Dwell observations on the Toyota COP.
 

For some time, I'd been experiencing an occasional misfire. Review of my datalogs showed that this was caused by a dropout in the crank trigger signal, as indicated by the fact that when it occurred, my RPM reading would spike to an impossibly high number.

Simple enough. Adjusted the position of the sensor relative to the crankwheel, tweaked the decoding circuit a tad.

The problem went away... almost. Every now and then I'd still get an occasional hiccup. I was just about ready to tear back into the crankwheel system, but decided to do a bit more logging first. Good thing, that.

The new logs showed that I was no longer getting sensor dropouts, just the tiniest little burble in RPM that you'd expect from a bona-fide ignition misfire. I'd have realized this earlier if I put two and two together and noticed that the tach on the dashboard was no longer dropping out when it happened. So this is now definitely a separate issue- one relating to the ignition system itself.

First order of business: new plugs. And this time, I gapped them down to about .035", as opposed to the .045" that I'd previously been running.

Didn't help. Well, maybe it did, maybe it didn't. The misfires were so infrequent at this point that it was hard to tell. But they were still there.

This morning, I decided to put my fancy new inductive DC current probe to work. Dragged out the scope, hooked channel 1 up to the brown / yellow ignition trigger, and put the current probe on channel 2, measuring the +12 supply into the #1 coil.

Up till now, I've had my "Running Dwell" set at 2.5ms, as commonly espoused by those who came before me in documenting the behavior of the Toyota COP. But wait- the Megasquirt automatically adjusts your specified dwell up or down depending upon battery voltage! And while you can adjust the ratio by which this is done for injector duration, you cannot adjust it for dwell, at least, not on the MS1. What's worse, the ECU's only knowledge of system voltage comes from its own power supply, it can't know what's going on at other points in the harness, such as at the coils.

So, here's what I saw with a warm, idling engine, with 2.5ms specified dwell:

http://img40.imagefra.me/img/img40/2...nm_9e8f5ed.gif

For those not accustomed to reading scope charts, the green line shows the voltage on the line between the ECU and the coil. The yellow line is the current flowing into the coil primary. For the green line, each major division on the vertical is 1 volt, and for the yellow line, each major division on vertical is 2 amps. (My current probe was set to 100mv / 1A, and the scope is set to 200mv / div). Each major division on the horizontal is one millisecond.

So you can see that although I'd specified a 2.5ms dwell time on the MS, it was derating this to approximately 2.1ms based on what it believed to be the system voltage: 14.24 to 14.35 volts as indicated by Megatune. At this dwell, the coil primary current made it to about 8.5A, and was still rising sharply when the coil fired, meaning that the coil was nowhere near saturation.

I then entered 4ms dwell into the MS, and got this:

http://img38.imagefra.me/img/img38/2...km_f2a4ab2.gif

It's now giving me 2.9ms actual dwell (suggesting that the correction is proportional, rather than fixed) and perhaps more interesting, we can see that after about 2.5ms, the coil itself has activated an internal current-limiter, clamping the primary at just over 11A. I didn't realize that they did this- probably goes a long way toward explaining why folks have been able to run them on the stock ECU without the coils self-destructing.


I continued playing. Next, I set the MS to a dwell of 3.2 ms:

http://img40.imagefra.me/img/img40/2...lm_972598d.gif

Actual dwell is now about 2.7ms, and we're just barely hitting that current limiter at ~11.1A.

Hmm. Let's try giving the MS a 3.0ms dwell value:

http://img38.imagefra.me/img/img38/2...jm_5144133.gif

That did it. Actual dwell is now 2.5ms, and the coil primary is peaking right at 11A without hitting the limiter. I could probably run just a tad more dwell without putting the coils at any risk, and likely will.

So, this makes me feel a lot better. It will be interesting to see if this fixes the problem, but if I had to bet Hustler's mom's life on it, I'd lean towards "yes." An extra 30% coil current should translate to an extra 30% potential ignition energy, and that's gotta help.

It also confirms the measurements which were taken long ago by others, with the obvious caveat that you have to make sure that your ECU is actually doing what you tell it to.

Dear Joe,

Please help me fill in my dwell voltage correction curve while I digest your information.

My current curve looks like this:

EDIT: rereading, it seems like my curves should be okay as is. at 12v I'm adding 28% vs. your 17% number. Maybe I need to back it out.

Note to self:

If car misfires, take it to Joe P's house.

--Alex

On 029v, open "msns-extra.asm", go to line 7068. There's your dwell Vs system voltage table. Correct to your heart's content, run "compile.bat" to generate the new binary (I run "wine ./casm08z.exe msns-extra.asm S L I M" on Linux), correct your therm values with easytherm, upload the new firmware. Easy as 1,2,3,5,8,11...

