How do we measure propeller Rpm ?

[midwoud1]

  Curious to know how other windmillers measure Rpm of their windgenerators.

  Analog, Digital, Bike speedo-meter, Laser.

  It's important to see cut-in ,stall speed , max.speed ,TSR vs windspeed.

  Mppt and loadcontroller is usefull but not everyone has it.

  Rgds Frans.

[SparWeb]

In the past I have settled for tapping one of the AC wires, filtering it a little, and letting my datalogger (PICaxe) count pulses.  This isn't very reliable, on my WT generator, because there seems to be a lot of harmonics in the AC.  I operate a permanent-magnet motor conversion, so the harmonics are exaggerated with all that iron in there.  I expect an axial PMA with air-core stator has much less of this noise and in that case the AC will work better than in my experience. 

Hugh Piggott probably still has a good diagram of a tachometer circuit on his Scoraig Wind website.  I have built it and it does work, but I made some mistakes, so I do not have it working any more reliably than the straight AC tap that I mentioned before.

I've use the bike speedo before, too, but only when doing bench-tests of generators before putting them on the tower.  The speedo gives a good reading but obviously you can't record it without a pencil... 

What I really should do it put the same kind of magnet/reed-switch combination on the actual wind turbine, permanently, with wires coming down the tower to read the signal from the reed switch.  No AC harmonics and very low power requirement.  How come I never get around to doing simple things like this, but waste days building tachometer circuit boards instead?   

[madlabs]

I use a hall sensor and a magnet on the motor shaft. Trasmitted via RF link. Very simple and accurate.

Jonathan

[Flux]

There are a lot of ways that work, the choice may be influenced by personal preferences.

Those interested in cycling may well choose the bicycle speedo method ( I wouldn''t have a clue about such things).

Jonathan's magnert and hall sensor is a good method and is made easier with a radio link. If you have to run cables it may need some care with noise pick up but can still be done.

If you have the rectifier near the battery and so have easy access to the ac voltage then measuring frequency seems the easiest way. Digital multimeters can measure frequency but are tricky to handle and need careful filtering.

I prefer the 2917 tacho chip but again it needs a bit of care to get good results. I would use some form of isolation for the signal ( small transformer of optoisolator) and even then you will need a simple r/c filter to leave you with the alternator fundamental frequency.

Although I haven't tried it I suspect a hall current sensor in the rectifier positive to the battery will give you enough ripple to work a 2917 tacho.  This ripple will be 6 times ac fundamental but you can choose the scaling components to read direct.

If you are using a microprocessor somewhere in the system then Sparweb's pulse counting can work from frequency or current ripple but you may need a scope to sort out the necessary filtering to be sure you are looking at the correct waveform.

You can of course use any of the independent tacho devices coupled to the turbine but as frequency is available from any alternator I prefer to use it.

Flux

[Frank S]

I think what Frans is asking is what system are others actually using not how to design one.

[midwoud1]

I think that Madlabs has a good application for Sparweb (iron core )
Wireless RF link need 12 Vdc up in the tower TX part.
I made a digital Rpm counter ,puls scanner and 7 segment Led displays.
It has a IR sensor counting holes in  a small disc .
Bench tested and callibrated.
A lot of work but it gives a good reference.
12 Vdc in the nacelle and the signal wire with coax to the readout in the shop.
Like Flux said anti noise.
A second counter is Analog .
With the diagram of Hugh Piggott using Ac from 1 phase near the rectifier (inside)
Using the LM 2917  dil 8 , no problems with harmonics because of open coils.
Read out with an  analog meter 500 ua ( Vu meter ) callibrated with the digital meter.
The signal of the 2917 goes in a microcontroller , with a relay board.
I use the Arduino to drive the pitchcontrolled blades and thats working perfect.
With the controller I can activate an actuator in steps of milliseconds.
Feather in and out and hysteresis.

Thanks for sharing   Rgds  - Frans -

[SparWeb]

Reading AC directly is fraught with trouble.  I just got this tiny "pocket" oscilloscope.  It was clipped across two phases of my turbine's AC when I took the reading, about 1 hour ago.  The wind is rather light, so it's just about the cut-in speed.



