Do Flat Tyres Make Your Speedo Lie?

There are some engineering questions that may not have huge importance in the world, but which become the subject of intense idle speculation. A good example is the question of whether a lower tyre pressure on a motor vehicle would make a difference to the indicated speed. There are several contrasting intuitive theories as to what should happen, so [mechatronicsguy] has taken the time for a bit of experimentation in order to find out what really happens.

At stake were the change in effective radius from a flattened portion of the tyre, the so-called tank tracks effect in which the entire circumference of the tyre is still traversed, and the prospect of a change in circumference due to the different pressure. The test wheels were made from foam, and were found to give a different reading when compressed. This might solve toe problem, but of course real car wheels have radial wires to give them stiffness. When these were simulated on the foam wheels with packing tape, the difference evaporated. Later this was confirmed by GPS-measuring a real car with deflated wheels.

All this makes for a fascinating read, because after all, there’s sometimes no substitute for a real-world test.

Header image: Gerlach, Public domain.

45 thoughts on “Do Flat Tyres Make Your Speedo Lie?

    1. Hi Cecil,

      I did consider that, and you’re right, there’s a whole industry of “indirect TPMS” systems that use the ABS wheel encoders. However I think they must be looking at either Fourier domain to check for resonance, or using tyres that behave differently to mine. Since going from ridiculously flat to ridiculously overfull changed my speedo scale by less than 1%.

      I’ll post a part 2 writeup shortly, where I check ratio of wheel speeds captured directly over CANBUS. I have two datasets captured on different roads, sadly, but having a tyre at 65% normal inflation changed my wheel speed ratio by less than 1%. There may be an effect there, but it’s a damn subtle one.


      1. You logged reported speedo speed vs GPS speed, which is not the same as individual ABS sensor data. Depending on the exact data you’re reading, speedo speed could be intentionally averaged and inflated by up to 10% to comply with the law (speedo must never under-read but may over-read).

        Two cars in our household have ABS-only TPMS and it will pick up a <20% drop fairly reliably.

        When you're travelling in a straight line it's fairly easy to imagine a small difference in one wheel's speed is easy enough to spot. A lot of modern cars have steering angle sensors and body accelerometers for the safety systems, so it can tell pretty relaibly when it's going straight.

        1. >When you’re travelling in a straight line it’s fairly easy to imagine a small difference in one wheel’s speed is easy enough to spot.
          Did you read my part 2 post? I tested that this afternoon and found that taking the tyre down to 65% of normal pressure only affected the wheel speed ratio by maybe 1%.

          Although I wouldn’t be surprised if indirect TPMS systems only worked with a subset of tyre designs.

          1. Or maybe they can use a difference less than 1% over a very long distance to trigger the system for gradual deflation, and possibly a sudden change in ratios for a puncture.

            With such a system, it wouldn’t be possible to detect deflation that happens evenly in all tyres, which happens when they have been inflated the first time with air, because the oxygen escapes within a few weeks.

          2. @SparkyGSSX
            “which happens when they have been inflated the first time with air, because the oxygen escapes within a few weeks”

            Tosh. Utter BS. Oxygen diffuses out of a tire SLIGHTLY faster than nitrogen would, but the difference is minute and certainly not in the order of “a few weeks”.

            It’s only those upselling you “premium nitrogen” filling on your tires that I’ve heard make such claims and they are not to be believed.

          3. “Did you read my part 2 post? I tested that this afternoon and found that taking the tyre down to 65% of normal pressure only affected the wheel speed ratio by maybe 1%.”

            It’s not just “wheel speed vs steering angle” – you have to convert steering angle into yaw rate, which requires a model (so it’s more like a Kalman filter on steering angle and vehicle speed). If I yank the steering wheel sharply and then back, the vehicle’s dynamics don’t change significantly since the car hasn’t had time to respond.

            There’s a paper on a simple version of this (“Monitoring Method of Indirect TPMS Under Steering Situation”). In the end the result is bigger than you’re seeing, but not by a ton, probably around 5%. Data processing and filtering is cool.

          4. “Or maybe they can use a difference less than 1% over a very long distance to trigger the system for gradual deflation,”

            In an “idealized case” (straight line constant driving) a sub-1% difference is easy to detect. In real world driving you need a car dynamics model since wheel rotation and tire radius constantly change anyway.

          5. I think you’re convincing yourself the problem is harder than it is – indirect TPMS is super common and works well, so clearly it’s possible and manufacturers have cracked the problem well enough that it’s been mandatory on cars since 2014.

            I mean, Wikipedia even explains exactly how they do it:
            “First-generation iTPMS systems are based on the principle that under-inflated tires have a slightly smaller diameter (and hence higher angular velocity) than a correctly inflated one. These differences are measurable through the wheel speed sensors of ABS/ESC systems.”


