Piggott Windturbine and charging regulation issues [Done]

[sperate]

Hello everybody,
you can find here a link to my actual project about going off-grid : Eolienne Piggott.

This is a windturbine made according to the plans in the Hugh Piggotts Book.

The electric generator is made of a car rear wheel hub, two steel disks, 24 neodymium permanent magnets and home made copper windings.
The stator (the coils) is molded in resign, and so are the two pieces of the rotor (steel disks + magnets).
Everything is mounted on a steel frame made of steel corners welded together.
The output is 3 phase at about 24 volts and can deliver 700W of power at about 10m/s windspeed. It is coupled to a three phase bridge rectifier and then connected to batteries.

For more details you can visit Hug's website : scoraigwind.com or this site : otherpower.com.

The problem i am working on is, when the batteries are full and there is still wind, you can't do as with solar pannels and unconnect the turbine from the battery. Because it would result in overspeed leading to premature wear of the bearings and even destruction of the wooden blades.
You could stop the turbine by shorting the generator but it's nonsense because you loose power while the turbine is stoped.
What can be done is diverting an amount of energy from the turbine to a resistive load (in order to heat water or whatever). But this need to be done according to the wind power developped on the turbine.
There are only two regulators designed to work this way made by Xantrex [http://www.wholesalesolar.com/products.folder/controller-folder/xantrexC60.html] and Morningstar [http://store.otherpower.com/bos/morningstar-tristar-ts-60-diversion-load-controller.html].

The project is to develop something like that.
Futher will come the schematics ...

sperate.

PS: Hope you understad my english :-)

[solutions]

You should have a way to furl the blades or at least cock the turbine perpendicular to the wind so it does not overspeed, ideally not even turn.  I don't care how much resistor you load it with, your blades will get destroyed by the mechanical loading on the blades when you are using an electric brake. You designed it for 700W, but what happens when Mother Nature throws 10kW of wind at it?

http://www.youtube.com/watch?v=T2x7u4GAqPc

[sperate]

It's a small windtirbine, simple but solid and i have no interest in motorting the blades or the yaw.

The fact is that i've stopped it with the electrical brake and there's no problem with that.
It is actually rapid but smooth because of the fact that the power of the brake is proportional with the speed of the blades.
So it starts fast and hard and finishes slow and smooth. I could take a video if someone wants to see.

I've even made a security brake, by mechanically inverting the normal operation of a triggered ground fault circuit-breaker and triggering with the voltage between two phases like this :
Emergency brake

When the voltage between two phases is high enough, the mosfet is triggered and the breaker puts the wind-turbine in short.

About the 10kW, in regard to the rotor diameter, it would mean that the incident wind speed is about 140m/s (witch it pretty much !!).
More seriously, if we consider gusts of wind at about 150km/h or 40m/s it means we have to negotiate with power peaks at about 3kW.
But these calculations are average because of two facts :
 - first, the lambda coefficient of the blades witch intervenes by changing the blade tip speed/wind speed ratio, if the blades start tu furl at very high speed for example.
 - second, the battery or the dump load clamps the voltage and don't let the rotor over-speed. If the power, thus the voltage, is really to high, the emergency breaker stops it.
And at that time (when the turbine is stopped) the blades catch a really smaller amount of energy than when the rotor turns.

The last but not least point that is complicated for me to explain in english, but that you will find in Hugh's literature, is that the design of the tail is made to furl all the rotor and put it perpendicular to the wind in the case of very strong winds.

Regards,
sperate.

[sperate]

Here's the schematic i'm working on with LTSpice to try and fix the diversion load regulation system :



Sorry but it is not a final version (connections missing, debug architecture).

The problem at that moment is that everything works fine until I close the regulation loop.
When I close it, LTSpice bugs.
I'm going to try with PSIM as soon as i get it.

Any comments, suggestions or questions are welcome !

[Ichan]

Wind and hydro power generation is always on my interest list, but haven't yet have time to dig into it.

Some newbie questions:
> What is your opinion on using car alternator for the generator?
> Is there any DIY controller like the Xantrex and Morningstar that you mention?
> Your load regulation schematic, is that to replace the controller above or as an add on?


-ichan

[sperate]

> What is your opinion on using car alternator for the generator?
There are several problem with car alternators :
 - speed must be above 1000 rpm to have some current injected in the battery. You have to rewind-it or use a multiplier.
 - it is not permanent magnet motor, so it has a winded rotor that needs current to create an electro-magnetic field (witch is a waste of power) and it has also carbons witch are wear parts.

I've seen some videos over the net about replacing the rotor of a car alternator with permanent magnets rotor .. and then also replacing it's stator .. with pieces that only work for certain normalized models I suppose. In this case why not by a PM motor directly.

