Direct heating controller for wind

[Janne]

Hi,

Lately I've been working with a direct heating controller to dump power from a wind turbine directly into a resistor bank. This controller is being built for a 240V 3kW exmork wind machine. The principle behing the controller is to use a RPM / power table to control the pulse width going to the load resistor.



The wind machine was coupled to an 11kW induction motor, which is being driven by a VFD. This setup was first used to gather voltage / speed / frequency data, since the information did not seem to be easily available. The generator is 8 pole, and unloaded AC voltage between phases is pretty much spot on 1V / rpm. Currently the setup is being used to load test the controller. Water heating elements visible in the background are used for testing.



(Larger version of drawing; http://pics.ww.com/v/Janne/Electronics/P1060168.JPG.html)

Here's the functional blocks of the controller. First off, the 3-phase AC is rectified, and filtered to a pair of capacitors. The DC voltage is then chopped with the FET, L1 is used to smoothen the current / voltage transition speed. Switching frequency is 20kHz. The load resistor is sized so, that with full duty it should load the turbine more than sufficiently at rated speed. Assuming of course, that my calculations with efficiency etc. hold water  
A picaxe microcontroller is used for determinig the duty cycle of the resistor. The power for the micro is derived directly from the main DC-bank, with the help of a 200VDC linear regulator + a swithed mode PSU that will accept DC input voltage. Speed input is derived from the generator's AC frequency, with the help of an optocoupler. The pulse width for the PWM generator is then loaded from a lookup table, which has been pre-calculated with an excel spreadsheet. The control part is pretty much the same as in my MPPT controller(which for some reason is still yet to be installed. Summer is clearly too short ). Diary about that can be found here; http://fieldlines.com/board/index.php/topic,130334.0.html



The controller itself. If it's home built, it's supposed to look like it, right? . The cover also has an LCD display to show rotor rpm duty cycle, and mode of operation. The switch selects between manual and automatic control, automatic mode works with the lookup table and manual mode just takes input from the potentiometer also located in the cover.

So far the controller is working ok in the test bench with a few problems. The inductor needs to of larger physical size, it's getting too hot in operation. I only added it afterwards from my leftover parts bin, initially I tried to get away without any filter inductor, which was a bad idea to start with. The electrical interference was horrible. Same story with the diode, now with the added inductor the diode also gets to do some work = the small UF4007 diode is getting overwhelmed. The isolation between low and high voltage sides is also not proper, so that needs to be fixed too if a nicer version of the unit is built.

Then of course there is the fun of live testing it. There is no telling how (if) the furling on the machine will work, when it's being loaded with a better power curve, than originally intended. Propably it will need some tuning too. As a backup for the controller, we're also going to install a voltage tripped backup load, just in case the magic smoke escapes from the thingy. Some form of mechanical latch, details about that are yet to be figured out.

[BrianSmith]

Running the FET with that high of a PWM frequency(20kHz) is going to make it much hotter than if you can run it slower.  Everytime you switch it on or off, you force it thru a high resistance area.  The time it takes transitioning thru that linear region will be when it generates the most heat internally.  If you have a really good FET driver circuit that hits the FET gate hard and switches it on and off very fast, it will help.  If you find it is getting too hot under heavy loads, you might try slowing down the switching frequency of the FET and see if that helps any. FYI. 

[Janne]

Hi,

The heating on the FET seems not to be a problem, in fact most of the heat seems to come from the current going through the on-state resistance(~0.5ohms). There is a TC4420 6A FET driver driving the FET gate.

One thing not mentioned, is that at least in the test bench the thing causes some disturbance to the nearby FM radio. I will investigate if it will be better after the resistors will be mounted inside a metal tank, and the output cables switched to shielded type.

[ghurd]

Hi Janne,

What part number is the 200V Regulator?

I am wondering if a cap in parallel with the load would quiet the RF.
I am not sure of the implications regarding Q1 and L1.

Possibly a cap from Drain of Q1 to ground?  (that seems like it could be a very bad idea on many levels, but I do not know)

The controller is NOT supposed to look like that if it is home built.
You are making the rest of us embarrassed!   
It looks great to me,
G-

[Janne]

Hi Glen,

Thanks for the suggestions. There is already an RC snubber(56ohm + 1nF cap in series) across the drain and source of the fet, but i could still play around with different values.. Cap in paraller with the load, that I didn't think of but seems worth a shot at least.

The 200V regulator is made from good old discrete components. There is a 220k resistor + 200V zener in series, with the node connected to the base of a 800V rated NPN transistor. Collector of the transistor is connected to the positive supply via a 1k power resistor, and the emitter goes to the 200V output. There is also a 10µF 400V cap at the emitter. Basically, after the output exceeds 200V the zener diode will rob the current from the base of the transistor, hence current to the output decreases.