Here's where I get into trouble.  To measure RPM from this waveform, should I be measuring, say, positive pulses?  That's a common way to go about it, but this wave crosses above zero during one of those little ripples, then drops, and then comes back up in the main rise in voltage, plus one more blip before going back down for the negative cycle.  Tapping into this I will have to count 3 positive zero-crossing for every complete cycle...  not 1 as you would expect from a pure sine wave.

I was able to deal with this with my old datalogger, in only need to factor the pulse counts to match bench-tests I'd done in the shop... until I started experimenting with capacitors connected across phases and they damped the ripples enough that sometimes (but not all the time) I had to revert back to 1 pulses per cycle.  It was very difficult to make heads or tails of the data logs until I realized that this could be going on.  I didn't have the oscilloscope at the time.

[boB]

You will of course want to low pass filter the high frequency ringy dingies from that relatively low frequency
phase to phase waveform.

A nice isolated way to take a waveform like this and convert to pulses that can be counted is to
drive an opto coupler LED through a current source and then pick up the pulses from the detector
side of the opto.  Then, process as (or if) necessary  (divide by # of poles, etc.)



boB

[Flux]

The main trouble comes from ringing at the zero crossing point as boB said. if you filter it enough  with the filter cut off low enough to be partly attenuating the fundamental you will remove the nuisance harmonics.

I would actually say your waveform was below cut in, when the rectifiers conduct the ringing at the waveform peak will vanish as the battery clamps it, it does still leave the ringing at the cross over point.

I tend to agree with boB that using opto isolators does help for several reasons, being current driven the led will not respond much to noise at the crossing point as the diode is below bias point. It also gives you isolation from common mode noise on the ground  that you get if you try to look at the waveform from one ac line to ground.

Flux

[midwoud1]

Wave image PMG  Hugh. 800 Watt . 1 phase.
28 Vdc battery load  8 Amps.
Wind 7 m/sec.
No perfect sinus , but the LM2917 accept it .

[Flux]

Yes that is typical of what happens when the rectifier is conducting. The flat tops are actually the battery clamp volts and the notches are the commutating points of the rectifier. They are away from the zero crossing point so the 2917 won't be confused.

If you look at the volts before cut in the waveform of an axial machine will be close to a sine wave and completely free of the ripple at the crossing point.

The ripple on Sparwebs alternator is almost certainly slot ripple and the frequency will be a function of the number of slots or a multiple of it.

Flux

[midwoud1]

Hi Flux .
Thanks for the info.
I see a better sinus shape with low speed below cut-in.
Will try later with disconnected rectifier and right blade angle to avoid runaway.
And keep  the wave on the scope display.

  - Frans -

[Mary B]

Info on the scope? If it is affordable I could use one.

Reading AC directly is fraught with trouble.  I just got this tiny "pocket" oscilloscope.  It was clipped across two phases of my turbine's AC when I took the reading, about 1 hour ago.  The wind is rather light, so it's just about the cut-in speed.



Here's where I get into trouble.  To measure RPM from this waveform, should I be measuring, say, positive pulses?  That's a common way to go about it, but this wave crosses above zero during one of those little ripples, then drops, and then comes back up in the main rise in voltage, plus one more blip before going back down for the negative cycle.  Tapping into this I will have to count 3 positive zero-crossing for every complete cycle...  not 1 as you would expect from a pure sine wave.

I was able to deal with this with my old datalogger, in only need to factor the pulse counts to match bench-tests I'd done in the shop... until I started experimenting with capacitors connected across phases and they damped the ripples enough that sometimes (but not all the time) I had to revert back to 1 pulses per cycle.  It was very difficult to make heads or tails of the data logs until I realized that this could be going on.  I didn't have the oscilloscope at the time.

[OperaHouse]

I hope everyone notices the resistor from the transformer to the diodes.

[fabricator]

I noticed it but there is no value for it, did you ever consider putting together kits for that circuit Frans? there's tons of pocket scopes on ebay http://www.ebay.com/sch/i.html?_sacat=0&_from=R40&_nkw=pocket+oscilloscope&rt=nc

[midwoud1]

Hi  Fab and Op.
Yes I know about that blank resistor .  I haven't put it in.
And the system is working good. I made a big elco near the Rpm meter.
It can be to avoid spikes with low rpm.
Thanks for the Ebay link about pocket scopes .
I ask one for Santa ....ho..ho..ho.
It looks ,Sparweb has one , very nice tool.

