Home made speedo

[discomikey]

cheers, that will be on the cards next, first off i need to find a suitable stepper motor to use as a pump control for a different scheme!

[xychix]

basically this could be converted into a speedo that can be calibrated on the fly.

place tomtom. drive 40Mph (or whatever) push button and new callibration is done.. Woul thereby be suitable for all tyresizes.

Just dunno if a arduino can store variables in it's programming storage or if that needs an extra (fragile and more costly) storage card.

With a storage card however you could store everything.. gps? gyroscope info? water sensors? temp ? miles per gallon? ...

[FridgeFreezer]

I sense mission creep...

Should be able to write to internal memory to store a calibration value, although it can be a bit fraught with difficulty compared to external storage. I2C or SPI EEPROMs are very cheap.

[CwazyWabbit]

I'd agree with FridgeFreezer, you see serial EEPROMs absolutely everywhere for this sort of storage, they are cheap and relatively simple to communicate with. As an example the TD5 speedometer (as well as most electronic speedometers) store their calibration data and total mileage in a serial EEPROM

[xychix]

http://arduino.cc/en/Tutorial/EEPROMWrite

there seems to be one onboard on the Arduino UNO. I'll see if there is one onboard of the arduino nano / mini... and yes they have

http://arduino.cc/en/Main/arduinoBoardNano

http://tronixstuff.com/2011/03/16/tutorial-your-arduinos-inbuilt-eeprom/

now only i2c to LCD is needed and a sensor to sense rotations on axle or something similar.

for MPG a diesel flow meter would be required I assume.

[xychix]

BTW. how does the speedometer work? Is that a mechanical turning wire or is that also electrical?

Is it a resistance to 12V or a pulse sensor?

[Bowie69]

A lot of the aftermarket ones are very simple, a magnet attached to a prop/driveflange with a reed switch and two wires from it, one pulse per revolution.

Factory original RRC ones are driven by a cable to a little box bolted to the chassis, this is 4 pulses per rev.

[xychix]

it would be easier if the original one is still in place I guess. than that signal could be used. The pulse would then be a 12v (up to 14.8V) pulse I assume?

What is the normal feed to the switch?

As I want to keep the original as well I won't be changing that.

[Bowie69]

The magnet operates the reed switch, that's how they work: http://en.wikipedia.org/wiki/Reed_switch

Personally if I were doing it, I would go the magnet and reed switch every time, even one of an old bike computer would do, and it would return whatever voltage you fed it, if you have an arduino running I guess you have 5V to use if you wanted?

[xychix]

The magnet operates the reed switch, that's how they work: http://en.wikipedia.org/wiki/Reed_switch

Personally if I were doing it, I would go the magnet and reed switch every time, even one of an old bike computer would do, and it would return whatever voltage you fed it, if you have an arduino running I guess you have 5V to use if you wanted?

if the original speedo is feeding 12V I'd keep that and tap a line into that 12V line, lower that to ~5v (classic voltage devider ~R500 / ~R700 should do.) so the original speedo can keep working and I can use a arduino as well.

You all got me enthusiastic, I shouldn't forget i've had my series now for 16 days and there's a sh!tl0ad of other things to fix before a fancy speedo...

[wood-gee]

Time to grave dig my own thread! although I had envisioned it as a kind of mini project thread, so I hope that excuses.

Finally, after a lot of learning about programming, I've got a digital speedo in the landy that works! After lots of failed starts, both from a software and hardware side, in the end I've used a hall effect sensor inside the speedo head, triggered by the speedo magnet. I'll upload some photos when I take it apart to put proper connections on in case anyone might be interested.

[CwazyWabbit]

I'd be interested to know/see how you did it, not got anything to use it on but I'm a curious fella (at least I think that's what she said to me)

[discomikey]

thanks for bringing this back up, i now know a lot mroe about programming, and just need to hack a speedo about when i get the chance, and i have a spare series transfer box to try and fit a TD5 speedo drive into.

[simonr]

This might be of use:

https://www.sparkfun.com/products/12923

It contains an Arduino Uno plus a 64x48 OLED Display in a very neat little package. Cheapest I've found so far is around 25GBP for the display & 15GBP for the programmer with the USB plug. The display is very clear & bright.

