Help me figure out.....SOLVED! (okay, not completely)

[EWflyer]

Microsquirt project, fuel-only, small motorcycle - Kawasaki EX-250, Revision 2.890 software loaded.

Finally beginning to beat the "reset" problem. It now runs okay with ignition input from the Coil Negative line. I've got it wired according to the Microsquirt manual directions: OPTO+ to Coil Negative line and OPTO- to 12 volt power. The part of the setup that seems to be curing my reset problem is a version of the "Peter Florence Tach circuit for V2.98 and higher" (referenced here: http://www.megamanual.com/v22manual/PFl ... ch_mod.pdf)

My microsquirt is willing to work (no processor resets) with the Peter Florence Tach circuit built out of a 4.3 volt zener, a 470 Ohm resistor, and a 220 Ohm resistor.

Now that it's willing to run well enough to get good information from datalogging, I've got another problem to work through. If you've got the time and expertise to help interpret what's going on in this datalog I'd be very grateful. When I ride the bike it starts off running okay or fair at best (the VE and AFR tables are probably horrible) but about three or four minutes into the ride it surges, loses power, and slowly dies.

The datalog shows the Wideband Oxygen sensor reading "peaking" repeatedly just before the bike quits.

I'd surely appreciate the help from anyone who's willing to cast an experienced eye over my datalog and MSQ.

EX-250.msq

Kawasaki EX-250 MSQ File

(25.37 KiB) Downloaded 86 times

wed7datalog201008251722.msl

Datalog from Wednesday 25th

(1.04 MiB) Downloaded 85 times

[EWflyer]

Here's the last datalog of the day. I think it shows that my version of the Peter Florence Tach circuit isn't perfect and will require some more experimentation to find the right combination of resistors, capacitor, and zener. But I'm feeling pretty optomistic now because at least I'm not still getting great big processor resets.

Now my problem appears to be reduced to Tach resets.

In this datalog there are two problems that are obvious to me:

Problem #1 - During engine startup the microsquirt gets a couple very erroneous tach input reading, the RPM trace spikes up several times to as high as 53716 RPM. That's not good, how do I correct for that?

Problem #2 - I can see the RPM trace line suddenly drop to zero several times. The "tachCount" in MegaTune goes back to zero. When these events occur I can tell that the microsquirt drops the DutyCycle to zero also (just as it should with a zero RPM indication) so that must mean the injectors don't fire which explains the Wideband O2 sensor reading going suddenly lean.

So I've obviously still got some work to do on my Coil Negative ingnition input signal.

wed8datalog201008251816.msl

Last datalog of Wednesday 25th

(1.44 MiB) Downloaded 79 times

[EWflyer]

I managed to figure out my problems.... The bike runs great now. I'm moving on to getting it tuned by datalogging and using VE analyser in MegaLogViewer.

[24c]

I managed to figure out my problems..

Great to hear about this, what was it? Was it a noise issue?

[EWflyer]

Yes, the coil negative signal is very noisy.

I've come to realize, and some of the documentation on the various Mega and Micro websites even mentions it, that each and every application will be different (sometives even on the same make and type of vehicle). That said, here's my current solution to the coil negative input problem:

Basically, I've got it wired up as the MicroSquirt wiring instructions direct you to do for direct coil sensing.... OPTO+ is connected to the coil negative line and OPTO- is connected to 12 volt power. When I was trying to figure out what to do next I researched through the schematic drawings of the OPTO circuit for every version of MegaSquirt and Microsquirt.

I understood the concept of connecting OPTO- to 12 volt power. It reduces the amount of heat energy being dissipated inside the MicroSquirt. So now you've created a 12 volt bias and are using just the coil's "flyback spike" effect for OPTO circuit triggering. I understand that this is good for the MicroSquirt's long term health, but by not providing a ground for the capacitors built into the MicroSquirt's OPTO circuit it either messes up or totally negates the noise cancelling effect they might have had.

So I had to configure some combination of external resistors, capacitors, and zener. There are suggestions/hints to be found on the Mega and Micro websites, especially in the MegaSquirt board assembly guides. Things like the "Dave Cap modification" or the "Peter Florence tach circuit modification." So there are bits of information to draw from.

