Amanda's dumpload controller

[dinges]

Dear diary,

Last weekend I started work on Amanda's dumploadcontroller. Schematics can be found here:

http://www.fieldlines.com/story/2006/6/22/04750/5496

What I especially like about this controller, apart from its functionality and possibilities, is the fact that it can be built with standard parts. I had just about all the parts (apart from a few of the ICs, which are very common anyway) lying around. No fancy etched PCBs are necessary (I use perf.board, which I like working with due to flexibility), no SMD parts (though some will be used in the main-board, for layout reasons), easy to understand design (once you've got it figured out, that is  )

This weekend the alarm-PCB and the temperature-PCB got finished. That is, as far as it can be finished, some parts are still missing (ICL8069 & LM335); the meter needs to have two wires, which will be attached after a new scale has been made with temperature indications.

The alarm board was tested and seems to operate satisfactory, as far as can be judged without the rest of the boards present. When an error situation is present one of the leds lights up, as does an (annoying) buzzer. The buzzer can be turned off, but the LEDs will continue to indicate the fault condition. Excellent!

The temperature board was pretty simple to build, not many parts. A bit of extra reassurance that the batteries won't be tortured to death when charging. When temperature exceeds a threshold, the equalization charge will stop. (I don't think the temp.sensor has any function during normal charging/dumpload operation)

Next weekend construction will start on the main board. The frontplate can already be seen in the top picture, it's the board most left of the three. This should be easier/simpler than the alarmboard, as far as my construction method (perf.board) is concerned.

All will be built in a 19" rack I still had lying around. It's perfect for the job and portable too. I expect the dumpload controller to take up only 3 slots (of the 20 or so). Not sure whether I will implement all 8 FET channels, I think one could dump about 100A x 8 = 800A that way... Anyway, some of the remaining slots will be filled with the FETs (one channel per board/slot) with battery-pole plugs (banana plugs, which can handle 16A max). For really heavy dumping I intend to use something else, a stand-alone unit with the FETs in them and just the control wire going to this dumpload controller. The main problem being the finding of suitable connectors (the banana plugs/pole plugs can only handle 16A), otherwise I think I would dare to build all FETs in this 19" rack. Still, 16A at 24V is still almost 400W PER CHANNEL ! 8 channels in all, so about 3200W... Not too bad, me thinks. Now, if I only can find some good 32A connectors (4mm banana-plug type) that would be even better.

Anyway, the system can later still be expanded by using bigger FETs and construction that's more suited to high currents. There exist FETs that should be able to handle 100-150A. That's a total of about 1000A for 8 channels. At 24V this equals 24kW.... (and at 48V it's 48kW)

What can I say. The electronics are there in the 19" rack and it's a simple matter of adding FETs to suit the dumping power capability one needs.

When it's all finished I expect to have about 7-8 free slots left. One of them will be taken up by a small dumpload controller for 12V, probably Ghurd's design or else the homepower circuit. This board will be completely independent from the rest (Amanda's controller).

In another (2) free slots I intend to build a small 24V power supply to charge the batteries from mains, when no wind/solar is present. This charger will also be operated by Amanda's circuit, i.e. turn off when the voltage of the batteries gets high enough that it wants to start dumping. Instead of dumping, it will turn off the charger.

Then there are about 5 slots left. I'm sure that they will find a use too, some day.

Only modifications I have made to Amanda's design: each board has its own 5V power supply from the 24V bus. Each board has a polarity protection diode (to prevent boo-boos resulting in magic smoke) and a fuse.

I must be off now, dear diary, but I promise I will tell you about my next adventures with this dumpload controller soon.

[inode buddha]

Wow, where can I find the plans for all this just in case? I had wiped my 'puter recently and I'm still getting back together. (RH/Fedora Extras). It might well be years before I can actually try this, but instead of banana plugs -- did you consider the 1/2 turn locking jacks from Tweco

Yeah, it's welding supplies. But look at the page for their cable and panel lugs...

[dinges]

Hi,

The link to Amanda's diary entry and schematics is the first link in my original post.

