Super cheap DIY anenometer (for Arduino)

[gildasd]

This is for a proof of concept, the idea is to measure variations in wind speed, not absolute value.
This thing has got to live a couple of weeks at the most.
If the result are ok, then I'll get something better: https://www.adafruit.com/products/1733

For now, the "sensor" is DC motor lifted out of a 3,99€ toy car.

Even at low speed (turning between fingers slowly) it produces more than 50mV.
It's cheap, it's crap, it's what I got (and can afford).

I intend to connect it like this:

But with a 4V Zener to ground if it starts generating a lot of volts.

The construction is going to be cheap ass (starting from the bottom):
An alloy plate.
A bit of epoxy paste to hold the motor.
The hood is half a plastic Easter egg.
3 more half easter eggs as vanes.


Question:
Am I missing something obvious?

[tautech]

But with a 4V Zener to ground if it starts generating a lot of volts.

Am I missing something obvious?

Have scope?
Check for unexpected spikes that might fry an Arduino.

Schottky diode to VCC?

[gildasd]

But with a 4V Zener to ground if it starts generating a lot of volts.

Am I missing something obvious?

Have scope?
Check for unexpected spikes that might fry an Arduino.

Schottky diode to VCC?

Only an analogue scope at home, but I'll test on the Uni's scope before I plug in.

Did you mean Schottky diode between VCC and the ground?

[Seekonk]

The zener is probably rectifying some RF energy.  I would place a ,1uf in parallel.  A 4.7K load resistor would also be a good ides.

[alsetalokin4017]

Your cutaway sketch shows the motor connected to Ground and to an Analog input. If your motor makes any spikes or sustained voltage over 5V this will probably damage your analog input. You may be able to protect it using a Zener diode but you might need some more complicated snubbing arrangement just to be completely safe.

But your graphic appears to show the motor connected to Vin instead of Ground. I don't think that is going to work very well for you.

[gildasd]

The zener is probably rectifying some RF energy.  I would place a ,1uf in parallel.  A 4.7K load resistor would also be a good ides.

Something like this?
I'm not doing high frequency sampling, 1Hz our less. So if the cap robs some of the signal, I won't even notice...

[MK14]

The zener is probably rectifying some RF energy.  I would place a ,1uf in parallel.  A 4.7K load resistor would also be a good ides.

Something like this?
I'm not doing high frequency sampling, 1Hz our less. So if the cap robs some of the signal, I won't even notice...

By putting everything after the resistor, ALL the components will tend to be protected. Even if something silly happens, like accidental connection to a battery or something.

The problem with a lone zener like that, is that if the power is switched off, the port pin is now susceptible to excessive voltages again.

You could put Schottky diodes in (instead of the 1N4148), for even better protection. But what I showed is probably fine.

You could make one of the 1N4148's I showed, a 4V zener, if you want to limit the voltage, because of the ADC. But the double diode protection, should do fine, anyway.

[Andy Watson]

The zener should be the other way up.

Have you considered that a "toy motor" might introduce too much coggin?

[gildasd]

The zener is probably rectifying some RF energy.  I would place a ,1uf in parallel.  A 4.7K load resistor would also be a good ides.

Something like this?
I'm not doing high frequency sampling, 1Hz our less. So if the cap robs some of the signal, I won't even notice...

By putting everything after the resistor, ALL the components will tend to be protected. Even if something silly happens, like accidental connection to a battery or something.

The problem with a lone zener like that, is that if the power is switched off, the port pin is now susceptible to excessive voltages again.

You could put Schottky diodes in (instead of the 1N4148), for even better protection. But what I showed is probably fine.

You could make one of the 1N4148's I showed, a 4V zener, if you want to limit the voltage, because of the ADC. But the double diode protection, should do fine, anyway.

I might be saying something stupid here, but looking at the datasheet of the 4148, the forward voltage is 1V.
That being that, does it mean that your circuit  limits the signal to 1V? I was hoping to max out at 3V at 20m/s...

