Syvecs Forums

Is there a place for reporting bugs?

I found a strange issue in the firmware which I have interpreted as a bug, however I could just be misunderstanding the problem.

I was trying to get my crank and cam signals optimised and setting thresholds. Everything was working and I thought I would switch on the filter for the crank sensor to see how that would clean up the signal, I am not sure actually what to expect from this filter so I was just trying it.
I then found that I was getting what seemed to be a rev limit at around 3000 rpm, it was actually a sync error and for some reason the cam signal was now reading an anomalous value greater than 5v when I rev around 3000rpm. I chased this problem for a while before I switched off the filter on the crank sensor and the problem went away.

So it seems like a firmware bug to me, both the sensors are magnetic two wire sensors without a pull-up and so should only be generator voltage from the magnet.

That's not a bug, that's a misunderstanding!

If you just turn the input filter on it slows the response sufficiently that with more than about 6 teeth on a crank pulley you are going to screw up the signal from the sensor.

Sure, I am willing to accept that it is my misunderstanding

However, why would setting the filter on the crank sensor effect the cam sensor reading?
It also didn't seem to change how the crank sensor signal looked when viewing the sync log.

Crank and cam sensor errors can be hard to distinguish for any ECU. Typically it's expecting a tooth within a certain window and will complain if it doesn't see one, or sees one too soon, but given enough errors from very few teeth it can get very confused and so some crank sensor problems will show up as a cam error and vice versa.

If you've got problem, post up the exact details from the beginning and we can see what we can do.

You can use the syncscope to diagnose problems like this with ease, and also use the Live display to show the differences between the filter on and off.

Typically a filter increases the time to drop to low volts and rise to high volts, moreso at the peaks. If you have set your voltage threshold at a point where these delays cause the edge detection to be shifted in time, you will have issues. The filtering is absolute in time, so at 6000rpm the effects will be twice that of 3000rpm which is why your are running into problems only over a certain RPM.

Thanks, that makes sense though I'm not sure how much use the filter is if it uses a constant time.
You would think that it would have to vary with rpm but I'm sure there is a good reason for that.

The car is running fine now, I just experienced this problem when I turned on the filter which is probably because I have a 36-1 trigger wheel as you said. I must admit, I don't really understand why it would effect the cam signal in the sync log. I assume the sync log is an interpreted output and does not display a raw scope output.

The sync log is raw.

The filter is an option and is not a crutch for a bad sensor or wiring setup. It is based on a constant time as it works nicely for getting rid of garbage that doesn't vary with RPM. If you need to get rid of very brief spikes use the digital filtering but keep in mind that a 36-1 wheel at 8000rpm is going to generate a half waveform in 100 microseconds or so, so you are only going to want to filter down in the 15-20 microsecond range. This filter is used to just ignore things rather than alter the input.

The individual sensor filter is (I think) based on a time constant to give classic 63% of full response at 1khz, which would mean with 36 teeth you're going to start getting some serious interference at just a few thousand RPM.

It is however useful for cam sensors where you are generally going to have just a single tooth (on a 36-1 setup) and it's at half engine speed.

I can't comment on what you're seeing exactly without looking at the actual .cal file or seeing some sync logs.

Ok, like I said the car is running well. I was just moving the magnet closer the the trigger to try a get the signal stronger and was slightly concerned that my pulses did not have flat tops.

I will switch the filter back on and try and reproduce the problem, then I'll post the logs up. So just to recap, I enabled the filter on the crank sensor and the sync log would show the cam sensor signal saturate the display at around 3000 rpm.

I'll find some time tomorrow to reproduce.

Why are you expecting flat tops with a reluctor sensor?

What problem are you trying to solve with the filter?

What do you mean by "saturate the display"? Do you mean lots of edges? Do you mean it reached max volts for the sync log?

You should be detecting your edges around mid way between max and min where the slope is steepest and most repeatable.

It would indeed be a bug if enabling the filter on the crankshaft sensor in fact filtered the camshaft signal. It would be very helpful to post a .cal file at this point so it can be sanity checked.