New Generator

[SparWeb]

Hi gang,

August has been a busy month for my RE projects.  I was going to post several things, but if I tried to post 4 simultaneous threads, I bet you guys would start accusing me of being a forum-hog!  There are some things worth sharing, though, so some blurb and photos...

After only 2 weeks running the first generator on my windmill, (axial-flux, cast-stator, dual rotor), I managed to burn it up in a moderate amount of wind.  The tail should have been furling, but I had too much weight on it so I got what I deserved.  The 8-foot diameter prop didn't help - too much input power for too little output power.  There was a recent thread about a series of burned-out stators, in which a combination of resistance heat and friction heat combined to flex the stator out of plane and the subsequent overheating just ran away with itself.  This is exactly the same situation I had, though at a smaller scale.  I seem to love repeating other people's mistakes.  I found that no particular phase of the stator was affected.  The damage was concentrated in one quadrant of the circle.  High current (approx 20A in 18 gauge wire) softened the stator, and once it got hot, it expanded to one side, rubbing the rotor, causing friction, producing more heat.  The resistance of all phases has risen by 50%, so the wire was overheated, too.

Fortunately, I had a 3 HP motor-conversion just sitting around waiting to be used.  It was a "winter project" last year, and last week I built the mount to get it on the tower.  This motor-conversion has not undergone any testing (other than a little "cogging torque" test) so I don't really know anything about it.  I re-used the 8-foot prop, which is probably a bit small.  Tonight is the first serious amount of wind it's seen since being raised 2 days ago.  It's furling in the 30 mph wind, turning at about 300 RPM, and producing up to 12 Amps.  It's after dark, and this wind came up suddenly, so that's all the data I can get for now.

Last weekend I came across another 3-HP motor, this one a Baldor very similar to the one I'm currently flying, but much lighter, so if I choose to convert it, this may make a good "second generation" replacement.  And, as I've already proven, it's good to have a backup!

My other discovery is a priceless array of VRLA batteries, apparently victim of age, not abuse, still with a charge in them!  1760 Amp-hours!  I was just beside myself when I realized what I'd found.  I was then confronted with the challenge of moving these behemoths!  450 pounds each - oof!  Fortunately they were dumped 1/2 block from the building where I work.  You should have seen me driving down the street in the shop's forklift.  At a blistering 5 kph!  They are now safe in the back of the warehouse waiting for me to figure out how to get them back home...

I estimate I have about 1 week before my little bank of Trojans is topped off by the new motor-conversion generator, and I'll need the new battery set...

Lastly, I've been trying to measure RPM with the "HZ" setting of my multi-meter.  This has not been very successful, and the solution I was going to try for the axial generator became moot when the stator died.  So now I'm back at square one with this new generator, and the problem may be even more intractable with this one.  Say the mill is turning at 300 RPM.  Since this is a 4-pole motor-conversion, the expected frequency is 100/60*4/2=10 Hertz.  Not very high.  The meter seems to be reading some harmonics, because it jumps between 10, then 20, then 15 Hz.  Can this be "filtered"?  Do others see this, too?  Has anyone tried switching in some appropriately-size capacitors (or inductors) across the phases to clean that up?  I have built in a switch that allows me to "dump" the current; I could also rig the switch to "filter" so that the filter circuit would only run momentarily while I take a speed reading.

If none of this works, I'm faced with wiring up a LM2907 tacho chip or hooking up a bike speedometer and running another wire down the tower.  More complicated but I'm already committed to putting a temperature sensor on the generator, so what's one more little wire?

This is so much fun...

Steve

[Flux]

The problem measuring frequency on the ac side of an alternator generating power is not unique to you, I find it amazing that others don't seem to have a problem, I question a lot of the answers.

Usually a filter consisting of a 47k series resistor and a shunt .22uF capacitor will clean up the waveform enough. The 2917 chip is an absolute pig in this respect and requires careful filtering. When in the presence of a pwm buck converter chopping the dc it becomes a real challenge.

