thirdgen

iTrader: (26)

So I have a 99. I got a boomslang harness, and have a simple question. If I were to put a MS on there, I need a CAS from a 97? Is this only an issue when using MS? What about people who use a Hydra? What other sensors do I need to use when bypassing the stock computer? The CAS is just for a timing signal, right? So how does a stock 99 get a timing signal? and why can't a MS read this signal?

DammitBeavis

If you have some control over how the ECU reads the crank trigger and cam sync signal, then the '99's sensors are superior to a CAS.

You can make the '99 outputs similar to the CAS by grinding off two of the crank trigger teeth (@10deg BTDC IIRC), and grinding off the double trigger teeth on the intake cam gear.

I don't believe that the MS needs the cam sync signal since it isn't sequential injection, and the ignition is wasted spark, so you could probably just use the crank trigger wheel if you can configure it to read the hall effect sensor's output.

Joe Perez

iTrader: (8)

Dammit, Beavis...

First off, the limitation is a software one, not an electrical one. The code for the MS1 CPU cannot accommodate unevenly spaced crank trigger signals, and the teeth on the NB wheel are not evenly spaced- there are two at TDC, and two at seventy-something BTDC. So to make an NB crankwheel work with an MS1, you grind off the two teeth that are at TDC.

The MS2 software is much more advanced in this regard, and it can use the NB crankwheel signal as-is. In fact, there's a preset in the trigger configuration window specifically for NB sensors. You just select that and you're done.

Now, all ECUs require a cam signal of some kind in order to do wasted-spark, regardless of whether you're running sequential injection or not. The ECU has to know which crank pulse corresponds to firing #1/4 (before TDC) and which corresponds to firing #2/3 (before BDC). The stock cam sensor (both NA and NB) gives a pulse before #1/4 TDC, so that the CPU knows where to steer the next crank pulse that it gets.

In order to run sequential injection, you actually need to go a step further and get rid of one of the two cam pulses, or at least be able to distinguish between the two. The ECU needs to get a distinct cam pulse which it can use to determine which cycle each cylinder is in in order to trigger that injector at the correct time.

With the stock ECU, it makes this determination based upon the fact that on the NA, one CMP signal is longer than the other and on the NB, the #1TDC CMP pulse is a double-pulse, whereas the #4 TDC pulse is a single pulse. The Hydra may know how to deal with this, I honestly don't know. The MS1 requires a single cam pulse to make this determination, it can't measure the length of the cam pulses.

So, to answer the OP:

If you install an MS1, in standalone, you'll need to cut two teeth off the crank wheel. You can't run an MS1 in parallel on an NB without a CAS.

If you install an MS2, you're good to go with no modifications and no need for a CAS.

DammitBeavis

Sorry, it's been a while since I did a non-sequential. I was thinking of the missing tooth wheels like the TECII use. They don't need a cam signal since the missing tooth defines which pair of cylinders is approaching TDC. But yes, for pulse-per-cycle crank triggers, you would need a sync pulse. Sorry for the mistake. Either way, I'd rather use a crank trigger of any sort rather than a CAS. Too much spark scatter from belt stretch.

thirdgen

iTrader: (26)

Great! My idea was to run MS II in parallel, and the answer is...I can. Thanks for the responses guys. So what is the difference between 3.0 and 3.57? I read that the 3.0 is easier to mod? I just want to buy the right stuff, that's all...

AbeFM

iTrader: (3)

As a general rule, stay away from parallel set ups. Lots of people do them, but they are riddeled with problems, they are a slow start, and a much bigger pain (and slightly more costly) to do. If you have a good reason for it (the most typical is smog) then maybe it makes sense, but if you can set up the car to switch back and forth I'd say it's the way to go.

I'd say, in an ideal world, a missing tooth wheel is the way to go, my feeling is (I plan to do this myself someday but have yet to try it) the stock NB sensor would work FINE for this, just add the wheel and remount the sensor, use the same input circuit.

However, to just get up and running with as few mods as possible, and to keep the car so you can jump from MS to OEM ecu's, buy an MS-II and do everything in there, change nothing on the motor.

AbeFM

iTrader: (3)

I knew I forgot something. More or less, the 3.57 is just prebuilt and surface mount. Which means it's much harder to mod. I would shy away from it as much of what's on the MS-board you're going to want to change. If you're very comfortable soldering onto surface-mount-devices (SMD's) then you could potentially get it, but odds are you're better off with the 3.0. Otherwise they are basically the same.

