patsmx5

iTrader: (16)

I had pure hell for 3 weeks dealing with it, ask Abe. .020 gap helped a ton. Now I get a misfire every 15 minutes. However, it comes and goes with the weather it seems now a days. I "can't believe" after 2 or 3 people said to gap down your plugs, you didn't try that first.

Joe Perez

iTrader: (8)

I did. First try was gapping my set of IFR8Hs (which were on the shelf) down to about .028". That made no difference.

Yesterday, I gapped a spare set of BKR7Es down to .020" exactly. Not .019", not .021". (I've never in my life spent so much time dicking around with a set of plugs.) Unfortunately, I made several other changes at the same time, including some capacitor values. Anyhoo, with that setup I got the misfire twice yesterday, once on the way down to Abe's place, once on the way back.

The plan from here on out is to start "de-modifying" the setup, and see which factors if any affect it. I need to put my old (aka new) VE table- the one where I spoofed the "Req Fuel" number down to 3.5 to increase useful resolution in the VE table itself. Gonna do that at lunchtime, along with some larger caps in the filter- right now I've got .001uf on the CKP and .01 on the CMP. I wish I'd have made them both the same yesterday but I didn't have two of either value on hand... Gonna try .033 first probably, and I've got some .0005s on hand if it turns out I need to go down instead of up.

cjernigan

iTrader: (13)

I would **** and moan about a missfire, if I had one anyway. I never deviated from the DIYautotune mods whatsoever. Wouldn't everyone be having this problem, including MSPNP owners if it was in the input circuits?

Joe Perez

iTrader: (8)

That's what is driving me absolutely batshit about this whole affair. I initially build the MS exactly to DIY specs, plus the .1uf capacitor across JS8 and GND.

After having the problem initially, I figured that the CMP signal being totally non-isolated (the CPU pin 11 going straight to the sensor itself) might be causing some issues. So I built a circuit using a opto-isolator (similar to the primary input) and replaced the "DIY" CMP circuit with it. No real improvement. For those who care, this was the circuit:



Then, I built a copy of the circuit Abe is using, which replaces both the CKP and CMP input circuits. This is what I built originally, the IC is a dual comparator, NTE943M (LM393):


It still had the problem. Abe suggested removing the caps altogether, which made it far worse. I replaced C2 with a .01, and that seemed to help somewhat- gonna try upping both C1 and C2 a little and see what that does.

Let me ask you this- is your CAS an optical or hall-effect type? Mine, curiously, is a hall-effect sensor (I'd thought all the 1.6s had optical sensors) and I'm wondering if there's some correlation- ie: Are the Optical CAS's more stable for some reason?

cjernigan

iTrader: (13)

I'm using a 94-97 magnetic type CAS. The 1.6s should all be optical. I think you can clean the internals of a 1.6 CAS, swear I read that somewhere.

Ben

iTrader: (33)

I have had random misfires. Adding grounds took care of damn near all of them. I still get one occasionally, typically at high speed cruise. There does not seem to be any rhyme or reason to when, it's all completely random. The tach wiggles during the misfire event.

I don't have the cap filter installed. I was assuming that it would take care of the remaining, occasional misfires. But I haven't had one misfire since going to HR a couple days ago.

Joe Perez

iTrader: (8)

Yeah, that's what I thought. Last time I had the CAS out, I pulled the cover and inspected it. It's definitely a magnetic type. I wonder if there is some correlation here- could the optical sensors be cleaner somehow? (Anybody got an optical CAS they wanna loan me?)

Well, mine definitely are not random. It never happens in cruise, only in boost. I cannot for the life of me figure out how this makes a difference. And when it happens, my tach needle doesn't just wiggle, it drops like a stone. I honestly did not know the needle could move that fast!

I've been running HR since the beginning. Started at 10d2, ran 10F for a while, and now I'm at 10G. A few days ago I installed 029y4, and it made no difference whatsoever. That reminds me, time to update the sig.

patsmx5

iTrader: (16)

No you didn't. You said you run .028 in your first post. Then patsmx5, cjernigan, and Savington suggested running less gap and that in all 3 cases it helped with their random misfires.

In the post I quoted and shortened above, you now say you tried .020, along with several other mods and the situation improved, but you can't pinpoint what helped because you changed several variables at once. So indeed you did not(try gapping to .020 first).

Joe Perez

iTrader: (8)

Ok, you win. I'm an idiot.

I'm not arguing with you- seriously. I just can't get my head around how too large of a plug gap can cause the MS to think I'm spiking to 43,000 RPM for one sample...

Sorry, just been feeling really frustrated with all this. Believe me, I will ultimately take the time to go through each variable one by one- I'm just feeling really pressed for time right now as I *need* this to be working perfectly by Sunday. I've spent too much time slacking off and not addressing the problem, and it's now biting me on the ***.

patsmx5

iTrader: (16)

2 minute explanation: Suppose you ran .100 gap. It would be hell on the coils. They use more power and sink more heat. It would put more strain on the output circuits of MS. (I think, Abe or someone smart confirm/deny this). Gotta go to Cal3.

Braineack

iTrader: (62)

one acronym: cops.

Joe Perez

iTrader: (8)

Shouldn't. Inside the igniter, the output of the MS drives the gate of a FET. The drain/source junction is what carries the load of the primary coil winding.

Besides, the primary current should, in theory, be purely a function of dwell time, without any regard whatsoever for what's happening on the secondary winding.

I failed miserably at my first attempt to build a COP setup. Just couldn't get the damn things to spark. If anybody has four known-working COPS they want to sell me, with a guarantee, I will pay you $100 for the set. I'm tired of the eBay crapshoot.

