Duty Cycle Spike?

[grippo]

Wes,

Your datalog is good data. It does look like something is wrong in the code. The compiler we use does its arithmetic somewhat differently than most others and I have run into a problem with it at various times. The significance of the 32 is that the pw is limited in the code to 32 ms because anything over that is unreasonable (except prime pulse can go to 43 ms). In reality, the calculated pulsewidth is going off the chart and I doubt you have any ve table entry that would explain this even if it were due to a map or rpm spike. When the value goes over the table limit, its set to the table limit. (for example, if map went to 400 and your largest map in the table is 200, then the ve for 200 would be returned - no extrapolation.

Let me look at the arithmetic tonight and I will either find something and send a fix or send you code that will capture the event and latch some data which you can look at at any time just before you turn the car off.

[wes kiser]

Thanks you very much, I love this fuel injection system, and this is the only thing about my setup that is bothering me. If it fixes it I wil send you a paypal for a 6 pack of the beer of your choice

[grippo]

Wes,

I will be just as happy as you when its fixed. A thought occurred to me while looking at the code - there is no battery voltage recorded. If there were something weird with the battery voltage (not a weak battery, just some sort of spike probably from the alternator - happens a lot) this might cause the injector open time to go negative. Since its an unsigned int it could act like a huge number> 32000. This would explain everything and is easily fixed by 1 or 2 lines of code.

I will put this fix in the code and send you a version to test. In the mean time, if you test your car again, record the battery voltage. It may show the problem.

[FixItAgainTony]

Hello Wes,
It looks like we are seeing something similar. Attached is a better view of one of my earlier log files.
There are two spikes:
Time / PW / DCyc / BattV
1721.031 / 32 / 192 / 21.4
2036.049 / 32 / 197.5 / 21.5

I have a few other logs, but they look about the same - voltage spike to 21.5 is the usual signature - the other stuff does not always get caught in the log, but car does miss a beat for a moment.

The BattV is what caught my eye - it will go high ~21.5V in the log, on the next sampling of data, it is still high (~14v) and coming down - perhaps it really did spike? I have access to a portable scope .... I'll take a look and go from there. There are a couple of 22V Zener diodes in the circuit - perhaps my set up is seeing some sort of voltage spike that is getting clamped ?

I have also attached my MSQ file. I probably do not have the trigger quite right - I am triggering off of the coil, but will be going to EDIS and have an output drive circuit for the SAW I was testing - I figured the trigger did not matter too much as I am running Fuel only right now.

MSII V3 PCB firmware 2.33 fuel only
High drive for Fuel Pump
Flyback installed / port injection low impedance

- Charles.

[grippo]

Talk about timing - I hit the submit key and my post appeared to hang - you must have hit your key seconds after I did. This is exactly what I was taliking about and glad to see real data for it. The battery spike should be fixed in the car, but I am still going to put a check in the code to prevent this because this can happen to anyone as their alternator wears out.

[Nick Jones]

Hi,

Just spotted this thread and wondering if I am seeing another variant of this problem. Also posted under 'Tach Spike withe EDIS'

Running MS1 2.2 and MSnS-e on a Triumph straight 6.

Initially this was a fuel only setup with points ignition. It used to have a fairly dirty tach signal with a few sharp spikes over general low level noise but it ran pretty well provided the rev limiter was disabled as the sharp spikes would set it off.

Now upgraded to EDIS-6 and re-enabled the rev limiter. Generally runs really well but was 'hitching' intermittently under load. Ran a log today for the first time since the install and discovered that although the tach signal is generally much cleaner, I still have spikes, which cause the rev limiter to act. Disable the rev limiter and the hitching goes away - but what is causing the spikes?? I can't see any reason why the cause should be the same as before but it seems a big co-incidence!

Have checked wiring and earths, wiggling the wiring with the engine running does not replicate the problem.

Also, sometimes the effect appears to be to quick for the log to catch although PW and DC may show a notch. With the rev limiter off the effect on the engine is almost undetectable.

Screen shot attached shows a small sample of the log.

Nick

[mops]

somebody at some stage mentioned that one could install a large capacitor on the MS power supply line... what would be a correct size for that ? possibly with large capacity there we could eliminate voltage spikes alltogether....

oh and BTW....
for last 3 weeks I've been troubleshooting erratic behavious of my MS2 unit. last night i got really p*** off and took it out of the car to try diffeerent firmwares. I flashed 2.34 over 2.33 and all my problems were gone ! all working like a dream now, however i did few other thing that might have affect it....

1. i pulled out pretty twisted wiring harness (possible interference there)
2. installed ignitor driven by vb921 (so vb921 now only provides signal current, not 7.5A of interfering high frequency pulses)
3. unscrewed MS from the car's chasis (maby some grounding issues there ? my ms is grounded straight to the battery with dedicated wire
4. funny think i noticed about db37 connector... when I had my MS inside casing, it seems like the connector doesnt go in all the way... it's limited by the aluminum backing plate around db37 connector.... I removed that plate and made sure the connector is all the way in.

seriouslt, in the future designs of MS I think high current (more interference possibility) wires should be grouped in one connector (i.e. injector, coil and grounding wires), and all the small, low voltage signal lines along with their grounds should be wired in another connector, to preven possibility of electrical interference....

I mean currently in DB37 connector... nearly 30 wires tightly packed together, some of them running huge pulse currents (like 7.5A off VB921... i cant even comprehend how a single db37 pin can handle that ! ) alongside things like VR sensor or O2 sensor wires...... I mean that's just asking for trouble....

[wes kiser]

Good call on the battery voltage. Every time there was a pulse spike, there was also a battery spike. On the pulse spikes, the batt voltage spiked 18 (probably spiked higher but the peak may not have corresponded to a sample). I also caught a 21.9v spike, but oddly not corresponding to a pw spike. Is there anything, besides a bad alternator that can causes spikes this high? I imagine that increasing the lag factor on the battery voltage sampling would help this, but if I have something spiking to 22 volts that can't be healthy and should be fixed, right? I would think there is no way noise in something could cause a spike this high?

[Bernard Fife]

Wes,

You can install a car stereo power filter on MegaSquirt's 12V supply. This is never a bad idea on a MegaSquirt install, it cleans up the power supply spikes a lot. They are quite cheap (~$5) and common, you can get them most places that sell car stereos.

Lance.

[wes kiser]

Thanks, I may try a car stereo filter. So the spikes can be that high without a component actually being bad? I gues what strikes me so odd is the voltage looks very stable, not noisy, except for the occasional spike. If this isn't that out of the ordinary I will just hook up the car stereo cap and be happy.