Joe Perez

iTrader: (8)

Correct. Logically, a missing tooth crankwheel provides the same information as a constant-tooth crankwheel plus a camwheel with two pulses per cam rev. With either system, you get one "reference" event for each complete crank rotation, that allows the ECU to identify the absolute position of the crank.

To do fully sequential ignition, the system also need to know the absolute position of the cam. This cannot be achieved with a missing-tooth crankwheel. You can use either a missing-tooth camwheel (by itself) or an even-tooth crankwheel and a camwheel which gives only one even per cam revolution, which is to say one event for every two crank revolutions.

The stock NA CAS and NB cam sensor provide two events per cam revolution, which from the MS's point of view is the same as one event per crank revolution. The events are slightly different in duration, however. The stock ECU is programmed to recognize this, doing the same with an aftermarket ECU requires some work, and is only really necessary for full sequential. In the most basic config, the MS overlooks that difference and treats the cam events the same. The MS1, in particular, cannot measure secondary trigger event duration, it just resets a crank counter every time it sees a cam pulse. The MS2 is a bit more sophisticated and could in theory decode the camwheel properly, though I've never tried it as it's not terribly important to do so.


With an NB, I don't see the point. You've already got a perfectly good crank sensor, and you don't stand to gain much from replacing it with a more complex missing tooth wheel.

Schematically, they are identical. As Abe said, 3.57 is surface-mount, which makes it easier to mass-produce in a factory environment. It's only available pre-assembled. The 3.57 also does not have the little prototyping area on it like the 3.0 does. It's intended for applications like what DIY has done in building the MSPNP. Internally, the PNP consists of a 3.57 board, plus a custom daughterboard which, in addition to hosting the big yellow ECU connector, also contains all the mods that most of us build onto the main board.

AbeFM

iTrader: (3)

Well, one good thing about the OEM '99 set up, with only 4 teeth, there's a lot more CPU time left over for running all the other gizmos. Sure, you're timing errors get larger, but I've yet to see a comparison of the additional teeth and what it does to accuracy. Some day I might runboth just to see how they compare.

One advantage to the missing tooth, though, the car will start faster. The MS-II's handling of the NB miata is functional, but not perfect. I would say it's plenty good to run the car, though. They just don't do a switch from wasted fuel/spark to sequential, and even in wasted mode they still "think" in sequential, which means 2-3 full revolutions until the motor starts. Not a big deal, you just get used to holding the key for a full second. Other than that, it starts right up and runs fine, just like a normal car.

DammitBeavis

Actually, fewer teeth does not necessarily give you more error.

The trick is to have the tooth just before the ignition event, with just enough time for the ECU to perform it's calculations. With slow ECU's you want your crank trigger around 70deg BTDC since the max advance is normally around 40deg. That gives the ECU 30 crank deg to do it's work, and the calculation should be spot-on. With faster ECU's, triggers at something like 60deg BTDC are ideal. Unless you have a super-fast ECU, it can't start it's calculations much later than that anyway no matter how many teeth you have.

And unless you're running an engine with very low rotating mass like a motorcycle, the crank accel/decel in those 20 or 30 degrees is negligible anyway since there won't be a firing event during that period.

rb26dett

Damn it, Beavis! Yes it does. You've considered 2 from four things that cause this, but not the third and fourth. For that I'll grade you 49% on your final EMS exam ;-)

My heavy **** engine with HEAPS of inertia gets around 7000rpm/second delta RPM, the most insane engines get around 36000, my logs are poor, I can probably beat that, possibly by as much as two times.

Do the math and post again for your final 51% and a chance at a passing grade ;-)

Fred.

Joe Perez

iTrader: (8)

Hehehe.

On the MS1, it does not matter how many teeth your wheel has. Only two of them are declared to be trigger teeth for the purpose of driving the countdown to ignition- the others just increment a counter.

On the MS2Extra, there is some rather interesting intelligence at work. Let's say you have a two tooth wheel, and each tooth is at 60° BTDC. When that tooth comes around, the CPU starts a timer and counts down the calculated number of microseconds it believes it will take for the correct firing position to come around. If it wants to fire at 18° BTDC, then it's got to rely on counter precision alone for 42° worth of rotation, and pray that the crankshaft's angular velocity remains constant during that time.