Jim

... 21, 34, 55, 89, 144

Thanks, Rev. Never occurred to me to look at the source.

Dwelltv is obviously the batt correction factor. Very curious as to what dwelltf is. I see it getting called twice, once is a straight STA, the second is an STA after a DEC. Curious to see where those wind up getting called...

You're seeing 8.5A @ 2.1 ms @ 14.3 V which is pretty much spot on with what I've measured here:

https://www.miataturbo.net/forum/t11744/

When the internal high current driver current limits at 11A, it will dissipate a lot of power, so it should be avoided. In the above tests I did, the current limit varied with temperature so a bit less dwell will keep things safe.

The firing capability of the ignition system is pretty much a function of the voltage at the plugs, which is proportional to the current. I will guess that the spark voltage at 8.5A is plenty for firing plugs with a 0.035" gap, so something else is causing your misfire, or if the extra dwell fixes it, perhaps the extra ignition strength is hiding another problem.

Joe, when were your misfires occurring? 11A is none too shabby, but 8.5A through .035" should be strong enough to light off the charge at any 'moderate' manifold pressure. We scoped a fairly hot OEM system on Friday that was hitting only 7A, and I think the factory gap on that car (SRT8 Challenger) is something ridiculous like .049".

MS2 and later has a dwell correction to battery voltage table.

dwelltv is the voltage. dwelltf is the factor. So, with the following table:

051T,068T,085T,102T,119T,136T (which is 51%, 68% 85%, 102%, 119% and 136% of 12 volts)
250T,124T,084T,064T,051T,044T (which is x3.9, x1.9, x1.3, x1, x0.8, x0.7 of the running dwell)

if your voltage is 6V (51%), your final dwell will be running_dwell*3.9, etc.

Jim

based on Jason's curve. I went ahead and changed my curve to this:

any thoughts?

It would need to be scoped. I would imagine, if using the oem power supply to the coils, that you are under dwelling them.

Maybe I understood something wrong but my curve looks like this:

I did this because I thought - that if you were running at ~14.0V most of the time - you want that 2.6ms aka 100% at that spot - wrong?

Greets

I was going off this data:

http://pages.sbcglobal.net/jcuadra/t...OP%20dwell.gif

It runs fine on the original curve that was in place.

But as Joe scoped above - even at 3.0ms running dwell set in the MS (i.e. at ~14.0V] it gives him a total of 2.5ms actual dwell...

What do your ignition settings look like? (There the need for you msq again... :doh: ]

Well if we look at the code:

Like Reverant says, at 14v you'd take 51/250*4 = .816

so 2.5 x .816 = 2.04ms, pretty close to what Joe got.

same with his 3.0ms values x .816 = 2.448, again pretty close to Joe's actual findings.

am i wrong? seriously I'm dumb correct me if I'm wrong.

Don't get me wrong - I have as much knowledge about this as a 3month old...

I think the main question is - what is the 100% point - 14v or 12v - because otherwise we did the same thing - we took the graph and looked up the values und made them to percent.

the code straight up says:

O.K. - what have you put in your ignition settings in MS2 now. (Pleeeeeease post your msq....)
I understand that 12v thing now, or better I think I understand.

Would that mean, Maximum Dwell Duration should be set to 3.0ms?

Greets

the way I read it, here: http://msextra.com/doc/ms2extra/MS2-...on.htm#moreign , you'd set the max dwell time to 2.5ms (like I have) and then just use the battery compensation to compensate for voltages. I just can't tell if it'll never spark above the max dwell time or not, but i see no other way to set it other than using "time after spark."

MY spark settings like like the 9494 DIYPNP values but with 2.5 set as max dwell duration. So far it's been running without a hitch.

You are absolutely right, but the problem lies in the fact that you're referencing voltage at the ECU for the correction factor, which is a problem because there's less voltage at the coils than at the ECU. The only way to properly set this up is with a scope, otherwise we're just guessing!

And my 'best guess' is that your original settings in post #2 were probably the most correct.

I guess if it ain't broke... it's running hella strong, and I'll be dynoing my MS-II unit vs. my MS-I unit on the same spark map on April 10th.

Yes, I believe Brainey has it right.

I do like the MS2s approach to dwell correction much better than the MS1s, insofar as that it's easily user-configurable. Hacking the code isn't particularly difficult (particularly with MS1, where it's a nice, monolithic package) however I really hate forking software needlessly. A year or two down the road, I (or the new owner of the car) will put new code onto it, forget about the fact that the old code had been manually hacked, and wonder why they're getting misfires all of a sudden.

