When is the speed limit, not the speed limit…?

Jemma Hawtry 

Speedometer at 70mph. But is it?
Speedometer at 70mph. But is it?

I sometimes like to just go and drive for no particular reason – just for something to do, when daytime TV gets too much to bear (Re-runs of The Professionals, anyone? There’s a limit to the number of times I can stand a Capri haring around sounding like its got a bad case of emphysema.) 

Any AROnline readers who have read my previous articles will know that I am one of the ‘chuck it out’ brigade. I prefer the technology in a car to have a reason for its existence, not to be something that is put in just because it’ll beat the competition. 

We get cars with radar-controlled cruise control for heaven’s sake and soon cars will be able to drive themselves at the touch of a button (and having seen the driving of some people in the fog recently, that’d be a very good idea). However, to be fair a lot of this stuff is window dressing so we can be convinced to pay over the odds (£1000 for an air-conditioning system that comes off the boat at £250 for example…). 

Anyway, given all that, you would think that the most basic parts of the car concerned are accurate to the Nth degree wouldn’t you? Well, you’d be wrong. 

I have many interests in my life – one of them happens to be computers. I’m one of those people who bought the original smartphones (you know, the ones the size of a housebrick, with the battery life of an anemic amoeba) and I love to tinker. I recently found a programme on my Android phone (mention ye not the iGullible) that recreates an aircraft head up display, complete with speed, inclination angles and a settable artificial horizon – so I thought I would try it out. It uses the ‘phone’s GPS receiver to produce course and speed data and the accelerometer in the ‘phone to produce the rest. Best of all, it’s free! There are even programmes that, through the use of a Bluetooth dongle, will allow you to realtime interface with your OBD-II engine – and, in the paid version, will provide you with fault codes. 

The car I have is fitted with cruise control, which is very useful in an area where the Traffic Police are rabid, and I usually set the cruise when in a 30mph limit to exactly that… except for the fact that the most basic part of the car is actually wrong. 30 mph on the speedometer of the car is 28mph on the GPS – 32mph on the speedometer reads at 30mph on the GPS. This piqued my curiosity…  so I took the car for a brief blast up the local piece of dual-carriageway (the A12) and, lo and behold, at a stated 70mph, I was actually doing 66mph. Furthermore, as my speed increased, the difference between actual and stated speed increased – to get a GPS speed of 70mph I had to be doing 78mph on the car speedometer! 

Car manufacturers have managed to computer control engines, brakes and even suspension for years. They have installed Voice Warning systems, automatic lighting (since the late 1950s no less) and all sorts of other gubbins – ooh, look a button for the heated steering wheel – but they can’t even provide us with an accurate speedometer! All it would take is a board with a single chip GPS receiver on it linked to the dashboard readout – it’s not beyond the wit of a BMW engineer… 

Now this might seem utterly unimportant at first glance but, if you look deeper, it’s highly significant and has an effect on a lot of different facets of driving. 

Buyers are given a slew of performance data, on which to base their assessment of a vehicle and influence their choices.  That’s fine, if any of it was correct. 0 – 60mph is all well and good, but it’s no use whatsoever if it’s really 0-54mph now is it? Then you have the fuel economy data – which is generally measured using presets – the best economy being at around 54-56mph, for which you are given figures, and again for simulated motorway work or urban driving…  but, if the figures at 55mph are really measuring 50mph then we have another problem, they are downright wrong which, under advertising law as I understand it, is termed as misleading the customer… 

However, that only deals with the problems when the speedometer reads (s)lower than the GPS ground speed. What could happen when the speedometer speed is higher than the GPS ground speed? What happens when you are in a potentially fatal accident and the Police’s calculations say you were breaking the speed limit – you don’t have a leg to stand on, even though you’re speedometer read 40mph the accident calculations say you were doing 45mph – that discrepancy could be the difference between 3 points and a prison sentence. All because an important system hasn’t been, and in a lot of cases can’t be, configured properly. 

