standalone question on 99
07-23-2008, 08:33 PM
#21
Given the support, the capabilities, and the cost, there's no good reason not to go MS - there's not much the other systems do the MS doesn't - unless there's some feature you want, go ms. And, since it's open source, well... I didn't like the input circuits, I made my own. I didn't like the way it idled, I wrote my own idle routine.
The board comes assembled or not. It's your choice, but you learn a lot more about how it works if you do it yourself, so I'd recommend it. Besides, you're going to have to do minimal modifications on it anyway, so you might as well know what's what so as to be more comfortable.
The board comes assembled or not. It's your choice, but you learn a lot more about how it works if you do it yourself, so I'd recommend it. Besides, you're going to have to do minimal modifications on it anyway, so you might as well know what's what so as to be more comfortable.
Reply
0
0
01-03-2009, 11:37 AM
#22
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
Told ya I'd post when I found more info. I found a response to error vs. trigger wheel teeth. This is from Ian, one of the guys that knows how the guts of the Autronic systems works. I'm not sure how much of this is applicable to other systems:
What follows is the scope trace of the coil output with a fixed timing value and an acceleration of 4100 RPM/sec with 4 pulses of a one pulse per cycle trigger wheel (2 crankshaft revolutions). The first pulse is fairly retarded, the 2nd is close but still a bit retarded, and the last two are dead on.
There are a few other things that need to be considered before everybody starts fitting multiple extra teeth to all their cranks.
All Autronic ECUs do a Load measurement at the same time as the Cylinder Pulse that occurs at the Cylinder I/P Lead angle. When the Trigger Events / Cycle value is the same as the number of engine cylinders (a normal configuration) then a Load measurement is taken at the same relative point on every cylinder. For example if the Cylinder I/P Lead angle is 60 deg then a Load measurement will be taken at 60deg BTDC on every cylinder. This means the Load value will be most stable because manifold reversion pulses, etc will also be occuring at the same relative times on each reading.
If you increase the number of Trigger Events / Cycle then the ECU will also be taking additional Load measurements. If you double the Trigger Events / cycle there will be two Load measurements per cylinder NOT at the same relative positions. If you have a 4 cyl engine with 60deg Cyl I/P Lead there will be alternating Load measurements at 60deg and 150deg BTDC effectively on each cylinder. If there is a significant reversion effect in the manifolding this may appear as an unstable Load reading (which some other ECU's seem to have as standard). This will also affect Manifold Rate fuel trims if the cal is using that in a Trim table.
Extra Trigger Events / Cycle could also have a similar effect on Ignition timing. A lot of people are aware that if you set a flat ignition table and watch the timing with a timing light when you suddenly accelerate the engine with no load you will see an Autronic retard the first pulse and then be correct for the rest. This is with one pulse per cylinder. If you do the same test with other ECU's and one pulse per cylinder you will see the timing retarded the WHOLE time the engine is accelerating. If you fit three pulses / cylinder to these engines for the other ECU's the retard effect nearly goes away.
The Autronic has a very good predictive timing algorithim that works with one pulse / cylinder. Engines do not run continuously smoothly even though they appear to. They slow down on compression strokes and accelerate on combustion strokes. Introducing extra Trigger Events / Cycle could destabilise the predictive timing in a similar way to the Load calculation. Using a camshaft driven Cylinder pulse has the same effect (seen as timing scatter). All Autronic CPU's up to SM4 v1.05 (& all SMC /SM2) do the prediction the same way...
All Autronic ECUs do a Load measurement at the same time as the Cylinder Pulse that occurs at the Cylinder I/P Lead angle. When the Trigger Events / Cycle value is the same as the number of engine cylinders (a normal configuration) then a Load measurement is taken at the same relative point on every cylinder. For example if the Cylinder I/P Lead angle is 60 deg then a Load measurement will be taken at 60deg BTDC on every cylinder. This means the Load value will be most stable because manifold reversion pulses, etc will also be occuring at the same relative times on each reading.
