Such a turbine is the Samrey range of turbines that are using 120 volts, 3 phase down the tower to the place of use.
http://www.samrey.co.uk/products/listcat/25
A friend is installing one over the summer in a site in Scotland and should show some interesting results.
A thought on Maximum Power Point Trackers (MPPT) for resistive loads, heat loads...
From what I have read on MPPT for heat loading, a diy'er could add an extra coil to the stator for the purpose of supplying voltage and current to a basic linear voltage detector. This detector could then add or take away loads as the voltage increased and decreased with wind speed, aiming to increase efficiency by stopping stall of the prop. I have a basic circuit in a Radio Shack book that could be adapted, it uses op-amps and the output goes to a LED bar, as voltage increase more LEDs come on, my thinking is replace the individual LEDs to relays, relays to loads and hey presto a simple load matching device for heat loads. Possibly better suited for controlling a dump load so that the prop does not stall causing the dump load to cut in and out due to immediate attachment of a single, maximum load.
Not sure if that makes much sense to any one, any thoughts.
RegardsJon M
www.otherpower.co.uk "I am certainly not a perfectionist, the thing is to know where you can take short c[ Parent ]
instead of driving an led one could drive a phototransistor or optocoupler so that the contolled element could switch either a logic circuit or drive a relay depending on the need.
the bar driver chips can be stacked to get 20,30 or more steps easy enough as well
so basically one could sense the high voltage generated, feed the bar driver chip and get stepped outputs to do whatever control one could dream up.
the bar driver chip can be set to cascade (lighting each led seperately or progressively, i would think the seperate mode would be most useful)
those chips are readily available and cheap, circuits abound for them and seems to me that one could be adapted to provide a useful control
for instance
i have no idea what a system might end up costing to produce for sale, but i am fairly convinced that a very versatile system could be designed relatively inexpensively for a limited production or one off product for ones personal use.
maybe one should just go for it and design/build it, perfect it, and then if there is a need to mass produce it, worry about how to do it with less parts and simplify the design?
there is just a buttload of surplus parts out there available, seems to me this is a project that has merit, if not for any other reason than it needs to be done in order to determine just how much more efficient the end result would be over that which is commonly done now where a turbine/alternator is built and connected directly to a battery.
the home built alternators have evolved to a very high state in my opinion, i am not sure one can get much more evolution without the use of some of these technologies.
i do know that these technologies have proven their worth in automotive alternators where a typical alternator (14volt) produces on average perhaps 1.5 kwatt at ~45% efficiency in its stock oem configuration being raised to as much as 4.5 kwatt and ~71% efficiency. the technology has moved both the output power capability and the efficiency radically upwards.
it might be that a windgenerator could benefit from the same or similar technologies and get higher output, or higher efficiency, or a broader power band? maybe some of each? and depending on the goal easily tailored/programmed to attain those goals most important to the individuals needs at a specific time and be able to alter to fit changing needs.
i am convinced of one thing, that being the ability to match the blades to the alternator and both to the load being supplied. that alone should have significant value to a large portion of those building these machines.
bob g[ Parent ]
something I have been kicking around for a number of years is the use of a bar/graph driver chip, the ones that drive an array of led's in something like 10 steps, which range can be set with a single resistor
I think that Amanda had already posted that concept:
But just getting the gist of what you're proposing; consider using an LM3914 like my dump load controller. This will give you 10 switched outputs from cut-in to full load
I can see that used with SS Relays with settable trigger points for a bank of separate load resistors, but I don't see it working to drive the voltage input of the PWM circuit in an optomised non-linear manner ...but if Amanda could elaborate more, then maybe I'll have a better understanding. I tried to search for details of Amanda's dump controller, but couldn't latch onto to anything?
Stew Corman from sunny Endicott [ Parent ]
http://www.fieldlines.com/story/2006/6/1/84528/29208
........oztulesFlinders Island Australia[ Parent ]
Whilst this would not be my first inclination, it is a do-able proposition, and would allow for non-linear responses to load conditions.... an analogue jump table so to speak.
You could parallel the outputs to load resistors and so effectively load up a higher impedance voltage circuit proportional to the values of the "resistors in parallel". ie load up a high impedance voltage circuit with the incremental resistances (loads) to ground via the switching transistors, and so proportionatly lower a control voltage as we switch in loads.
This would give you a control voltage to mess with the dead time control of say a tl494, or with the reference voltage for I sense for a current controlled device (4990 etc)(let the error amps run wide open and just control current) or even just invert it for the error amps in the voltage controlled and current controlled chip types.
This would certainly allow for programmed non-linear steps in an analogue pwm control system.... now I think about it, it would probably be one of the simpler ways to implement analogue control....
I might have to make one of these, its been an idea of mine since i first head about dump loads.
Watching this space....
It may be more useful to monitor the wind speed rather than the generator output as a source for the switching steps.