[gildasd]

The zener should be the other way up.

Have you considered that a "toy motor" might introduce too much coggin?

I removed the gear box, the cups drive the rotor directly.

[MK14]

The zener is probably rectifying some RF energy.  I would place a ,1uf in parallel.  A 4.7K load resistor would also be a good ides.

Something like this?
I'm not doing high frequency sampling, 1Hz our less. So if the cap robs some of the signal, I won't even notice...

By putting everything after the resistor, ALL the components will tend to be protected. Even if something silly happens, like accidental connection to a battery or something.

The problem with a lone zener like that, is that if the power is switched off, the port pin is now susceptible to excessive voltages again.

You could put Schottky diodes in (instead of the 1N4148), for even better protection. But what I showed is probably fine.

You could make one of the 1N4148's I showed, a 4V zener, if you want to limit the voltage, because of the ADC. But the double diode protection, should do fine, anyway.

I might be saying something stupid here, but looking at the datasheet of the 4148, the forward voltage is 1V.
That being that, does it mean that your circuit  limits the signal to 1V? I was hoping to max out at 3V at 20m/s...

They are reverse biased, so only come into play at around -0.7 V and suppyVoltage+0.7V

i.e. voltages around 0 up to supplyVoltage, should be left alone.

If the above is NOT the case, then I have drawn the diode(s), the wrong way around. Easily done.

They are to protect the MCU input from damage, from too low or high voltages. I.e. Below ground or above Vcc (Supply voltage).

[gildasd]

I get the general idea, thanks!

[tautech]

They are to protect the MCU input from damage, from too low or high voltages. I.e. Below ground or above Vcc (Supply voltage).

Exactly why I mentioned Schottky diodes in reply #1.
Schottky diodes will prevent more than 0.3 V influence to the power rails, therefore staying under the usually recommended + or - 0.5V for all IC inputs.

[pmbrunelle]

I think that it could be difficult to mount the rotating assembly well aligned with the electric motor.

I think I would try some epoxy, and then applying power to the motor (in the low-drag direction), so that the rotating assembly self-stabilizes on the electric motor while the epoxy cures. Obviously, the motor must be oriented vertically...

[Kleinstein]

Those cheap motors are usually not rotating really easy. Chances are better with a little fan from an old CPU cooler. If you are lucky you can even remove the coil and iron part an leave only the magnets and the hall chip.

For protection the resistor and zener are Ok, the 1µF cap is rather large and should better be non polarized. For additional protection there should be a second resistor between this circuit and the µC.

[MK14]

They are to protect the MCU input from damage, from too low or high voltages. I.e. Below ground or above Vcc (Supply voltage).

Exactly why I mentioned Schottky diodes in reply #1.
Schottky diodes will prevent more than 0.3 V influence to the power rails, therefore staying under the usually recommended + or - 0.5V for all IC inputs.

I agreee.
But also the internal (IC) diodes, are suppose to be slightly weakened (either by design or they are like that anyway), to ensure that external, low cost diodes can be used.
Sticking with the datasheet is best, so 0.5V it is.

[MK14]

Those cheap motors are usually not rotating really easy. Chances are better with a little fan from an old CPU cooler. If you are lucky you can even remove the coil and iron part an leave only the magnets and the hall chip.

For protection the resistor and zener are Ok, the 1µF cap is rather large and should better be non polarized. For additional protection there should be a second resistor between this circuit and the µC.

Maybe not. Because when the power is off, and Vcc=0V. Voltages below the 4V of the zener, can damage/destroy the cpu. Dave has done at least one video, explaining some of the details about this (it shows an MCU mysteriously being powered up).

There are millions of ways of protecting stuff. Opinions vary.

EDIT:
As long as the 4K7 resistor, is connected, in a way which limits the current. I guess voltages below 4V, but above 0.5V or so, would be more of a datasheet violation and/or undesirable. E.g. Might cause latch up on power up. Rather than actually damage the MCU.