If you increase the number of Trigger Events / Cycle then the ECU will also be taking additional Load measurements. If you double the Trigger Events / cycle there will be two Load measurements per cylinder NOT at the same relative positions. If you have a 4 cyl engine with 60deg Cyl I/P Lead there will be alternating Load measurements at 60deg and 150deg BTDC effectively on each cylinder. If there is a significant reversion effect in the manifolding this may appear as an unstable Load reading (which some other ECU's seem to have as standard). This will also affect Manifold Rate fuel trims if the cal is using that in a Trim table.
Extra Trigger Events / Cycle could also have a similar effect on Ignition timing. A lot of people are aware that if you set a flat ignition table and watch the timing with a timing light when you suddenly accelerate the engine with no load you will see an Autronic retard the first pulse and then be correct for the rest. This is with one pulse per cylinder. If you do the same test with other ECU's and one pulse per cylinder you will see the timing retarded the WHOLE time the engine is accelerating. If you fit three pulses / cylinder to these engines for the other ECU's the retard effect nearly goes away.
The Autronic has a very good predictive timing algorithim that works with one pulse / cylinder. Engines do not run continuously smoothly even though they appear to. They slow down on compression strokes and accelerate on combustion strokes. Introducing extra Trigger Events / Cycle could destabilise the predictive timing in a similar way to the Load calculation. Using a camshaft driven Cylinder pulse has the same effect (seen as timing scatter). All Autronic CPU's up to SM4 v1.05 (& all SMC /SM2) do the prediction the same way...
For the purpose of this thread the v1.06 and onwards have an additional parameter called the Digital Angle Filter. When set to Fast it operates the same as earlier SM4 & SMC / SM2 CPU's. When set to Slow the prediction is averaged over more Cylinder Pulses to minimise scatter.
Also when using dwell the v1.06+ has a wider tolerance than v1.00-v1.05 before applying dwell extension. It is a complex scenario because the dwell pulse may be initiated several Cylinder pulses before the timing point to achieve the desired dwell time. If the engine speed increases suddenly (after initiation of the dwell pulse) and the output is triggered at the correct position the dwell pulse will be shorter than planned. Depending on the coil minimum dwell requirements the short pulse may undercharge the coil resulting in a missfire. Under certain circumstances the SM4 (and the old dwell boards) will extend the dwell to a minimum value to ensure a spark, which will be slightly retarded, rather than risk a spark missfire. In pre-v1.06 CPU's the tolerance was found to be a bit conservative so it (the tolerance) has been widened in v1.06 and later to achieve better timing accuracy when accelerating.
For the 6 tooth Subaru crank the Autronic uses the 65deg teeth for all timing. The Subaru uses the 10deg teeth for cranking position. When cranking it turns the dwell on with the 65deg tooth and off with the 10deg tooth. Maybe the Subaru ECU couldn't calculate the engine position accurately at cranking so it needed an extra tooth at the cranking ignition point. That then raises the question of ECU's that require cylinder pulses at the cranking position - how good are they if they need a tooth there?
Also when using dwell the v1.06+ has a wider tolerance than v1.00-v1.05 before applying dwell extension. It is a complex scenario because the dwell pulse may be initiated several Cylinder pulses before the timing point to achieve the desired dwell time. If the engine speed increases suddenly (after initiation of the dwell pulse) and the output is triggered at the correct position the dwell pulse will be shorter than planned. Depending on the coil minimum dwell requirements the short pulse may undercharge the coil resulting in a missfire. Under certain circumstances the SM4 (and the old dwell boards) will extend the dwell to a minimum value to ensure a spark, which will be slightly retarded, rather than risk a spark missfire. In pre-v1.06 CPU's the tolerance was found to be a bit conservative so it (the tolerance) has been widened in v1.06 and later to achieve better timing accuracy when accelerating.
