The evolution of "Tach Noise Filtering" function

[tmbryhn]

Indeed. I often expand the clearance or add a load resistor across.
Altough I have experienced consistent trigger miss at a given RPM using a resistor of too low value, say 1k. The problem occured at a higher RPM as i tried resistors with increased values. Used a larger resistor (2k2) and the problem happened at a higher rpm. Just installed a 10k and it let me rev to the limiter.
I have not figuered that one out yet. Guess I got lazy since it worked, lol!

[muythaibxr]

On the nippondenso 24 tooth systems it even filters crosstalk on 2nd trig from the 24 tooth wherl... that is until you drop below 1400ish rpm, then 85ms isn't long enough. So I wrote a special crosstalk filter that cures that.

[tmbryhn]

Nice! Yes, because of the adaptive threshold that is.
I wish that time constant could be altered like some of the other features by adding external components around the MAX and running it in another mode. Of course, you can alter the threshold level by feeding it with an analog signal on one of the pins, thus raising the level when needed. Would require an additional micro or some analog circuitry sensing the input amplitude and then stimulating the MAX.

I get a period of 85ms @ 706 RPM for a single tooth/rev. pattern.
706/60 = 11.76 revs/sec
1/11.76 gives a period of 0.085 sec.

Is there something about that tooth pattern that I'm missing? I thought the Nippon' 24 was just a 24-tooth wheel + a single tooth for sync per rev. on an individual sensor?

I had a similar problem when I set up an ECU for a friend running 4-1 trigger wheel. The engine cranked just to slow for the period to time out and the adaptive threshold would reset to the low threshold level of bout 20mVpp. So I made him upgrade to a 36-1
Did similar math back then, and that quickly revealed the problem.

[muythaibxr]

The wheel turns at cam speed, so you have to double your rpm.

[tmbryhn]

Ahh, I forgot the Nippon 24 is distributor/cam based.
Then it all makes sense. Your calc (1400) matches with mine (706) then ;-)

I have not gotten out my scope to measure on in the car, but do you have an idea of why I got consistent sync losses at a given RPM when shunting the VR+/- together with a relatively low value resistor (1k), and that the RPM point for when the sync error occurs is raised with increased resistance? Sync is perfect up until the given RPM point. It's like hitting the limiter.
I have read somewhere in the MS3-pro manual that you can add a 10k ristor across the two VR-terminals if you have sync problems. I reckon this is for loading the sensor so that there is a smaller amount of energy needed to get clamped by the internal diodes of the MAX circuit?

[muythaibxr]

I don't know for sure, but I think the analog signal gets clipped with the 36-1 and 60-2 wheels (voltage too high) and that causes issues for the conditioner that adding a resistor fixes.

Someone more experienced on the details of the electrical side of things can comment.

Ken

[Peter Florance]

I don't know for sure, but I think the analog signal gets clipped with the 36-1 and 60-2 wheels (voltage too high) and that causes issues for the conditioner that adding a resistor fixes.

Someone more experienced on the details of the electrical side of things can comment.

Ken

That is a lot of it. The clipping screws up threshold vs. amplitude ratio.

[Matt Cramer]

I don't know for sure, but I think the analog signal gets clipped with the 36-1 and 60-2 wheels (voltage too high) and that causes issues for the conditioner that adding a resistor fixes.

Someone more experienced on the details of the electrical side of things can comment.

Ken

Right. The clipping messes up the adaptive function on MAX9926s because it compares the size of the pulse in the missing tooth to the clipped input and not the actual voltage. Putting a resistor in parallel with the sensor pulls down its voltage to the point that the adaptive filtering can cover things correctly.

[jsmcortina]

New firmware released, see : http://www.msextra.com/forums/viewtopic ... 91&t=61354

James