I think the car industry has spent too much time prettying up our new cars with the intent of getting more money out of us and, in the meantime, has spent far too little time applying the advances of technology to the basic systems of the vehicle itself… 

Maybe, instead of trying to charge us £350 for a piece of two-core speaker wire and a USB port, they should try re-assessing the basics of the machines they are producing, so core functions are fit for purpose…

Keith Adams


  1. I think the questions you raise here are quite easy to answer.

    First of all a car’s speedo should never under-read – there is a matching law and I think you’ll not find a car that does unless something is broken or wrong parts are fitted. The law gives a range within this – the car must over-read the speed, so that the driver is always on the safe side. There is a max. value – I cannot remember the exact value now – but how much the car actually over-reads depends on a raft of variables.

    I’ll ignore GPS for a while.

    The speedo is nothing more than a rev counter for the gearbox output – in other words how often the driven wheels turn in average – so, if you know the exact speed a wheel does when doing a full turn, you’ll be able to build a very, very precise speedometer. This is used in all sorts of sports, like cycling or classic car rallying (tripmaster).

    The problem is that the effective radius of a wheel is not constant – a new tyre might have 1-2% greater diameter than a worn one. Higher air pressure might add another 1% over a slightly under-inflated tyre. Then there is the choice of wheels – with today’s cars you can choose between many optional wheels of different sizes, each with a slightly different diameter – up to 5% difference is a typical value between the largest and smallest variant. Add up all these factors and you’ll find that your speedo can vary between nearly exact and almost 10% out – which is as good as it is. The alternative? Well, would you trust the average driver to recalibrate the speedo after a change of wheels?

    I found from personal experience, as well as from old tests, that BMC’s late 1960s cars had more accurate speedos than most modern cars. Why? There was no choice of different tyre and wheel sizes to accommodate. All the cars in our classic fleet are less than 1kph from GPS at constant low speeds (50kph) and less than 3kph off at constant, higher motorway speeds.

    When you read manufacturers’ data or road test data, you can be sure that the information has not been acquired using the car’s own speedometer, but through way more accurate testing equipment and so, given that the speedo over-reads, the car’s owner would get slightly favourable values when trying to replicate the measures.

    Now for GPS… Sadly – and I’m using GPS a lot – this is not accurate enough in many circumstances to really replace a good speedometer. There are too many places where GPS does not deliver speed readings even close to the speedometer – mountains, built-up areas with tall buildings, a rainy forest.

    However, even under good conditions, as GPS does not measure speed itself, but the position, a guaranteed error in the speed can only be given when doing a constant speed. While accelerating a GPS will always under-read, while braking it will always over-read. The typical units in ‘phones or sat-navs do so by quite some margin because they update their readings only once per second. There are units out there that do this more often, so the error can be less, but the principle is the same.

  2. I must admit that my barge has a bang-on speedometer. I know this because I was using the same app for about *cough* 4 months last year when a wheel sensor failed killing the speedo, ABS, TC, fuel gauge etc. (damm multiplexing!) The limitations of a GPS speedo are noted even in short tunnels or by tall buildings – you lose the reading and that’s not good.

  3. I think the best system would be to have a GPS-calibrated speedo – each time you engage cruise control and the car maintains a relatively constant speed for a period of time, it compares that to the speedo reading.

    It doesn’t take me long to figure out the % error based on my GPS when I change my car then I know whether to drive at 55mph or 51mph through the M1 roadworks. You can always tell when the person in front of you thinks they’re driving at 49mph through average speed cameras but, in reality, it’s nearer 44mph.

    There’s also the optical error from the gap between needle and dial which depends on your seating position. There’s about 1/2mph difference in my car between the optional digital speedo readout on the driver-info screen and the usual dial.

  4. The Construction and Use Regulations do, as mentioned above, require the speedometer never to under-read. It can over-read by a maximum of 10% and the manufacturer usually programmes in a deliberate over-read of a few percent to allow for component wear.

