MicroSquirt.com read only forum (Jan.15,2018)

grippo,

I understand what you're saying, and it makes a lot of sense. But couldn't this problem be avoided by using the "Falling edge" instead of "Rising Edge" or vise versa, or am I not understanding this option correctly?

How would I go about putting a bias of .5 or 1 volt into the signal?

Thanks,

Kyle

If you look at the waveform you can see that the rising edge is near vertical so it crosses the 0 V line at a avery clearly defined place, that will be the same each time, while the falling slope, since it is started near the 0 line, the exact 0 crossing could occur anywhere over a fair space of time, giving you sloppy timing. If it is sloppy enough it will be rejected. But regardless of which edge you pick, the small wiggle on the flat part of the curve can cross the 0 axis right near the middle and then again when you have the large falling edge. This would give 2 detects which would be rejected because they are too close together.

Now remember I am not saying with certainty that this is the problem, it could be your configuration (msq), or many other things.

I don't know off the top of my head what resistors to change, but there is an extensive writeup about it in the megamanual or at microsquirt.info. Bruce also posted some stickys.

Kyle R wrote: How would I go about putting a bias of .5 or 1 volt into the signal?

You can very easily introduce a bias. Take a diode, like a 1N4003 or similar, and put it inline with the VR+ signal. In other words, disconnect the VR signal from the VR1+ input on MicroSquirt and introduce the diode in series. Have the band end of the diode point toward the VR sensor (away from the VR1+ uS input). What this does is introduce a -0.6 volt drop in the circuit, meaning that the input waveform will have to go 0.6 more negative for the same output response (compared to the un-mod circuit). If one is not enough then add another diode in series (band also pointing towards the sensor) this will increase the drop to 1.2v or so.

Test this with first the single crankwheel mode like you did earlier, and make sure there is a stable cranking RPM.

Great, we'll give that a try^^

Here is the MSQ file that we got off the microsquirt. If anyone could take a look and see if anything is obviously wrong, we'd appreciate that.

We also did some data logging tonight, here is one of our cranking attempts. We've also had it where we get one intermittent tach pulse and trigger, we are going to try to reproduce that tomorrow as well.

Thanks,

Kyle

We put the diode in place and it seemed to fix the problem, so hopefully we can get it running this afternoon.

With this diode in place, will the signal get messy at higher rpm? The engine will spin about 11,000 RPM

Thanks,

Kyle

It could move and it could hurt you, depending on which way the signal moves - up or down. On the other hand it could help by time compressing the signal so there is much less of a flat spot.

Kyle R wrote:We put the diode in place and it seemed to fix the problem, so hopefully we can get it running this afternoon.

With this diode in place, will the signal get messy at higher rpm? The engine will spin about 11,000 RPM

Thanks,

Kyle

The diode addition should not affect high RPM operation, unless, like Al indicates, the flat spot jitter peaks expands into the offset region. I would suspect not, the 0.6V offset should be enough to keep you safe, but the final say is when you push to the 11K point...

- Bruce

A team of undergrads at UMaine has a similar problem. We have the same sled, a 2007 Yamaha Phazer, and can get spark from "Single Crank Wheel" and "Single Cam Wheel", but not "Rising Cam with tach" or "M-0 Wheel". We've added a diode in series with the VR+ line and tried up to three resistors on the VR2+ line. The addition of the resistors allowed tach signals to come through, but we still don't have spark when using the "M-0 Wheel" dual spark setting.

Unlike Kyle, our cam sensor ouput is not square. We looked at it with an oscilloscope, and it appears to ramp up from zero, then fall suddenly. Since its the same snowmobile, we're not sure why it looks different. We are assuming that this is the problem. Is there anyway to transform the signal so that it is square, or is it a problem with the sensor? Any other suggestions would be greatly appreciated!

Hi Kyle,

I have had some success with this exact setup, I'll PM you some ideas.

Geoff

The signal doesn't have to be square. Most VR signals will be very spiky. The main thing is that either or both of the rising or falling edge zero crossings occur in a fairly vertical and consistent fashion, and (as pointed out in this thread) that the signal not hover around 0V for any length of time.

Here's what the signal from my VR sensor looks like (or at least looked like with a previous, working, 8-1 trigger wheel):



In this case, IIRC, I had some 12V zeners clipping the signal, so it normally would have been pointier. Notice that the rising edge here makes the better trigger point, since it crosses zero more vertically than the falling edge. In some cases you may need to reverse the mechanical polarity of the sensor (switch the VR+ and VR- lines) in order for uS's bias to best condition your signal for zero crossing detection.

If you don't have a scope, you might want to drag in someone from the EE dept and show them the VR portion of the schematics.