Design Considerations with MPPT

[kitestrings]

Greetings,

Neilho and I are planning our next wind collaborative.  The scope of this project has changed quite a bit due primarily to two factors: 1) the immergence of Midnite Solar “Classic”, and 2) plans to use a smaller tower (Rohn 25G).

I generally like the overall approach Dan, and others, have taken with his 17-footer.  I feel like I'm in the middle of Goldilocks here (not because of my hairline) but, the 17-footer is ‘too big'; the 12-footer is ‘too small', and I'm looking for the ‘just right'.  We're currently looking at something in the 14-15' rotor diameter, similar to some of Dan's earlier designs with some updates incorporated.  We'd like some early feedback on rotor/stator layout as I've been attempting to model both 12 and 16-pole (smaller magnet) configurations, and in particular, I'd like to explore what might change knowing that MPPT will be employed.

Here's my first cut at the 12-pole layout:

Rotor- 12-pole, dual
24 – 1.5” x 3.5” N42's

Stator- 9 coil, 3/phase, wye (star)
I'm assuming a cut-in of about 100 rpm, I'm coming up with about 82 turns to reach just under 54VDC as a target voltage. 
I know this isn't exact science, but I like to try to understand the math, so for anyone willing to offer a sanity check, here are the details:

Telsa (estd) = .7T
Tot flux/pole = 2.03 mWb
RMS /ph = 22.1 VAC
Line volts = 38.29 VAC
DC volts = 53.6 VDC
Freq = 10 Hz

I was further assuming single strand #14 gauge wire, which would put the coil resistance at approximately .22 ohms; or about 1.3 ohms phase to phase.  With normal axial flux losses, and an assumed constant TSR of 6.5 – and no MPPT – I'm guessing it might look like this:

So we're interested in general feedback.  Specifically, I'm wondering what we should consider differently knowing up front that we'll likely use the Classic.

Here's where it gets interesting – with the Classic managing the upper limit Voc is reportedly critical, hence the need for a ‘clipper' or reliable load diversion.  Halfcrazy suggests a 6-7 volt (higher) differential to the input battery voltage allows the Classic a bit of room to maneuver, but the unit also reportedly can boost as well as buck.  So I'm wondering why you wouldn't move in the other direction (lower cut-in)?

I know there are a lot of variables, and we're used to tweaking things based on real-world results, but we appreciate the collective input and expertise here.  Regards,

~kitestrings

[Flux]

I don't know enough about the classic to answer all your questions but I can answer some of them from my own mppt experience.

The mppt removes a lot of restrictions forced on you by direct battery charging. You can go for a slightly lower wind speed at cut in without paying the penalties in high wind. For direct connection I would aim for something rather higher than design tsr at 7mph or you will hit stall far too early. With mppt you can cut in at a lower speed and start producing power from something like 5mph upwards with a bigger machine.

Not sure what your intended tsr is but cut in lower than 100 rpm may be possible to advantage.

You are no longer bogged down with the stall issues so the higher the alternator efficiency the better it will work. If you can afford to throw more magnets at it it will work better. With magnets being a one off cost and the classic not being cheap I wouldn't be afraid to use plenty of magnet. You can use the full alternator efficiency with no penalty. Similarly you don't get any benefit from line resistance so again you mustn't skimp there ( it does make things rather expensive but it is a one off cost)

The maximum voltage is absolute and that is the point where I have no knowledge. If it can boost as well as buck then there is a great deal in favour of using the lowest possible cut in voltage. Remember with mppt you will be tracking prop speed directly with wind speed so you can expect something like 4 or 5 times cut in volts at full power at furling point. If you should loose load then expect about an 80% overspeed . You will struggle to keep within the voltage limit off load in a decent wind. Any attempt to raise the cut in volts will make this problem far worse and at best all you can hope to gain is a bit less line loss.

If you run anywhere near the maximum voltage then a clipper will be virtually essential, voltage sensitive relays and dumping aren't likely to be precise enough unless you stay very safely below maximum volts.