          6. “I think you’re convincing yourself the problem is harder than it is – indirect TPMS is super common and works well, so clearly it’s possible and manufacturers have cracked the problem well enough that it’s been mandatory on cars since 2014.”
            The main point of my testing was to check whether low tyres would affect measured speed enough to cause a significant error or a speeding ticket. The results for that appear to be a resounding no.

            Indirect TPMS seems like it’s possible, but extremely fiddly to do well, with possible drawbacks such as being blind to tyes on the same axle going flat together. Also I’d argue that the fact that they had to go to “second generation” iTPMS indicates that the first generation did not work well at all. I have no doubt that the effect is large enough to measure when you use Fourier methods and Kalman filtering. But my point was that the change in radius effect is one or two orders of magnitude more subtle than people think.

          7. Oh, it’s definitely not a visible effect from a speedometer point of view. Wheels just don’t deform that much.

            Using Fourier analysis it’s extremely easy, but best to re-calibrate it at full so you’re just looking at shifts over time than anything absolute.

            “Indirect TPMS seems like it’s possible, but extremely fiddly to do well, with possible drawbacks such as being blind to tyes on the same axle going flat together.”

            No, the speed ratio stuff is actually only blind to all 4 wheels going flat (the Fourier analysis still picks that up). What you were looking at were just wheels on the same axle, and you don’t need to do that (one of the simpler measures is the diagonal difference, using FL+RR vs FR+RL since they have an equal center of mass in the baseline). There are any number of variations which can pick up each wheel individually. Just think about it – both front tires going flat introduces a front/back asymmetry, both left tires introduces a left/right asymmetry, diagonal tires going flat introduces an asymmetry under load (when turning).

            “Fiddly” is a bit of a stretch. In extremely controlled conditions (a straight line, constant speed run) you’ve got huge signal/noise (although as you note the effect is small in absolute terms). The main reason it’s fiddly is the same reason anything on a car is fiddly – because you’re twisting and turning the vehicle and submitting it to large forces. Yeah, the later revisions of iTPMS do better, but it’s the same argument for direct TPMS – those early sensors broke all the damn time.

      2. wow. what a waste of time and expense. WTF wouldn’t they just put the sensor on the transmission like every well built car in the world? Sounds dumb if you ask me. If i change the tire size, the transmission still reports the exact same, but I can easily drive by one of those speed cameras that show your actual speed, make a mental note of how fast over or under it actually is and it’s done until i change to a different wheel/tire size.

        talk about wasting a ton of time and money to make something worse.

  1. Some dragster tyres are a good example here. At stand still, they’re quite flat, and when loaded by the massive torque at take off, this compresses them more. Then towards the end of the strip, when going whickedly fast, the forces cause them to tighten and expand massively radially. This effectively gives them an extra gear ratio that comes in progressively along the strip.

    Hopefully nobody is driving a road car with flat tyres at 300 mph, so in daily life this might be a bit moot.

        1. On certain American cruise missiles landscape scanning is collision avoidance while it follows a GPS waypoint system. But of course this is the unclassified version.

  2. I’ve seen firsthand that it really doesn’t matter, the tire won’t live very long. When its gone then you’d better hope you are on a steel rim; because aluminum will crack

  3. The obvious observation is: Iif a vehicle has different diameter tires than the speedo is calibrated for, then it will produce an erronous reading, too low for bigger tires and too high for smaller tires.
    If all four tires are flat and the vehicle is basically riding on its rims, then obviously the effective tire diameter is radically smaller than spec and the speedo should read a lot higher than the true vehicle speed.

  4. Your speedo lies anyway – Gav even spots this in his analysis, where the displayed speed is greater than the internal speed reported on the CAN bus by exactly 10%. Every car I’ve had has done this, and I remember an old project in Silicon Chip that was intended for calibrating the speedo on pre-CAN vehicles by changing the pulse rate of the speed sensor… only for users to discover that it gave incorrect odometer readings. Turns out the odo is *always* precisely calibrated, which proves that the incorrect speedo readings must be deliberate.

    1. It’s required to be by EU law (and most of the rest of the world has similar regulations afaik)

      See: UN Regulation No. 39 – Rev.2 – Speedometer and odometer
      5.4. The speed indicated shall not be less than the true speed of the vehicle. At the
      test speeds specified in paragraph 5.3.5. above, there shall be the following
      relationship between the speed displayed (V1) and the true speed (V2).
      0 < (V1 – V2) < 0.1 V2 + 4 km/h

      (Note < should be smaller than or equal to, but Hackaday doesn't show the symbols correctly)

      So they normally just certify the car with the largest size tires allowed as the "tires normally fitted" and call it good. That way if someone does ever mount smaller tires, they'll still get a reasonable speed indication.

