Friday Nov 9

[DanB]

Last week Shawn got his alternator mostly together.  Pictured above he's very carefully putting the 3" x 1.5" x .75" N40 grade NdFeB magnets on the rotor, using an Aluminum template to get them spaced correctly.  The rotors are 18" in diameter, cut from 1/2" thick steel.  There are 16 magnets on each rotor.  This is very similar to my 17' Wind turbine, but it will hopefully include most of the improvements I had in mind for that machine:  larger spindle/bearings and hub, larger magnet rotors, stronger stator brackets, bigger hardware.

George and Shawn fit the stator to it.  This is a temporary assembly just for testing.  We'll take it all apart again soon for painting.

We attached Shawns alternator to the rear axle of Rosy the truck.  I would've liked to use the tractor for this but the slowest I can run the PTO is about 150 rpm, this alternator needs to be tested at slower speeds.

During testing we got up to abut 150 rpm, the engine sped up - the alternator slowed down and there was a funny smell.  Turned out that the torque on the alternator was so great that the passenger side wheel broke loose and starting turning - so we had to chain that wheel to the front axle.

To get output data we loaded up my batteries and measured the power into them from the alternator.  I put loads on the batteries to hold them down to 50VDC.  The line between the alternator and the rectifiers is 50' long of 10 gage, so there are some losses there.  To recap... this stator has 12 coils, each wound with 68 turns using two strands of 15 gage.  I expect it's about 50% efficient at 3KW based on an estimation of the resistance (considering the weight of the wire we used).  The airgap is currently set right at 1" and the cutin speed is at 72 rpm.  I'm surprised the cutin is this low with such a wide airgap actually - I think stepping up to N40 grade magnets, and  having a bit more space between the magnets improves this quite a bit over my 17' machine.  To do this over again I might consider winding with a few less turns and slightly thicker wire.  As it is - I think we'll need to open the airgap a bit more to let the speed up for a 17' blade.

We also put this 10' alternator together just for testing and fit it to the PTO on my tractor.  This one is fairly standard, 12 N35 grade magnets per rotor, and the stator is wound with 140 turns of #17 gage wire.  The airgap on this one is 11/16".

Thats what we came up with on that one.  I also (for fun) watched foot pounds on this.  It seems about 50% efficient at 700 watts -which is consistant with what I figured based upon the resistance.  Of course at lower output it's much better.  This would run probably all day at 500 - 600 watts on the tractor, it gets warm but not bad.  At 1000 watts it gets quite hot very quickly.  Of course - testing like this there's no wind going by but I still think these ones need to furl around 600 - 700 watts output or else...  That said - we've not had a burnout with one of these yet except once when the installation didn't allow the machine to furl.

[Nando]

Hmmm !! they look good

Have You measured the un-loaded output voltages ?.

If so can you show it, as well as, the internal measured resistance of each winding ?

With electronics You may get around 85 % efficiency.

Did You ever think about using the will mill power to set the furling point of the mill ?.

I, indeed, hate to use the generator as a power energy "HEAT SINK"( self ballasting)

Nando

[DanB]

Hi Nando -

the unloaded output voltage should be linear with rpm, so yes.  The larger alternator is hitting 50VDC at 72 rpm.  The smaller one cuts in at 140 rpm.

Here's the graph that shows 'input' and 'output' on the smaller alternator when it's clamped down to 50V.  On the low end the scale had lots of wobble and the angle between  the scale and the 12" 'arm' that we welded to the alternator was a bit less than 90 deg, so on the low end it's 'optimistic', but at higher rpm I expect it's fairly accurate with regard to power in.

[electrondady1]

hello dan,

i don't think even henry ford would have imagined the way you are using the model a truck!

wonderful to see they are still useful in any way (not shure if my 85 astro van will still be ticking in 2055.

i'm interested in how you estimate the efficiency .

the output is measured in relation to a potential output?

but how is the potential output estimated?

[DanB]

we welded an arm to the alternator so that we can hook up a spring scale 12" out from the spindle - so we can measure torque in foot pounds.  torque (foot pounds) * rpm/5252  = hp (horsepower) and 1 hp = 746 Watts, so this way we measure the power into the unit.   Divide the power out by the power in and that tells us % efficiency.  It's kind of rough data - again, on the low end of rpm things wobble and good readings are hard to get.  I also expect the spring scale (which reads up to 150 pounds) is not very accurate < 10 pounds, it's not a terribly scientific instrument.

