Speedometer calibration ?

Ok, if your speedo is off 2~3 mph (not the odo, just the speedo), that "ruins everything", lol?

Yeah, that can ruin your whole day.

 

So, since my speedometer says 68 when my actual speed is 65, then when my odometer reads 52,307, will I only have driven 50,000 miles? (Please forgive my maths, if incorrect. It’s not my strong suit. It’s more a question of “does this affect the odometer, too?”)

 

No. The odometer does not carry the same intentional overreading error that is built into the speedometer. A bunch of class-action product liability lawyers made sure of that a decadeor so ago by fleecing down several car makers for alleged overreading odometers, claiming warranty fraud. I received the legal notices on two brands of cars, even though my particular units didn't have the alleged problems.

If you put a ScanGauge or Torque or other OBDII-port monitor on your car, it will show you the unbiased speed known to the internal computers. At least on Prii (and likely on many other car models too), this internal reading doesn't have the intentional error built into the speedometer display.

 

The solutions that I found on the web (change tires size, change gears, give up, ...) are not satisfactory at all. So here is the cheap and easy solution that I propose:

 

This solution seems to apply a correction factor proportionally across all speeds.

My one question would be does this hack also interfere with the input to the odometer? Because if it does there would be a legal liability on the installer of this device installing it knowing it will have the odometer under-report the miles/kms driven. This would in effect be odometer fraud.

This hack would not work In our region as the speed displayed is simply 2 km/h more than the true speed, so a true speed of 20 km/h is displayed as 22 km/h; 30 km/h is displayed as 32 km/h; 50 km/h is displayed as 52 km/h; 80 km/h is displayed as 82 km/h; etc. So as annoying as it is, the mental arithmetic is quite easy, just add 2 to the displayed speed to know what the true speed is. That is my solution to this problem.

 

In what way is an existing OBDII-port monitor product, as I mentioned in the post above yours, not satisfactory? A Bluetooth dongle on the OBDII port, coupled with a smartphone app such as Torque or one of its competitors, should quickly and easily show the speed known internally to the car's ECUs, which we already have reason to believe is quite accurate.

This proposal is making unverified assumptions about the electrical architecture within modern cars. And it looks like 20th Century assumptions being applied to 21st Century products. Until you verify that there is a practical signal line where you can insert this frequency converter device, we don't know that your approach can work on models as new as Prii. If it turns out that the ABS module is outputting numeric data onto a high speed digital data bus (maybe the CAN bus?) that eventually is routed by the speedometer, then your frequency converter device will be a wild goose chase. Though it should work for some older era designs, pre-Prius.

And yes, we already know that numeric speed numbers are available on a digital data bus. See above for OBDII readers.

Your rationalization for using small power-of-2 integers for division is decades out of date. Full precision arithmetic units are readily available in cheap microcontrollers. Full blown computer processors with long floating point arithmetic units are available as close as the nearest smartphone.

And the displayed example, for converting a 74 MPH pulse rate down to 64 MPH, is absurdly wide for the North American market. That is a Euro-market adjustment. For North America, you'd be dropping 64 just down to 63 or 62, or a stretch to 61. And your device lacks the digits to take fractional steps in between those ratios.

===============

Elsewhere, I'm seeing an Arduino project plugging into the CAN bus to read and display speed. A slight software modification of this should allow arbitrary adjustment factors of greater precision.

 

Indeed, that's why at the end of my video I clearly mention that now I need to test it and see whether it really works or not. And should it fail, there is still room for modifying the code that I put in my microcontroller.

I agree. Yet I want to make the circuit as cheap as possible by using the most basic electronics. There is no problem with doing so.

You missed the point. I used such a big gap between the true speed a the speedometer readings so that we could easily see on the oscilloscope that the circuit was working properly by outputting the signal with a longer period. It would have been difficult to see any difference if I used a gap of 2 or 3 mph. This was just an EXAMPLE!

I want the speedometer to display the correct speed, not some extra device. Otherwise I would simply use my phone's GPS.

 

I bet that it won't affect the odometer, at least for my car's model (prius 2010). The reason is that I had some fuse that was blown, resulting in the speedometer displaying a zero speed. And yet, the odometer was still counting miles. So it seems that there are two different lines. But, as I mentioned in my video, I need to test it. I hope to be able to do it soon.

 

“Give up”, in other words just forgetting about it, means that if your speedometer reads 50 mph you’re actually going 48~49 mph.

Is this SERIOUSLY so terrible, lol? And if you MUST go 50, just press a little harder on the gas, till the speedometer reads 51~52 mph?

 

Fun hack. One of the first questions I had was, as you were proposing how to gather the calibration data, although you gave your rationale for choosing 64 as a sample point, were you going to collect at some variety of other sample points as well, just to check your assumption that the correction needed is linear? As dolj suggests, it might not be. It seems like it would be ordinary diligence to be aware of that assumption, and design the experiment to be sure it is checked.

