[LAV1000]

I think you still have something wrong in your vr-conditioner or circuit. On my tests, I have had perfect 0-5v square wave out of the max9926. Not oddly shaped one like yours. Max should just pull down the signal wire, and the pull-up resistor raises the voltage to 5v when maxx doesn't ground the signal output. So where that more than 5v comes from is really strange.

Could be the calibration of his scope probes.

[noisymime]

As expected the voltage doesn't go above 3.2v which almost doesn't saturate the atmega pin.

Than I removed capacitor c25 which is a 0.01uF cap:

There you go problem fixed. Voltage is a bit high though (5.52v).
I don't understand why there is a capacitor at a digital input signal? The cap will result in a drift in trigger as the rpm increases because it wants to charge.

The cap is there as part of a low pass filter, which is pretty common on these types of crank/cam input setups to clean up noise on the line, but the behaviour you're seeing definitely indicates something strange going on. I would expect the square wave to easily be reaching 5v rather than what you're seeing.
The total resistance on the input to 5v should be 2k (1k on the pullup resistor plus another 1k protection resistor, R59)and combined with a 0.01uF filter, this results in a LPF that has a rise/fall time of ~46uS to 90% of 5v. What you're seeing in your image is far, far longer than this (At least 100uS based on your second image, and that's only to ~3.3v).

Are you running with the JP4 jumper on when you've tried this?

[melvinvdb]

Tomorrow I will start the car to see the results.
When using a 1nF capacitor the square waveform is ok. When using 2nF it is already pretty bad.

The resistance on the 5v input is 10.5Kohm to ground and 10k to 5v.
The VR conditioner board has a 10K pullup resistor R4.
With the conditioner board disconnected:
Speeduino PCB:
5v input to ground: 3.6Mohm
5v input to 5v: 3.6Mohm
Conditioner board:
5v output to ground: 20k
5v output to 5v: 10k (because of R4)

JP4 is NOT enabled as it is a VR sensor that doesn't require a pullup resistor. Also JP4 doesn't have any effect on the 5v input circuit as JP2 is set to VR.

Just to be clear: with the 5v input circuit I mean the circuit going to D19 on the Arduino. That is the circuit not pulling up nicely.

[melvinvdb]

Update:
No more sync issues
The car is revving perfectly fine now.
I don't know why a 10nF capacitor won't work and a 1nF capacitor does work on my board.
I've also tested with a different conditioner board but same result.

[el_fela]

With the MAX9926 based conditioners, you definitely want the trigger edge to be FALLING, not rising.

The cabling could be a concern, particularly given it's unshielded. VR sensors really should have a shielded cable between them and the conditioner as they can be very sensitive to noise.

From the manual:

"VR conditioners need to have Trigger Edge in TunerStudio set to the correct option based on the conditioner you are using. Rising for MAX or LM conditioners, and Falling for most others."

This is VERY confusing ! What is the right way? Please let us know

[pazi88]

With the MAX9926 based conditioners, you definitely want the trigger edge to be FALLING, not rising.

The cabling could be a concern, particularly given it's unshielded. VR sensors really should have a shielded cable between them and the conditioner as they can be very sensitive to noise.

From the manual:

"VR conditioners need to have Trigger Edge in TunerStudio set to the correct option based on the conditioner you are using. Rising for MAX or LM conditioners, and Falling for most others."

This is VERY confusing ! What is the right way? Please let us know

I did do small writeup about this topic: viewtopic.php?f=18&t=4440&hilit=max9926

The technically correct trigger edge with max9926 is rising as the manual suggests. But how the speeduino crank trigger filtering is done, you can get quite significant amount of timing retard when using rising edge trigger with high tooth counts (like 60-2). So to get least amount of possible timing retard, the falling is better option. And match VR-wiring to work with that.

[el_fela]

I did do small writeup about this topic: viewtopic.php?f=18&t=4440&hilit=max9926

The technically correct trigger edge with max9926 is rising as the manual suggests. But how the speeduino crank trigger filtering is done, you can get quite significant amount of timing retard when using rising edge trigger with high tooth counts (like 60-2). So to get least amount of possible timing retard, the falling is better option. And match VR-wiring to work with that.

NICE!!! I'm using a hall sensor + speeduino vr conditioner (JP4 on, hall jumper on) with a 60 - 2 trigger... I' loosing sync at 5700+ RPM. Anything I can do to make this better? Just need to reach 6500 !

[Chris Wolfson]

Check the construction that holds the sensor in place. if it vibrates, you may loose sync at some rpm. Same for the trigger.

Post some pictures of what you have a problem with. People in this forum are heroic and ingenious, but their abilities of telepathic diagnosis are still limited.

[PSIG]

Assuming you don't have other issues such as what @Chris mentioned; Hall sensors are typically routed directly (not through a signal conditioner), and as @pazi88 said using the falling edge of the signal. Alternatively, the crank/cam input filters (resistors and capacitors) and pullups may be altered for better response, or the filters even eliminated in clean systems. It all depends on what is limiting your signals. Can you 'scope the sensor and conditioner outputs to look for clues?

[el_fela]

Thanks guys! The signal looks super clean, but I haven't scoped it at 5700 RPM yet... Also the way I'm holding the sensor is kinda shaky , so I'll check that.

Assuming there is no noise, how can I alter filters to test?