[Flux]

Yes they look just about as I would expect except for the first point on Shawn's graph.

N40 does make a fair difference, seems about right for cut in with that gap to me.

For Nando I make the resistance of the big one just under 1 ohm and the 10ft is about 3 ohms.

I tend to agree that anything above 700W for any significant time is pushing your luck even up in the wind. With wind ratings you can probably hit the 1kW peak without trouble if the sustained power is kept to about 600W.

Flux

[Nando]

The input power seems not right, I may be wrong though. ( linearity of the torque detector).

The peak efficiency is 1150 / 3250 = 35.4 %

This wind mill with a charge controller should be able to produce better than 1700 watts.

It seems that the open voltage should go as high as 130 + 30 for the Resistive losses + 30 Volts stator - rotor gap = 190 to 220 volts out.

So the total power of 1500 to 1700 watts could be delivered to the load of 48 volts.

This mill using a two ( 2 ) phase MPPT controller capable of the high voltage could do a good charging profile.

Looking at the initial voltage generation straight line and projecting it to the vertical line of the 450 RPM, it seems capable of 1750 watts -- this a estimated guess assuming low stator-rotor gap voltage .

I always recommend that good generator data is taken, without any load.

It is an excellent point to observe behaviors and possible solutions to problems that may be present in the future.

Nando

[electrondady1]

dan your using a * to represent a math function but i dont't know what it might be.

if you use the same spring for all your tests it would work out as far as comparisones between your different generators.

and so by opening or closing the air gap yu can tune the geni ?

going for a specific power output, say cutin, at a given rpm?

[Volvo farmer]

"So the total power of 1500 to 1700 watts could be delivered to the load of 48 volts.

This mill using a two ( 2 ) phase MPPT controller capable of the high voltage could do a good charging profile."

This is good news indeed! over double the reliable power out of the same mill, for free! I'm planning on erecting my 10' axial flux windmill in the near future, though mine will be for 24V. I was going to run it through a regular old C60 with a dump load but I'd much rather use a two ( 2 ) phase MPPT controller and harvest the extra power. Where can I get one?

 

[Flying Z]

The asterisk * is used for multiplication.

[Countryboy]

* is a common symbol for the math function of multiplication.

Multiplication is usually shown as X or *.

[Nando]

If you have a MX60 you have a 2 phase MPPT charger.

You may need a ballast controller to limit the voltage to 100 or so volts, the limitation of the MX60

Nando

[Volvo farmer]

I do have an MX60 for my solar, and I would gladly buy another for my wind turbine. Outback does not recommend the use of their MX60 with wind turbines like the one I have, and the ones Dan builds. Why do they not recommend it? They could surely sell more MX60s if they thought they would reliably improve the performance of wind turbines.

The MX60 is rated for 150VOC on solar, so I don't understand why a ballast controller would be necessary to limit the voltage to 100V.

But here's the crux of the matter as I understand it. I've sat in a chair and watched my MX60 work. Every three minutes, it sweeps the incoming power source (solar in my case) and tries to find the maximum power point. This process takes a 45 seconds to a minute. The thing still charges while it is sweeping but not at the MPP, it is searching for the MPP and charges at the search value. Once it finds the MPP, it uses the values it has found for maximum until the next sweep comes along.

Wind is a far more variable resource than solar. Ever notice how your hat blows off your head when you least expect it? Yet it's pretty hard to get a sunburn without staying out in the sun for a while.

The MX60 was designed for solar, it was not designed for wind. I was at Dan's house back when he was on 24V and saw his ammeter swing from 10A up to over 40A within a minute, his relays clicked on and his dump load shed the extra power until the gusts died down. I only watched it for five minutes but it was obvious to me that a device like the MX60, which takes 45 seconds or so to sweep the MPP, was not appropriate in a variable power situation like wind.

Now if someone makes a device that can sweep the MPP in a second, every thirty seconds, I believe it would be appropriate for wind power, but the MX60 ain't it. Until that device comes along, MPPT for wind power is exactly like a flux capacitor. Theoretically, I could save up all my extra solar energy and whiz it off into the future in a Delorean and harvest it when I need it, but that is not the current state of the technology.

[Flux]

The MX60 has the converter technology capable of doing this job, but I agree with you that the MPPT bit will not cope with wind conditions.

If you limit the volts safely and try it on wind I have no idea what it will track like.

As far as I know , if you want this luxury you have to build it at the present time.

