Baseline MSQ for '99-00, including Abe's Board
09-04-2009, 05:04 PM
#62
Senior Member
Join Date: Nov 2007
Location: Belgium
Posts: 999
Total Cats: 73
.Congrats on getting it running though!
I made the same mistake of swapping the coils. Take a look at them. One of them has a blue dot painted on it. The corresponding connector has a blue piece of tape on it. I only found that out after going through the same ordeal as you
.Where did you get the 6µs spark hardware latency from?
On a sidenote, I'm still playing with injector opening time. Currently using 0.92ms (vs 0.88 last week). I'll probably end up at .90 or so. This is on stock injectors.
Last edited by WestfieldMX5; 09-04-2009 at 05:44 PM.
Reply
0
0
09-04-2009, 05:26 PM
#63
Elite Member
Thread Starter
iTrader: (3)
Join Date: Aug 2006
Location: San Diego, CA
Posts: 3,047
Total Cats: 12
Don't forget there's battery voltage correction factors and all sorts of stuff, stuff which is modeled as linear but really isn't in the MS code.
Still, having a good number is better than not.
Really? The blue tape is OEM? My car had this, I assumed it was some PO who'd messed with it. That's awesome. :-)
Your car takes 7 *revolutions* or 7 crank teeth? The MS code looks f... you know, I forgot, but I once knew. It does want to see both sets of cam teeth, though in theory you need, maximally, 3/4 of a turn.
Best case:
Crank pulse, 1 or 2 cam puleses, then you are synced on the next crank pulse (i.e. you know engine speed (ROUGHLY), and exact position.
Middle case one:
Two crank teeth, 1 or 2 two cam, then crank. You have to wait the extra 1/4 turn here. Depending on the position of the cam pulse, you will know speed and might be able to fire before the next tooth since you have position and velocity.
Middle case two:
Three crank teeth. You now know crank position and velocity (you didn't see the cam, do you know which teeth that was), you can fire when ready. The downside here is on full sequential, you don't know which cylinder you're on.
In middle case two, if you're bank fire, you can run. If you're sequential, you can start batch fired till you figure out where you are (should be within 1/4 turn anyway) or you can wait till the cam pulse.
Middle case three:
Two cam pulses, two crank pulses, you know position and velocity
Worst case:
One cam pulse, because now you don't know if you saw THE one cam pulse, or if you started right in the middle of the two. So basically, you ignore that single cam pulse, and you're back to middle case two except you had to wait a few degrees of turning.
Still, having a good number is better than not.
Really? The blue tape is OEM? My car had this, I assumed it was some PO who'd messed with it. That's awesome. :-)
Your car takes 7 *revolutions* or 7 crank teeth? The MS code looks f... you know, I forgot, but I once knew. It does want to see both sets of cam teeth, though in theory you need, maximally, 3/4 of a turn.
Best case:
Crank pulse, 1 or 2 cam puleses, then you are synced on the next crank pulse (i.e. you know engine speed (ROUGHLY), and exact position.
Middle case one:
Two crank teeth, 1 or 2 two cam, then crank. You have to wait the extra 1/4 turn here. Depending on the position of the cam pulse, you will know speed and might be able to fire before the next tooth since you have position and velocity.
Middle case two:
Three crank teeth. You now know crank position and velocity (you didn't see the cam, do you know which teeth that was), you can fire when ready. The downside here is on full sequential, you don't know which cylinder you're on.
In middle case two, if you're bank fire, you can run. If you're sequential, you can start batch fired till you figure out where you are (should be within 1/4 turn anyway) or you can wait till the cam pulse.
Middle case three:
Two cam pulses, two crank pulses, you know position and velocity
Worst case:
One cam pulse, because now you don't know if you saw THE one cam pulse, or if you started right in the middle of the two. So basically, you ignore that single cam pulse, and you're back to middle case two except you had to wait a few degrees of turning.
Reply
0
0
09-04-2009, 05:38 PM
#64
Senior Member
Join Date: Nov 2007
Location: Belgium
Posts: 999
Total Cats: 73
Acutally more like 7 cycles. So about 14 revolutions or at 350ish cranking rpms about 2.5 seconds. Way too long for my tiny Odyssey PC680 battery :(.