    Most modern cars have mulitplex wiring and the speed is provided by the ABS system which feeds it onto the other systems that require it, engine management, traction control, ESP, speed sensitive stereos, auto door locks etc., etc.


  5. Umm, where to start?

    Performance figures are taken using ‘professional’ equipment and not the speedo – so a 0-60mph time will categorically NOT be for a 0-54mph dash. Other readers have already pointed out that a car’s speed reading device can over-read by up to 10% but never under-read – a legal requirement in (at least) the UK – also the effects of tyre wear on speedometer readings. I also agree that GPS speed readings in a car can be enlightening.

    However, a car isn’t most economical at 54-56mph. The use of 56mph in fuel economy tests is only because the original 1970s EEC Directive referred to fuel efficiency at a constant 90kph (as well as at 120kph – 75mph, illegal in the UK – and a specific ‘urban’ cycle). All Government fuel efficiency tests are carried out in a laboratory as per the EEC Directive which passed into UK Law as something like “The Passenger Car (Fuel Consumption) Order 1977” off the top of my head. There’s nothing scientific or sexy about the use of 56mph or any of the other fuel efficiency tests!

    Generally, the slower a car travels, the greater its fuel efficiency – anyone remember the “83mpg Metro” (at 30mph) advert?

  6. I thought everyone knew that Rover speedometers are set at 10% below actual speed. That’s why all the chaps wearing their flat caps get overtaken all the time.

  7. This varies between different cars and ages. My old Triumph 2500 saloon had a speedo which always read exactly right – I checked that with a GPS. This caused trouble because I was used to an over-read and went too fast.

    The over-read was terrible on a Nissan Almera hire car (the worst I’ve driven) where 70mph on speedo was actually 60mph. However, an E46 BMW 3 Series had only about a 2mph over-read at that speed.

  8. A vehicle speedo must always under-read by law as laid down by European Legislation. Anyone wanting to look into this in more detail needs to refer to
    75/443/EEC as last ammended by 97/39/EC and Regulation 39 for Speedometer and Reverse Gear.

    These provisions clearly state how your vehicle’s speedo should operate. The way that speedos are tested for type approval is as follows:

    1) The vehicle must be driven at 3 speeds, 40km/h, 80km/h and then 120km/h in a north and south direction on a selection of wheels and tyre sizes after they have been agreed by both manufacturer and, in my case for example, the VCA (this can also be done in mph).

    2) The true speed is measured by calibrated equipment and the driver must put the needle/display directly on the speed being recorded.

    3) You then take the true speed of the north and south runs and average them out. Then you subtract the vehicle speedo indicated test speed (V1) from the average speed (V2) to give you a number. V2 is then divided by 10 and you then add 4km/h to that number. This number gives the tolerance that the speedo is allowed to work in.

    Thus, for example, if the indicated speed was 120km/h and the two runs gave you and average true speed of 114.3km/h (V2) this would then subtracted from the original 120km/h = 5.7km/h

    V2 is then this is divided by 10 and 4km/h added = 15.4km/h. The speedo was therefore under-reading by 5.7km/h and so it is within the specified limit.

    There are all sorts of myths about who sets them as what and Rover did this, BMW do that, Nissan do this and Mercedes does this. The bottom line is that they can’t give you a true speed reading because each manufacturer calibrates their speedo to suit their car and also a range of tyre sizes and it also takes into consideration tyre wear.

    It’s not a deliberate ploy to screw figures etc., it’s what each and every car manufacturer has to do in order to gain Type Approval for their vehicles. I know, I do it for a living!

