. Not to mention the Failin' Miata fuel rail.

'Course, I didn't have the right o-ring grommet things, so I made some:
http://abefm.smugmug.com/photos/243230559-X2.jpg

The final ones were a bit better. Straighter. The diameter is a bit small... Not sure I should sweat it.

Finally, for comparison, yes the Evo 550cc injectors (from a 2005, supposedly) were under $100 and fit great!
http://abefm.smugmug.com/photos/243230642-X2.jpg


Really, though, I just wanted to use the phrase Failin' Miata. All their shit breaks, what the hell?

So, with these peak-and-hold injectors... weirdness happens. If you follow the instructions in the megamanual, you will quickly end up with 0% current setting. Why it still runs, I don't know. I have a guess that the opening time is too long.

Anyway, everything was hunkey dory, got an awesome idle (well, after a huge scare getting the motor up to 230+ degrees, because my table for the water temp sensor wasn't right - this was all at idle, who knows, I guess it's ok though)..... But at ~4k, things go dead lean. Super weird lean. I had the VE table looking like a step function.

It would make the SMALLEST of differnces, if any. Then, I put the current to 10 % and it gets way better. So now it's at 25% which I'm sure it too much. Need to figure all this out, it probably will invovle more work than I want and an oscilliscope.

Luckily, I have a supply of 50, 25, 20, 15 and yes: 1 milliohm resistors. :-)

So the moral of the story, if the numbers seem unreasonable, go with your gut - the manual isn't always right.

Now I get to do a compression test, then maybe this weekend I'll tackle trying to get the injector numbers figured out. If I was smart I'd pull-and-flow test them, they seem to click weird in test mode.

Ok, am I going crazy here? The limited advice on the forums seems like total guesswork. I've got a very simple circuit for an input, and I MIGHT be having noise issues.

They are telling me to take out my transistor inverting follower and use their opto-in.

http://abefm.smugmug.com/photos/244097764-L-0.jpg

But I see them as functionally the same, as long as I keep my voltages below, oh, 3000 volts. And the bottom circuit is already built.

I guess I could do it like this:
http://abefm.smugmug.com/photos/244124058-XL.jpg
But it is actually any better than just adding a bigger cap to the circuit I've got?

I think I'm just missing one tooth every great once in a while and the software sucks at dealing with it.

<edit> Maybe I should swap the green line "jumper" for a ~470 ohm resistor, just to give that cap a fighting chance.

Well, progress, of a sort.

Since I was dead tired last night, the only* work I did on the MS was to put on bigger capacitors. I put them on, the motor started right up, so I went to bed. On the way to work, I thought everything was awesome, while it was cold. Less stumbling at low RPM.

Then things warmed up and it went to total shit, barely could rev past the 5k rough spot.

So, an order of magnitude bigger caps doesn't work. Guess I'll try smaller. :-) I should get it on the scope and look at my traces. Maybe the sensors chatter somehow, though if they are true hall effect I don't see how they would.



*I lied. Someone mentioned grounding and power not being good enough, so I ran two 14 guage grounds to the block in addition to the OEM harness grounds, AND I added a 14 guage +12V switched directly to the main relay block.

I liked that input circuit they listed on Pat's thread at the MS2E dev board where they run the crank signal "backwards" into the LED on the opto circuit then run 5v into the other end. I liked the first one from the Neon better than the second. It should work nicely stand-alone or parallel. I'd try it but mine is working well and I really don't want to screw with it.

When your stumbling at revs over 5k, are you loosing sync?

Wow! Yes! It's working!

Ok, first off: Yeah, using the grounding of the circuit to complete the circuit on the opto is pretty cool. Definitely the "right" way. That guy is on crack telling me not to. :-) I think I drew it up more or less right there. Too much capacitance.