Mayby the picture makes it look nicer than it is .Though, i have to admit, Ii kinda like the looks, but with the lid closed well shut

[ghurd]

For better looking pictures, stand farther from the controller!
G-

[joestue]

Janne,

As you already have already invested in a 6A gate driver there is no reason not to go to a higher switching freq to cut down on inductor losses (really trading them for eddy current losses) I'm running 10Arms through 27 turns around a CRT yoke core. (wire is made from about 1/2 of the litz wire from the same crt) and it hardly gets warm (wire is about equivalent to a 14 awg wire) and that's at 70-90Khz.
A 15 ohm gate resistor is switching the fets in about 200 ns (irfp450's)

There's a couple options with the noise, but first off you need to get the construction up to the task.
The source needs to be bypassed with several uf of mylar caps, rated to about about half the current the supply consumes, feed these caps with an inductor wound on a ferrite core, from those electrolytic caps.
secondly, feed the gate from a coax cable connected directly to the gate and source, if you are driving it without any resistors i'd expect that 6 amp driver to drag the source above/below ground and oscilate several times before turning on, and the gate driver needs to be bypasses with surface mount capacitors, unless you limit the current to a reasonable value such as one amp or less.
last but not least the ground plane.. wait, where is the ground plane? i don't see one? nor filter caps across the load?

[Janne]

Joestue,

I thank you for your suggestions.
Please correct me if I'm wrong, but I'm not sure if increasing the switching frequency would help much in here. With the current inductor value it's not even near working at a continous mode, it takes about 1 /10th of a period for the current in the inductor became limited only by the load resistor. My intention with the inductor was only to lower the dv/dt value in the load to something more sensible, than the whole 550V at fet switching speed.  I've only little experience with switching mode psu's so it's not always obvious to me what's a good idea and what not.
The fet gate has a 5ohm resistor in series, and the gate signal seems like a neat square wave without ringing. Without the resistor i saw spikes of around 20V in magnitude on my CRO screen, so the gate drive seems to be doing ok for the moment.

The load does not have any caps on it, since I though resistors don't mind being driven with (near) pwm.. At least so far they seem to be doing ok.
I have also a 3.3µF plastic film cap paraller with the Electrolytics, can you elaborate if it's ok to use this type of capasitor instead of a mylar one?
The noise problem seems to be under control  at least to a some degree now with the inductor and the simple R/C snubber. Though I still needed to add shielding to the cables feeding the load resistors, as on full operating voltage I ocassinally got glitches to the LCD display.

Ground plane?? Do you mean the dirt under my test bench 

[joestue]

Makes sense.
Keep in mind that dv/dt is still only limited by the ratio of load inductance to the inductor. A snubber across the load may be needed to kill the noise..
generally most epoxy filled film caps in a square polyester box are mylar, doesn't really matter too much what it is, but it needs to have low inductance and not overheat.. which it shouldn't ever in this application.

[BrianSmith]

Janne,

I took a look at the schematic and I don't see why you need the inductor for your heating element resistors.  The resistance is going to limit the current more than anything else unless you have a really big inductor.  Also you might want to move the diode across just the inductor, not the inductor and the load (if the load is just some resistors for heating and not anything real inductive like a motor).

Now if you are running other things off the + side of the rectifier, you might want to put a diode in series with the digital stuff with a decent sized cap on the digital side to keep the load from making your digital supply drop off.  That way if your heating load drops the + node on your rectifier for a bit when the FET is on, the diode will isolate it so long as the cap on the other side of the diode is big enough. 

If you have a battery on + node of the rectifier and not just a cap, then it probably isn't a problem.

[Janne]

I hear you. If the inductor is on saturation mode, it won't limit the the dv much at all, and also of course the inductance dimishes with raising current. I've ordered some larger powdered iron toroids, to make a bigger choke and hopefully run things cool enough after that. The cores are coming with USPS, so it might take a while before they arrive.. It would sure help if there were a good supplier of magnetic components nearby, that would ship to small customers as well . I'll also try a snubber directly over the load resistor, to see if the things can still be easily improved.

Other than the choke problem, the thing seems to run quite ok now. Before putting it to the test, I need to figure out some form of over voltage tripped relay as a safeguard.. Do you know any off the shelf solutions for this one? The latching voltage would need to be around 400-450VAC, at about 30Hz. Many sorts of voltage supervisor relays seem to be available, but they're all rated at 50-60Hz, and of course the suppliers will not confirm them working at lower frequencies.

[BrianSmith]

Janne,

I will look around for a simple circuit to try.