No, I think I'll just stick with my original solution, which is to up-rate the specified dwell. It works fine, introduces the fewest variables, and is easily copied by folks who don't own scopes and don't like to mess with compiling their own code.

Definitely interested in results of that.

So, Joe, do you suggest I stick with the dwell curve that was defaulted, or try the new values based off Jason's optimal dwell curve?

That's what the 10,000 uF capacitors will help.

And of course, for those of us running in wasted-spark mode, the problem is amplified, as the total draw on the wire supplying the coils is twice what I observed on a single coil. 22 amps is a hell of a lot of current for what I recall to be a 16ga wire. If these cars fell under the purview of the NEC for cabling vs. current, we'd be in violation.

The currents are short transients though.

Are the two noticeably different?

The gist of my findings here are basically that Jason's data is correct insofar as 2.5 ms dwell, with the caveat that, at least in my system, they are correct for whatever supply voltage is actually being encountered at the coils (which I should probably measure), and must not be derated based upon the ECU's observation of system voltage.

I should note also that in my setup, there are a pair of 4,700 uf caps in parallel across the coils' supply, located at a point between the old igniter connector and the back of the head.

I wonder if this accounts in some way for the knee in the current profile that I observed in my analysis?

The MS-II default curve is:

It runs fine and I have my dwell set to 2.5ms. The manual suggests it'll be fine for 99% of users.

Based off Jason's dwell curve, I thought it might be more ideal to go to:

but it's definitely going to be dwelling less.

So far as I can tell, my own observations, Jason's observations, and the MS-II default curve all match up. If you dwell the coil for 2.5ms at what is most likely 12 volts or so (as measured at the coil) you get peak energy.

Since voltage as measured at the ECU is always going to be higher than actual voltage at the coil, I'd suggest sticking with the MS-II default.

Joe the knee in the current waveform is due to partial saturation of the magnetic core in the coil.

gotcha chief.

If that were the case, I'd expect the slew rate to go down, not up.

For all I know, it could just be an artifact of my (relatively cheap) current probe. This is the first time I've done this sort of thing with my new Chinese unit as opposed to the $2,500 Tektronix one we have in the lab.

'Tis not an artifact. I'm looking at a trace right now showing the same behavior. Completely different coils on a completely different car.

When a core saturates, inductance drops.
V = L * di/dt

For a given applied V, if L goes down, di/dt (rate of change or slope of current) goes up.

I get the same exact slope with a $30k setup at work. :D

Hmmm. Well, I admit that I suck with magnetics. Anything involving Q falls into the realm of FUD so far as I'm concerned. We have a guy here at work who absolutely revels in it, but he frightens me.

Several years ago, I did this exact same test with my then-current setup, involving the stock '90-'93 coils, using the expensive Tek scope from work. Here's what I saw:

http://img34.imagefra.me/img/img34/3...lm_3f63c21.gif

You can see that the current rise rate is slowing down as primary current increases. Knowing no better, I simply assumed this to be the natural behavior of an ignition coil primary. In this capture, the stock Mitsu igniter is in place, so this may simply be an artifact of that device (a "soft" current limit, as opposed to the Toyota's hard limit) though of course this is only speculation.

The leveling-off continued as dwell was increased beyond what's shown in this image, until a knee point was reached which I judged to correspond to the saturation point.

The reducing slope in the current rise above is due to the primary resistance. Classic L/R waveform. In the Toy COPS, primary resistance appears very low.

ok here. i finally remembered. goes in for emissions tomorrow.

Going slightly off topic here - does anybody know what's the dwell time for the 2001+ stock ECU? Is it the same as the 99/00?

thanks

her shann

In my dwell thread, I posted the ideal dwell numbers for the 01 coils. The factory dwell will be slightly shorter than that.

Jason,

thanks - found it! Looks like the dwell time is just slightly longer than the 99/00, which means I will probably benefit with a dwell reducer circuit for the Toyota COP.

her shann

Just to update this thread: The MS-II and MS-III dwell setting is based on 14v, not 12v like MS-I. Therefore Zaphord's correction table is perfect and then use 2.1ms as the dwell, NOT 2.5ms.

Sorry to resurrect this thread but can I just confirm the ms2 settings for Toyota COPs should be:

Dwell battery adjustment
9 156
10 140
11 128
12 116
14 100
16 88

Cranking dwell(ms) 3.5

Max dwell duration (ms) 2.1

Thanks,
Barry

barry, did you ever get a confirmation on those numbers as correct/incorrect?

Well I have been running them and the car is running great, so I guess they are fine.
Barry

how will this differ if say I have standard ignition not COPS?