For the 6 tooth Subaru crank the Autronic uses the 65deg teeth for all timing. The Subaru uses the 10deg teeth for cranking position. When cranking it turns the dwell on with the 65deg tooth and off with the 10deg tooth. Maybe the Subaru ECU couldn't calculate the engine position accurately at cranking so it needed an extra tooth at the cranking ignition point. That then raises the question of ECU's that require cylinder pulses at the cranking position - how good are they if they need a tooth there?
Reply
0
0
01-03-2009, 04:05 PM
#23
Wow, well, the guy knows the insides of the autotronics, you have to say that. Of course, really what he's saying all comes down to how THAT computer handles it. For your computer, I'd go with it. :-)
FYI, though, some European stuff, audis, porsche, etc, use several HUNDRED crank triggers - they read the starter gear on the flywheel. Go figure. Of course, they also have to other references, one is a cam, and the other I'm not sure - both significantly lower. I've also heard that the 300zx uses some obscene number of pulses, over 100 (maybe 150?) on the cam.
Really, you can always choose to ignore extra pulses, till simply the computational load of ignoring them becomes to high. However, I tend to agree, certainly reading them in the same spot and some math should get you there - the extra pulses are only for checking your work or looking for something weird (I've heard tell of pre-ignition sensors based on the engine slowing down slightly or at the wrong time.)
Er, more to the point, is there anything you need help with? :-)
FYI, though, some European stuff, audis, porsche, etc, use several HUNDRED crank triggers - they read the starter gear on the flywheel. Go figure. Of course, they also have to other references, one is a cam, and the other I'm not sure - both significantly lower. I've also heard that the 300zx uses some obscene number of pulses, over 100 (maybe 150?) on the cam.
Really, you can always choose to ignore extra pulses, till simply the computational load of ignoring them becomes to high. However, I tend to agree, certainly reading them in the same spot and some math should get you there - the extra pulses are only for checking your work or looking for something weird (I've heard tell of pre-ignition sensors based on the engine slowing down slightly or at the wrong time.)
Er, more to the point, is there anything you need help with? :-)
Reply
0
0
01-03-2009, 04:53 PM
#24
Junior Member
Join Date: Feb 2007
Location: 11368 miles from where i would like to be
Posts: 269
Total Cats: 92
All nissan CAS units I've seen have the 360 tooth pattern for sort of rotary decoding style operation. It's for perfectly fast sync up though, not accuracy. Matt Cramer from diyautotune reckons they read all teeth, but the documentation I've seen suggests otherwise.
Pics :
diyefi/RPM-engine-position-sensors - Photobucket - Video and Image Hosting
Pics :
diyefi/RPM-engine-position-sensors - Photobucket - Video and Image Hosting
Reply
0
0
01-03-2009, 06:41 PM
#25
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
Nope, all is working fine for me, Thanks.
I'm installing a built motor and I needed an extra high-speed input for traction control. So I had to combine my crank and cam signals into one input. While I was reading up to find out where I needed to move my cam pulse to, I ran across that and thought I would share.
FYI, I ended up grinding all the sensor teeth off the intake cam gear, then drilling and tapping a hole at the new location. I put a M6x1 bolt in from the back and ground the end down to the shape of the original teeth with a dremel.
I'm installing a built motor and I needed an extra high-speed input for traction control. So I had to combine my crank and cam signals into one input. While I was reading up to find out where I needed to move my cam pulse to, I ran across that and thought I would share.
FYI, I ended up grinding all the sensor teeth off the intake cam gear, then drilling and tapping a hole at the new location. I put a M6x1 bolt in from the back and ground the end down to the shape of the original teeth with a dremel.
Reply
0
0
01-04-2009, 04:59 AM
#26
total ms noob here, let me get this straight: ms2 will work with the nb crank wheel/sensor and its possible to get it working without installing the n/a wheel and cas?