The effect on the odometer, again as dolj points out, could be the deal-breaker for the idea.

One minor terminology point in the video: once you're no longer working in decimal, the point is no longer a "decimal point". Some will say "binary point" and others will use the general term "radix point" where you don't need to know the base.

 

At 64 mph, my speedometer indicates 69, so yes, this is annoying.
When you buy a watch you want it to display the correct time, right? And I'm sure that pilots on commercial aircrafts want their altimeter to be accurate, especially when they have to land in bad weather with zero visibility. The purpose of having a measuring instrument is for us to get information, not to be fooled because some dumbass in the government decided so.

 

Yes, I agree that I have to check that. At this point, my circuit is just a prototype to test the idea. It won't be a problem if the velocity dependence on the frequency is of the form V=CF+k instead of V=CF, I will just have to reprogram it. If the relationship is more complicated, I may need to use a more powerful microcontroller.

I'll check that!

You're right, this completely slipped out of my mind.

 

That is a very unusually high difference for the North American market. Do you have the right tire size? Or a vehicle originally built for the EU, UK, or other foreign markets where such offsets are more typical?

That isn't how they land in such conditions. Ordinary barometer shifts would kill them. There are specialized landing systems for such purposes.

 

I believe so: 195/65R15 (It's a gen 3, 2010)
I bought it used, but from the car's history it comes from the US.

 

Hi Val,
I am in New Zealand and own a 2020 Prius Prime with also a speedo error of approx 6 to 7% over indicated. As everything on this later model is done by the numbers, the inaccurate reading is far reaching. Example, when I follow the Toyota NZ response to set my cruise control at 5 to 6 KM/H above the desired speed in order to drive for example at 100 KM/H it triggers other distracting issues such as a overspeed warning. The open road limit here is 100 KM/H and now my vehicle thinks it is going at 106 KM/H when my 2 GPS units I use for checking actual speed both agree I travelling at 99 KM/H.
It angers me that a company like Toyota would suggest that I 'speed' rather than fix the problem. In fact they state a tolerance of 10% error reading would still be acceptable. Unbelievable.
As a side issue this discrepancy will only get worse as tires wear down.

Anyway I am very interested in finding out if you can fit your fix to your car and get the desired results. It should work on later models also I hope.
Btw, your video can't be 'tagged' for notify on YouTube as it is rated for Kids access or something like that.

Hope to hear about your results soon and if positive I'm a buyer for sure.

TinkerNZ

 

What did your GPS say about that speedometer accuracy of your previous cars? Or like most people who bring up this topic, is this the first car you ever checked?

This has been normal automobile industry practice for a very long time. There are even international agreements covering the topic, depending on which markets they align with.

Speedometer - Wikipedia

Compared to what you already have, the change with tire wear will likely not be noticeable.

 

If the biggest problem in your life is a speedo reading a little high, you’ve got it made.

 

Not quite a one-to-one comparison.

Sure, having your speedometer say 69 when you're actually only going 64 MpH is "annoying".

On the other hand - having an inaccurate watch can be deadly. (Think a railroad worker being on the tracks when an express train comes through because his watch was 5 minutes slow.) For most of us though, a watch that's off a couple of minutes an hour really won't assure that we get to that appointment on time anyway.

And - speaking as a Commercial pilot with an instrument rating - an inaccurate altimeter can also be deadly. Even for those not flying commercially. That's why the pitot-static systems on aircraft have to be tested and calibrated by a qualified technician periodically.

So though I think you have a neat idea there - let's not take something that's merely annoying and try to justify it by comparison to things that are genuinely deadly. It's not as if the "precision" of 99.9% of driving done in the world - even commercial driving - makes an exactly-accurate speedometer a necessity. Besides, at least here in the USA, "speed limits" are regarded as merely "suggestions" to most drivers.

Pretty much every car I've driven - and I drive a LOT of rentals - reads a mile or two low at 60MpH.
It doesn't annoy me in the least.

I'd really rather see the inaccuracy in fuel gauges addressed. THAT could actually be important. My 2017 with an 11.4 (US) gallon tank was supposedly half full by the dash readout - but it took 7.4 gallons to fill it up. (and I didn't "squeeze" any extra in after the shutoff!) That's a much larger discrepancy than the 7-8% error on an instrument that really has little to do with safety or practical matters.

 

If I remember right, Elektroingenieur quoted the repair manual explaining that the speedometer doesn't base its displayed number on the speed pulse, but on data retrieved over CAN. It just redistributes the speed pulse for use by other accessories that use it (like auto volume adjustment in the audio amp, and so on).

That's probably the biggest fly in the ointment for this idea.

 

.