Ben

iTrader: (33)

Joe, if you're getting insane RPM spike readings, it's not grounding, gaps, or filters. That's the ignitor on its way out. The symptoms of a failing ignitor include misfires with tach drop out.

If you want to send me your COPs, I'll run them in my car for a while.


Ben

Joe Perez

iTrader: (8)

First, an update:





Today at lunchtime I replaced C1 and C2 with .033uf, turned up the boost, and went for a drive. Made a solid 180-185kPa (with onset over-shoot to 203) and not a single misfire!

Gotta do some more driving to really convince myself that we're solid, but this is the first time I can recall it not giving me a single misfire- I'm really leaning towards calling this one resolved. There is much rejoicing.

I believe tonight I shall re-install my IFR8H plugs, set to .030", and see what, if anything, that does.

I was going to throw them in the trash anyway, so PM me your address. Unfortunately, not having the proper connectors I first soldered directly to the pins and then, like a total idiot, assumed that everything was perfect and poured epoxy into the shells before actually testing them. Still want 'em?

I've got a spare igniter shell, maybe I'll try building one. Should be just a couple of N-channel FETs...


I'm serious, though. $100 for a set of COPs with the caveat that the seller guarantees me they will work when properly installed.

patsmx5

iTrader: (16)

Yea, I started to say your caps were huge.(probably should have, but I got off on a tangent about plug gap instead.) I run a 1 pico farad. I did run a 1 micro farad per Abe's suggestion, but had bad misfires. While back Abe argued with others on a ms forum that .1 micro wasn't too big, but now he say's they are.

AbeFM

iTrader: (3)

Damnit, I keep forgetting the glue thing. Nonetheless, bring them over - and I'll knock them into mine, with aligator clips if I have to. Make sure they work and work well. My feeling is the odds are really low there's something wrong with your coils. It took me forever to get my wiring sorted, and I've seen other people have similar issues.

And for the love of god, don't run a 030 gap! You're only making it worse, and you ARE damaging the coils the harder they have to work (though not the MS). The 0.4 hp you'll get out of a bigger gap isn't worth one misfire per pull or a shorter lifetime. Small gap, work on COPs, drink beer.


Two things, 1) Yes, I think it's too big! Live and learn. Two you're in a parallel set up, so you already have a werid filter on your signal (through truthfully not that big a deal, it's there. 3) If I were you, I'd try 10 pf. Or 2.2.

Joe Perez

iTrader: (8)

I'm just having a hard time getting my head wrapped around the idea that a bunch of people are apparently very happy with the .1uf cap between JS8 and GND (ok, the resistors in that scenario are very different, but it's still .1uf) and that he's now very happy with no caps at all. With no caps, mine barely ran at all (it was misfiring at idle) and with .001uf before that, I had pretty regular misfires under boost.

Yesterday I put a .01 in C2 (CMP) and that seemed to improve things somewhat, then today I put .033uf in both C1 and C2 and didn't get a single skip.

This evening on the drive home I'm going to try .0047 in both positions and see what that does. I've also got some .0088s, some .001s and some .022s to try.

patsmx5

iTrader: (16)

Well, first of all I'm running MS II with stock 99 sensors. I tried running about 20 (no joke) different input circuits Abe thought would work. I later find out he didn't actually run them, but they were all variations of what he used, and he thought they would work or be better in some way. After 2 weeks I talked to Arga and he sent me a very simple schematic of his input circuits. I've been running them since.

Joe Perez

iTrader: (8)

I believe we have success. Perhaps a title change would be in order.

This evening before leaving work I removed the .033uf caps and installed .0047uf in C1 and C2. These are the J-code high precision ones, no more ceramic disc -20/+80 BS. Took the very long way home, hammered the **** out of it again and again (175-180 kPa all the way to seven grand up some long hills) and it was simply perfect- not so much as a hiccup and no log abnormalities.

In the interest of being perfectly scientific, I stopped about a mile short of home and replaced the .0047s with a pair of .00047uf caps, keeping in mind that I originally had trouble with .001s when I first built the current input circuit last week. As expected, I hit a hard misfire between 4000-5000 RPM every single time I got into boost, and the log claims that I peaked at around 32,000 RPM.

Didn't have any .0022 or .0033uf caps with me, but tomorrow just for kicks I may try to find a pair and run them at lunchtime to see what happens. Or I may not- I'm actually very happy with I've got now, and having a comfortable margin away from my closest known failure point (.001uf) pleases me.

While the .00047uf caps were in, I also made sure to make a few runs past the "misfire window" while not in boost, and I didn't have a failure. So there's no question that whatever is happening is somehow related to being in boost. That above all else has been driving me nuts, and I finally had an idea that may explain it.

We're all familiar with the tendency of inductive devices to create a big voltage spike when they are de-energized, right? I mean, that's how ignition coils work. Well, what the heck are fuel injectors, then? They are really big inductive devices, wired in parallel, and connected directly to the MegaSquirt

Think about it- if you want a "hotter" spark out of your coils, you increase the dwell time. Well, how do you get a bigger kickback out of your injectors? Increase the "on" time. And that, for any given RPM, is the primary difference between boost and not-boost. Yes, there is a clamping circuit, but all that power still had to travel through the harness and halfway across the PCA in order to get clamped in the first place. 4000 RPM at 60kPa = ~3ms, but 4000 RPM at 160 kPa = ~9ms. More dwell time!


patsmx5, I'd be curious to see the circuit you ended up with. I suspect there are a lot of running around very happy with undocumented mods.

patsmx5

iTrader: (16)

Can't find the schematic. I have it somewhere. I know it's on this computer, but it's hiding. I'll send it to ya or post it when I find it.