Still struggling with ignition issues [mostly solved]

[grippo]

What about the mask settings? The description in the help says that percentage mask is the percentage of the tooth to use as mask before expecting the next tooth. Since I only have one tooth, 1/4" wide, and the circumference of my pickup area is about 9.8", I would expect to need a mask percentage of about 3500% or more. I found, though, that the maximum allowed is 200%, and when I set to this I was only getting spark once every two or three revs.

Thanks,
Paul

The % mask means % of the last time between tach pulses on the particular channel you are on. So it can never be near or greater than 100% or you will guarantee missing the next tach pulse. The idea is to avoid coil ringing noise so the processor refuses to accept any tach pulses on a channel until mask % of the time betweenh tach pulses. The time mask is an absolute time which you could use if you knew the coil ringing time. I would keep the mask around 50% and not use the time mask.

[750essess]

Sorry I thought I read you where at BTDC not ATDC.

[SQLGUY]

I re-measured and re-marked (with a Sharpie) both 1/4 TDC, which was very close to my initial marking, and 2/3 TDC (1/4 BDC), on the VR rotor and on the pickup plate.

I then verified with the scope that the VR inputs were indeed 180 degrees off from each other. I also verified that their pulses were in phase (both went high first, then low relative to the VR- on pin 33) and of very similar amplitude and shape. The signals are also very clean, at least during cranking, when referenced to pin 33.

Still, to get my calculated, 0 degree advance, spark to fire at TDC for the two channels, I had to use the advanced offset option. I ended up with -120 for the base trigger offset and -55 for the advanced offset. This gave me spot on TDC firing for both the 1/4 coil and the 2/3 one. At this point I was able to start the bike and rev it a bit. I've attached my msq, just in case.

I'm pretty sure I'm not hallucinating. I do know how to use a scope, and the one I have is a decent one (Tektronix 2235 - 100MHz 2 channel). So why am I seeing what I'm seeing?

Thanks,
Paul

[grippo]

I will try running your msq and see what I get. But I just want to confirm this is NOT and ODD fire engine. Is that correct ? All 4 cylinder's TDCs are 180 deg apart in firing order ?

[SQLGUY]

It's a tranverse four. Cylinders 1 and 4 are TDC when 2 and 3 are BDC. Firing order is 1, 3, 4, 2.

Here you can see a better view of the engine (cylinders 1 and 2, anyway):



And a cutaway drawing of a SECA 650 engine. Mine is the same, but for displacement (larger bore and stroke).



And here you can see the pickup plate and rotor with 1/4 TDC (index mark at right when lined up with black spot next to slot on rotor) and 2/3 TDC (black spot at left on plate when lined with that same black spot near the slot on the rotor). I found TDC by putting a screwdriver in the spark plug holes and cranking the engine around to what seemed the top center of movement. There's probably a few degrees of error in my markings.



Thanks for the help,
Paul

[grippo]

I ran your msq exactly as is using a 300 rpm input signal on both VR1 and VR2, except they were phase offset by 180 deg to match the sensors being 180 deg apart. The outputs were nowhere near 180 out of phase - they were separated by 235 deg and 125 deg rathere than 180 and 180.

Based on your picture (the first is easier to follow) I interpret that when the large welded on pointer is pointing down, so that the bottom (1/4) sensor sends a signal, then the engine is about 90 deg ATDC for 1/4. This is based on CCW rotation facing the picture. In the first pictur (on page 3), the pointer is toward the left (approx due west) and you said the engine was 1/4 TDC at that point, so a 90 deg ccw turn(to due south) would put the engine 90 deg atdc when the bottom 1/4 sensor triggered. So based on that, a base trigger offset of -90 deg should be used, and since the engine is not odd fire, the adv offset for #2 should be 0. When I reran with nothing changed except those 2 numbers, I got the following on the scope:

----------------V--------------------------------------------V----------------Ignition 1 or 2 (= 1/4 or 2/3 TDC)

-----------------------------^--------------------------------------------^-------------VRIN 1 or 2
----------------|--90 deg--|-------------------------------|--90 deg--|---------

This is exactly what I would expect. At the time the VRs are trigger you are 90 deg past TDC and the spark outputs occur 90 deg before the VR triggers (actually 270 deg after, since the trigger has to come first).

[SQLGUY]

but that's not what actually happens on the bike.

I had welded on that tooth with the intention of running with -90 and 0. If I run with -80 and 0 I get the #1 cylinder firing near its TDC, but the #2 is about 50 degrees retarded. Again, #2 is on the VR1 channel and #1 is on the VR2 channel.

Any other ideas why pratice would be so different from theory here? Should I try reflashing and reprogramming from scratch?

I had seen some other threads where people were overdriving the VR inputs and needed to add input resistance. Could that be part of the problem here? IIRC, my VR signals run about 8V p-p during cranking.

------------------------------------------------------------------------------Udpate----------------------------------------------------------------------------------

I re-flashed and re-programmed from scratch (2.884), no change in behavior.

I then tried flipping to "Falling Edge." That did allow me to get correct firing during cranking on both channels with -70 for the base trigger offset and 0 degrees for the advanced offset. As a confirmation that this was the problem, I switched back to "Rising Edge" and, sure enough, I was getting an additional 50 or so degrees offset between channels. So, while I don't understand it (the VR pulse is pretty narrow and quite symmetrical), I'll stick with "Falling Edge".