If that is correct, why are so many members like cjernigan and patsmx5 using the cas or some other sensor besides the stock nb one?
If that is correct, why are so many members like cjernigan and patsmx5 using the cas or some other sensor besides the stock nb one?
Reply
0
0
01-04-2009, 05:18 AM
#27
Junior Member
Join Date: Feb 2007
Location: 11368 miles from where i would like to be
Posts: 269
Total Cats: 92
1) they have ms1?
2) abe scared them off?
3) the nb code isn't fully working yet?
Abe had some issues which were probably mostly his own fault :-p (sorry) but there may have been code issues too. Hard to know, he's using it now anyway AFAIK.
Fred.
2) abe scared them off?
3) the nb code isn't fully working yet?
Abe had some issues which were probably mostly his own fault :-p (sorry) but there may have been code issues too. Hard to know, he's using it now anyway AFAIK.
Fred.
Reply
0
0
01-04-2009, 12:54 PM
#29
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
You'll definitely get more accurate timing with your signals coming from the crank too. In fact, if I had a NA I'd convert it to NB sensors. On my DSM with a CAS (same configuration as a Miata) there was nearly a 2 degree timing drift from 1500 to 3500 from belt stretch.
Reply
0
0
01-04-2009, 01:18 PM
#30
Junior Member
Join Date: Apr 2008
Posts: 349
Total Cats: -2
The Autronic guy unwittingly points to why a 36 or 60 tooth strategy is often considered superior. The ECU sees acceleration and deceleration much sooner. Retard on acceleration is non-ideal, but won't hurt anything but power. On deceleration, a four-tooth sensing strategy can get you significant advance. Think about how quickly an engine decelerates on an upshift, and how much boost could be in the chamber when that spark goes waaaay early. Things can get ugly.
Reply
0
0
01-04-2009, 02:16 PM
#31
Junior Member
Join Date: Feb 2007
Location: 11368 miles from where i would like to be
Posts: 269
Total Cats: 92
You'll definitely get more accurate timing with your signals coming from the crank too. In fact, if I had a NA I'd convert it to NB sensors. On my DSM with a CAS (same configuration as a Miata) there was nearly a 2 degree timing drift from 1500 to 3500 from belt stretch.
Good post by the other guy with regards earlier signals.
Fred.
Reply
0
0
01-04-2009, 03:41 PM
#32
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
The Autronic guy unwittingly points to why a 36 or 60 tooth strategy is often considered superior. The ECU sees acceleration and deceleration much sooner. Retard on acceleration is non-ideal, but won't hurt anything but power. On deceleration, a four-tooth sensing strategy can get you significant advance. Think about how quickly an engine decelerates on an upshift, and how much boost could be in the chamber when that spark goes waaaay early. Things can get ugly.
My example was using a clutch-kick during a rallycross in the DSM. Fast full throttle accel with no load followed by a hard decel. In that car it would be even worse because of the CAS, the timing belt, slack in the CAS drive, and the momentum of the cams. I run that thing right at the edge of knock too. 110+*C intake temps, 120+*C coolant temps, 22psi of boost, and as much ignition advance as I can run without lighting up the J&S display. The knock sensors will pickup some noise when I do that, but I can imagine there's a lot of drivetrain bits clanging together during that kind of operation as well.
Either way, I'm beginning to feel that it might all be academic. When I began I was told I was using the ideal setup, now that I start to see the holes in the theory, I have to wonder why this stuff has held up so well.
I've been running pulse-per-cylinder on a bone stock miata bottom-end @200WHP for 2 years, then ~240WHP for another 2 years. I just pulled the motor to install my new built engine. The bottom end seems fine after 5 years of autocross in SM2 and 30000 hard driven boosted miles. Ringlands look good, no slack in the bearings, and the piston heights are all perfect. I can only assume that there is some law of physics or some invisible engineering wizardry holding things together for me.