Dwell battery adjustment is something that could be added to your initial post.
I used Jason's data and came up with this:
cranking / running : 5.5 / 3.3 ms
battery correction:
6.0 500
8.0 248
10.0 180
12.0 130
14.0 100
16.0 80
Dwell battery adjustment is something that could be added to your initial post.
I used Jason's data and came up with this:
cranking / running : 5.5 / 3.3 ms
battery correction:
6.0 500
8.0 248
10.0 180
12.0 130
14.0 100
16.0 80
Last edited by WestfieldMX5; 09-04-2009 at 05:53 PM.
Reply
0
0
09-04-2009, 05:39 PM
#65
Elite Member
Thread Starter
iTrader: (3)
Join Date: Aug 2006
Location: San Diego, CA
Posts: 3,047
Total Cats: 12
Ha. Sorry, I totally forgot your question.
Timing light at low speed (lowest it is stable), you know now the position of the engine. When it sparks, there's some overhead in turning on those transistors, etc. Let's call is 5 us. It's 50 ms per revolution at 1200 rpm, or 138 us per degree. When you set your timing for "4.0 degrees", it's really 552 (= 138 * 4) us after the 0 mark. In reality, it takes 5 us, so, you're firing 557 us past center.
Now you rev to 7,000 rpm, and you'll see the 24 us/degree. So the computer waits until 4*24 = 96 us after 0, but that extra 5 us is really 0.2 degrees off.
Huh, I must be off in my math, I doubt I could SEE 0.2 degrees with a timing light.
Well, anyway... The point is, you rev the motor up (fixed timing), and add latency until the timing mark doesn't move with revs. Then you know that this error is gone. Maybe I was able to see 0.2*, my guess is I could see 1/2-1/4 degree....
Anyway, if your chain is long (input filtering or output - if you have lots of capacitors on your inputs it'll add lag) this will let you remove it. And you find it by looking for your timing to retard as revs increase, then you need more latency.
Timing light at low speed (lowest it is stable), you know now the position of the engine. When it sparks, there's some overhead in turning on those transistors, etc. Let's call is 5 us. It's 50 ms per revolution at 1200 rpm, or 138 us per degree. When you set your timing for "4.0 degrees", it's really 552 (= 138 * 4) us after the 0 mark. In reality, it takes 5 us, so, you're firing 557 us past center.
Now you rev to 7,000 rpm, and you'll see the 24 us/degree. So the computer waits until 4*24 = 96 us after 0, but that extra 5 us is really 0.2 degrees off.
Huh, I must be off in my math, I doubt I could SEE 0.2 degrees with a timing light.
Well, anyway... The point is, you rev the motor up (fixed timing), and add latency until the timing mark doesn't move with revs. Then you know that this error is gone. Maybe I was able to see 0.2*, my guess is I could see 1/2-1/4 degree....
Anyway, if your chain is long (input filtering or output - if you have lots of capacitors on your inputs it'll add lag) this will let you remove it. And you find it by looking for your timing to retard as revs increase, then you need more latency.
Reply
0
0
09-05-2009, 02:17 PM
#68
Elite Member
Thread Starter
iTrader: (3)
Join Date: Aug 2006
Location: San Diego, CA
Posts: 3,047
Total Cats: 12
Yep. It's all software. I doubt mine's more than 2 seconds, but it's much longer than before.
They really should be able to start much faster... all the info is there, the OEM can do it, but what can I say, it's not a priority for them. And they've killed the community spirit - write a fix for their code, they'll call you an *******, tell you to **** off, then a couple months later quietly sneak in their own version.
Anyway. I've been working on the car more than the software lately, and letting the software go as "good enough". Anyway, time it, if it's really 4 seconds I'll time mine and we can compare notes.
They really should be able to start much faster... all the info is there, the OEM can do it, but what can I say, it's not a priority for them. And they've killed the community spirit - write a fix for their code, they'll call you an *******, tell you to **** off, then a couple months later quietly sneak in their own version.
