 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Windmills : The MPPT Project.
Page 1 of 7   |
|||||
| Author | Message | ||||
Gizmo Admin Group  Joined: 05/06/2004 Location: AustraliaPosts: 5116 |
Hi All. After the discussions on this thread about building a MPPT to suit the F&P, I figured it needed its own topic, so here is. I feel with the combined know-how on this message board we could come up with a workable design. Like any machine, its nothing more than the sum of its parts, so lets see what we can come up with. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
| Gizmo Admin Group Joined: 05/06/2004 Location: AustraliaPosts: 5116 |
Talking about using a hardware or software block for the brains of the MPPT, I myself would like to see a software based system, using a PicAxe. I like software, its easy for me to work with and make changes to the way things work without turning on the soldering iron. But I can understand why others would prefer a hardware based system. So how about blocks? A block for the switching ( Inductor, MOSFET, driver and protection circuits ), A block for the MPPT's power supply ( to run on 12/24/48 supply voltage ), a block for current measurement ( Hall sensor and conditioning circuit ), and a block for the smarts ( PicAxe, PIC or hardware ). Each block is self contained in that it has its connections to the outside world that would work with the other blocks, no matter what they are, ie the MOSFET driver circuit will take a pulse from either a PicAxe or hardware smarts circuit. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
vasi Guru  Joined: 23/03/2007 Location: RomaniaPosts: 1697 |
Hi Glenn, I wanted to reply on last post of @warpspeed, to vote for software/hardware combination but, not with my english ... Every post on this thread will go also in my inbox, like it was with the last thread. Hobbit name: Togo Toadfoot of Frogmorton Elvish name: Mablung Miriel Beyound Arduino Lang |
||||
| 9c12m Newbie  Joined: 04/09/2007 Location: AustraliaPosts: 28 |
Hi Folks, Wish I could contribute ideas but have seen some scant information on the subject at the following links: http://www.redrok.com/electron.htm#mppt http://www.ecoinnovation.co.nz/pdf/smartdrive_what_is_it.pdf page 8 9c12m |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
O/k I can see that the software approach is popular, but the hardware approach is still intriguing. I am playing around with a few hardware ideas at the moment for a possible control system. The basic concept is as follows: A linear hall effect current sensor such as the Radiospares 650-576 (about $55.00) would be used to measure dc output current. This can measure up to 57 ampere turns of direct current. So one turn through the sensor could measures 0 to 57 amps. Two turns through the sensor 0 to 28.5 amps, and for example ten turns 0 to 5.7 amps, and so on. By altering the number of turns looped through the sensor, the current measurement range can be very easily adjusted to suit just about any system. This method of sensing current completely isolates the measurement from the voltage on the current carrying wire, and it solves a lot of problems with noise and grounding, as well as being completely overload proof. Output from this Hall sensor then goes into a circuit to detect if the output current is increasing or decreasing. The actual amount of current is not important, only the direction of any changes. 
A voltage comparator gives a logic high if the output current is increasing, and logic low if the output current is decreasing. If there is no perceptible change, the logic level may be fluctuating at random. Output from the voltage comparator clocks an edge triggered flip flop. This is arranged so that whenever a a change from increasing to decreasing output current is detected the flip flop changes state. It can be seen, that while current continues to increase the flip flop remains in whatever state it is in. As soon as the current peaks and then begins to fall, the flip flop changes state. This will cause the system to hunt backwards and forwards seeking peak output current. If it is going the wrong way, the flip flop changes direction. It will locate the efficiency peak, and then rock back and forth very slightly right on the peak. Output from the flip flop causes a large capacitor to either charge or discharge very slowly through a resistor. The voltage on this capacitor directly controls the Pulse Width Modulation (PWM) duty cycle. A common commercial switching regulator control chip will do all the rest fairly readily. This is not a proper schematic, just a concept drawing. There would need to be some extra refinement such as maximum current limit, and under voltage shutdown to conserve power when there is no wind. Sorry for the poor quality of the circuit, but I have just started using Linux, and all my Windows base CAD programs don't work with Linux. So I have to draw by hand. Cheers, Tony. |
||||
Gill Senior Member  Joined: 11/11/2006 Location: AustraliaPosts: 669 |
Thanks 9c12m, those links are good tor those not so sure of what MPPT is about. To be honest I don't think my understanding would pass detailed scrutiny either but like most we know enough to search for better ways to improve our mill efficiency. I'm pro the software approach. Though I should be more specific and say PICAXE. I think even those with no previous experience could get one running OK whereas PIC and other processors requiring use of a programming device would perhaps need to source pre-programmed chips from a group kitty. was working fine... til the smoke got out. Cheers Gill _Cairns, FNQ |
||||
Bryan1 Guru Joined: 22/02/2006 Location: AustraliaPosts: 1450 |