From there I still have the problem that, once the engine catches, timing seems to be deviating from what it's doing during cranking. The timing light shows no coherent firing once the engine starts trying to run, and I get plenty of backfiring and misfiring. I've tried a spark map that's all 0's, but still this behavior continues. That's where I'm stuck now; the engine will try to start but will not run at this point.

------------------------------------------------------------------------------Udpate 2----------------------------------------------------------------------------------

By playing around with the spark map, I got the bike to start and eventually to rev up somewhat. Just now I got it started and had it running at 4000 RPM steadily for a minute or so. During that time I was able to look at the timing and compare it to what Megatune was showing. Megatune was showing full advance from my map (37.5 degrees). The timing light was showing maybe 20 degrees advanced. Moreover, when I let the throttle back down slowly, timing which Megatune was showing as 7 degrees advanced was actually well ATDC.

So I am still dealing with things that don't make sense. Why is uS apparently adding a bunch of negative offset to my timing when the engine goes from cranking to running? How can I compensate for this?

Thanks,
Paul

[grippo]

"I then tried flipping to "Falling Edge." That did allow me to get correct firing during cranking on both channels with 0 degrees for the advanced offset."

Was your base trigger offset also -90 or was it still -270 ?

I checked on the scope and added 30 deg to the table advance and the scope ignition outputs both moved back accordingly. So I don't think the processor is adding anything that isn't supposed to be there.I also checked that there was no difference between cranking mode and past cranking, and there wasn't and shouldn't be, because you are using calculated trigger (meaning table advance) while cranking.

I don't know what to tell you now. I will talk to Bruce about it tomorrow.

[SQLGUY]

(At least, never with the current iteration of my fabricated reluctor).

With rising edge I was running -120 on the base trigger and -55 on the advanced offset.

With falling edge I am running -70 on the base trigger and 0 on the advanced offset.

Both of these give me TDC firing on the two channels during cranking.

Any thoughts about adding resitors to the VR inputs? What p-p voltage is expected on the VR inputs before attenuation might be needed?

One other thought: is it possible that the VR2 input is triggering on falling edge regardless of the configured setting? It wouldn't surprise me if there's 27.5 degrees of rotation between the rising egde and falling edge of my tach signal.

I looked at the source a bit. I haven't found anything yet that might explain what I'm seeing, but I do see that cranking ignition is a separate block of code from running ignition; I also noticed that there are a lot of ignition code sections that apply to crank wheel dual spark setups but not to dual tach dual spark setups (DualSpkOptn = 1). Considering that everyone else seems to be using a 36-1 wheel I am beginning to feel that I'm really in uncharted waters trying to make dual tachs work. For about $70 I could get a suitable-to-be-welded to my reluctor 36-1 wheel from England. Might I be saving myself future headaches by going down that path instead?

Thanks,
Paul

[grippo]

(At least, never with the current iteration of my fabricated reluctor).

With rising edge I was running -120 on the base trigger and -55 on the advanced offset.

With falling edge I am running -70 on the base trigger and 0 on the advanced offset.

Both of these give me TDC firing on the two channels during cranking.

Any thoughts about adding resitors to the VR inputs? What p-p voltage is expected on the VR inputs before attenuation might be needed?

One other thought: is it possible that the VR2 input is triggering on falling edge regardless of the configured setting? It wouldn't surprise me if there's 27.5 degrees of rotation between the rising egde and falling edge of my tach signal.

I looked at the source a bit. I haven't found anything yet that might explain what I'm seeing, but I do see that cranking ignition is a separate block of code from running ignition; I also noticed that there are a lot of ignition code sections that apply to crank wheel dual spark setups but not to dual tach dual spark setups (DualSpkOptn = 1). Considering that everyone else seems to be using a 36-1 wheel I am beginning to feel that I'm really in uncharted waters trying to make dual tachs work. For about $70 I could get a suitable-to-be-welded to my reluctor 36-1 wheel from England. Might I be saving myself future headaches by going down that path instead?

Thanks,
Paul

It appears there is a delay between when the pointer passes the sensor and it is actually received by the processor. I have heard of such delays that are inductance caused and a function of rpm, but nothing of the magnitude you are seeing. The 20 deg discrepancy at cranking is on the order of 10 ms.

If on your 1 min at 4000 rpm the rpm was stable and there were no missed tach inputs then that means your vr circuit is good and I wouldn't mess with the resistors.

I have never seen the processor trigger on the same edge regardless of your input setting. What can happen is that someone changes the setting but doesn't cycle power or at least stops the engine from running, so that the change never takes effect.

Cranking and running are 2 different modes for both fuel and ignition. However, you set Calculated for cranking trigger, so there shouldn't be any difference when you go past cranking rpm. I tested this when I tested your msq and there was no difference.

I don't think you are in uncharted waters with dual tach mode - it is working on at least several engines. Getting ignition right is the hardest problem for everyone except those who luck out and have a system that is completely perfect for the default settings and the hardware. We are working on a solution for this, but it will take time.

That said, it is my opinion that an ignition that only updates engine position every 180 deg will not give precision timing on a small, light engine that can accelerate very quickly. So that is a reason for going to it. But I don't know if that is going to cure your problem.

Some things you might try: change alpha-beta-gamma to last interval; try swapping vr sensors; scope vr1 vs ign1 and vr2 vs ign2 . The scoping needs to be at those points in the circuit where the signals go to and come out of the processor - but not on the processor itself - too risky to short things out.