In fact, the engine that I'm putting in now came from the guy that just sold his turbo kit on this board in this thread https://www.miataturbo.net/forum/t28960/
Hell, he was making ~500HP with a Motec and a stock NA CAS setup on this longblock. If that didn't generate some scrap metal I think I'll be safe with a crank trigger at 350HP.
Reply
0
0
01-04-2009, 03:58 PM
#33
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
I was actually told (while tracking down an issue with injection phasing) that I should set my I/P lead with the engine in the middle of the rev range rather than at idle if I was going to use the CAS. That's when I found that by setting my I/P lead at 1500 I was about 2 degrees off. He said that all the CAS setups (hall and optical) had drift and scatter issues, but they were pretty consistent once the engine was up to operating temp. He said that you could see the difference in the timing maps after a dyno tune if you compared the same engine using a CAS vs a crank trigger.
If you want to see it yourself, set your timing to a fixed 10deg and grab a timing light. Point it at the crank and watch the timing drift and scatter at different speeds. Then point the light at the cam gear and you'll note that it's rock steady (there will still be a bit of scatter from the slack in the drive gear, but no drift). Perform the same test on a vehicle with a crank trigger and you'll get opposite results, the cam timing will drift and jitter, but the crank will be rock solid.
Reply
0
0
01-04-2009, 05:19 PM
#34
total ms noob here, let me get this straight: ms2 will work with the nb crank wheel/sensor and its possible to get it working without installing the n/a wheel and cas?
If that is correct, why are so many members like cjernigan and patsmx5 using the cas or some other sensor besides the stock nb one?
If that is correct, why are so many members like cjernigan and patsmx5 using the cas or some other sensor besides the stock nb one?
Most of the people running something else either did it before I came along and told everyone what I tried, or they heard from other people it couldn't be done.
I'm looking for someone with a bit of electronics skill (can use a voltmeter, solder, etc) to help me test a board - it's a plug-and-play board for '99 + cars. Yes the MS-II code sucks, so does MS-I, but you will get better timing out of the II and if you do it right you don't have to cut any harnesses, just plug in a board to the OEM connector, and don't have to buy any sensors, either, though a GM AIT is highly recommended.
If you want to see it yourself, set your timing to a fixed 10deg and grab a timing light. Point it at the crank and watch the timing drift and scatter at different speeds. Then point the light at the cam gear and you'll note that it's rock steady (there will still be a bit of scatter from the slack in the drive gear, but no drift). Perform the same test on a vehicle with a crank trigger and you'll get opposite results, the cam timing will drift and jitter, but the crank will be rock solid.
Reply
0
0
01-04-2009, 06:08 PM
#35
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
There goes all my arguments. I was going to point out delays in the sensors, and then I was going to say watch cam phase in the logs... But that should work, so if you've done that and that's what you see, I agree. Although I doubt camshaft momentum has hardly anything to do with it, I'm sure fighting the valves is much, much more significant a force. Then again, they should "average out"?
My experiment was with two similar but different engines, not one with two different setups. After being told that it was true, seeing that evidence was enough to convince me. However it was hardly scientifically rigorous and there are quite a few variables. Maybe my timing belt was weak, or tensioned wong.
I'm actually planning to convert the DSM to crank trigger, but if I'm wrong and wasting my time then I'd like to know before I go through the trouble.
Reply
0
0
01-04-2009, 06:55 PM
#36
There's a miata trigger right there on the wheel, seriously, don't bother. If the autotronic won't see the sensor, it's $1.37 in parts to make something which will spit out something clean. I even have half assed plans for a board you could steal and make but I wouldn't bother with them.
If the computer seriously has issues with the 4 teeth, cut two off. I think some people have done this on the MS-I
If the computer seriously has issues with the 4 teeth, cut two off. I think some people have done this on the MS-I
Reply
0
0
01-04-2009, 09:47 PM
#37
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
There's a miata trigger right there on the wheel, seriously, don't bother. If the autotronic won't see the sensor, it's $1.37 in parts to make something which will spit out something clean. I even have half assed plans for a board you could steal and make but I wouldn't bother with them.