If you can build the buck converter reliably you can make it track the alternator characteristics near enough by analog means, that doesn't use any form of MPPT tracking. Once set the alternator characteristic is fixed.

The converter is a challenge and I have previously proposed a scheme where the alternator is wound to track the high wind end and a boost converter is used for the low winds. This is not the ideal scheme, but it gives results with a comparatively small and simple converter. You can still have over 1kW reliably and still work in low winds. You could probably have 1.5kW safely in high winds if you were prepared to live with the necessary high speed and noise, but I would sensibly rate that 10ft machine at 1kW. That would cover the normal wind regime of most areas at good efficiency.

Not a lot of point in doing an AirX type rating for 3 hrs a year.

Flux

[Nando]

The software used by the MX60 is a bit RUSTY since the microprocessor has not been replaced with a newer one with much higher speed and capabilities.

The MPPT algorithm has a process that demands stopping the charging process to define the peak voltage to estimate a MPPT curve, which is time consuming.

A better process is to define ( using the PWM frequency and duty cycle) plus a forced step condition to observe the incoming supply voltage level that with the current determination levels can automatically direct the processor to maintain a superior and maximized MPPT curve without stopping the charging and by just defining the forced step condition period, which is different for Solar, Wind or Hydro, one has an ideal MPPT charger controller.

The voltage limitation of 150 volts ( the FETS may be 200 Volts rating) and some internal parts do not have the 200 volts limit force a voltage limit of 150 volts, still the MX60 does not have a survival circuit to protect itself from higher than limited upper voltage.

The power level limitations is due to the voltage ratio conversion, like 100 to 24 volts, efficiency is lower, maximum efficiency occurs when the input voltage is close to the output voltage.

One solution for the MX60 with high voltage supplies is to have a converter in series with the MX60 to bring the voltage to a MX60 best efficiency and good power transfer.

Like to have a converter from 400 volts to 100 volts and the MX60 to operate from 100 volts, this converter should have a support info for the MX60 type converter to have a good MPPT curve -- still MX60 parameter "hunting" by stopping PWM charging operation is not a good idea.

Nando

[BigBreaker]

Wind MPPT is different than solar MPPT and, in a way, easier.

For starters it is pretty easy to determine the optimum volts and amps the turbine should be producing for a given RPM.  It is also nicely self correcting.  If the turbine is over speeding due to light loading than the power out of the turbine will seem too low.  The increasing load necessary to correct this will reduce RPM and increase power.  It works to self correct in the other direction as well.  Also this matrix of RPM and power will only need to be calculated once and stored in a  table - no search necessary.  It is characteristic of a turbine, not the configuration and state of charge for the battery bank.

The next question is how to determine the best charging volts and amps for the battery.  This is the tricky battery charging problem that is never fully solved.  At a basic level though, you can present the battery a constant voltage and let it soak up the amps it can take.  A FET gated capacitor could provide a near constant voltage for that purpose.

Now the tricky part.  You need an input load determined by the turbine's matrix and then convert the volts and amps at that load into a constant voltage source for the batteries, and/or dump load.  Here you need a fancy switching system with feedback, not unlike a PC power supply (or it's DC equivalent).  Unlike a power supply you need the input load to vary with the turbine RPM.  As RPM increases the power supply can't clamp the volts

That's as far as I have gotten.  Nando, you have stated that field based conversions are needed rather than switched convertors at this power level.  I understand switched electronics but have only begun internalizing transformers and inductors.

Cuk convertors look like a good start.  I'd add FET based rectification and a integrated dump load while I was at it as well.

[willib]

flux writes..""The converter is a challenge and I have previously proposed a scheme where the alternator is wound to track the high wind end and a boost converter is used for the low winds. This is not the ideal scheme, but it gives results with a comparatively small and simple converter.""

what if you could design an alternator to track the middle band of wind ( charge the battery directly, no boost or pwm)

at the high end when the alternator is outputting more volts than the battery you switch to a simple pwm cicuit , that has two PWM circuits attached , one charges the battery the other dumps the excess power to a dump load(heaters)?

i know what you are going to say ...

with your scheme the alternator is in a failsafe mode if something goes wrong the alternator will not overspeed, right?

but we rely on electronics everyday , the electronic ignition in our cars sometimes fail but not that often ...

[Nando]

Well, it is interesting your statements, though your path for your algorithm is troublesome to do a good MPPT design.