Rgds  - Frans -

[SparWeb]

The pocket scope I have is from "SEEED"  (note the 3 e's in the name)  I think I paid about 70 dollars for it.
They are selling a newer version of the product now, a few more features but similar.
A hobbyist like me is not worried about accuracy, or fidelity, and MHz/GHz audio/radio frequencies also don't interest me.
Once I got my head around that, I realized I don't need a 800$ 20-pound Tektronix scope. 
All I want is a visual with enough detail to perceive what effects my efforts are having.
This little device is good enough for me, for now.

[midwoud1]

SparWeb   Thanks.

Another way with an opto sensor and chopper wheel on the mainshaft.
Not practical , ( long wires and coax signal)

Fab. Making the circuit in kit form. It's a few dollars components.
       And 2 hours stress to make it for a beginner.
       I think everyone can make it.

- Frans -

[SparWeb]

Frans,
I recognize that diagram from the datasheet for the LM2917.  Many similarities with the diagram from Hugh Piggott.  Have you made changes to Hugh's diagram?  It has some edits that I do not have on the copy I saved some years ago.

The LM2917 diagram does not show any filtering, required for direct-reading AC, so it is only appropriate with a magnetic pick-up or reed switch, or even the optical gate you mention (and reject for good reason).

Overall, my experience with tachometers on the generator AC is spotty, but that's almost entirely due to my inexperience with filtering and just guessing at the waveform.  If I had just bought / borrowed / stolen an oscilloscope 5 years ago I would have the problem licked by now and understand it more thoroughly too.

If youcan design the AC low-pass filter with more skill than my "cookbook" ability, and have access to an oscilloscope to verify your results, then I'm sure you will succeed.

[midwoud1]

Hi Sparweb.

The difference between our systems . I have open coils and you have iron core.
In my case there is a clean zero crossing point, like Flux mentioned as important.
As we see on the scope images.
A lowpass RC filter can be usefull, but I'm not good at it to calculate.
Or what Flux said take the signal between the rectifier and + battery.
The 2917 can handle the higher pulse frequency because it can do with car ignitions aswell.
I did not change much in Hughs diagrams.
C2 is in my diagram a 1uf elco.
And I added a 500 ua analog meter , an adjustable potmeter 47K ,an Elco 3300 uf.
The output goes to an interface and than to a microcontroller.
Do you want to switch something with an Rpm counter ? 
I can give more info about it.

  Rgds  - Frans -

[Flux]

The limit resistor between the transformer and clamp diodes will be needed if you use anything other than a tiny transformer with high primary volts. In theory the thing self destructs if the diodes conduct and there is no limit resistor.

In real life a tiny 240v transformer will have a lot of internal resistance and if fed with the fairly low voltage of a turbine it will most likely be ok.

I don't like the idea and would prefer to have a defined limit resistor and it can be quite high as the input impedence of the 2917 is very high.  In fact I believe that the 2917 can tolerate something over 20v peak on the input and if you use a 12v transformer with 240v primary you could safely omit the clamp diodes.

If you have trouble with noise at the zero crossing point then incorporating this resistor is a good idea as it lets you filter the input with a capacitor across the clamp diodes.

Flux

[fabricator]

Hi Flux .
Thanks for the info.
I see a better sinus shape with low speed below cut-in.
Will try later with disconnected rectifier and right blade angle to avoid runaway.
And keep  the wave on the scope display.

  - Frans -

Frans, are the ground points in the schematic bonded to the Neg of the 12 volt power supply for the unit?

[Flux]

Fab, I can probably answer this one if Frans doesn't see it.

You are right that the ground connections are 12v negative. This 12v supply can be independent or the 12v neg can be battery negative and the 12v can be dropped from the main battery with a regulator.

Having the transformer to isolate the input makes the power supply very flexible If you use a totally isolated supply you may or may not find it better to connect the 12v neg to battery neg depending on the noise situation about the place and the waveform of the turbine ac.

[fabricator]

Got ya, thanks Flux.

[midwoud1]

Hi ,

All the groundpoints are connected to Neg. 12Vdc.
I have an independent regulated powersupply, Vdc .14 volt.
2 volt extra for the voltageregulator 7812. Clean 12 volt out.
The reason for it , I have both a digital counter and an analog counter.
The digital draws 350 Ma and the analog 2917... 50 Ma
And the turbine charging voltage on the batteries has a little AC component.
Batteries 2 x 12 Volt . 230 Ah.

  - Frans -