If the computer seriously has issues with the 4 teeth, cut two off. I think some people have done this on the MS-I
If the computer seriously has issues with the 4 teeth, cut two off. I think some people have done this on the MS-I
My miata (autocross car) has been working fine off the OEM NB crank triggers on an Autronic standalone for a couple years now. I converted it from a TEC-II with a 60-2 crank trigger to an Autronic SM4 in late 2006. I chucked the 60-2 wheel in the conversion, ground off two of the OEM crank teeth at ~10*BTDC and also ground off the double teeth on the intake cam gear so that there is only one sync pulse. There are no problems there, works great and always has.
My CAS experience has been with a 4G63 in my DSM (rallycross car) which is in the same configuration and works exactly the same way. Even the CAS pattern is the same. It's using an Autronic SMC standalone but I was thinking about converting it to a crank trigger system using leftover miata parts. That's the engine I experienced the CAS timing drift on.
I was looking for confirmation that somebody else got the same test results in the crank vs. CAS trigger experiment so that I could confirm the conversion on my DSM will result in a more consistent ignition system.
The reason I was messing with the miata triggers this time around are not really related to this thread at all, but I ran across the quotes I posted from Ian while researching the trigger patterns. My intention with the miata (if anybody cares) is to combine the cyl and sync pulses into one signal wire to free up my sync input for use as a wheel speed input for traction control.
Reply
0
0
01-05-2009, 02:35 AM
#38
Ah! I thought you were saying you wanted to have more accuracy on the engine speed to judge wheel slippage based on that, which I thought was insane. :-P
I can check the ... shoot, no timing light! But I have seen my own cam bounce around, and my friend's crank bounce around, and I'm crank timed while he's cam. Not a great data point, I'm afraid. I don't have a scope at home right now, so I can't watch this for you.Joe had both set ups on his car, but I don't know if he can go back and forth in software.
BTW, the FreeEMS (ok, so it's not done yet) has somewhere around 80 inputs, so there's less worry about running out of pins. :-)
I can check the ... shoot, no timing light! But I have seen my own cam bounce around, and my friend's crank bounce around, and I'm crank timed while he's cam. Not a great data point, I'm afraid. I don't have a scope at home right now, so I can't watch this for you.Joe had both set ups on his car, but I don't know if he can go back and forth in software.
BTW, the FreeEMS (ok, so it's not done yet) has somewhere around 80 inputs, so there's less worry about running out of pins. :-)
Reply
0
0
01-05-2009, 01:12 PM
#39
Junior Member
Join Date: Jan 2007
Location: EXTREME Southern Illinois
Posts: 281
Total Cats: 0
I have some spare inputs left, however the computer only has 5 'high speed' inputs capable of reading and interpreting toothed wheels. 2 for cam/crank, 1 for vehicle speed, and 2 for variable valve timing control. I want to use 4 of them for traction control by tapping into the ABS sensors, so that only leaves one for reading engine position. I could probably use 3 with one for each front wheel, then one on the speed sensor in the gearbox. The speed sensor SHOULD be an average of both rear wheels, but I'd like to have the option to try it both ways.
Reply
0
0
01-05-2009, 02:01 PM
#40
Curious, why would you need more inputs for variable valve timing? I guess if each cam was independent you would need a third channel, otherwise the cam and crank would have all the info you need. And I bet you could do the speedometer through a divider to get vehicle speed and still have reasonably accuracy.
Still, it seems like a fun project. BTW, traction control reading the front wheels? You're not talking ABS, right? If the front starts to slip, what do you do with the rear to maintain traction? :-)
Still, it seems like a fun project. BTW, traction control reading the front wheels? You're not talking ABS, right? If the front starts to slip, what do you do with the rear to maintain traction? :-)
Reply
0
0
