Idea for simple MPPT heating controller

[elt]

Hi all,

With lots of help from the folks here (thank you!) I made a PWM dump controller for my mill. (here)

I read recently of folks looking for heating controllers and I've been wondering what the difference is between a dump controller and heating controller are ...I don't see the need for a different (digitally controlled) circuit but I am not an Electrical Engineer.

As is, the micro in the dump controller monitors the voltage and increases the duty cycle on the PWM to keep the battery voltage from rising above a certain point.

With the same hardware and just reprogramming the micro the same board can work as an MPPT heating controller? The power function would be voltage (from the ADC) squared times the PWM duty cycle. Maximizing that value by varying the PWM then achieves the maximum power point and dithering the PWM to stay on top of the curve does the tracking.

I'm not saying that programming he MPPT for wind is trivial; I know better. I'm just thinking that the same digital controller circuit should work for heating as well. Actually, the voltage ratings on some of the components need to higher but couldn't the same circuit work?

To get more current, can you just add another standalone controller or should one be slaved to the other synchronize the PWM?

Thank you for your help!

 - Ed.

[Ungrounded Lightning Rod]

With the same hardware and just reprogramming the micro the same board can work as an MPPT heating controller?

Should work if you replace the battery with a capacitor so the mill speed and voltage can vary and the cap smooths the voltage for the ADC.

The power function would be voltage (from the ADC) squared times the PWM duty cycle.

Yep, that's a good approximation of voltage times input current, provided the inductor is running all the way down to zero current between cycles.

Maximizing that value by varying the PWM then achieves the maximum power point and dithering the PWM to stay on top of the curve does the tracking.

Note that the dithering must be VERY slow - many seconds per cycle.  The "impedance" you're trying to "match" is the torque/RPM/angle-of-attack relationship, not the electrical impedance.  So the mill must be allowed to change speed and stabilize as the dither cycle progresses to get a good reading of the actual power for a given load.

[Ungrounded Lightning Rod]

To get more current, can you just add another standalone controller or should one be slaved to the other synchronize the PWM?

I'd say slave 'em.  Otherwise they'll separately try to adjust the mill's speed to identify the max power point and thus confuse each other.

[Flux]

Should work fine. Heating shouldn't be difficult, it is inherently easier than battery charging in many ways.

You will need to hold the load off to get the thing to start up ( mill volts will get you that). You basically want volts to rise with wind speed and load to track speed cubed. Left to its own devices a resistive load will load with a square law so you need to lower effective resistance as the speed comes up, otherwise the prop will run away if you start with ideal load at cut in.

A basic track will be fine. If you can get the thing to dither and track true mppt then that is a bonus but wind is so variable that I am not sure you will be able to instantaneously track it. I like the idea of programming a basic curve that matches the load near enough, if you can get it to learn a better curve then all well and good.

A very rough curve will do, props are very tolerant of a bit of change in tsr

Flux

[elt]

> Note that the dithering must be VERY slow

Well I think that's the trouble that I've had with my MPPT booster board. Slow and small steps mean that the mill might run away if you load too slowly or stop if you unload too slowly... I think I need something smarter than a simple dither, something proportional to the change in speed (volts.) I want to try a PID algorithm but haven't gone that far yet.

>> To get more current,[...]

> I'd say slave 'em. Otherwise they'll separately try to adjust the mill's speed

That makes sense.

Thank you very much.

 - Ed.

[domwild]

Elt,

Thanks for that. Been following the MPPT threads for a long time and find it intellectually stimulating.

From memory, didn't you change your MOSFET driver setup on DumpBoard 319as.gif to a simplified driver setup by using a single driver chip?? I am most likely wrong.

Thanks.

[Flux]

I don't think you can track wind with a scheme that relies on continuous tracking alone. Sooner or later it will get so far off track that it will stall out or run away.

I see the way forward as being a control that confines the thing to a load curve that keeps the alternator in complete control. The track doesn't have to be very good to do this and a programme with 2 or 3 points will be good enough. If you get those points right then I doubt that you will see any benefit from dithering or any attempt at peak tracking. If your points are a fair bit off you should be able to learn a better curve and substitute that and once it had learned the correct curve it won't need to change it unless something changes. For heating there is no need to change once you got it right. For battery charging you may see a little benefit from updating the curve for changes in battery voltage but even then a good mean value would be so near that you would probably not see any improvement.

Solar is a different thing. Wind machines are tied to a specific relationship between prop rpm and wind speed. Once you have settled this relationship it doesn't need altering continuously. A proper tracking scheme is just an easy way to programme a load curve without having to get the points anyway near right in the first place.

Flux

[elt]

No, you're right. I did use the Microchip driver on the last build; I thought that the original post had more info on the circuit so I pointed to that one.

But I'm looking at something else even simpler. The TD220 has FET driver and two regulators in one 8 pin device. The $2 cost is about the same as the three chips and less if you have to by a high voltage LM317HV for use on 48 volt mills. (48 volts is appealing for heating because you get twice the watts per controller.)

I'll personally stick to 24 volts for now. My mill is for standby/backup operation but I figure that while it's standing by, it could be making heat for me. But it won't generate so much power that I'll need more than one controller and I'm already invested in 24v inverters. When I get to the point of wanting to test high current, slaved controllers I'll most likely just offer to give two to someone already doing heating and able to use them.

 - Ed.

[elt]

Hi Flux, I've been struggling trying to understand this ever since you wrote:

> Left to its own devices a resistive load will load with a square law

My issue is that if the mill was running at MPPT and the wind increased, how do you make a loading adjustment to get back at MPPT?

I'm guessing it would work this way:

Suppose that the mill is in a steady state with a constant wind and (assume for this example that) it as loaded at the MPPT value making speed-cubed power.

If the wind speed increases but the loading hasn't changed (yet) then the blade will attempt to move toward the speed-squared curve. Voltage will go up so the power will go up however it will be making less power than optimum. (The blades will be overspeeding.)

Now if I loaded up the blades by changing the PWM based on the new voltage, that would be too much of a load (since the blades are spinning faster than their MPPT RPM) and the mill would stall.

So the loading adjustment once the mill moves off from the MPPT point is not based on the new speed-cubed but rather the old speed-cubed plus the delta-speed squared?

Is that right?

You wrote:

> I like the idea of programming  a basic curve that matches the load near enough,

I'm with you there. I'm willing to call MPPT "more power tracking" than if you did nothing and I'll be happy with whatever extra I get.

Thank you very much,

 - Ed.