Boost converter, pwm and mppt

[willib]

was just experimenting with a dc-dc boost converter

i was able to boost 3.04V to 25.59V dc

i put the Fet on a vector-board with diode.(finally)

hooked up pwm and off i went.

the input voltag was from one phase of my new mini-gen rectified with a bridge rectifier and a 6.8uF cap across the bridge

i fed this to the boost converter , and fed the output to two, six volt(sla) batteries and one 12V sla ,the rest voltage is/was 25.44V

the fact that i was able to boost such a low voltage to such a high voltage is very promissing

the pwm is running at 19Khz , for a inductor i used on of my coils .

the input to the boost converter is the output from mini gen(one phase)

3.04v @ .64A

the outout from the boost converter was charging the three batteries i mentioned earlier

25.59v @ .07A

it seems to me that an mppt algorithm for a windmill only needs to vary the PWM/current to get max rpm

if RPM go down decrease pwm to try to get RPM back up ,

if RPM go up increase pwm to get more current till RPM go down

keep repeating till batteries are full

if they are not full wait for next wind gust and repeat

the psudo code assumes that the alt is more powerfull than the blades

and may not be complete

[stephent]

Howdy willib.

Just remember, using that algorythm won't load the genny down as much (rpm wise) as a normal controller.

Furling might be delayed? If it uses furling.

But post pics or schematics I follow your stuff quite a bit..

You do interesting stuff.

[willib]

this is the circuit diagram

the Fet gate is controlled by varible pwm

when i turn the alt by hand , i can adjust the torque on the alternator (alt) by the pwm

if i turn it up , the load on the alt is increased

if i turn the pwm down , the load on the alt is decreased making it easier to turn and i am able to turn it faster

i also added another 6V battery to the battery array being charged , bringing the boost output to 31.95V @ .04A

[willib]

this may look better

http://www.otherpower.com/images/scimages/2965/boost_circuit.JPG

[stephent]

That 5uf cap is taking a hammering. (ripple currant) Might want to check it after it's running a bit.

[willib]

its a huge AC cap ,i scrounged it

370V its as big as a soda can

it says 5uF but under that it says 60uF ?

so i'm not sure which it is

[stephent]

run cap? like in metal box/container?

or plastic?

[willib]

its in a large Aluminum case , slightly wider than a soda can same shape though

i think i got it out of a electric hot water heater

[SamoaPower]

"it seems to me that an mppt algorithm for a windmill only needs to vary the PWM/current to get max rpm ..."

At first, I thought that you were just putting us all on. After thinking a bit, it dawned on me that you've simply redefined MPPT to mean MINIMUM Power Point Tracking.

After all, everyone knows that maximum RPM occurs at minimum loading.

That's great! For all those who want to make lawn ornaments and toys, who don't want to be bothered with all that electrical jazz, this should do it. Just think, no more transmission line loss, no chance of stator burn-out and great fun watching a fast-spinning machine.

Bravo willib!

[willib]

its kind of hard to discribe what i mean, but if you had read farther down , i belive i did write " if the RPM go up then increase the pwm"

[maker of toys]

I think Samoa is implying that you need to define an optimum RPM, so that your PWM routine knows if it should widen or narrow the pulses. . . because, in any given wind, the rotor rpm will stabilize at the highest value when the torque loading is at the lowest value. . . . by any other name, the minimum power point. also sometimes known as a run-away condition.

and with the power available in the wind dependent on wind speed, the optimum RPM also varies with wind-speed. (based on the obvious (to me) observation that you want the blades to stay somewhere near design TSR, with the aim of maximizing the efficiency with which you couple power out of the air mass while providing sufficient torque to couple that power into the coils of the alternator.)

so, with no explicit definition of optimum rpm in your code, the pwm will continue to reduce pulse-width, seeking more rpm, be stymied by the plateau that happens when the blades reach their maximum unloaded speed in the existing wind. . . .and wouldn't ever trigger the 'increase pulse width' branch of the code.

<rant severity = "minor">

what people seem to forget is TORQUE is a force that actually does work. RPM (in simple terms-- the rest of you engineers, dont jump on me too much) is merely a RATE at which torque is applied. . . . you can have all the RPM in the world, but unless you produce torque, you apply no power. RPM is not a force, and can do no work.

in fact, power is sometimes defined as "torque * RPM"

</rant>

[jimovonz]

willib, increasing the duty when rpm increases (and vice versa) will give you a constant rpm (presuming the alt is any reasonable match to the blades - stalled from the get go if you don't select some 'cut in' rpm for the controller). If your intention is to maximise current into your battery then you must base your control on current. Having said that, I think that it is beneficial to also consider rpm to improve the controllers response time (an algorithm based solely on current measurement would have to change duty relatively slowly to get around the problem of the inertia of the rotating mass). I have had some success with an algorithm based on the following table:

I   RPM   Duty

==============

+    -    NC

-    +    +

+    +    last

-    -    -

NC   = No Change - leave the duty cycle the same as in the last iteration.

last = Do the same as in the last iteration (increase if previously increased and vice versa). Start with +.