Many people use the single zener method, so there are not necessarily black and white answers. A lot of it is opinion.

[gildasd]

Those cheap motors are usually not rotating really easy. Chances are better with a little fan from an old CPU cooler. If you are lucky you can even remove the coil and iron part an leave only the magnets and the hall chip.

For protection the resistor and zener are Ok, the 1µF cap is rather large and should better be non polarized. For additional protection there should be a second resistor between this circuit and the µC.

The two I have have low enough drag for my purpose - and they are identical - of making two similar anemometers.
They were unusable, for this purpose, with their little gear box.
I'll think about your circuit.

[gildasd]

I think that it could be difficult to mount the rotating assembly well aligned with the electric motor.

I think I would try some epoxy, and then applying power to the motor (in the low-drag direction), so that the rotating assembly self-stabilizes on the electric motor while the epoxy cures. Obviously, the motor must be oriented vertically...

I was thinking of using a lathe to centre the egg...
Pour a bit of epoxy in, let cure, drill hole smack in the middle.

[gildasd]

This is how they look.
Making the protection circuit now.

[gildasd]

The recommended circuit does not work right now.
I get a steady 1.93V at the MCU port pin and a hand pin makes the signal goe down to 1.926V.
The same spin with no ptotection produces +0.045V...
Weird.

Edit;
Works much better with the 4.7kO resistor after the protection circuit.
Negative voltage does still get through...

[MK14]

With a quick look at the circuit, it seems fine.

D1 1N4148, seems to be the only easy way, a faulty component, could produce a "stray" voltage. As it is the only easy source of Vcc.

You could try removing D1 and/or testing D1 with a multimeter or something.

But I would guess it is mis-wired, or one of the components is faulty/leaky.

[gildasd]

With a quick look at the circuit, it seems fine.

D1 1N4148, seems to be the only easy way, a faulty component, could produce a "stray" voltage. As it is the only easy source of Vcc.

You could try removing D1 and/or testing D1 with a multimeter or something.

But I would guess it is mis-wired, or one of the components is faulty/leaky.

Glue on the pin of a diode - arrgh!
That being said, the output is more sensitive to small anemometer movements with the resistor after the protection circuit.
I'm thinking of putting a 1µF dielectric to smooth the signal further, as to reduce the needed sampling time on the ADC...

Added pictures:
1) Donor toy car.
2) The slot for the wiring get electrical tape to avoid shorts.
3) Heat shrink on the capacitor pins to make the fit tight.

[MK14]

With a quick look at the circuit, it seems fine.

D1 1N4148, seems to be the only easy way, a faulty component, could produce a "stray" voltage. As it is the only easy source of Vcc.

You could try removing D1 and/or testing D1 with a multimeter or something.

But I would guess it is mis-wired, or one of the components is faulty/leaky.

Glue on the pin of a diode - arrgh!
That being said, the output is more sensitive to small anemometer movements with the resistor after the protection circuit.
I'm thinking of putting a 1µF dielectric to smooth the signal further to reduce the sampling time on the ADC...

That makes sense. I'd expect the motor signal to be very noisy. You effectively would be making an RC filter.

[gildasd]

With a quick look at the circuit, it seems fine.

D1 1N4148, seems to be the only easy way, a faulty component, could produce a "stray" voltage. As it is the only easy source of Vcc.

You could try removing D1 and/or testing D1 with a multimeter or something.

But I would guess it is mis-wired, or one of the components is faulty/leaky.

Glue on the pin of a diode - arrgh!
That being said, the output is more sensitive to small anemometer movements with the resistor after the protection circuit.
I'm thinking of putting a 1µF dielectric to smooth the signal further to reduce the sampling time on the ADC...

That makes sense. I'd expect the motor signal to be very noisy. You effectively would be making an RC filter.

Thanks for the help!
Is MK14 a reference to Griffon Engined Spits?

