Sokool

iTrader: (1)

Running a MS3X on a 1.6 running full sequential. The car has undergone a complete overhaul including a custom engine harness for this build. I built the ecu for this project. This is my first time attempting to start the car.

Tunerstudio shows a fluctuating RPM signal during cranking of around 250 RPM and then back to 0 RPM. Composite log shows good CKP signal but no CMP and sync error 38. I have continuity between pin 4 at the CAS plug and pin 32 at the MS3X connector. I configured the MS3X card as outlined HERE for VR sensor. This engine was previously running batch on a MSPNP so no idea if the CMP signal of the CAS was previously in good working order. Is there a good way to test a CAS without an oscilloscope? Not sure my DVOM could keep up with the pulses.

I have attached a Composite log and Datalog of cranking. As well as my .msq

Please tell my I'm dumb and have missed something simple.

Joe Perez

iTrader: (8)

Posting from phone while being held captive in the trunk of a Caddillac Deville, so forgive brevity.

If the cam signal wasn't working previosly, the engine wouldn't have run at all. Even without sequantial injection, CMP is required for wasted-spark.

Sokool

iTrader: (1)

MS1 has provisions for CMP?

Joe Perez

iTrader: (8)

Of course. It's JS8, which connects to pin 11 of the CPU.

In order to run wasted-spark ignition, the CPU needs to at least know which CKP pulse corresponds to #1/4 TDC, if not absolute engine-phase. It can derive this information from a missing tooth on a crankwheel, or from a second input which gives one pulse per crank revolution. The CMP signal in the Miata's CAS fits the latter description.

Gotta go- the car has stopped and I can hear people shouting in Russian outside.

Sokool

iTrader: (1)

Shows how much I pay attention. It was late last night and was quite frustrated. Anyone else with some thoughts? Do I have the MS3X card configured properly?

Matt Cramer

Why?

If you haven't swapped a VR sensor into your CAS for unknown reasons, set it up for Hall / optical sensors by installing JP7.

Braineack

iTrader: (62)

probably not. did you turn the pots correctly?

You need to follow this method:

I don't see any CMP activity on your composite log so I'm assuming you dont have JP7 installed?

Otherwise, double check that you have Pin 32 of the Expander db37 going back to 2G on the harness connector.

Sokool

iTrader: (1)

Because I was going off of the wrong assumption when I biult the ECU over a year ago. None of the writeups were in circulation then.

After my own research last night I came the same conclusion. I will do the appropriate adjustments and report back.

Joe Perez

iTrader: (8)

Yep. That'll get 'em every time.

richyvrlimited

iTrader: (1)

No CMP is required for wasted spark?!? If it was your old MS1 and trigger wheel setup would never have run

The CMP is only required on an MX5 with the OEM CAS in the context of its wheel pattern being coded into the MS ECU - or at least fudged on a MS1

Joe Perez

iTrader: (8)

Did you read what I wrote? I said that CMP *IS* required for wasted-spark, even on an MS1. (It's also used to sequence the injectors, even when run in two-channel batch fire mode.)

It sounds like the OP is homing in on the problem, so this thread is kind of solved. But for everyone's general reference, a quick primer on how wheel decoding works in general:


For Crank-only (or single-channel Cam only), you have several basic options:

You can run a crank wheel with an even number of teeth, with one or more teeth missing. This gives the ECU enough information to know when the #1 and #4 pistons are physically at TDC, which can then be used to drive wasted-spark ignition and two-channel batch injection. It can't run full sequential, because it can't distinguish between #1 TDC and #4 TDC.


You can run a cam wheel with an even number of teeth, with one or more missing teeth. Because the cam rotates at 1/2 crank speed, that missing tooth is enough to tell the ECU precisely where #1 TDC is, and this, you can run full sequential. (Note that this configuration is extremely uncommon.)


If you have only a single wheel with no missing teeth, be it on the crank or the cam, then you can only run single-channel, meaning you will need a mechanical distributor. This, in fact, was the original operating mode of the Megasquirt way back in the beginning, as the points inside the distributor were re-purposed to act as the "wheel".


