AbeFM

iTrader: (3)

I had this same issue, I moved the circuit over to the board and it didn't work as well. Man I was frustrated. Then I didn't know if it was the VICS solenoid.....

I would definitely prescribe some probing.


Why not find out the biggest cap that doesn't let you free rev to redline, then use half that?

We ought to just come up with a way to mount these. My bungee to the AC box was ideal, but now without it I'm mostly happy with what I have, but a bracket with thumbscrews is classier than a rubber band

The_Pipefather

iTrader: (5)

Abe, how is your comparator circuit better than the stock MS one? Is it a sort of upgrade for us folks running the stock MS board?

AbeFM

iTrader: (3)

My comparator circuit (excepting that on my car I don't use the caps since I didn't need them. Joe must have the dirtiest circuit ever, or maybe it's some weird faked zero crossing thing or some other weirdness) is straight from the NB miata.



I just probed it out. It took some time to do, but it's pretty straightforward. Logically the circuit makes sense.

Basically, the output is tied to the input, so when the input switches state, the output will ignore it 100% until it's maintained for a certain amount of time. This is ideal for getting rid of short duration noise signals.

It also has the effect of moving the threshold. Instead of calling everything below 2.5V "low" and everything above 2.5V "high", it depends on the state you're in. If you're in the low state, then something isn't considered high unless it's 3V, and if you're in the high state, it's not considered low unless it's under 2V. And in both cases, it needs to maintain that level for an extended period.


Joe:
I'd still suggest much probing. Baring that, I would suggest two classes of changes.
1) Tweak the first pull up resistor. Either it's overpowering the current syncing abilities of your OEM sensor, or it's not strong enough and sensitive to noise. As a first stab, look at the pull ups in your NA OEM ECU, and see what they use!

2) Explaining how the feedback works makes me think this is a good idea. Maybe allow the non-inverting input to see a bit more of the output than you're doing. That should definitely make things less sensitive to noise. While trying the OEM values here would be a good place to start, you could also just blindly half the top resistor (swap the 13k for 6k), or you could figure out where that number is (curious it's not symmetric, it looks like (ballpark) 1.56V (38%) when high and 3.72V (48% high) when low...

Anyway, while they seem reasonable, you could push those out a bit, or do them individually by bumping the 18 or the 22 over.

My own circuit (due to parts availability) uses a 12k not a 13, making my swings just a bit wider.

AbeFM

iTrader: (3)

Just an FYI to people. There's no indication on here (that I see?) of which crossings are electrically open and which are connected. Check my (less pretty) drawing for these details.

Joe Perez

iTrader: (8)

Yeah, I need to fix that drawing. It looks better in the schematic editor.

AbeFM

iTrader: (3)

No comment on my treatise on changing values?

The_Pipefather

iTrader: (5)

I didn't understand a word of that. Maybe I should have phrased my question in a different way: What does it do, and how is it different in operation to the stock MS circuit?

EDIT: Read through it one more time, so its some kind of filter for the CKP and CMP signals?

Joe Perez

iTrader: (8)

Sorry, I took the afternoon off, cleaned and waxed the car (I needed to actually succeed at something) and then tinkered with the MS a bit.

It's running again.

Basically, the goal for today was not so much to get it perfect, but to get it running in time for Sunday. I simply cannot miss The Puke, as it'll be my last big group run before moving east.

First, I installed a pair of .022uf caps. That fixed the problem. Then, I installed some .047uf caps in series, to bring the total down to an equivalent .015, and the problem stayed fixed. However it seemed while I was driving it that it didn't have quite as much pep as before.

Hmmm.

Wait- everything I'm doing is adding a finite delay to the signal. IOW, the amount of delay does not vary with RPM. Meaning that as RPM increases, the amount of the delay as a percentage of degrees crank rotation increases. My timing is automatically retarding itself in proportion to RPM.

****.

Got out the timing gun, set the MS to fixed 10, and free-reveed it in the garage while watching the mark. Sure enough, it's dead-on-***** 10° BTDC at idle, but as I increase the RPMs the timing retards. Looks like I was loosing maybe 5-7 degrees total, hard to say since I wasn't watching the tach.

I even fail at failing...

Tomorrow I'll get someone to hold RPM at a high value while I observe the timing. My thinking is that for now I can compensate this out by simply adding advance to the spark tables in proportion to RPM, maybe one degree per thousand or thereabouts.

Joe Perez

iTrader: (8)

Yeah. Apparently I have the noisiest CAS in the history of mankind. The circuit as a whole adds low-pass filtering to the input and hysteresis (resistance to change) to the output. The idea being to eliminate the effects of a noisy signal.