As you switch in different pwm lengths you effect the voltage output (dependent on system impedance) which is what you are sensing as your sample.... seems like a cat chasing it's tail.
If you make a simple unloaded wind sampler (tiny turbine with no load), and use that as your reference, you stand an excellent chance of trialling staged switching which will be independant of your sampling.
This should give a simple bullet proof jump table, which can be tuned to an individual turbine through trial and error, finding the best pwm rate per windspeed, and selecting the resistances accordingly. This should beat the elusive algorithym and sampling problems associated with digital learning curves which seem not to have been simply solved yet for wind turbines, rather than solar cells.... (which seems to have been solved quite well).
Digital running off wind speed should be simple as well I guess, this should be a simple jump table too.
It is when you effect your sample with your outputs that things seem to get sticky, and require tricky programming to overcome.
just a thought
..........oztulesFlinders Island Australia[ Parent ]
It may be more useful to monitor the wind speed rather than the generator output as a source for the switching steps. As you switch in different pwm lengths you effect the voltage output (dependent on system impedance) which is what you are sensing as your sample.... seems like a cat chasing it's tail
As you switch in different pwm lengths you effect the voltage output (dependent on system impedance) which is what you are sensing as your sample.... seems like a cat chasing it's tail
In effect , we are attempting to "seek" to the best performance level ...when the load is low at low WS, the Cp is low and corresponding RE#, L/D ...so it starts to speed up, since at "no load" a typical turbine will rotate at approx 2x TSR of loaded condition. As it speeds up, the RE# increases, L/D increases both from RE# and from improved AOA, Cp improves and the voltage is higher at the new rpm. Higher voltage across the a fixed resistance load is higher output power. There is some equilibrium rpm that balances the power output to the blade performance at each and every different WS presuming a steady wind. It will be evidenced by varied TSR measurements as a function of WS also. Increase the power too much and it bogs down the blades until a new equilibrium is met. Once full load is obtained ...you get what you get at higher WS.
That is not to say that the optimum PWM is necessarily a linear relationship with output voltage ...but it makes a decent first approximation. The key to implementing the PWM input is to determine at what WS the max design ouput is reached to deliver 100%PWM independent of furl setting parameter. Could be at 15mph, 17, 18.5 ....anywhere up to furl at say 25mph. If the performance curve is far from linear with WS, then the modified PM trigger with perhaps the 10 steps provided by Amanda's LM3914, could be a solution where each chip output is tweaked with a trim pot. Let's try KISS first. Setting 0% PWM setpoint is trivial ...no power at 7 or 8 mph anyway.
Any and all comments welcome
Stew Corman from sunny Endicott[ Parent ]
I still feel that measuring the power at it's source (WS) and making real world decisions (from pre tested points at varied wind speeds ) is going to be hard to mess up.
We want max power for a particular wind speed, no more, no less. When we measure the wind, twiddle the trimpots to get optimum power out, we don't need to know tsr, cp, or any other factor, just what is the best pulse width at a particular wind speed. This from testing is hard to beat . Test it at ten points, and it will represent a fair approximation of the best you can get out of the system at ten different wind speeds.... from whoa to go.
This would be as close to kiss as is possible, and still be fairly true.
.........oztulesFlinders Island Australia[ Parent ]
'It may be more useful to monitor the wind speed rather than the generator output as a source for the switching steps.'
I did explain that adding an extra coil in the stator would be a way of doing it, so it would be independent of the load.
That said,
Im now thinking that if the extra coil could be used to sample RPM by using the relationship between voltage of a coil and RPM. Then feeding the voltage into an op-amp that would, add load when the turbine went past designed TSR and took load away when it went into stall.
And keep the linear voltage detector for dump load controlling so that dump loads can be increased incrementally.
Im keen for the KISS principle to be used on a DIY MPPT and think any introduction of reference tables or algorithms may hinder this in the short term.
regards Jon M
Put simply, if we measure wind speed, do our figures and say we should get 1kw in the shaft at that speed, then we set the pwm width to give us 1kw out (less losses). no algorithms, just preset pulse width at 10 steps as tested.
In practice we could run the turbine, manually adjust the pwm to get the best power out at a particular wind speed, and that would be a point in our pwm 10 step table. Do this for 10 different wind speeds, and you have the jump table that the resistance steps will mimic. This should be simple to implement. just simple analogue.
I'm either plain wrong with this, or I am not explaining myself properly.
Put simply if we measure the wind, manually adjust the pwm for best results, and do it at 10 different wind speeds, surely we must have a workable (albeit notchy) mppt.
To keep it simple, we need to measure a fixed thing (instantaneous wind speed), and make a fixed decision.(pulse width from tested 10 points) It don't get simpler than that.
Trying to measure a dynamic relationship between volts, current, tsr, ws, blade loading etc, leads us to the present problems that no one seems to have solved easily...