Si

[wood-gee]

So a little teaser as I'm intending to get the final soldering done tomorrow or Wednesday after work. I must apologise for the quality of the pictures; took my phone on a surfing holiday and it turns out expensive phone cameras don't like sand on their lenses; who knew!

For those of you that haven't seen it before, this fuggy picture shows the inside of a series speedo; in this case one with only an odo.

An old school speedo like this works by the speedo cable rotating, which then rotates a magnet in the speedo head, which is shown in the picture below. The magnetic force generated by the magnet rotating then drags round a cup, which is connected to the needle which sits around, but not contacting the magnet. You may just about be able to see this in the photo, I sadly forgot to annotate it, it's the shiny thing that the camera doesn't like in the previous photo. An interesting side note on this I read somewhere suggested that this relationship isn't brilliant at slow magnet rotation speeds, hence why speedos like this aren't brilliantly accurate at slow speeds.

What I've used for my speedo is a hall effect sensor, which is like a digital switch, which is actuated when in the presence of a succificently strong magnetic field. I did a lot of um-ing and ah-ing about where to trigger it from, and had been planning to use a geartooth sensor (a variation on a hall effect sensor that could trigger off steel gears or splines) up until recently, when I decided to try this. The nice thing about doing it in the speedo head is it requires little modification, keeps the electronics high above water level and avoids needing to run a wire from the gearbox somewhere.

Anyway, back to hall effect sensors. What's great about them being a digital switch is that they doesn't get issues with debounce, which is when a mechanical switch 'bounces' on and off it's contacts when depressed, which can lead to issues with the received signal. Another really handy thing about the workings of this speedo is the use of the bar magnet. This means there will be a really clear 'magnet present' and 'magnet not present' section to it's rotation, and a clean graduation between the two, which should mean the best chance of a good clean pulse train. I found that the magnet on the speedo is sufficiently powerful to actuate the sensor, although it has to be quite close. I also found that only one end does actuates this (I presume down to poles, I should probably know why, but frankly probably wasn't concentrating in that lecture). What this means is that if we only look at the rising edge of the resulting waveform, we should get a resolution of 1 pulse/rev or applying ratios to get a distance:

[i have 7.50r16s and have assumed that they're a full 31inches diameter; should result in a slight over read which I'm hoping will give me a safety margin for speed cameras, intend to do a measurement at some point to calculate % under read from this]

=Tyre circumference / (diff ratio/speedo ratio)

=2.47/(3.54/2.2)

=1.535 m - should be way more than enough, missing a pulse would result in approx 0.1% error in a mile

As indicated in the above picture, I then mounted the hall effect sensor above the magnet. With the next set of pictures I hope to demonstrate the detail of how I did this, and will post my code and a bit about how it works to make life easier for anyone else wanting to have a go.

[wood-gee]

As a note to you disco monkey, I can't remember what people said a TD5 speedo drive outputs, and am feeling too lazy to go back a few pages to check, but if you were just counting pulses I'd argue that this is a better way to do it; takes minimal modification, and the series speedo head stays fully functioning, so you can verify your digital speed value against this; in my case I got a few of the ratio's the wrong way round and that made it really easy to spot that it wasn't right, as well as keeping your odo working for insurance purposes.

*edited for meek spelling

[wood-gee]

As promised, here's the final detail on how I got this working. As mentioned, I put a hall effect sensor inside the speedo head. I did this by making a little aluminium plate which picks up off some of the mounting holes in the speedo head which the sensor bolts to, shown here:

I stuck a D-sub connector on the other side to connect to it, seemed a good idea as connector screws on so it shouldn't come undone when pulling on and off the dash (an issue I've had by using header pins on my LCD screen). Also lots of pins spare on it to allow for future improvements!

This then attaches to the speedo head as show below:

Another view on the position of the sensor relative to the speedo head

Made a hole in the back of the speedo case for the connector to stick out of - positioned where the voltage regulator use to be as I swapped it for a solid state one a while ago. The sensor and connector can be seen attached to the case below:

Issues I've found from this:

• D-sub connector is a nice idea, but positioning where I have means it fowls the heater duct, either needs a low profile connector or to be in a different place
• I originally removed 2 of the 4 screws holding the head together, put one back in. When the speedo is fully tightened up to it's shell without all 4 in the pressure from the tubes used for the warning lights push the two parts of the speedo head apart, prevent the needle from running straight on it's bush, and making it drag noticeably. I addressed this by putting 1 screw back in and shortening the tubes slightly.