Hadn't considered those plugs (over here they're made by Dinse and commonly called 'Dinse plugs') for the 19" rack, but they're in the back of my mind should I one day have a system big enough to require something that big For the 19" rack they're a bit too big.

Banana plugs (modified) exist in a 32A version too. Think the 16A will be enough, for the 19" rack. An external dumpload-module (actual load+FET) would require something more substantial, probably not with connectors but using lugs or something like that.

[inode buddha]

My bad, sorry I didn't read closely. I did check the links and save the files after I posted. I guess I'll call that a "mini-slashdot", where one posts without reading everything. Sort of like talking out the arse. (Blushes)

On the other hand, the engineering that happens here is truly impressive in some ways; its real-life people with real-life problems and budgets, and skills to solve them. Not just research. As for myself, I will be happy to build my first genny this year.
Things are kinda hairy right now, I just moved. Thx.

[oztules]

Nice construction techniques Peter..... now those extra slots could contain a nice boost converter, and then you could "match the load" and dump it at the same time

This might also solve your dilemma with the induction conversion .... wind it heavy and boost it.

..............oztules

[willib]

Oz , did ya ever do a schematic for the new boost converter , or did i miss it?

[oztules]

Hi Willib,

 If your referring to the l4990 version, then the answer is not at this stage. I somehow lost some interest in it as it was 16v startup. I wanted something more universal.

The tl494 seems to be easier to play with and well documented, and easy to procure... (any pc supply)

I will do some more with it when I get the mill started... and then finished, so I can get a true idea of the dynamics involved with the windmill to battery interfacing. It's fine to talk about it, but doing it in the actual environment is far more instructive, and more likely to show up shortcomings. The current tl494 is fine for 500 odd watts with both coils, but I have yet to feedback the windmill rpm/current to tell it how much to boost.

When that one is perfected in the real world, i will fiddle with the l4990 version, for by then I should have a better idea of the feedback regime required to match the load.

This is all longhand for being a slackarse at this time....

..........oztules

[willib]

"The current tl494 is fine for 500 odd watts with both coils, but I have yet to feedback the windmill rpm/current to tell it how much to boost."

any ideas how you would go about that?

[oztules]

Yes Willib,

Flux recommends using a current sense to read the current going into the batts, and using that current to set the output max from the boost.

I think he means that (for a 24v system) if I set current for say 20A, then at lower rpm, as much as possible will be provided from the boost circuit (using 50%duty, cut in will be 12v output from the mill, boosted to 24v.), as the mill gets up past 24v, it will start to contribute current directly from the rectifiers. As more current comes from the recs, the boost will drop off trying to maintain a max of 20A into the batts. Finally the mill will put 20A into the batts directly, and the boost will think it did it.... so will shut off the pwm trying to drop the current.

As the rpm drops (wind slows down), the current from the recs will drop below 20A, and so the boost will try and keep it up at 20A and so pwm will fire up.

The long and the short is that we design the boost for constant current, but measure it where both systems contribute. This way the boost only tops up what the mill cant do to a max of 20A into the batts from all sources. At below 24v, the boost will provide all it can from the watts the mill is generating. Hopefully in the 90% range for conversion.

As the mill goes towards max, it may generate 100A, but the pwm will take no notice of the extra power and just idle till needed.

It won't be until I build the mill that I will be able to see this in action, and get a feel for the settings that work the best, without stalling the prop at lower speeds (the rate of change of the op amps will need to be adjusted to be smooth over the range, and not be to stiff.) Flux incorporated a seperate op amp for just this reason, as I think he feels the pwm error amps are a bit to touchy.

It is possible that we can use a digital tacho chip to provide outputs to a logrythmic resistor array to give us mppt on the first step of the generator ie between the 12-28v we can measure the rpm, and in say 10 steps, switch in different resistors into the pwm to give us a power curve that will match the ideal drain off the mill.

After that point, the mill follows the power curve fairly closely anyway.

So as it stands, the fixed current control is already in the tl494 board, just needs an interface board to control it automatically from the rpm for mppt, or the current to be sensed from the right part of the battery lines for Flux's design.