Hiya Guy's, In the previous thread the suggestion of toriods cameup and on the rfi boards I got there a heap of them. In this link is a pic of one of the small boards and if they are suited I'm happy to strip the off the boards and supply them for the project. Tomorrow i'll measure them up and edit this post with the size. As far as the software approach goes you can't beat the 16f88 pic chip which can hold a bootloader so a programmer isnt needed just a max232 to get it to talk with a puter. Cheers Bryan  |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Designing the magnetics is a bit tricky, there are a number of conflicting requirements not easily met. I have something here that should work up to about 130 input volts maximum, and around 15 amps output maximum, at 20 Khz. I spent several hours today working on magnetic calculations trying to figure out what we could possibly use. These particular chokes are wound by one company, and supplied to another company (that I used to work for) in fairly large numbers, to manufacture a commercial product. What I don't know is if the company that winds them is prepared to supply them to us in one off quantities. As you can see they are coffee mug sized, and not exactly small. But if we can get them, they will be machine wound, complete, and ready to go, exactly as in the picture below.  Cheers, Tony. |
||||
| vasi Guru Joined: 23/03/2007 Location: RomaniaPosts: 1697 |
Warpspeed, use QCAD, you can install it via the software provided for this. But the schematic is ok. Hobbit name: Togo Toadfoot of Frogmorton Elvish name: Mablung Miriel Beyound Arduino Lang |
||||
herbnz Senior Member Joined: 18/02/2007 Location: New ZealandPosts: 258 |
Tony Looks good to me if we could input up to 150 volts this means we can run FP with less current. 110 volts DC is max volts before we need reqistered personal to wire. Would cores found in old computer power supplies be big enough. Great work I do like your ideas probably would favour pic but Glen's dual approach good hardware is simple. Herb |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Vasi, I downloaded the demo version of QCAD and had a very quick look at it. Not enough time now to spend on this, but I will definitely have to get myself something. QCAD runs on Linux and should do the job. Thanks for the tip! Herb, Some sort of simple microcontroller is undoubtedly going to be the neatest solution. I have some Motorola microcontrollers and a fairly fancy programmer for them here, but nothing really ideal for this particular job. First I will see if these ready wound toroids are available to us, if they are, then I will concentrate on the power switching side of things. Gill may volunteer to investigate some control system algorithms. Sorry, but old salvaged toroids from dubious sources are simply not going to work. Something has to be properly engineered to do this. The size,core material, and number of turns are very important to success. There needs to be sufficient inductance to support the high input voltage, but too many turns on a core just saturate it at high dc current. The core loss is also a worry, and too high an ac flux density will easily raise the core to destructive temperatures. Insufficient inductance or magnetic saturation produces excessive peak currents. So there are a whole lot of factors to optimize in the design of a suitable toroid. The chances of some random component working are just about zero. This toroid initially looks fairly suitable, but I will need to test it in an actual circuit and measure some temperature rises and establish some sort of sensible limits on voltage, current and frequency for it. Assuming at this stage that this choke would be able to supply 15 amps at 14 output volts that is only 210 watts. At higher dc output voltages, there would still be 15 amps, so the output power would increase significantly with higher battery voltages. Large twelve volt systems are the problem. The idea is that if you needed say, sixty amps output, you could use four MOSFETS, four toroids, and four catch diodes in parallel. That would give 840 output watts. With luck we can increase the output power to any level by simply paralleling standard modules. There is also a 150 Amp version of the same Hall current sensor available. Cheers, Tony. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Just received a reply from Dyne Industries on the 9829 machine wound toroid. They are not a stock item but would be wound on request. The price is a real shocker at $42.00 each, with GST extra! For five or more, they are quoting $35.00 each. I was afraid of something like that, so now I will start looking at other options. Cheers, Tony. |
||||
| KiwiJohn Guru Joined: 01/12/2005 Location: New ZealandPosts: 691 |
Hi Tony, colour me dumb but I dont understand how a buck regulator can increase current? |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
It reduces the voltage, and increases the current by the same proportion as the duty cycle. So if you were converting 140 volts down to 14 volts the output current would be ten times higher than the input current, (if there were no conversion losses). Cheers, Tony. |
||||
| herbnz Senior Member Joined: 18/02/2007 Location: New ZealandPosts: 258 |
Hi Tony the more I look at the hardware version I see it deserves a try. However the first stage is to tap into your expertise in the switching requlator. I see now why my odd attempt to built these has ended in failure, the inductor is much mor critical than I realised. I assume the placing of the 12volt power supply on the windmill side is to save power if no output from mill. John In my understanding the increase in current as voltage is decreased is by the duty cycle first charging the inductor in short time because higher volts. then discharging via the diode to the load at the lower voltage therefore more current. of course old story power in = power out + losses |
||||
GWatPE Senior Member Joined: 01/09/2006 Location: AustraliaPosts: 2127 |