So, the news: I went ahead and installed 0.001uf (1nf) resistor, and BINGO! Still started, and drove fine! I had *one* hiccup on the way home (15 min trip), and one on the way back. Not sure if there's some correlation (time, temperature, etc). But better than 20+ per trip, sometimes 5 or six in ten seconds.

Now, the weird news. I had put the cam sensor back to the stock location, and I *thought* it was starting slower. When I got out of my house from lunch (having completed one trip on the set up and changing nothing), it wouldn't start. No sync.

Well, I remembered last time moving the cam sensor helped, so I did (wiggling the wires doesn't seem to do anything), and it started up (not a particularly fast sync, but not a slow one).

So, in short, my circuits are these:
http://abefm.smugmug.com/photos/241407267-L.jpg
But I'll have to double check the resistance values, and the caps are 1nf.

Overall, I'm pretty happy. A day or two of messing with it, getting a feeling for it, and I'll be adding boost. Time to look at redoing my tables, I guess. :-) I'll play with different sized shims on the cam sensor. I have a feeling what helps it start is bad at runtime, but that's not based on any actual facts. :-)

Frank - please post what you're using right now, exact values, etc.

I'm running the same thing I sent you a few weeks ago. Both cam and crank circuits are the same.

No pull-up on the line to the sensor (stock ECU handles that)
signal input -> 10k resistor -> transistor base
Vcc -> 4.7k-> emitter
collector -> ground
the line to TSEL (J10 for the cam) is pulled between the emitter and the 4.7k

Attachment 214787

Attachment 214788

Attachment 214789

It's really strange you can get away without any capacitance. There's a chance the caps in the OEM ECU are doing this for you, but I'm pretty impressed. Then again, without another transistor to buffer out the noise, maybe I'm more susceptable to it.

I should give that a shot.

Thinking about the code, it doesnt' care on the leading/trailing edge of the cam pulse (so inverting it doesn't matter) since they happen well within the window of one crank tooth and the next, irrerspective of which edge you look at.

Maybe I'll go back to that part of your set up. And maybe I'll try to figure out if I want the risingor falling edge, though last time I messed with it I found the one I was using worked best.

Abe, how is your idle valve behaving?

Jim

That reminds me, I need to make a public apology to Neogenises. I told him multiple times the signals needed to be the same polarity but it doesn't look like it matters.

Ha. Yeah. Well, thanks for rubbing that in my face, Rev. :-P Here I was all prepared to tell you about the only issue with my car, when you ask me about the idle. :-)

To get that out of the way, the idle is significantly worse with the idle hooked up than not. It's noisey, way noisey. It used to work when I had it on ~70hz, but it was far and away the loudest sound in the whole car below ~1800 rpm. So I moved it up to 160 or something (you only get two choices for freq in MS-IIx). It got queiter, and I never really got it tuned. Then I switched off PID and it's been idling well, except that it idles at like 550 rpm when headlights and fans and heater are on.
They changed the way the controller works in b17, and I can't get it working nice like I had it in the earlier version. It's just much slower, it STARTS to work when you max the Proportional Gain, but to get it right I think I'd have to be at like 210% gain, not 100% (the max). Arga, are you running beta 17 or 18 yet?

Anyway, the real news:
Remember how I said my car magically wouldn't start? Well, did the same thing this morning, but I couldn't fix it. I put the cam shim back in, didn't help. It seems like it's not firing? I was checking timing yesterday (took the ground lead, put in a length of wire with like 30-50 wraps, put it around my timinglight pick up, and then let it see ground) which worked (though I'm not sure how valid this is, timing wise). Anyway, put it all back, I thought things were cool... But now no start. Oddly, sometimes it says synced. Most of the time, actually.

Here's the weird part: RPM are reporting way low, and dropping. Instead of 250, typically what I see, it's 180 and drops to 50 pretty quick.

So, something is way wrong. I guess I need a scope. Something is wrong, period, but I don't know what, since Arga has no filter caps at all.

Crack the stock ECU open and reverse engineer what they have.