Though the wind mill has parameters that can me measured and and stored as You say:

>Also this matrix of RPM and power will only need to be calculated once and stored in a table - no search necessary

With the proper algorithm this step is completely un-necessary.

CUK converters are not as efficient as a BUCK or Boost converter -- read literature and notice the efficiency penalties of the design.

The best charging voltage is defined by the battery itself and the selected charging algorithm, and it is not as tricky as You think.

FANCY SWITCHING, I do not understand what you mean by that, you need the same type of switching that any wide voltage range power supply uses, like the PC power supplies.

For effective MPPT, the power source MUST be allowed to move up / down and the charger to transfer the power, should have a ratio of at least 1,3 to 1 (input/out) for the MPPT to properly do its job.

A practical algorithm is one that can do its job using different power sources.

A charge controller that needs to define the parameters of RPM and voltage of the the mill, will have to be calibrated with a new wind mill system, this is an expensive process, so the solution is to have an algorithm that can take any power source like Solar, Wind or Hydro and just telling the charger what type, automatically the algorithm reference to the type is inserted in the general algorithm for good operation.

For this reason alone, the algorithm should be free of storing data for any new type of mill, solar or hydro.

Thee algorithm, already exists and the basic process is simple:

Define the input voltage by analyzing the PWM duty cycle.

Read the charging current.

Then vary the output voltage up and down to partially load and unload the wind mill, then observe the within the loading/unloading times the way that the duty cycle changes and how the current varies, the result is a bell curve that the microprocessor defines the direction of the variations to increase or decrease the charging current.

Once the battery voltage has raised to a reference, the next charging voltage profile is in placed continuing the charging and detecting the battery upper levels to start reducing the charging current and automatically turning on the Ballast controller section to keep the wind mill from over speeding.

Now the charger section: when the voltage of the mill is higher than the battery, the BUCK converter starts charging within the MPPT profile.

The operating input voltage range is defined by the parameters to cover, within the capabilities of the power source.

What when the power source is lower than the battery bank ?

Most of this type of considerations, the power may be low compared to the upper or peak end of the power source -- so a definition of using a simple BOOST converter with MPPT or add additional switching elements to attain the desired results using in this case the main BUCK charger which is operated as BOOST converter --

If I were to do it I would prefer to add the smaller BOOST converter.

Nando

[BigBreaker]

I like the approach of using an RPM table for input load determination because it minimizes search errors and avoids having to sense RMS current in a low frequency environment.  RPM can be sensed without feedback and without actually interacting with the wild AC coming off the turbine.  That feedback can come from the battery or turbine load reacting to the wind speed, direction, battery load, voltage conversion or "MPPT" charging.  A simple optical or magnetic sensor on the rotor will do.  Avoiding that feedback and not having to sense both tiny and huge voltage/current frequency is worth something.  It's also easy to set the matrix to stall the blades at high speed by raising the load to suboptimal levels.  Lookup tables don't break, they don't divide by zero, they don't need to converge.  They just work.

I don't have nearly the experience you do on conversion.  I'll take your comment that Cuk is a dead-end.  It does buck and boost simultaneously which I like.  I think the power conversion needs to be customized beyond those circuit topologies, though.  Somewhere, yet to be discovered, is a nifty circuit that will do wild AC to 12-48V DC and a dump load in one go, no rectifers.  Extra credit for 120AC/60Hz from either the wild AC or the batteries.

I'm picturing a triad of FET gated oscillators in three phase configuration as the "front end" conversion from wild AC to three phase 120V/60Hz.  You'd want poly phase so that the wild AC waveform always made a decent fit with one of the oscillators.  The oscillators become the source for all the system loads.  The loads would include a dump load, battery charging and mains conection running out.

A micro senses the wild AC and gates it to "kick" the oscillators to make up frequency, and voltage drops from the loads.  The micro can also pull up/down the loads based on turbine RPM to maintain optimum power.  Finally it would parse out power to the three end loads: battery, dump (likely a heater) and mains.  Alternatively the power battery can be run through an MSW inverter to drive the oscillators rather than draw off charging power.  This would be needed to service the mains when the turbine can't do it alone.

Am I completely loco?

[Nando]

LOCO !!, I do not dare to make an statement on that !!

You are trying to circumvent capabilities with a solution that may be highly borderline.

Current sensing is easy with an Allegro Hall effect IC, they have from 5 amps up to 200 amps, I use the 5 amps and the 100 amps for some projects.