This is one area where I haven't done much work, My unit isn't likely to survive an open circuit and we try to make sure it doesn't happen, so far it hasn't but I have no idea how the classic works, I assume it doesn't go open circuit like various solar mppt units.

I anticipate a precise clipper will be nearly as complex and costly as the mppt converter.

I think that if you get it right you will see a fair improvement in light winds even though conventional machines are believed to be efficient there. The improvement in higher winds will be dramatic to the point where if you are in a good wind area you may not need to go as big as 15ft.

Keep the alternator efficiency as high as possible and stator heating will be no issue at all, you will have to furl to keep within the classic rating not because of stator limitations. I see no reason why you can't still reach over 80% alternator efficiency at full load and you can do this with the prop producing a cp of 0.35 or more.

I see over 1kW from a 6ft prop and the limit is my input voltage to the converter. It would probably do 1.5kW if I pushed my luck and I didn't make much attempt to build an efficient alternator. With 15ft I am sure your only limit will be the classic.

Good luck, keep us posted on how well it performs you should see a vast improvement, you will have to plan your load schedule carefully to see the full benefit otherwise you will be dumping lots of power.

Flux

[Rob Beckers]

Kitestrings, we've been using Omron measurement relays (driving a contactor) to protect inverters from overvoltage in case the other means (dump load etc.) can't keep it under control. So far this has proven very reliable. These relays are robust. I have attached a drawing showing the principle. It is fail-safe, in that when the power goes down (to the Omron and/or contactor) it will disengage the inverter.

In your case you have a charge controller, the principle is the same though. The Omron has 3 different input ranges; the version shown can do 20 .. 200V, 30 .. 300V, and 60 .. 600V, just by selecting another input. Either AC or DC. Once set it's pretty precise and repeatable, and it will disconnect in 100ms (fast enough to catch a turbine that's speeding up unloaded). We normally use it on the 3-phase AC side, since disconnecting AC with a contactor is a lot cheaper than doing it for DC (under load). The Omron version shown is for 120V, there are also 24V DC version of the same thing.

Hope this helps!

-Rob-

[imsmooth]

I use the Aurora MPPT inverter 3.6.  I built my own shunt controller.

http://www.mindchallenger.com/wind/wgpage16.html

It is self-powering, so it does not use any mains power.  It is set to start dumping power into a non-inductive resistor load at a pre-set voltage.  Unlike a contactor, it will still allow you to draw power from the turbine even as it dumps excess power to keep the voltage below the inverter maximum.

[kitestrings]

Thanks for the responses.  I regret taking so long to follow up.  Trying to sugar (maple season), but we got buried in heavy snow (30”) and high winds here.  Just now getting back to ‘normal'.

Eventually a matching Clipper from Midnite Solar will be available, and in the interim some folks seem to be finding workable diversion, or shorting schemes on the AC side to protect the Classic:

http://midnitesolar.com/smf_forum/index.php?topic=7.0

Our overall intent has been to better leverage the low to mid-range velocity, but more frequent winds that “pay the bills” so to speak.  I'd prefer to maintain load at all times, furl early and reliably and divert to a load outside of the stator.  Even with sophisticated electronics there's no substitute an increase in wind-wept area.  Still based on early feedback with the Classic, maybe we're a bit too big.  We'd like to stay within a comfortable working range of the Classic.

Flux, you make some compelling comments to consider:

Higher design TSR at cut-in (I think Hugh suggests ~8.5)
Lower cut-in rpm target

I follow that there will be little or no value to line resistance, so I interpret this to mean we should use as much copper as space will permit – perhaps going to #13, #12, or an equivalent pair?

I'm a bit confused by your statement to “throw more magnets at it.”  We were planning 1.5”x3”x.75” N42's in a 12-pole format.  We're you thinking we'd be better to go to 16-pole?  There are examples of 12-footers with 16-pole designs; but also 14-15 foot units with 12-pole alternators.  There are known trades-offs in rotor size, weight, magnet cost, but as you say it is a one-off cost (and the heart of alternator).