Anyway a prefer the solution of building my own generator with basic materials.
 - copper wire
 - two steel disks (some people made it from car breake disks)
 - 24 NdFeB permanent magnets
 - resign
 - hub from a car rear wheel

You can find in attached documents an extract of Hugh's book that precisely explains why the car alternator is a bad idea.

> Is there any DIY controller like the Xantrex and Morningstar that you mention?
A didn't see any-one at that moment

> Your load regulation schematic, is that to replace the controller above or as an add on?
Yes it is to replace the above controller, as i didn't found any DIY schematic.

[solutions]

Try a 10M resistor across M1.  When you turn M1 off, the node floats and may be the cause of LTSPICE going nuts...maybe

Posted on: March 18, 2012, 08:21:27 20:21 - Automerged
sperate: you are new here....you are supposed to share books, not just a few pages from them. Give more than you take is a good rule to follow. thanks.

Hugh Piggott's "A Wind Turbine Recipe Book" 2009 (metric) edition:

http://www.mediafire.com/?apda7dbkh7kaelh

[sperate]

I've managed to make it work with the "alternate" solver in the LTSpice options and by using different parts for the error amplifier and the comparator.
Here is the old explicated schematic and the updated schematic i am working on :



Next step will be to feedback the error amplifier to avoid super-high gain in regulation loop.

[cncfred]

Hello

 before the TimerBlox IC LTC6992 from Linear considered that would
 I use my control.
 Is an adjustable PWM IC

 greeting

[sperate]

Very interesting chip.
I tried to use the TL494 (PWM + error amplifier) but the problem was that it has a minimum duty-cycle which you can not go lower.
Almost all power supply regulation chips are designed to always maintain a small amount of duty-cycle.
But the LTC6992, depending on the reference (especially the LTC6992-1), does not have this minimum duty-cycle.

Thanks for the track, i've ordered samples to Linear.
But anyway i'll continue in the first time with op-amps and comparator in order to build something with easy to find, common components.


[At that time nothing new about the design .. didn't had time.]

[sperate]

Hello,
here's the latest schematic and simulation results with different generator voltages corresponding to different incomming windspeeds.



As you can see there's a lot of instability which leads to oscillation on vfeedback.
0.5Volts on  Vfeedback means about 1.5V on the battery volatge.

I think that these oscillations come from the filtering of Verror.

Sperate.

[sperate]

Hello,
good news, the regulation theoretically seems to work !

Here are the latest schematic and simulation results :

 

Look at vbatfilter, despite the increased currents and tensions coming from the generator, the battery voltage hangs to about 28.8V.
Response time to steps is about 10ms.

Next steps :
 - reduce ripple on Vbat with increased capacitance on C4.
 - tweak filters on Vbat and on Verror.

But to do that i'll switch on PSIM software because switching simulation is sooooo long on LTSpice !

Anyway, I'm happy with those first results and a physical prototype can already be started in parallel of the simulation.

Sperate.

[Ichan]

Hi Sperate,

I just look into Morningstar product brochure, it has 3 control function: charge, load, and diversion - your schematic, seems to be only the diversion controller?

Other thing, what is your suggestion for anyone who do not want to (or have time) to build the generator? What option available to be used as generator (with some modification perhaps)?

-ichan

[sperate]

Hello Ichan,

The design concern is only the diversion load.
Because it is simple to make a charge controller which disconnects the batteries when they are full.

What could be interesting is to add an MPPT option, but this is more software than hardware.
My main concern is to make a solid and cheap DIY wind turbine battery charge regulator.
There is no other way than load diversion to regulate with a small windturbine.

About another way with the generator, i'm not a specialist but you could take a look at that for example :
 - http://cgi.ebay.fr/GENERATRICE-12V-660W-3-ponts-de-diodes-/200746290102?pt=FR_YO_MaisonJardin_Jardin_EnergieRenouvelable&hash=item2ebd694bb6

Sperate.

[Ichan]

Cheap and working is an excellent combination, would like to follow your progress.

I did some search for the generator, looks like china make some good product - understandable they have plenty of rare earth magnet sources.

No relation with, this one take my attention:

http://www.ginlong.com/wind-turbine-pmg-pma-permanent-magnet-generator-alternator.htm


-ichan

[sperate]

Ok,
that's the kind of stuff i was looking for to show you.

Have you took a look at their regulator (controller) ?
http://www.ginlong.com/download/201112/User_manual_GCB-20kW.pdf

They're only doing On/Off on the diversion load with preseted hysteresis voltages.
It's very simple but that's because they do not have batteries to charge.
As the inverter has a wide voltage and power input range, their main concern is just to limit this input power (thus voltage) to protect the inverter and maybe the turbine also.

Sperate.

PS : I've switched on PSIM, learned how it works, and now i'm trying to use it to make stable regulation loop.
Simulation is much faster than with ltspice, many thanks again to bunion an carboot for this software.