    Here’s what the legislation states with regards testing of the vehicle etc.:

    4.3. The accuracy of the speedometer equipment shall be tested in accordance with the following procedure:

    4.3.1. the vehicle is equipped with one of the types of tyre normally fitted; the test shall be repeated for each of the types of speedometer specified by the manufacturer;

    4.3.2. the load on the axle driving the speedometer equipment must correspond to the weight complying with 2.6 of Annex I to Directive No 70/156/EEC;

    4.3.3. the reference temperature at the speedometer shall be 23 ± 5 °C;

    4.3.4. during each test the pressure of the tyres shall be the normal running pressure as defined in 2.3; (+ 0.2 bar)

    4.3.5. the vehicle is tested at the following three speeds: 40, 80 and 120 km/h, or 80 % of the maximum speed specified by the manufacturer, if this is inferior to 150 km/h;

    4.3.6. the test instrumentation used for measuring the true vehicle speed shall be accurate to ± 1.0 %; the surface of a test track when used be flat and dry, and shall provide sufficient adhesion.

    4.4. The speed indicated must never be less than the true speed. At the speeds specified for the test in 4.3.5 above and between these speeds, there shall be the following relationship between the speed indicated on the dial of the speedometer (V1) and the true speed (V2):

    0≤ V1 – V2 ≤ V2/10 + 4kmh

  9. There seems to be a misunderstanding over the use the terms under-reading and over-reading in both the article and comments.

    I thought “over-read” was the correct term for an instrument showing a higher value than the true value. However, while describing the same – using more technical terms and the correct definitions – Nic Fasci used the term “under-read”.

    I’m therefore now confused not about how speedos need to be set but just about the use of “under” or “over”.

  10. @Alexander Boucke
    When I use the term “under-read” I am viewing it from a true speed point of view. Hence, at a true speed of 50km/h, for example, the speedo reading will be under 50km/h.

    However, if you view it from the other perspective, at a speedo-indicated 50km/h, the true speed will be under so you could say that the speedo is over-reading from the actual true speed of, say, around 47 or 48 km/h!

    Confusing isn’t it? Welcome to the wonderful world of legislation and interpretation.

  11. Nic Fasci :

    Confusing isn’t it? Welcome to the wonderful world of legislation and interpretation.

    Sure… Try finding out whether you are obliged to wear a seatbelt fitted to a classic car which is old enough not to be subject to the requirement to have them fitted here in Germany… :).

  12. Alexander Boucke :
    Sure… Try finding out whether you are obliged to wear a seatbelt fitted to a classic car which is old enough not to be subject to the requirement to have them fitted here in Germany… .

    Is that not one of those rules where, “if it is fitted you need to use it, if not then all is fine”?

    I guess it’s a bit like spotlights – if not wired in, then they are not tested for MoT purposes.

  13. @Paul T
    It’s not that simple: the law states that ‘compulsory safety belts need to be used’. Hence, given the wording of that provision, it is a valid question if the belts fitted are optional (front seats on cars pre-1st April, 1970 here in Germany)…

    Incidentally, with regards to the spotlights, that would be a fail here in Germany.

    I accept that EC regulations for new car homologation are a wee bit more complicated than these simple questions.

  14. @Various
    All well and good – but the speedometer on the Safrane is electronic, not a geared drive from the transmission, which means it shouldn’t be an issue and yet it still reads low. The tyres are manufacturer size at the correct pressures, yet it still reads low.

    Modern SOC GPS systems use a minimum draw in their resting state, so why aren’t they fitted as standard?

    The point about performance figures still stands because it’s all very well saying that you fit all sorts of electronics to give exact figures but, if the speedometer on the car is reading wrong, then you won’t get the figures you should at the speed you should get them…

  15. Jemma :@Various All well and good – but the speedometer on the Safrane is electronic, not a geared drive from the transmission, which means it shouldn’t be an issue and yet it still reads low. The tyres are manufacturer size at the correct pressures, yet it still reads low.

    Modern SOC GPS systems use a minimum draw in their resting state, so why aren’t they fitted as standard?

    The point about performance figures still stands because it’s all very well saying that you fit all sorts of electronics to give exact figures but, if the speedometer on the car is reading wrong, then you won’t get the figures you should at the speed you should get them…

    They’re all electronic nowadays and they take their “drive” from either the gearbox or the ABS system. The true speed is there in the car if you were to connect to the OBD system but, for the reasons that I’ve stated above, the signal from the speedo “drive” is then filtered and calibrated to meet the legislative requirements laid down by the European Comission.