[discomikey]

Sorry IF this seems a bit stupid, but isn't that just sensing the speed, not actuating the speedo head? I messed about with a 10 motor on the back of a spare head today using a scale extricate transformer as the power source but it could only manage 30mph, what is the revolution of the speedo drive per propshaft rev?

[wood-gee]

Yeah, sorry if what I've said has got a bit muddled over the course of this thread.

My eventual plan is to both sense the speed of the landy and drive the speedo using a servo motor. What I've detailed above is only the sensing part of that, my logic being that by having both the original speedo and my sensor running I can make sure it's reliable and doesn't do anything strange or buggy before then removing the original speedo altogether, (plus I seem to take ages to make any progress, so it's also just because I'm being slow).

The ratio of prop to speedo is 2.2 I think.

I was thinking of using a servo rather than a motor, however servos usually only do around 180 degrees, so either use a sail winch servo, which do more, or gear a normal servo. This means you could still use the speedo drive as input, plus my understanding is that servos don't need feedback, whereas you might need closed loop control if you wanted a motor to be fairly accurate, which seems like work that could be avoided.

[discomikey]

I did wonder whether to use a stepper motor directly mounted in the speedo head, connected to the needle directly rather than through the magnet assembly, however this would bring the need for a digital odometer too.

Some stepper motors can include closed loop control iirc.

My other option was to butcher a discovery rev gauge and use the actuator for that, it is driven by pulse, and as a result I think it could be driven via a microcontroller.

I will be using a simple pid type control methodology either way so it will have to be closed loop I think.

P.s. is that 2.2 speedo cable rotations per prop rotation?

[wood-gee]

2.2 prop rotations per speedo rev, if you google series speedo ratio it comes up on a lot of forums, but basically a worm gear drives the speedo drive pinions, hence the reduction ratio.

I'd argue that you could potentially keep the odo etc, and just drive the needle, although I think you'd need to mount your motor off axis to clear the speedo head, although this would work well with using a gear ratio. The main challenge with that I think would be supporting the needle, a small enough motor to fit and gears, but equally I give you the right to laugh at me when I come back and say that it wasn't possible.

Yeah, that's what drew me to servos over stepper motors, steppers still need control, often people mention 'zeroing' them off the end stop first, tbh doing it with PID would be cool, all about dat step response.

Disco gauge is a good idea and I'd be very interested to see if you managed it, if it's just driven by pulses I'm sure you could do that easily with an arduino (I have a feeling that's what you said your using?)

Out of curiosity, have you tried reading the alternator pulses before to get RPM?

[simonr]

There are plenty of RC servos with sweeps up to 360 degree these days - mostly intended for robotics.

Although steppers or DIY servo is fairly easy - nothing beats the simplicity of driving an RC servo. Hard to beat on cost / size also.

Si

[discomikey]

I'll have to dig out my old discovery binnacle and have a go with it.

[discomikey]

To save boring you with the maths, to use an unmodified (ish) speedo head and just drive the input with a motor, therefore keeping the standard odometer, the motor needs to be able to cope with 2,200rpm, assuming max speed is 100mph, or 1550rpm assuming max speed of 90mph.

This is running 235/85R16 tyres, and 3.54 differentials with a speedo drive of 2.2 prop rotations per speedo rotation.

[discomikey]

So to me the options are as follows.

Find a motor capable, and create a coupling system to fit to the back of the speedo head, control with PWM, and calibrate using sat nav, this is less neat, but retains the standard odometer.

Or, pull speedo apart, drive needle directly with a stepper motor of some sorts, control with PWM, and calibrate as above, odometer function will be lost, but a screen can be employed as above for this, maybe can turn into a multifunction screen too, showing mpg etc. Neater, but more expensive and time consuming.

Either way, I will need to try and graft a td5 sender onto a series transfer box