      1. Indeed a speedometer is not a callibrated device and will always show a deviation that is legally defined.
        (GPS is only accurate over an interval and can not accurately show current speed as the measurement frequency of location is generally only 1Hz.)

        This is one of the reasons in Europe every car comes with a document that lists all the legal tire sizes that are possible. You can not legally use other sizes because the speedometer would be deviating too much from the real speed.

        Another thing, which I didn’t see in the comments yet, not all speedometers measure the speed of the car at the wheel, even if that data is available through the ABS system.
        Some cars measure the car speed by measuring rotation inside the gearbox.

        1. It most definitely *is* calibrated, it just doesn’t show the true value. An uncalibrated sensor will have a random offset by definition, but as you state the offset is quite carefully defined.

    1. Hello,
      interesting that this gives so little difference on your car!

      Would you add the tire dimension with tire model and weight on the tire (not car weight) to the data?
      A narrow wall “sporty” tire could have less effect than a “balloon” offroad one.

      Also there are reinforced sidewall tires, runflats and so on that could affect the result.

      1. “interesting that this gives so little difference on your car!”

        It’s not really surprising – if you look at most research papers on iTPMS it requires quite a bit of data processing and sensor fusion.

        You definitely can’t just get it from relative speeds and steering angle as there’s quite a phase lag between yaw rate and steering angle and you need vehicle speed to do the conversion anyway. Obviously I can yank my steering wheel back and forth when I’m stopped and I don’t turn at all.

        For cases where the vehicle doesn’t have a yaw rate sensor, there’s a Kalman filter on (steering angle, vehicle speed) which predicts vehicle dynamics and then the predicted yaw rate is used. But obviously because gyros are dirt cheap plenty of vehicle just… use a yaw rate sensor. Even those typically need filtering though since there are usually offsets.

    1. Because the bottom line is you still have a flat. Too many variables. A good tolerance analysis of all components is necessary. Tire heat, sidewall construction, TPMS reporting period, Tire cent. data, etc.

  5. The entire perimeter of the tire rolls along the road if inflated or flat.
    The “simulation” using foam is faulty because the foam can compress causing the circumference to change.
    Reporting that outcome as a “fact to investigate” should be promptly dismissed. Find an accurate way to test.

  6. I have a Jeep Wrangler 2005. It came withp215/75r15 tires from the factory. The wheels were upgraded and tire size changed to 31×10.50r15. The diffrence in height of the wheels is about 1.5 inch. The speedometer is now wrong ie- at 20 mph on speedo real speed is 22 mph. At 50 mph speedo the real speed is 57 mph. Dodge sells replacement gears for the gearbox to correct for tire changes.

  7. Some vehicles with traction control use viscous couplers in their differential which will be damaged if driven for too long on different sized tires or on a low tire. The friction material will be worn away excessively as the two sides are compensating.

    1. Not quite how that works. As the name implies, a viscous coupling uses a viscous liquid and a set of closely spaced plates. When a wheel slips, it starts spinning its plate(s) faster and moving the fluid at a different rate than the other axle, leading to the fluid trying to drag the slower plates to catch up. No friction material, but you can overheat and degrade the fluid.

  8. Unless you have slipping/traction issues, or overinflation to the point of tire deformation (both of which would be dangerous) then it won’t change your speed.

    Look at the flat tire in the top image, the radius where the rubber meets the road, and the radius where it leaves the road is the same as a normal tire, so, as long as it doesn’t slip, or loose contact with the road, it will still travel approximately the same distance per rotation.

    There will always be some minor variation, but it will be a small enough percentage to not matter I’m practice.

    1. Tires do deform continuously when inflated just like any object under load. Radials restrict that quite a bit, though, so the effect is incredibly minimal, like you’re saying.

      It’s a really small effect, though. If you read an early paper by Isermann “Indirect Vehicle Tire Pressure Monitoring with Wheel and Suspension Sensors” you’ll see a 25% underinflation results in a few tenths of a percent effect on wheel speeds. Harmonic analysis does a lot better.

    1. There’s also a joint paper from researchers at Audi & VW: “Comparative analysis of tire evaluation methods for an indirect tire pressure monitoring system (iTPMS)”. Focuses on the resonant frequency detection method, which has much better signal/noise in general (not surprising).

  9. I retrofitted a newer engine, complete with donor engine ecm into an older truck. I had to give the programmer of the ecm my rear end gear ratio, type of transmission and the overall tire size. Problem was I didn’t have rims and tires yet, had to estimate. This was in part to make the original speedometer read correctly. I was 1/2 inch off, in the negative if I recall. Speedometer is off by a few miles per hour, reading faster than I am going, according to speed radar signs around here. Less off the slower I am going, more the faster I am going.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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