Speedo/tacho error & calibration

[cros]

I think davetouring asked about this. Here's the most OTT answer you could ever hope for

k=7784 (miles) or k=4838 (km) on the face is the number of diff sensor pulses required to clock up 1 unit of distance. For a given speed, you need a pulses at (speed*k/3600) Hz. I used a frequency generator and got the % overread at a few speeds (x=mph, y=% overread):



The blue line is before I went fiddling, the purple is after "calibration". The error was flat past 120mph. With the original setup, you are going 53mph when the meter reads 60mph.



The circled trim pot is used to calibrate the speedo.

This might be a particularly overreading meter, but it's clear the things are cautious. With soft/worn/lower-profile tyres it'll overread even more. I saw something in the TIS before about a spec for this before.

A sanity check at the end was that running at 120 for 1 minute made the trip counter move on by exactly 2.

Ever notice the OBC limit chiming when the speedo has gone past the limit by 10%? This is why. The OBC is a quartz timed interpretation of the pulses and should be bang on. The speedo isn't.

I can't figure out how many pulses the diff gives per rev of the wheel (the figures say 9.25?). If it's 10 then every odometer is overreading by 8%. If it's 9 then you can subtract 2.7% off the error curves above and every odometer underreads by 2.7%.

The odometer can't be calibrated electrically. A stepping motor is run off the pulses and gearing sets the k value.

[Don't wreck your speedo and expect me to fix it or blame me if you get pulled for speeding vs. the clock readout...]

[cros]

Noticed that my "new" metric tacho was underreading by 200rpm, so had to open the cluster again.

Decided to play it safe and redo the speedo cal, in case the diff does give 9 pulses per rev.

Top is original, bottom is yesterdays, middle is the way I'm leaving it. It's in Km's for extra confusion.



Found out that European type approval rules state no underreading and a maximum of 10% + 4km/h overreading.

If you twist the rev needle enough it slips on its shaft and you can set it up how you like. An ECU controlled warm idle with a good ICV is usually accurate to +/- 5% for this. There is no obvious electric adjustment.

The speedo needle can twist too. Between this and the electronic adjustment, you can change the slope and height of the error curve.

Hope all this is of use to somebody...

[Jordan]

Quite useful to those of us that know electronics. Good info!

[cros]

Jordan - I saw you put E36 needles on yours. Was hard to get the old ones off? I pulled the tacho one so hard I though I'd snap the shaft, but it wouldn't move.

I forgot to say too, looking at the face of the speedo with the input socket/pins on the right of the PCB: the top is GND, next is diff input, next is service computer / cruise control output, and on the bottom is 12v input.

[Jordan]

Getting the e30 needles off the instruments wasnt too bad, but i find this varies greatly. Sometimes they are stuck on there, sometimes not. I actually destroyed a tach once because the needle would not come off the shaft. The shaft actually tore out of the mechanism

The e36 ones were MURDER to get off.... broke the bases on 2 of them, but it didnt much matter since you have to glue the e30 needle bases to them to make them work on the e30 instrumentation.

[StuBeeDoo]

Cros.....
Can you explain in simple terms how this mod is carried out please?
I've noticed that my speedo is wildly inaccurate. When I'm doing 30, it reads nearly 40.
Cheers
Stuart.

[cros]

What? Do you not have an AF/IF waveform generator lying around?

Seriously though, you could do it by using a handheld GPS or the rev counter to guage speedo accuracy. Lucky for you I'm in an essay writing mood

The results won't be as good and the setup will be way more tedious. If using the rev counter, first check its accuracy using a motor tach multimeter on the LV side of the coil, or mabye even just by checking the idle speed accuracy.

Use http://www.kabamus.com/garage/gears.html to find the road speed for chosen revs and gear. I'd recommend something like 2000rpm in 4th to get the speedo to a working speed and have the rev counter close enough to the assumed accurate idle. Note what your speedo currently reads at this. Mabye check a range of speeds to make sure you want to go further.

Given that your tyres won't be at 100psi and will be missing some tread, you should aim to calibrate at least a 5% overread to cover yourself.

Take apart the instrument cluster, pop out the speedo and note the position of the pot so you can undo everything if you screw up. Use a multimeter to measure the resistance, or just mark the position with tippex or whatever.

You probably don't need to turn the pot more than 1/4 of a rev and I can't tell you which way you need to turn it, so just turn it in any direction and put the cluster back together (don't bother with screws on the speedo, and just use 3 on the cluster). Plug the cluster back in and just balance it back into the dash. Go for a drive and see what your revs & gear gives as a speed readout.

Now you should be able to tell if you turned it in the right direction and roughly how far it needs to be turned. Wash, rinse & repeat until happy.

Check speed vs. gears & revs at a few different points to make sure your error is constant across the range. If not you may want to twist the needle on the shaft and start again.

When I looked at revs vs speed after recalibration, I got less of an error than the 5 - 10% I'd set up. This suggests 9 pulses/rev off the diff.

[As before, don't blame me if you break something or get pulled for speeding vs. the clock readout...]

[Demlotcrew]

Very good! And very interesting.

Andrew

[Brianmoooore]

By a remarkable coincidence, the last thing I did this evening in the workshop, was to count the number of vanes on the speedo rotors of an E30 and an E29 diff.
The rotors have nine vanes.

[cros]

Interesting!

I suppose the 2-3% 'error' from 9 vs 9.25 thing must be to take into account tyre deflection and wear giving a reduced rolling diameter. The odometer is probably fairly accurate so.

Noticed that the VDO chrome bumpered speedos use k=78xx for miles. Meaning they give fractionally more margin for the above than the Motometer ones.

[russ325i]

apologies as i am completely uninitiated but i have heard a rumour somewhere about being able to get the speed readout to display on the OBC? is this true? as surely you could use this to calibrate the speedo with less complication?

(awaits the shame of being completely mislead...)

~russ

[cros]

Don't think the E30 OBC will do it as such. You can set a limit and listen for the gong, but it's not a readout. True, it should be a good indication of speedo error.

The 8 series OBC test mode gives a live readout via the ABS sensors

About the OBC error: the diff gives 9 pulses per wheel rev. The speedo (& probably OBC) expect about 9.25 pulses per rev. That's a bit of builtin 'underread' that might null error due to tread depth and sidewall deflection. Mabye.

[Brianmoooore]

Was playing around with this on my fairly recently fitted OBC on the way back from Silverstone (Boo- hoo, Andy), last night.
Set the OBC limit to 80, and noticed it kicked in and out with the speedo at about 87.
The OBC was probably designed by BMW to be accurate, but this would depend on how closely the rolling circumference of my tyres match that expected by the designers.