Feeding the input from an opto isolator seems to work wonders but you need to watch the opto diode rating. Connect a reverse diode across the opto one and choose a series resistor to keep you to the maximum diode current. Connect a resistor in series with the opto transistor and capacitively couple the square wave out to the 2917 input pin.

Flux

[SparWeb]

"When in the presence of a pwm buck converter chopping the dc..."

That doesn't bode well, then.  That's on my horizon.

I already sized and tried a resistor-capacitor filter, really close to what you describe, Flux, which helped only marginally.  That was on the axial flux genny 2 months ago, which was a 16 pole device, so the frequencies, around 30 Hz, were more manageable.  This motor conversion runs up to only 10 Hz.

Time for a reed-switch, I suppose...

[Flux]

Multimeters are very variable in their nature, I have never seriously used them for frequency. I have a frequency meter with germanium transistors that was virtually a prototype and it is much better natured than the present things designed to be happy at Mhz.

I think you will have real fun at 10 Hz or below. The 2917 will work perfectly well but the ripple on the output will be dreadful. It should drive an analogue meter but trying to make any sense of it on a digital will be a challenge if you want a finite speed of response. Reed switches will probably be worse at even lower frequency.

If you can live with additional leads up the tower then an encoder using hall effect or opto with a multi tooth chopper or gearwheel would seem a better bet.

Life is never easy with wind power.

Flux

[Drives]

In my experience, my Fluke 87 has a real problem with reading frequency from a variable voltage/distorted waveform, but does read frequencies accurate down to a couple of Hertz.

Rather than reading the frequency from the voltage waveform, why not read it from the current waveform?

What if you connected a shunt between the rectfier bridge, and battery.  Then place your multimeter across the shunt, and read the current pulses for frequency?

This won't work if you are below cut in, but may prove useful once generating power into the battery.

The shunt will have to drop enough voltage for the meter to trigger, say .5 Volts at full scale?

Otherwise, feed gennie voltage to a high value resistor (10K?) in series with a single phase bridge rectifier.  Connect the output to a zener diode (12VDC?) connected to clamp at its rated voltage.  Hook the multimeter across the zener diode, and take your frequency reading.  Again, it won't work well at low voltage, but should read accurate once the zener is in conduction.  May need some filtering.

I am in my office for the first time in over a month, and had some random thoughts during lunch.

[Flux]

Dean

I did wonder about suggesting using the current ripple to measure frequency. It would be fine for single phase. For 3 phase it will be a 6 pulse ripple as long as the phases are exactly balanced. If things are not well balanced it may get very confusing unless there is some sort of dc restorer circuit to restore the mean level of the ripple.

We used to check for failed diodes on brushless alternators by looking at the exciter field ripple. it was easy to see a fault on a scope but devising a circuit to show it automatically was rather a challenge.

At least the 6 pulse ripple would make it far easier than the low fundamental frequency of the motor conversion.

Flux

[SparWeb]

Some great ideas.  The ultimate goal is to have a reliable speed measurement for performance calculations, and possibly to trigger star/delta switches.  I'm going to try a higher capacitance value on the filter, but failing that, I would prefer an encoder disk or toothed wheel of some sort.  There is lots of room inside the motor housing to fit a disk on the shaft.

Thanks!

[Chagrin]

I never head a problem using a multimeter to measure frequency -- a cheapo to boot. I did have the benefit of a oscilloscope during my construction, however, and was able to get everything set up to where I had a pretty clean waveform.

Oscilloscopes are quite enlightening as to how coil spacing / magnet spacing affects your waveform. I'd highly recommend rigging one up.

[TomW]

Chagrin;

Yes, very useful device for actually seeing what is happening. No getting fooled by funky waveforms confusing a cheap meter.

For this stuff, even an ultra cheap, low bandwidth 'scope be fine.

I strongly suggest than anyone with access to a 'scope and one of these dual rotor units get them together for some insight into whats going on with the waveforms. Interesting stuff

Cheers.

