OT: Car speedometer accuracy

[Huge]

Yep. It was a mistake. I apologise. I misread your posting.

[Harry Bloomfield]

Harry Bloomfield explained on 31/10/2016 :

I should add, that it has always been my desire, to have that analogue speedo read the correct speed. It is just a matter of changing a flag in the dash firmware, problem is finding that flag.

[The Natural Philosopher]

However the final output is probably a moving-coil meter.

Which is unlikely to be better than 5% accurate.

[The Natural Philosopher]

well whether its designed that way or not, in every case where I have had some kind of reputable third party mneans of recording speed, it has been the case.

[Allan]

Get a couple of different devices (e.g. car sat nav, phone with GPS & speed app, handheld walking GPS e.g. Garmin if you have/can borrow one) and run some comparative tests. If the devices all read the same, and the car speedo is different, that would appear to be fairly conclusive.

[Terry Casey]

That should give an interesting result as the measurements on motorways are in kilometres!

[damduck-egg]

On some roads the Rozzers have had markers erected at 1/4 mile intervals over a measured mile for training/ calibration purposes. Here is the 1/2 mile one on the A33 not far from Reading.

G.Harman

[Tim+]

I'm not sure about this either. If this was the case there would be more variation between my recent cars which were/are all exactly 5% optimistic.

Tim

[Dave Plowman (News)]

In these electronic days it would be trivial to make a speedo accurate.

The +10/-0% at 30 mph dates back from crude eddy current types where it would increase the cost to make one accurate.

It just suits some car makers to specify their speedos over read to the maximum allowed by law. If it were simply tolerance, then the error would vary in different examples of the same car (or make). But it tends to be extremely consistent.

Nor can you guess by price whether a maker will fit an accurate or optimistic speedo.

[Dave Plowman (News)]

Oddly, BMW cars were one of the worst offenders. Jaguar tended to specify accurate units.

Not sure of today, though. Autocar used to publish details of true speed against indicated in all their full road tests. But I gave up buying Autocar years ago.

[The Natural Philosopher]

If you look at the statistical variation in components, you will find they follow a pretty accurate bell curve, or at least that was my findings on testing 1000 transistors exposed to the same light source.

Ok there were two that were simply 'out of spec', and the rest showed a perfect bell curve except it was truncated a little at the end where presumably substandard transistors had been rejected.

My point? Simply this. The manufacturer of the speedo units will be asked to supply "+- 5%". IN practice the majority will be within 1-2%, and he will reject any that are outside tolerance..

Given the state of the law, that is why in general all speedos read exactly 5% low, because the stuff is capable of better, but its not

*specified* to be better, than 5%.

And if you asked people to calculate what the speed limit for articulated lorries was, using their speedos, you would find most of them saying that they travel at 60mph. In fact the limit is 56mph =

90km/h. A 7% error...w.r.t. to 60mph

And those limiters are pretty darn accurate.

[Dave Plowman (News)]

That would be fine - except not all car makers do this. And those who do tend to be very consistent at the maximum error allowed, at 10%. So if they can make every speedo over read by 10%, it would be equally as easy to make them over-read by 1%.

[Harry Bloomfield]

A common misconception... Yes they do over read more as they wear, but not so much as you would expect.

[TimW]

One would guess if the radius of the wheel is around 250mm and the total wear over the lifetime of the tyre is around 2.5mm, so about 1% and that would translate directly to 1% error in the speedo.

I presume tyre pressure would make a difference too, maybe more.

TW

[The Natural Philosopher]

That is an odd one. Low tyre pressures do NOT change the circumference of the wheel with steel cored radials.

Just the ride height.

They MAY change the slip, if any, under acceleration and braking, though.

[Harry Bloomfield]

Brian Gaff used his keyboard to write :

Actually, no - hence the wide allowance for their calibration in the CU regulations. It is a hangover from when all speedos were mechanical.

Modern speedos use the pulse from the ABS system. Mine is around 50 pulses per wheel rotation, that is integrated in software, to then drive a stepper motor analogue needle. The software deliberately makes the speedo read that few percent fast. The same ABS signal is used to provide a digital speed display, but that is always spot on.

More crude, certainly not. Assuming all the satellites are well positioned and reception is reasonable - it is very accurate, but only on the straight and level, at a steady speed. Perfect for checking the calibration of your dash speedo, but its update will always lag your true speed except as above.

[NY]

How much does the effective radius of a tyre vary as the pressure varies? How much does the pressure have to decrease from the correct value for the radius (and therefore measured speed) to vary by 1%?

[Harry Bloomfield]

The Natural Philosopher expressed precisely :

and 1% of 70mph is?

Absolutely tiny! Not noticeable on a normal speedo, not really noticeable on a satnav.

[The Natural Philosopher]

Radius is not a valid concept in a wheel that is not round.

What counts is the circumference, and that is stable as it is controlled by steel cords in the tread.

[The Natural Philosopher]

I wasn't trying to deny that, My point was you don't have to *think* when you can *calculate*.

A point lost on art students who can't do either....