..............oztules

[dinges]

LOL. The ubiquitous TL494. Have plenty of those lying about, and no bonus points for guessing where they came from. (I must admit I came across an old PC the other day and I neglected to take it with me (PSU was still in it), thinking 'I have enough of this junk'). When I came home I slammed my forehead: I still need a good PC PSU for conversion to universal bench power supply... Moral: one can never have too much 'junk'.

Do you have a schematic for your MPPT, as the situation is at the moment? Or any good designs on the web for MPPTs? Am curious, though I have to admit I haven't delved into it that deep yet.

Have just had a few new brainwaves w.r.t. my 19" rack. I'll add a card with a LCD DVM-module so battery voltage can be constantly monitored. On that card will also be a low-voltage switch-off, that disconnects any load when battery voltage drops below, say, 22V. Plus another buzzer that goes off when battery voltage drops. Maybe with an external alarm connection, so one can attach a horn or something (that way the entire neighbourhood will be notified when battery voltage drops below 22V, LOL)

Plus an extra card to control an external 24V PSU. If there's no input from wind or solar and battery voltage drops, it should turn on a grid connected battery charger. Have some nice solid-state relays for this purpose in mind (switching the 220VAC).

Slowly, the 19" rack is beginning to evolve (in my mind) from a dumpload controller to a full control system for a future RE system. Good thing to have so many spare slots in that rack.

Peter.

[drdongle]

So why are you using a BNC on the Temp card?

[dinges]

The BNC on the front is basically there for testing and calibrating purposes. On the front, because of easy accessability. The real temp. sensor will be connected via another connector in the back (don't know what type yet; may very well be BNC too). Both connectors (on the back & the frontal BNC) will be paralleled.

It's the best, cheapest good quality connector I had available. Have plenty of them lying around. There's no real reason it should be a coaxial connector. It's just that my home-laboratory is more or less standardized around BNC, so easy interconnectivity.

Don't look for too much behind it

[willib]

how about using a hall effect sensor hooked in one phase of the alt , you can get accurate ( i believe) readings of Amps from that .

Jim did a diary on that a while back

and a hall effect switch could count the RPM , compare that count to a known timebase , out pops your RPM. digitally speaking

[oztules]

Hi Peter

"Do you have a schematic for your MPPT, as the situation is at the moment?"...... No

At this stage a led display not unlike Amanda's dump led display turning on different resistor bridges to the pwm . It will probably be driven by a tacho..... it may also be done using a wind speed indicator, etc.... plenty of ways here.

10 stations will give fair control over the initial range (12-28v). After that the gennie will do it fine.... providing I follow Flux's teachings re "Matching the load". In this way mppt will be achieved on the lower power ranges electronically, and at the higher power ranges, the gennie will do it itself without all the heating drama that has been bandied about lately. His system uses 1/4R of Danb's system, so can with stand far greater power levels for longer, and a better use of the prop.

The problem I see is that whatever I measure, it is dynamic, and as such will require real world testing and calibrating..... wind tunnel and 10 trim pots would be nice.

After that is perfected/or dumped, I will proceed to use the tacho result directly on the pwm, although getting a non-linear response curve will be challenging for me.

As I will be using only the lower power levels to play with the power electronics, it should be manageable with very small heat sinks, and small scale circutry... even for a multi kilowatt output..... probably inside your rack, with no fan.

I look forward to both Amanda's efforts (heroin of the electronics people here), and perhaps SamoaPower who seems to be following the mppt path.

Mine will most likely take different guises as it fails a few times from just not "getting it".  But for now it's in the ideas stage....which look terribly simple... so they must be wrong?...but I'll get it in the end.

In reality, I'm still getting to grips with the interactions of the wind, the prop, the alt, and the batteries impedance... that balance will be interesting.

The problem I see with all this is that each prop-alt combination will require different mppt, and so perhaps the design with 10 trim pots will be the best for universality on this board...... or I'll succomb to simplicity, and do Flux's system...at least his works.

His circuits are in "matching the load" (which I think Admin has carefully hidden so as not to confuse the innocent:) )

..............oztules