[sperate]

Hello,

Good news about the project !

The regulation loop works really better now.
Vout (The filtered Vbat) is clamped to 28.8 volts, no matter what happens on the incoming wind speed (thus power) or the outgoing power.
You can see the results of the PSIM simulations here.
I have also joined the psim schematic file in the rar archive for those who are interested.

To achieve a working controller, one of the tricks was to use symmetrical supply (+15V/-15V) for the summer and a PI (Proportionnal Integral) stage instead of my RC low pass filter.

If anyone has questions, feel free to ask me.

Good night,

Sperate.

[Ichan]

What is your comment if i say: "i think proportional alone is enough for this case - there will be no overshoot, does it?"

-ichan

[sperate]

I don't know if proportional is enough. What I know is that the integration makes the static error null.
About the overshoot, it doesn't bother me.

I'll test and answer more precisely.

----------------------

After verification, here are the results without the integrator :
Low Gain


High Gain


That's what I thouht, the static error is always present. Making the gain higher just reduces it but it doesn't clear it, and moreover the signal becomes very ugly.
The advantage of the integration is to have a very high gain in low frequencies and a relative small gain at higher freq.
So, great reactivity in the DC time domain (and no error on the set point) and lower reactivity for AC signals.



Posted on: May 19, 2012, 03:13:14 15:13 - Automerged
Hello,
News from the controller ..

First I've integrated current regulation to be able to limit the battery inrush current. For example, while finishing battery charge, a small amount of current is needed to let it charge completely.

Here are schematics, commented waves and PSIM file.

[Ichan]

Thanks for verifying it, appreciate your methodological answer.

I don't use PSIM, have you simulate it for ON/OFF control (with hysteresis)?

-ichan

[sperate]

Latest news :

It's been a little time since I've switched back to LTSpice to try and simulate in more realistic terms.
Here's the latest schematic, waves and spice file.

The results are encouraging. We can see on the waves that the voltage and current regulation work simultaneously.
The simulation isn't long enough to see the signals stabilised or even periodic but it would mean too much simulation time and my poor little 1 core, 2Ghz laptoop is crying.

The only part of the schematic that is not real is the sawtooth generator for which I plan do use a simple NE555.
The schematic becomes full of Op-amps, but that's the fate of analogics.
------------
Next, real design and prototyping.
I wonder how to make the control analog signals (voltage and current set points) with a microcontroller. I hesitate in using a DAC, a digital potentiometer or a pwm with low pass filter. I think the best is the digital potentiometer option but that goes contrary to my simplicity of supply willing
Any suggestions or ideas about that ?

@ Ichan

I don't use PSIM, have you simulate it for ON/OFF control (with hysteresis)?

Not at all but I could try.

sperate.

[Ichan]

I am waiting for the real application result 

Using microcontroller for this case had been thought from the begining, but I am afraid not having enough time these days.

For low changing signal like in this application then pwm + low pass filter should be enough as DAC, but I think DAC is not required in here - the mosfet driven by internal pwm of the micros.

Measuring the high current of the Ibat will be more problematic, with the high side current sensing as on your schematic you need to pay attention on the common mode voltage limit of the differential amplifier op-amps being exceeded. Hall effect sensor would be easier but sure more expensive.

If I will do this with micros then I will choose dspic33fj64gp802: 12bit ADC, 16 bit DAC, 16 bit PWM, analog comparators, and available in 28 spdip package.

-ichan

[sperate]

If I will do this with micros then I will choose dspic33fj64gp802: 12bit ADC, 16 bit DAC, 16 bit PWM, analog comparators, and available in 28 spdip package.

You mean doing all the regulation loop in software and using PIC's PWM to drive the diversion MOSFET ?

Measuring the high current of the Ibat will be more problematic

I know but in this architecture, the common mode gain of the whole differential amplifier is equal to the common mode gain of the second stage divided by the differential gain of the first stage. In theory. So we'll see.
Anyway current regulation is an option. The basis is voltage control, fixed at 28.8 (more or less) with a potentiometer I think.
In that case there's no µcontroller at all.

But I wanted to train myself on current sensing in order to log some data about incoming power on the battery or the whole system.

[Ichan]

You mean doing all the regulation loop in software and using PIC's PWM to drive the diversion MOSFET ?

Yes that's right.

Analog is a gray area between black and white, I respect people who master this area a lot - as I am not.

If you do and share your analog version then I can make sure it will intriguing me to do the digital version.

-ichan

[sperate]

Meanwhile I've found an interesting website with modified car alternator design.
They sell all the parts and a simple charge controller that also works with 2 level hysteresis comparator (see manual : http://www.flexcharge.com/flexcharge_usa/manuals/nc25a_manual.pdf).

http://www.windbluepower.com/