Hi Gizmo, I have my own cct design that I have installed on my custom ironless design. [obviously no cogging here!] peter has had the luxury of viewing the MPPT. From my experience the main hurdle to overcome is the cubic relationship between wind speed and wind energy. My early designs were similar to the design you posted. I had a lot of problems trying to use feedback to control the system. Instability can let the smoke out of most semiconductors. My MPPT allows the generator to start producing current in a 24 battery from a generator voltage of 6V at 1.5W to a maximum recorded 455W at 28.8V. My generator has 3 blades with a swept diameter of 2.1m. Passive generator tail side furling starts protecting the generator electronics at wind velocities above 15 knots. A Wind MPPT has an additional rapidly changing variable compared to a Solar MPPT. Ideally the generator RPM should track the wind speed to enable the rotor blades to operate at their optimum. Any blade has a sweet spot at which it perfoms the best. An Iron design may use the frequency relationship of the laminations. As the torque increases with increasing wind speed, the rpm increases as the iron begins to saturate. There is a danger that with too large a blade area, the rotor cannot be stopped above a certain wind speed, [even with the output shorted]. A MPPT should alow the blade to operate at this sweet spot most of the time. A MPPT machine will have a power output that follows the power of the wind gusts. My measurements have shown that the wind is a fickle beast. The gusty nature of wind does not make it an easy task for a MPPT. I had little success with a micro controller. I have found that a combination of constant current control at the bottom of the power spectrum and changing over to voltage controlled current limiting has given the best results. A simple fixed resistance current shunt in combination with a voltage controlled current shunt to control a simple PWM cct like an LM 3524. The cct is powered from the generator and hence does not contribute to battery consumption. [ < 50 mA ] The shunt has a 200mV maximum voltage requirement. The design uses a simple pot core inductor which is easy to wind. The design has no heatsinks and uses SMD components. My cct has been in operation now for over a year. I have not been a fan of high generator voltage unless the generator is noisy and has to be located a long way from the battery where wiring losses would be an issue. I charge a 24V battery and use the wind generated power and that from some solar panels to run domestic appliances with a standard 240VAC inverter. I spent the better part of a year, with probably a dozen individual prototypes that were modified, until I took a different approach that needed only 1 mod. I think you will find that designs will work at the bottom end or at the top end of the power curve. My initial design worked poorly and I could not replicate them. I have increased the generator power from 180-200W from the early designs to the 455W now recorded. The power envelope extends from 3 knots to 20 knots. above 20 knots the output plateaus. I have had to employ radio design type measures to reduce RFI and EMI with these designs. The rapidly changing high voltage modulated square waves will radiate and get into most TV or radio's. Electrical screening is essential. I will not offer any ccts here as my cct cannot be breadboarded. I look forward to your solutions to this problem. cheers, Gordon become more energy aware |
||||
| KiwiJohn Guru Joined: 01/12/2005 Location: New ZealandPosts: 691 |
OK, if you guys say so but for the life of me I cant understand an increase in current as if I recall correctly current is the movement of a specific number of electrons past a point and I cant see how more electrons can flow out of the inductor than flow in. |
||||
| Gill Senior Member Joined: 11/11/2006 Location: AustraliaPosts: 669 |
Pleased to be a part of the effort. I'm guessing what is needed is output data from the stator we plan to use? I'm thinking 7 coils in series, 3 phase Star, x 2 parallel? I'm thinking Voltage and Amps into a range of resistors over a spread of RPM? Is this what we need, Tony? John, I've looked at it and can't see it either. Still, If you listened to me we'd all be driving cars with stone wheels. "Trust the Force Luke" was working fine... til the smoke got out. Cheers Gill _Cairns, FNQ |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
The inductor needs to store considerable energy, several times the energy released into the load over one switching cycle. For instance, a design parameter might be that the inductor carries an average of fifteen amps, with a two amp peak to peak ripple current. In this example, the inductor discharges into the load over 90% of the cycle time, the current decreasing from sixteen amps to fourteen amps. The inductor charges back up during the MOSFET "on" time which may be only 10% of the duty cycle, (if the input voltage is ten times greater than the output voltage). An important component is the input reservoir storage capacitor of the buck regulator. The inductor charges up over a fairly short time period, but the input capacitor averages these intermittent high current peaks to some low averaged steady dc input current. So one requirement is that there needs to be sufficient inductance to store sufficient energy so that the current does not vary excessively over the whole of each switching cycle. This inductance tends to decrease as the magnetic core approaches saturation, so you need to be careful to keep the dc magnetic flux well below saturation. Put too many turns on your core and you will saturate it, not enough turns, and there will be insufficient inductance and energy storage. But it is a lot more complicated even than that, so far we have only considered dc current, inductance and dc flux. The voltage across the coil is constantly being switched, and the ac flux in the core needs to generate a back emf sufficient to oppose this voltage. The fewer turns and the greater the voltage, the more volts per turn are required, and more ac flux swing in the core is needed to generate it. It is ac flux that causes core losses and core heating. So volts per turn and operating frequency can have a huge effect on core heating. There are also a huge range of alternative grades of core material to choose from, there are eight grades from the manufacturer I am looking at. The characteristics vary quite a lot, so that complicates things even further. Magnetics design is about the most obscure branch of electronics there is, and there are very few people that really have a good grasp of all this. Cheers, Tony. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Yes, what sort of maximum rectified dc voltages and power are people seeing with typical F&P machines, I have no idea? Cheers, Tony. |
||||
| Page 1 of 7 |
|||||
 Print this page |
 |
The Back Shed's forum code is written, and hosted, in Australia. | © JAQ Software 2025 |