TomW

[Countryboy]

I'd be finding an engine hoist.  Last I knew, you could buy a cheapie cherry picker hoist for about $200 brand new.  They are usually good for 2000 pounds, and can lift one of those batteries high enough to load it in a truck bed.

That'd be an easy way to move the batteries.

Either that or find a heavy duty dolly cart - the piano dolly carts for moving pianos are about indestructible.

[RP]

Being 4 pole I "think" that would make it equivalent to an 8 cylinder engine.  What about a tachometer for a car?  They definately work with messy waveforms.  

(Thinking out loud here):  A half wave rectifier (single diode) from one phase should give 4 pulses/revolution.  This should be the same a 8 pulses / 2 revolutions.  AN automotive tach has to deal with the inductive kickback, ringing and sparkgap effects on the wave form.

One problem is it would have to a digital tach to be able to read the low rpms.  On the other hand if this is a 3 phase conversion and you use a full bridge on all 3 phases then you'd have 18 times the number of pulses (is that right?) and could make your own dial for an analog tach.

[RP]

"A half wave rectifier (single diode) from one phase should give 4 pulses/revolution."

Dang!  No, it would give 2 pulses in half wave mode.  So it would be equivalent to a four cylinder engine.  You'd need a full wave bridge on a single phase to be like a V8.

So all the other speculative math is off also...

[BigBreaker]

I'll reiterate my view that a single small magnet on the rotor and a small stationary coil pickup solves all these problems.  It also creates COMPLETE isolation from the stator currents.

ICs do not like noisy, high power signals.  You can opto-isolate, cap shunt, filter, etc all you want.  Nasty currents and noise can still punch through and ruine the entire control system.

[Ungrounded Lightning Rod]

Or drop down to your loal hunting supply or sporting-goods department and get a "deer hoist".  Those are a lightweight and inexpensive block-and-tackle arrangement that can handle a thousand pounds or so.  They're typically 4-to-1 but you can combine a couple of 'em for a better ratio.

Back in my hand-to-mouth days in the '70s I used two deer hoists, two sawhorse brackets, and some 2x4s to improvise an engine hoist to short-block a V8 Ford Galaxie.  Built a 20 foot sawhorse (with a-frame braces across each end and triangle braces from each leg to the crossbar), hung it half-over the front of the car (as if the car had driven into one end of the sawhorse), hung one deer hoist over the engine in-place and the other over the space in front of the car to land/launch the shortblocks.  Three-stage drill was lift with one hoist, "tarzan" to the other, then lower into place.

Worked like a champ with myself and my girlfriend as the only labor.  (Even when the cat decided to watch the proceedings from the a-frame of the car as the block was going in, leaving the girlfriend holding the block while I chased the cat out.  B-)  )

[SparWeb]

Still, I like your line of thinking...  Get a generic auto tach and "fool" it somehow.

[SparWeb]

For that, I'm equipping the tractor with a round-bale fork.  I'm custom making it, to get the tines spaced just right.  They will be spaced to fit under the mounting rails on the batteries, and spear the round bales, just fine.

Isn't it neat when two needs can be satisfied by one design?

[TomW]

Spar;

Indeed, multitaskers are great. The wife pretty much insists on kitchen tools being capable of multiple tasks to earn space. Nothing finer than using what you have to do what needs done.

Cheers.

TomW

[SparWeb]

You mean using the steak knives to open jars?

My wife prefers to use my pliers to turn bolt heads.  This is a self-reinforcing habit, as the bolt heads get rounder and rounder...

[SparWeb]

Following the previous discussion, I picked up some cheap auto instruments at Princess Auto.  I tried out the tachometer on the AC output and that seemed to work once above 50 RPM or so.  The gauge has a "4-6-8" switch on the back, hopefully one of those settings will be appropriate.  "4", I expect.  2 bangs per revolution.

I also got a temperature gauge and I think I'll be able to make that work, too.  

One concern with the temperature gauge will be ensuring common grounds.  The way these gauges are wired, the sensor has a voltage applied with respect to ground, but if it is 100 feet away from the read-out and power supply, then there can be some difference between each point's perception of "0 volts".