My current external solution is:

On the inbound coil negative line I've got the following components in this order: A 220 Ohm 1/2 Watt resistor followed by a .01 uf 400 VDC capacitor (grounded to one of the MicroSquirt's main ground lines) followed by another 220 Ohm 1/2 Watt resistor followed by a 5 volt zener (this is just a version of the Peter Florence tach circuit).

I don't really know why this works, but it does. It has totally banished all processor resets. Who knows, maybe parts of it are totally stupid, I don't know. I was surprised by how low I had to go with the resistors and the zener, especially since all the suggestions I'd gotten over the last few months always involved things like 4.7K resistors and zeners of 22 to 30 volts with never any mention of capacitors at all. As I've said many times here on the forum, I really don't know anything about electronics.

The only problem (very minor) is that at start-up the MicroSquirt sometimes shows strange, really high, tachometer readings (how about 56,712 RPM?). If there's a way to prevent this from happening, I haven't discovered it yet.

I also had a fuel delivery problem. I eventually realized that the fuel pump had trapped air inside of it. When the bike moved the air bubble would move as the fuel inside the fuel pump housing sloshed around. The result of this bubble-and-slosh problem were random, momentary drops in fuel pressure which would cause mysterious lean oxygen sensor readings. I've re-mounted the pump with a steeper angle. I should have never made this mistake because I actually have the maintenance manual for the Suzuki LT-R450 Quadracer that the pump comes from and it plainly shows the pump mounted nearly vertical with the intake line at the bottom and the high-pressure output and a "flood line" at the top. What was I thinking (or not thinking) when I mounted it nearly horizontal?

As I've said, the bike now runs great. What a relief.

[24c]

The only problem (very minor) is that at start-up the MicroSquirt sometimes shows strange, really high, tachometer readings (how about 56,712 RPM?). If there's a way to prevent this from happening, I haven't discovered it yet.

I am not an electronics person, but the only thing I can suggest is alternator noise, as sometimes they produce a higher frequency AC voltage, which might be coming into the ground circuit of the Microsquirt. If you could repeatedly get this high RPM, and say disconnect the alternator and it didn't happen, you'd be homing in on the culprit.

[EWflyer]

Here's where I stand right now: The bike starts and runs great up to 9,400 RPM. At 9,400 RPM I get tach drop-out.

Does anyone have any suggestions or information to help me reach higher RPMs, I need 14,500 RPM for the bike's redline.

Once again, here's my current setup:

I've got it wired up as directed by the MicroSquirt manual wiring instructions for direct coil sensing.... OPTO+ is connected to the coil negative line and OPTO- is connected to 12 volt power (creating a 12 volt bias to protect the MicroSquirt from too much current and heat). On the inbound coil negative line (connected to OPTO+) I've got the following components in this order: A 220 Ohm 1/2 Watt resistor followed by a .01 uf 400 VDC capacitor (grounded to one of the MicroSquirt's main ground lines) followed by another 220 Ohm 1/2 Watt resistor followed by a 5 volt zener (this is just a version of the Peter Florence tach circuit).

Is there some electronics "rule of thumb" or other common-knowledge reasoning I should follow in choosing the right combination of resistors, capacitor, and zener to deliver/condition the tach input to the MicroSquirt? What will allow me to get to the 14,500 RPM redline that my bike needs?

[24c]

I once stuck a diode in series on VR1+IN when I was running 2.890 on the GTS1000, and this got me past 5K hiccup and straight through to 11K, but I seem to remember John Pellew...Taimoshan? used a diode on his ground to get his Aprilia engined bike to rev cleanly upto the redline. You'd have to search the posts, as I think forum member 800vtwin was involved too. Sorry I can't be more specific.

Have you tried to datalog this? It might give you a clue what is happening and help others see too.

PS BTW you might be interested in what matt_gsxr is doing elsewhere.

[EWflyer]

Just a standard diode? A 1N001 diode. I've got a few of those on hand. Just that and nothing else on the line? (like a capacitor or resistor). I'm willing to try just about any recipe to get a reliable high-RPM result.

I'll research matt_gsxr to see if there's anything in his work that might help me out.