A 1µF just smooths the sine, a 10µF effectively make the output DC while only robbing 5 to 10 mV.

[MK14]

Thanks for the help!
Is MK14 a reference to Griffon Engined Spits?

No.
It is a 1970s, classic/vintage electronics hobbyist self build microprocessor hex display/keyboard, early microcomputer. Clive Sinclair related, by 'Science of Cambridge'.
With a phenomenal 256 BYTES! of ram.

https://en.wikipedia.org/wiki/MK14

[gildasd]

Thanks for the help!
Is MK14 a reference to Griffon Engined Spits?

No.
It is a 1970s, classic/vintage electronics hobbyist self build microprocessor hex display/keyboard, early microcomputer. Clive Sinclair related, by 'Science of Cambridge'.
With a phenomenal 256 BYTES! of ram.

https://en.wikipedia.org/wiki/MK14

256 bytes! What will people do with so much space?
Doing a burn test now. Anemometer has already detected that the source fan is not 100% constant speed.

[gildasd]

Makes for a very small circuit!

[MK14]

Makes for a very small circuit!

Yes, that is looking nice, neat, tidy and compact.
Neat/tidy (usually/often) means reliable/consistent.

[Mephitus]

Maybe I am missing the point here, but couldn't this be immensely simplified with a hall effect sensor detecting the RPM on a non-motorized fan? It would make for a safer and more reliable circuit.

[gildasd]

Maybe I am missing the point here, but couldn't this be immensely simplified with a hall effect sensor detecting the RPM on a non-motorized fan? It would make for a safer and more reliable circuit.

Yes in theory.
In practice, the motor is off a 4€ toy, the 3D printing using the Uni's printer, discount ping pong balls, CAPS de-soldered from PC main boards and scrap alloy machined on Uni lathe/mill...
Plus this motor is already natively max 4V, not the 12V that a PC motor produces, so it can talk directly to an Arduino ADC.
Total cost is 6 to 7€ per complete system plugged in.
This kind of anemometer maxes out aerodynamically at 90km/h, producing 5V...
So the protection is more for shorts, crazy seagulls and other dumb stuff (radar test, idiot turning the vanes the wrong way round).

If anything, the only possible problem is increasing the low speed sensitivity, while losing some top end, with different vanes.

[Mephitus]

Ahh, I see. That would make sense. As something that might help in future projects. I have found that printers are a gold mine for parts. Many times people even give them away for free. They commonly have hall effect sensors as well as IR interrupter sensors that could be used for projects like this. Hope your testing works out!

[gildasd]

Ahh, I see. That would make sense. As something that might help in future projects. I have found that printers are a gold mine for parts. Many times people even give them away for free. They commonly have hall effect sensors as well as IR interrupter sensors that could be used for projects like this. Hope your testing works out!

I have a late 1990's alarm system fished from a dumpster that is going to be a gold mine!

[MK14]

I have a late 1990's alarm system fished from a dumpster that is going to be a gold mine!

Can I send you a message, to put into your Mars probe project then, made out of a 1990's alarm ?
Also the protection circuit would need updating, as there is radiation? in space, but no wind, so you don't need the ping pong balls!

[gildasd]

I have a late 1990's alarm system fished from a dumpster that is going to be a gold mine!

Can I send you a message, to put into your Mars probe project then, made out of a 1990's alarm ?
Also the protection circuit would need updating, as there is radiation? in space, but no wind, so you don't need the ping pong balls!

No Mars probe, "just" testing superstructure optimisation for commercial shipping...
Send away!

[gildasd]

Changing the angle of the vanes has wielded some improvement.

The (highly scientific) testing method was:
A large table fan was set level with the anemometer and the distance was slowly increased till failure to register.
Version 1: 82cm (original, left)
Version 2: 88cm (new, right)

So a noticeable and repeatable improvement.
But to improve even more, I still think the vanes need to be bigger.