Now say you've got a 36-1 wheel, with one tooth every 10° of rotation. As before, we want to fire at 18° BTDC. Now, the ECU does not start the timer on the 60° tooth, it waits until a later tooth comes around, so it spends less time free-running on the timer. It'll still use a "reference" tooth to do the dwell calculation (figuring out when to start charging the coil) however for the actual moment of firing, it'll wait until the 20° tooth comes around and then only have to free-run on the timer for 2°.

One other benefit of a large tooth count is that you can do much more accurate timing during cranking. Rather than the CPU having to do any timing at all, it simply fires the ignition exactly on a particular tooth. With a 36-1 wheel, you can specifiy cranking timing in multiples of 10°, with a 60 tooth wheel, you can work in multiples of 6°.

rb26dett

Joe gets 75% so far, and a bonus 10% for his accurate and correct supplementary information :-) To hit 110 from 85, what is the final piece of the puzzle Joe?

AbeFM

iTrader: (3)

I think if you're building a lot of stuff from scratch, and don't mind the work, you're better off with the missing tooth approach. If MS handled it better, then a non-missing tooth and a cam wheel would be IT. Which is what many car manufacturers do. :-)

Runnnig a CAS is not recomended, getting away from that is one of the few changes mazda made - a company so unwilling to change thought it was the one thing really worth doing. And certainly, if you're pushing the motor to it's limits (and if you're not, what are you doing here?) you might as well take what you can get.

And, as a bonus, there's a recorded "prediction error" to the timing of the MS-II, so I have lots of logs which I should digest and post, but I imagine with more teeth this would be smaller - at LEAST by as much as 1/tooth count.

DammitBeavis

That is simply what I was told by Autronic when I was deciding if I should modify my my stock triggers, or use the 60-2 wheel that was already mounted on the crank pulley. I'll ask them for an explanation. I don't have the math, just the word of Ray Hall.

rb26dett

You are correct about the angular precision, however because of the large gap from the last sample, you have pretty much no idea what the engine is currently doing even if you DO know where it is. That should be a clear enough and simple enough explanation.

In practice though, you are right, if it's aligned well enough, it should be fairly good.

The bit Joe covered about using the closest tooth it the important part of the upgrade timing wise as according to Ken the RPM is only calculated once per rev anyway in ms2e. I was a bit shocked/surprised by that, but it runs my engine well, so who cares :-)

EDIT : and of course, the 25% Joe scored was for his cranking example. short term RPM variation while cranking is HUGE, esp with high compression.

Fred.

rb26dett

http://www.google.com/search?hl=en&s...00&btnG=Search

If you stand on the gas in your uber light weight machine at 900rpm you'll have around 100rpm error in your lookups which doesn't matter all that much I guess. At higher revs it becomes negligible.

The timing error will be more significant at low rpm too and that will matter a bit at least.

So in summary, it doesn't actually matter that much while actually driving, but there are four aspects to it :-)

Fred.

DammitBeavis

I would imagine a reasonable calculation would include rate of change data from previous triggers. Either way, I posted the question to the Autronic/Motec folks, and I'm bound to at least learn something new.

AbeFM

iTrader: (3)

I know from what logs I looked at back when I was checking out my triggers more closely, certainly low RPM (<2k) the errors were big, and idle they were ALL over the place. MS-IIe uses some predictions, though there's debate from the authors about which you should use.

The simplistic way to view it, if you've got a motor that's slowing over 1/4 of the revolution, lots of teeth will mean your last bit of error is in how long it takes it to cover 6*. 4 teeth means your error in how long to cover 90*.

10% of 6* is <1 degree, 10 % of 90* is nearly 10 degrees. The less often you update the rpm, the more you want to have a lot of teeth. How much does it matter practically? Sure, all this stuff seems to have no effect, but if you're trying to get <1* accuracy on spark timing, then you'd better be confident your average speed does't loose count over the 30 or 40 degrees you're going to cover, and that means knowing RPM within 2.5%. What is the second derivative of engine speed? Got me. :-)

(edit): The nice part about very few teeth is the errors in reading them don't play much of a factor. Of course, this can be overcome with averaging, but you're going to come to a point where you're getting computationally expensive

thirdgen

iTrader: (26)

So I was gonna start out by purchasing a MS II with a 3.0 PCA, this all comes pre-assembled? Is there any thing else I need to order from DIY autotune with this? I was also thinking about maybe not going with MS II, but maybe a different standalone like a Microtech LT-8. What do you guys think?