Your system uses a boost converter in low winds as I recall.  I'm curious in your experience what the practical limits of this are (converter input vs. battery voltage)?  Also, at 6' and 1 kW I would have thought you would have been well off the converter, or is the gain in the lower overall resistance that this affords you?

Rob,

Thanks for the hardware tip and schematic.  I think we'd be more likely to close in on a dump load for protection.  With the relay as shown it protect the Classic, but would unload the turbine – further complicating things there.

[Flux]

With the classic you are able to use the highest alternator efficiency. With 12 magnets things are good and will be much better than with direct connection but if you want to go all the way you could use 16 magnets and make the alternator even more efficient. You reach a point of diminishing returns and you may not feel inclined to spend the extra money for a small increase in mid wind output, but there will be some amount of gain if you do it, without having to throw it all away in some other way with the direct connection.

I am sorry I didn't make my last point clear, I only gave details of a machine with boost converter in the matching the load article as I hadn't had anything else running then and even that was I though too complex for most people to follow.

I have since built a 6ft machine with a buck converter, which operates more like the Classic and that was the scheme I was basing this information on. I would have seen benefit from using an even more efficient alternator than the one I used.

Flux

[kitestrings]

Okay.  One of the limitations is availablity of magnets in a size that would be optimum.  We could just go with 16 of the same dimension, adjusting the coil trurns, and acknowledge that we're overbuilt for the rotor size (14-15' say).  I assume the Classis will allow us to optimize loading in lifghter winds where we would otherwise stall miserably with a direct connected scheme.

This was a 16-pole layout that I was playing with awhile back:



Otherwise, the next closest choice would probably be 1 x 3 or 1 x 4.  Maybe statying a bit thicker, and sqeezing down the magnet rotor diameter it could work?  I guess we'll have to look at this layout.

Sticking with a 12-pole configuration, I assume the factors that would move us to the most effient alternator would be thicker magnets, or smaller gap, and thickest wire that will fit.  The lower cut-in will reduce the turns a bit, and that should help.  A 12-pole has one other distinction - it allows you to estimate the rotor speed by simply reading the frquency and shifting the decimal point (10 Hz = 100 rpm in my original post).

Thank you again for your help.  We're anxious to get started.

~kitestrings

[Flux]

Your 16 pole layout looks fine, you have space between coils and for any design I would fill it with copper but otherwise it's fine.

You could also manage with 12 poles by using thicker or bigger magnets. Even if you don't change anything you will be well ahead with the classic so I am sure you will come up with something. you have a bit more flexibility with voltage using the classic so you could add a few extra turns to make the coils fit together rather than having to change the wire size to keep the ideal number of turns for direct connection.

For a digital display you have a point about 12 poles for frequency, I prefer analogue displays for wind rather than a load of ranting digits so 16 poles with a 2917 tacho chip and analogue meter works for me, but you do have a point there.

For 12 pole you will benefit from thicker magnets but the best gap is about 1.5 times magnet thickness. If you use smaller gaps you get a higher flux density that lets you use less turns, but you loose some of the valuable space needed for thicker wire. There is a range of gaps over which it doesn't make a lot of difference but gaps too wide for the magnet loose you flux due to leakage and gaps too small waste winding space and the necessary mechanical gap for safe working stays much the same and it becomes too large a proportion of the total gap.

For the old type steel magnets with non linear demagnetising curve you lost out badly if you didn't work the magnet at BH max. This is far less the case with neo but the other factors I mentioned are really pointing to the fact that the best point is still to work your magnet at near BH max. For neo this means aiming for a flux density in the gap of 600 - 700mT. In theory this occurs when the gap is the same as total magnet thickness but for low leakage it works better to keep about 1.5 times one magnet thickness.

I am sure you are on the right lines, keep us posted on the results.