    I know what you’re saying and I do appreciate what you’re saying. The tyres that you have fitted with the correct pressures will be one of a series of tyre parameters in there. However, as your tyres wear, they lose about 10mm or so (if not more depending on your tyres) in diameter as you go towards the legal limit of your tyre wear and this means that the tyres spin faster. Thus, if you were already close to the true speed of the car, you’d end up travelling faster than you think you are as the signals are processed and you could end up with speeding tickets etc.

    Would you therefore rather have a speedo that looks after you or one that gets you a ticket?

    Incidentally, if you were to use a GPS system for the speedo, it would have to be calibrated to comply with the legislation as well.

  16. Err, Nic, a worn tyre makes the car slower, not faster.

    Jemma, I can’t see your problem about performance and consumption figures. Going by the car’s speedo instead of the true speed, the owners will get the warm fuzzy feeling that their car is faster from 0 to 60mph than the manufacturer states and, on top of that, also uses less petrol – if they happen to run exactly as per the test conditions.

  17. The system used today exists for very good reasons, as has been stated in other readers’ posts…

    I would rather have a speedo that over-read by a small but consistent safety margin than a GPS system that exposed me to the huge limitations of the GPS system.

    Official economy and performance tests don’t use the car’s speedo! LOL!

  18. Alexander Boucke :Err, Nic, a worn tyre makes the car slower, not faster.

    Quite possibly so, but the wheel spins faster thus giving the impression to the various sytems that the speed is increasing. Some tyre pressure monitoring systems use the ABS sensors to determine if the tyre has lost pressure – MINI for example (as I run a Cooper S JCW with a SC on board!) – thus, if my tyre goes down, it spins faster therefore it thinks that it’s flat. The same principle applies when converting signals from wheel speed to indicated speedo reading.

    When we do noise testing, we have to choose tyres carefully so that they give us an “average acceleration” across the range of tyres available – the same applies when testing speedos. A 14″ and 15″ could be covered by one calibration, a 16″ and 17″ need another and so on.

    Incidentally, on the Rover 75, we managed to cover 15″, 16″ and 17″ wheels on one calbration setting to make the reading legal. The 18″ wheels on the MG ZTs needed another calibration or pulses per mile to make it legal which is why fitting 18″ wheels to a Rover 75 makes the speedo inaccurate unless you start mucking about with profile heights to match a standard ZT wheel and tyre assembly for the rolling circumference of the tyre.

  19. I think that there’s another reason you can’t use GPS – I am pretty sure it does not compensate for the gradient you are travelling on (at least, not as yet).

    Simple trig will tell you that the distance travelled which the GPS will register will be shorter than the distance you have actually travelled when going uphill or downhill.

  20. @Ianto
    Buy a decent Android phone and put DoubleTwist on it – you’ll have a world’s better OS that isn’t ring-fenced by that all-knowing messiah Jobs.

  21. @Nic Fasci
    I’d just prefer it to be correct – around my area you can get a ticket for keeping to the speed limit if the local cop takes a dislike to what you are driving…

    It’s got so bad around my area that Sainsburys have pulled their deliveries out of the town – a major reason for the decision to do so was the abuse and trouble their drivers were getting from the local Traffic Police. They now deliver out of Ipswich.

    Note to self: Must not rant about police. Lol

  22. Hey Jemma! You seem to love technology but struggle to see why car speedomoeters might not be accurate all the time.

    Anyway, if your solution is to integrate some sort of GPS tracking gizmo into the car to give you your MPH readings to +/-0.5% accuracy, I can tell you that I definitely DO NOT want that feature in my car, thank you very much.

    Have you any thoughts as to why that might be?

Leave a Reply

Your email address will not be published.


This site uses Akismet to reduce spam. Learn how your comment data is processed.