Anyway. I've been working on the car more than the software lately, and letting the software go as "good enough". Anyway, time it, if it's really 4 seconds I'll time mine and we can compare notes.
Reply
0
0
09-05-2009, 02:19 PM
#69
Elite Member
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
I wish that was the only problem. My car shuts down after 10 seconds. When I say it shuts down, I mean the ECU shuts down, it will power off like you killed the power and it will no longer connect to Megatune. I have to physically reset the power to the MS to be able to start the car again (and have it die after another 10 seconds).
Jim
Jim
Reply
0
0
09-05-2009, 02:33 PM
#70
Elite Member
Thread Starter
iTrader: (3)
Join Date: Aug 2006
Location: San Diego, CA
Posts: 3,047
Total Cats: 12
Have a thread about that? That needs looking into! It sounds familiar, but I can't put a finger on it.
Ah. Could it be that black disc thing, the "fuse"? If you're sucking too much power, it'll over heat.
Pull up a thread for this. Does it matter the code version? If the motor is running? I know I've seen this, I just am drawing a blank.
Ah. Could it be that black disc thing, the "fuse"? If you're sucking too much power, it'll over heat.
Pull up a thread for this. Does it matter the code version? If the motor is running? I know I've seen this, I just am drawing a blank.
Reply
0
0
09-05-2009, 05:28 PM
#72
Elite Member
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
Have a thread about that? That needs looking into! It sounds familiar, but I can't put a finger on it.
Ah. Could it be that black disc thing, the "fuse"? If you're sucking too much power, it'll over heat.
Pull up a thread for this. Does it matter the code version? If the motor is running? I know I've seen this, I just am drawing a blank.
Ah. Could it be that black disc thing, the "fuse"? If you're sucking too much power, it'll over heat.
Pull up a thread for this. Does it matter the code version? If the motor is running? I know I've seen this, I just am drawing a blank.
Jim
Reply
0
0
09-07-2009, 12:33 PM
#74
Elite Member
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
Mazda DOES use blue tape for some parts of their harnesses. I stripped an RX-7 harness some time ago, and it was there, even under black tape (usually over crimped wires, ie two ground wires joined to one).
Jim
Jim
Reply
0
0
09-20-2009, 01:39 PM
#75
The 1-2 revolutions is something that should be taken care of in software, but it's how it is and you get used to it really quick.
What's funny, if there's fuel in the manifold, it'll catch, run for a cycle or so, then die till the MS "notices" it's running, then it'll run again. :-)
What's funny, if there's fuel in the manifold, it'll catch, run for a cycle or so, then die till the MS "notices" it's running, then it'll run again. :-)
Everynow and then, i do get a sync drop though. Its random.
Reply
0
0
09-27-2009, 04:55 AM
#78
Abe, they arent resets. They are loss of sync problems.
I datalogged while i was cruising for about 75-80 mph, and I got random "loss of sync" events where my car bucks for a split second.
The Loss of sync Reason is 32. Is there a way to alleviate this problem?
I datalogged while i was cruising for about 75-80 mph, and I got random "loss of sync" events where my car bucks for a split second.
The Loss of sync Reason is 32. Is there a way to alleviate this problem?