Flux

[kitestrings]

Thanks.  One final question (well okay there could be more ;-).  I recalll reading a post about the efficiency of power through the rectifier - we'll likely do post mounted diodes, as apposed to a bridge - but as I recall the efficiency there is inluenced in part by the frequency of switching.  It may be a non-issue with higher voltages using the Classic, but does this influence the descision one way or another?

[Flux]

I think I remember the discussion you are on about but I couldn't follow any of it at the time. You are looking at very low frequencies and I see no reason why conventional power rectifiers shouldn't be fine. You can in some cases get faster grades in sensible packages but normally fast grades come in dreadful packages suited to switch mode power supplies where other factors make stud mount near impossible.

We did manage to obtain large power diodes in dual sided compression packages ( Hockey pucks I think you call them) for inverter welders but they cost the earth. I really think that the frequency is far too low for there to be any issue with standard devices. I certainly have found decent stud mount devices to be way better than the potted bridge things that always seem to run hot and need much bigger heatsinks.

Flux

[Rob Beckers]

Rob,

Thanks for the hardware tip and schematic.  I think we'd be more likely to close in on a dump load for protection.  With the relay as shown it protect the Classic, but would unload the turbine – further complicating things there.

Actually, you'd use it in addition to a dump load. Without a really stiff load, such as batteries, it's hard to keep turbines under control with just a dump load (a fine line between putting on enough load to drive it to stall, vs. not burning up the alternator). Before you say that won't happen; I've seen it more than once, burning up inverters in the process. The relay is a measure of last resort to keep the expensive electronic toys alive...

-Rob-

[kitestrings]

thanks Rob,

Your point is well taken.  I understand now that you're using it as a fail-safe device.

We've lived with wind, off-grid, for over 25-years now.  And, we've installed and repaired countless others.  Anything that can happen, will - it's got to be one of Murphy's laws.

Some folks seem to be quick to short the windings as a method of controlling speed (e.g. the homemade clipper refered to earlier), or to shut down the unit.  I fundementally dislike it.  It puts stress and heat where we're otherwise working hard to limit it.  For shut-down a mechanical furling device is much simplier and effect IMHO.  Others may disagree.

One unrelated footnote -

I did receive a response from Halfcrazy.  He clarified that the Classic cannot buck and boost in the same application.  He recommends winding for about 10 volts above the the nominal battery voltage, or ~58-60VDC in our case.  I guess we'll set our sites on that.

Thanks for all the contributions.  I hope to post progress reports soon.

~kitestrings

[bob g]

i am curious, why with an mppt would one want to limit to only 10volts above battery nominal voltage?

it would seem to me that one would be better served, to run as high a voltage as the mppt can safely handle, and let the buck converter section do its matching magic.

the efficiency of the alternator would be increased substantially, as well as the overall system efficiency, would it not?  (not sure why i am asking this, actually the alternators efficiency would be significantly higher, which leads to higher overall efficiency)

i would lean toward not only the highest voltage possible, but do it with the fewest turns of wire possible, and with multistrand (more than two in hand) winding procedures.

this mppt thing is something that has been sorely needed in the wind generation game for a very long time, it is nice to see that finally folks are starting to embrace it, and over time learn how best to optimize their alternator designs to take best advantage of the technology.

in other words, i like it!

finally the holy grail, an electronic transmission!

bob g

[Flux]

Bob, I take it that this is the cut in voltage and that seems sensible for a reasonably charged battery and the converter fully phased forward.

At 4 times cut in wind speed it will be up to about 300V. i am not sure of the absolute voltage limit, but i wouldn't want to design for anything close to the limit as it will go way up on this if it goes open circuit.

Even working at 300v at full load the efficiency will be way above a direct connected machine bogged down to 60V.

Flux

[bob g]

ahhh, cut in voltage!

now that makes sense, for some reason i was thinking that the 10 volt over the battery nominal was the design goal for the alternator at its normal working parameter (mean windspeed/tsr)

thanks for setting me straight on that Flux



bob g