places to electronically monitor gear position

[mike_sekinger]

...never got to look at a car with an electronic

Edited November 8, 2005 by mike_sekinger

[Guest teigan]

okay, it's worth a try. i can convert the tach gauge resistance value to a digital input then compare it to a look up table in memory. can someone present the ideal figures to me for compiling this table?

[madmezza]

okay, it's worth a try.  i can convert the tach gauge resistance value to a digital input then compare it to a look up table in memory.  can someone present the ideal figures to me for compiling this table?

<{POST_SNAPBACK}>

Do it dynamically. You know the car's got 5 gears - start with a table initialised to something default, and drive for a bit in each gear. Have an algorithm that looks for 5 distinct steps and updates the table. In that way, if you ever change the gearbox/tyre size etc, it will automatically reset. Be sensible about how often it updates - otherwise it won't be stable.

That's how I'd do it anyway! In fact, when I finally get around to putting a uP in the car, that's how I will do it

[mchlrodrigues]

mine's a 1999 V8. So speed & rpm are both available on the ODBII interface. Thats what I used to log the data.

Will

<{POST_SNAPBACK}>

Additional Info:

On MoTeC stand alone ECU's gear position can be determined in one of two ways.

1. Comparing Rpm and wheel speed signals. Program in ratios for each gear and the number of pulses per revolution of the of the wheel. Final drive and tire dia inputs are requested for road speed.

2. Potentiometer input 0-5v, typically used on sequential gearboxes. A specific volage is specified for each gear with a tolerance to allow for voltage fluctuation due to vibration.

MoTeC engineers have told me that method 2 is more reliable. At low speeds, there are few wheel pulses to compare which apparently can result in the occasional error. Wheelspin is another problem unless one is monitoring wheelspeed on all of the wheels which requires a multiplexer (at least with a MoTeC ECU).

Microswitches, if used with an ignition cut are said to be the worst. A microswitch does not make immediate, single contact. There is a signal "bounce" as the contacts arc and close. The time interval is ~20ms. Nothing to us humans, but apparently a long time in computer terms. As a result of the intermittent contact over this period, the ECU becomes confused and multiple ignition cuts can occur.

[catty s2]

I once saw a photo of an Escort with a lcd display inlayed into the gear knob showing gear positions. How it worked i dont know. Why? Cos they run out of other things to do with the car probably. Gearstick position and engine revs should tell you what gear your in!

[Guest teigan]

i'd rather have a complete table ready to go. i currently only have dynamic ram on my computer and it needs to be battery backed if i want an updatable table. i want no live circuits when my car is off. i've already disabled the original clock, and the domelights only turn on when dark, because of a reed switch interrupt, when 3V are present on the solar cell.

i disagree that microswitches would be the worst accuracy. debouncing a switch is child's play. mounting the switch where it won't get dirty is the issue. i'm still not convinced speedo and tach data is going to be at all accurate. it's worth a try though, because i plan to have performance data downloadable via the serial interface in future, and i'd need to make the speedo and tach hookups eventually.

[scooter]

i'm still not convinced speedo and tach data is going to be at all accurate.

<{POST_SNAPBACK}>

Will your system count input pulses over a specific time to establish driveshaft speed and engine speed, or measure the time between individual pulses?

If you're using the first one, the key to maximising the accuracy is to have as many pulses as possible per sample so that at low speeds there's still a reasonable number of pulses to sample in your time period.

[Guest teigan]

the original method suggested to me was speedo vs. tach. the driveshaft would be better, but there is no ready sensor in place. i could use an optical shaft encoder, but those can only reliably handle 1700 pulses per second or 1000 rpm.

[molemot]

You could try using reed switches and magnets mounted on the gearbox; I would think this would be better than trying to mount stuff around the gear shift lever. I think simple solutions are best, so I would try to avoid software based interface cleverness if possible!

[USAndretti42]

Teigan, I sent you a PM with some data showing the wheel/engine speed ratio calculation. Let me know if you do not get it today.

[Guest teigan]

Teigan, I sent you a PM with some data showing the wheel/engine speed ratio calculation. 

<{POST_SNAPBACK}>

thanks. that's quite comprehensive. i'm sure others will be curious to see how fast their wheels are spinning.