In all seriousness, I'm starting to wonder if maybe my CAS is junk after all. Anybody have one they'd care to loan me?

(and yes, Abe- I will get around to scoping out the signals eventually.)

AbeFM

iTrader: (3)

Yes indeedily doodily neighbor. It takes some of those false signals and gets rid of them. If you're having issues where your logs show RPMs that don't make sense for one data point, you might try it. If not, accept well enough alone. :-)


I dunno about your MS-I-for-weenies, but the MS-II has a latency you can set. So, it's a few us, which is meaningless at low RPM and significant at high RPM. Maybe dig through the triggering screens and see if there's something there.

Ok, nothing to say here. But it bears repeating. I feel better in comparison.

Joe Perez

iTrader: (8)

Thank you, neighbor! We do indeed have such an adjustment- it's buried at the bottom of the "Advanced Code Options" screen, and like so many other things, not in the documentation.

I'll have to wait until tomorrow to try this- I don't think the neighbors would appreciate me revving the engine to redline in the garage at 10:45pm.

Joe Perez

iTrader: (8)

As you recall, I settled on ~.015uf (actually, .022 and .047 in series) for C1 and C2 in the comparator circuit. This seemed to be working well, not giving me misfires. So I closed up the MS and bolted it back into place last night.

Today, drove out to my place of business to do some timing observations. It is located in an industrial area where nobody cares about people redlining their engines in the back lot.

I locked spark at 10°BTDC and used a timing gun to observe the pulley mark as I increased RPM. I found that due to the latency caused by the capacitors, I was losing about one degree per 1,000 RPM. I calculated this to be a delay of 166 microseconds (hurray for math) so I keyed this into the "Hardware Latency" box, re-tested, and found the spark advance to be solid across the RPM range. Well, not solid exactly, the timing belt causes some jitter, but on average it was neither advancing or retarding as RPM increased.

Packed everything away, drove home, got onto the boost enjoying the extra pep of correctly advanced timing, and had the misfire.



Pulled the MS back out of the car.

At this point, I'm running low on capacitors whose legs haven't been snipped too short, so I tack jumper wires across the .047s and install a pair of .022s across the existing .022s, bringing me to .044uf, which I know from previous experimentation to be "safe." Put it all back together and drive back out to the office. I re-lock and re-observe the timing, and calculate that I now have ~350us of delay. So I go to type "350" into the latency box, and find that it maxes at 255. Aaargh- damn 8 bit variables...

Well, with 255us in the "Hardware Latency" box, I'm short by about 5 degrees at 6,000 RPM. So to compensate, I'm just gonna add some advance to the spark tables.

Clearly, there is still work to be done. I've decided that when I have some spare time on my hands I'm going to remove my whole input circuit, run some wires outside the MS, and build an external input signal conditioner. This will make probing and experimentation much easier.

On the plus side, I consider my primary objective for today to be accomplished, which was to get the car running well enough for tomorrow's canyon run. Operationally it's in reasonable shape, I'm just not happy about the means.

I'm honestly starting to suspect that my CAS is just garbage. I understand that a couple of other people have had a similar problem, so if someone has a spare CAS lying around they'd like to loan me, I'd be very grateful. It's either that, or completely re-do the system for a crank trigger...

AbeFM

iTrader: (3)

I'm starting to doubt your CAS as well. It would be interesting to see if you could jumper across (remove) the caps on one line, then the other, to see if one circuit is the cause of all your problems.

Also, 350us is a HUGE delay. Guess it doesn't matter, but it is big. I think mine's more like 5.
-Abe.
P.S. I do have a crank trigger and wheel borrowed from a fellow board member. :-)

Joe Perez

iTrader: (8)

Like I said, the next step is to make the input signal conditioner an external box so that I can easily probe and modify without having to remove the MS every time. Once that's done we can scope and experiment at will.
Yep, I was truly shocked when I actually measured it. Kinda reinforces the idea that my CAS is just garbage- no lower value of cap in the filter circuit has worked for me reliably.
And you'll be bringing it with you to Dalton's tomorrow at 9am, I presume? I still haven't gotten a "Yes, Joe, I'm going to The Puke" confirmation out of you.

AbeFM

iTrader: (3)

Yes Joe, when I see a 1.6 liter, I want to puke. And I will, tomorrow, ladyfriend in tow.

Got my boost squared away finally (i.e. **** the shitcrapshitfuckfuck EBC the MS fakes and use my MBC), and my brakes don't squeak as much.