Power conversion is just a process to define what to use and how to use it.

In your case the Voltage Range and the current as well, as the battery voltage.

High voltage, You do not read, You read the PWM duty cycle to define what the input voltage is - 12 to 15 volts driving the MosFet of the buck converter.

If you want to transfer power which is wild to a battery bank that is more or less stabilized -- THAT FORCES the need of using a power converter in this case a Buck converter.

Your triad of FET gated oscillators need a lot of more detailed information for me to give an opinion about the usefulness or practicability -- from what you say, it seems that is not good idea, but will wait for you to give a better detailed response.

For one it seems in-efficient

Nando

[Nando]

The MX60 MPPT technology is the old fashion type.

As I was told, in many cases the MX60 stop charging to read the voltage, since I have not used a MX60, I can not be more explicit, just the comments I receive.

The MX60 has too long time slots between MPPT point readings, which reduce the conversion efficiency and for systems not Solar, in many cases, is un-usable for Eolic ( Wind) or Hydro.

In one wind mill, the owner had to place a flywheel to stabilize the generator ( this is what He said) to be able to use the MPPT, limiting the voltage to 100 volts with a Ballast controller if the power was much higher -- he was charging 24 volts battery bank, so above 100 volts generated he was dividing the power into the battery bank and the ballast around 20 to 30 % of the time; he had good wind,

The ideal MPPT is the one that does continuous search while charging with variable slope, if using the Bell shape approach ( one of the best), the analog type is simple to perform, if using a microprocessor, the steps are a bit more complicated, since the PWM duty cycle and the current need to be defined within a time slot that may be constricted by the power source time constant ( time to produce detectable power changes).

The analog type can, most of the time, be implemented to any charge controller, it such is a PWM type.

The process is to generate a low frequency 50 % duty cycle to modulate the voltage feed back to the PWM,,(up/down) for a few % value, then taking the inverse of the PWM ( when the PWM is OFF duty cycle wise) and do a low path filter to present a voltage, then the current is sampled during the Up modulation, then stored for the down modulation, this 3 sources are summed to direct the setting of the feedback voltage of the PWM -- quite effective and one of the processes that acquire the highest MPPT harvesting levels.

Nando

[BigBreaker]

I know there are Hall sensors for the current but that doesn't solve the fact that the current is dependent on the load with a very tight feedback couple.  That isn't good at all for stability.

I think I understand what you are saying about the duty cycle.  My extension of a battery charging turbine uses a lot of FETs and PWM so I was initially confused.  You would use a hall sensor to drive a FET between the batteries and the rectifiers?  that makes total sense.

My idea about three phase oscillation has a simpler version.  Imagine a single phase turbine driving a FET gated inductor / capacitor pair in series to ground - no rectifier.  The oscillator is intended to hold 120V/60Hz.  A second FET connects the oscillator to the load.

Now the magic- those FETS are driven by a micro with some extra sensors.  The micro PWMs the FETs in order to load the turbine appropriately, care for the frequency, and waveform of the oscillators and allow a traditional mains load to pull power out of the oscillator.

[Nando]

To your statement:

>I know there are Hall sensors for the current but that doesn't solve the fact that the current is dependent on the load with a very tight feedback couple. That isn't good at all for stability

the Current is dependent on the load, OF COURSE, but It does not affect stability,

As a matter of fact it makes sure that the controller dose not over charge the battery bank.

If current does not exist, then the output voltage is set to the maximum and defined by the voltage feedback resistors.

Even with your idea, You will have to have the current to set the proper charging parameters, Some may use charging pulses with know joules levels, cumbersome and limited in usage.

I do not use the hall current detector to drive a FET, the summing of the parameters talked about earlier is used to set the PWM duty cycle and the charger voltage output, which is modulated continuously for the MPPT process.

Hard to envision when you say :

>Imagine a single phase turbine driving a FET gated inductor / capacitor pair in series to ground - no rectifier

The topology is not really indicated, -- a Schematic is needed to analyze properly what You mean.

Then the magic STARTS as You say, again a schematic is needed.

Nando

[dinges]

Looking forward to hearing how this 17-footer holds up, and whether the improvements make it into a reliable & durable genny.

Step by step...

[rpcancun]

3250 Watts at 200 rpm?

Wow!

Is that because the increase in magnet size?

I have 44 2x2x.50 mags, would it be better to make a 22 coil stator

or double up the mags and go with 11 coils with 2x2x1 mags?

Thx

Rob