Code:
unsigned char syncreason; /* this variable will hold reasons that trigger wheels lost sync e.g. 0 = no problem 1 = init error 2 = missing tooth at wrong time 3 = too many teeth before missing tooth (last) 4 = too few teeth before missing tooth (last) 5 = 1st tooth failed test 6 = nonsense input (last) 7 = nonsense input (mid) 8 = too many teeth before missing tooth (mid) 9 = too few teeth before missing tooth (mid) 10 = too many teeth before end of sequence 11 = too few teeth before second trigger 12 = too many sync errors 13 = dizzy wrong edge 14 = trigger return vane size 15 = EDIS 16 = EDIS space for more common reasons plus other special reasons for the custom wheels 20 = subaru 6/7 tooth 6 error 21 = subaru 6/7 tooth 3 error 22 = Rover #2 missing tooth error 23 = 420A long tooth not found 24 = 420A cam phase wrong 25 = 420A 26 = 420A 27 = 420A 28 = 36-1+1 29 = 36-2-2-2 semi sync failed 30 = 36-2-2-2 tooth 14 error 31 = Miata 99-00 - 2 cams not seen 32 = Miata 99-00 - 0 cams seen 33 = 6G72 - tooth 2 error 34 = 6G72 - tooth 4 error 35 = Weber-Marelli 36 = CAS 4/1 37 = 4G63 38 = 4G63 39 = 4G63 40 = Twin trigger 41 = Twin trigger 42 = Chrysler 2.2/2.5 43 = Renix 44 = Suzuki Swift 45 = Vitara 46 = Vitara 47 = Daihatsu 3 48 = Daihatsu 4 49 = VTR1000 50 = Rover #3 51 = GM 7X 52 = 36-2-2-2 tooth 30 error 53 = rc51 semi error 54 = rc51 re-sync error tooth 6 55 = rc51 re-sync error tooth 16 56 = rc51 re-sync error tooth 18 57: Nissan CAS: wrong pattern, e.g. slot length increases but should decrease 58: Nissan CAS: longest slot test failed */
Reply
0
0
09-28-2009, 01:00 PM
#79
Elite Member
Thread Starter
iTrader: (3)
Join Date: Aug 2006
Location: San Diego, CA
Posts: 3,047
Total Cats: 12
Well, huh. I've been meaning to do some logging on my own as I occasionally see bucking which I didn't in earlier software revisions. If it's software or not, it's hard to say, all I know for sure is many people complain of issues where they didn't have them before. There's a few things to do, in two categories:
1) Assume it's in the firmware
a) Turn off all the noise handling in case it's messing things up
b) Turn on all the noise handling, in case it's helping
c) Go to an earlier version of the code. I might try this.
2) Assume it's real
a) Check what capacitor value you are using on the board. If removing the cap makes it worse, there's a noise issue
b) Spacing between cam and sensor could somehow be on the edge, I'd moved mine around a bunch with spacers until I got it working better, but the new circuit worked better still
c) Play with the values in the circuit that reads the cam. It would be nice to know for SURE it was a missing tooth and not an extra one, if that's the case (same as with the missing cap) you might be able to make the filter less strong, to help the tooth signal get through.
There's also the chance what's really happening is it's picking up an extra crank tooth, then sees the cam pulse as missing. Check your logs for where this happens, post it, I'm curious about hte "timing error" there, right where it happens. Best of all is the tooth logger, I don't know if you want to run that as long as you'd have to (or if you even can) to catch an event, but that would really tell you what's going on. A circular buffer looking for errors would work but good luck getting them to code that.
I'd say pull the cap on that line, and if it helps enough, awesome. I imagine it won't. If it makes it worse, let me know. Then we can try swapping a resistor or two to ease it up.
Also, play with the software "fixes" for it, noise filtering on the cam signal.
1) Assume it's in the firmware
a) Turn off all the noise handling in case it's messing things up
b) Turn on all the noise handling, in case it's helping
c) Go to an earlier version of the code. I might try this.
2) Assume it's real
a) Check what capacitor value you are using on the board. If removing the cap makes it worse, there's a noise issue
b) Spacing between cam and sensor could somehow be on the edge, I'd moved mine around a bunch with spacers until I got it working better, but the new circuit worked better still
c) Play with the values in the circuit that reads the cam. It would be nice to know for SURE it was a missing tooth and not an extra one, if that's the case (same as with the missing cap) you might be able to make the filter less strong, to help the tooth signal get through.
There's also the chance what's really happening is it's picking up an extra crank tooth, then sees the cam pulse as missing. Check your logs for where this happens, post it, I'm curious about hte "timing error" there, right where it happens. Best of all is the tooth logger, I don't know if you want to run that as long as you'd have to (or if you even can) to catch an event, but that would really tell you what's going on. A circular buffer looking for errors would work but good luck getting them to code that.
I'd say pull the cap on that line, and if it helps enough, awesome. I imagine it won't. If it makes it worse, let me know. Then we can try swapping a resistor or two to ease it up.
Also, play with the software "fixes" for it, noise filtering on the cam signal.
Reply
0
0