With the NA Miata's CAS, the CKP signal does have an even number of teeth, but since it only has one tooth per ignition event, you can't just cut one of them off and call it a missing-tooth wheel. There needs to be at least twice as many teeth as ignition events in order to do that, since every event must correspond to a physical tooth.

The CAS's CKP line could, in theory, be used all by itself if you could also figure out how to get a distributor to run in parallel with it, but that's kind of silly.

Thus, in order to use the CAS to drive a wasted-spark ignition, we require both the CMP and CKP outputs. CKP provides the actual firing-event reference, while CMP provides information about engine-phase.

The CMP line actually supplies enough information to run fully sequential, since even though it has two pulses per cam revolution, one of them is longer than the other. The longer one fully overlaps a CKP pulse, whereas the shorter one overlaps only one edge of a CKP pulse.

The MS1 is not capable of figuring this out, since it can only "see" one edge of the pulses- it cannot detect their actual state in real time. (This is not a software limitation- it's a physical limitation of the actual CPU itself. The IRQ input is the only one capable of providing sufficient temporal resolution to deal with primary trigger inputs, and that pin is able to act only on falling edges.)

So to the MS1, the CMP signal just looks like one pulse per crank revolution, which is functionally equivalent to a missing tooth crankwheel.

The MS2 and 3, by comparison, are capable of reading the realtime state of all input pins, so it can figure out the difference between the two CMP pulses, and use this to set absolute engine phase thus supporting full sequential operation.

Sokool

iTrader: (1)

Jumped JP7 and set my pots to 2.5v and it fired right up. Thank you all for showing me the error in my ways.

richyvrlimited

iTrader: (1)

I did read it I genuinely thought the CMP wasn't required for wasted spark. You didnt have a CMP signal on your 36-1 crank trigger setup?

I thought that like with that setup you don't need a CMP signal for wasted spark and batch injection. I know you need the stock MK1 CAS's CMP as the code the MS uses is hard written to decode it using both signals.

Obviously I'm wrong, thanks for putting me straight

Joe Perez

iTrader: (8)

No, I did not.

You don't need a CMP signal with a missing-tooth wheel such as a 36-1. The missing tooth itself provides the same information that a CMP pulse would, insofar as wasted-spark and two-channel batch injection requires. The ECU knows that whenever it sees that gap, it is approaching TDC on cylinders 1 and 4. Since 1 and 4 both share the same ignition coil, this is enough information to sync.


The problem arises when you have a CKP signal that does not have an obvious distinguishing mark to indicate the position of 1/4 TDC. In the case of the NA CAS, the CKP signal is four evenly-spaced pulses per cam revolution with no uneven gaps. This is not enough information to do wasted-spark, since the ECU has no way of knowing which CKP pulse corresponds to 1/4 TDC versus 2/3 TDC.

The CMP pulse lets the ECU know which CKP pulse is which. Since there is only one CMP pulse for every two CKP pulses, the ECU knows that when it sees a CMP pulse, the next ignition event is going to be for 1 & 4.



The fact that the CMP pulses are of unequal duration is what allows the MS2 and 3 to further decide which one of the four CKP pulses specifically corresponds to cylinder #1's TDC on the compression/ignition cycle, however that's more information than we need to consider in the context of an MS1. Because of the physical limitations of the MS1 CPU, both CMP pulses appear to be the same to it.



Perhaps a picture will help:



The top is CKP, the bottom is CMP. This shows just a bit more than two complete engine cycles.

First, cover the bottom trace with your hand or a sheet of paper, and look only at the top trace. In order to fire the correct ignition coil, you need to know which pulse corresponds to which cylinder pair. How can you tell?

You can't. You need the information from the bottom trace. So uncover it.

Notice that even though the CMP pulses alternate between long-short-long-short, it is always true that the rising edge of CMP occurs a fraction of a millisecond before the rising edge of CKP.

CMP, in this case, is simply being used to reset the cycle counter. When CMP goes high, the ECU says "Ok, the very next CKP pulse I see is going to be for cylinders 1/4. After that, the next CKP pulse will be for cylinders 2/3." It just keeps repeating that process over and over again.