+/-  = Refers to a comparison to the last iteration (Duty col is how the duty should change in this iteration).

Updating the duty around 5-10 times per second based on average current/rpm readings seems to work for me.

[stephent]

Humm--it looks like whatever he's doing is working at 92% effeciency.

Anyone have any ideas on why that's working?

Especially since he's chasing max rpm. (under load of course)

He must have a secret weapon in that code he's just not telling about.

1. Folks, it's easier with 10 people working together on a problem sorting it out then 1.
   
2. It's just as easy to stiffle something with a quick remark, as encourage.
   
   

Naaaww, let's just whack him with another 2x4 upside the head cuz it "shouldn't" work.

[jimovonz]

I always admire willib's tenacity and willingness to get in there and try it out. He's obviously a do'er and his sharing with us is probably just as valuable a contribution as those that spout all the technical theory.

I think that willib was proposing a control algorithm (not implemented) and his results were from his DC-DC converter circuit with its input power not driven by the wind.

92% efficiency shouldn't be too hard to achieve at a 1.8W loading. Another story at 1kW.

[willib]

Stephent Dan & Jim & Samoa

i agree the MPPT idea meeded some more thought

but lets concentrate on the boost converter , today i shall try using a donor battery instead of turning an alt by hand

i think it works because i've selected a proper coil with proper inductance

after the tests i will know how well(or if) the efficiency holds up , i suspect the efficiency will not be as good with more current going to the recipient batteries

but in either case i was able to boost a very low voltage to a much higher one and that is something good for us all

[Flux]

Don't change your hand driven alternator for a battery unless you have a current limit circuit.

Flux

[stephent]

Jim...

Nice post, informative, helpful, encouraging.

But he (willib) has his hands on it, it's working somehow. (let's dissect this beast if willib is willing)

and willib.....ever measure the open circuit voltage from the genny at the same rpm you are turning/spinning it?

Maybe take load off and see what volts (rectified volts) come up no matter what rpm rises to?

Like maybe it's 3.04 x 1.4?

[willib]

the boost will not go on forever, as i'v found out , but i did not try varing the pwm frequency

a 6V battery(or alt) won't boost up to 36 V at the present pwm frequency (~19khz), but it wii boost to 25V

i'm sure if i used a inductor wound on a ferrite core that i could have got better efficiency

but i just used what i had on hand which was one of my new coils for the minigen , air core.

i was able to vary the donor batterys current in steps of .1A from 0.1A to 1A after that i increased the current in steps of .5A up to 4A

again it is kind of hard to nail down a reading because the recipient batteries voltage is rising while the donors volts are decreasing , but i tried my best

Output......Input

 I , Volts  - I , Volts

 0.02 , 25.46 - 0.1 , 6.18

 0.04 , 25.54 - 0.2 , 6.13

 0.06 , 25.54 - 0.3 , 6.10

 0.07 , 25.55 - 0.4 , 6.09

 0.09 , 25.62 - 0.51, 6.02

 00.1 , 25.61 - 0.6 , 6.01

 0.11 , 25.72 - 0.71, 5.96

 0.13 , 25.67 - 0.81, 5.95

 0.15 , 25.82 - 0.94, 5.84

 0.15 , 26.18 - 1.0 , 5.84

 00.2 , 26.54 - 1.5 , 5.71

 0.21 , 26.59 - 1.66, 5.68

 0.24 , 26.97 - 2.03, 5.5

at this point i should mention that the donor battery is a small 4AH 6V sla battery , which i recently brought back from uselessness

 0.26 , 26.95 - 2.40, 5.44

 0.26 , 27.00 - 3.0 , 5.27

 0.23 , 27.10 - 3.51, 5.07

 0.19 , 26.87 - 4.0 , 4.87

well thats my data

the efficiency peaked at 83% way back when the input current was 0.3A

but for an air core inductor i think it did not do too badly

the efficiency was still 51% at 7watts output , when the input current was 2.4A

... seven watts is not a even close to a kilowatt ,i realize .. and i know for sure Oztules has done much better

speaking of Oztules , where is he?

when the weather gets nice down under we dont hear fron him?

it still sort of amazes me that it was still boosting at 4.87V input , with a better inductor i'm hoping for a couple hundred watts

[willib]

i tried a new inductor , hopefully the pic will come out.

it is an E core ( with tape holding it together removed)

with two windings paralleled , 70turns of 0.045"dia wire Each

the resistance is 40 milli Ohms

the core is approx 2" tall

the efficiency was 72% at 5.78 Watts, small steps

question : is the resistance too high ,

is the inductance too low, high, What?

to recap the pwm is 19khz

as i turn it up the output current goes to 0.27A approx..

then drops and the highest i can get is 0.21A @ 27.5V output , the input is 4.85V @ 1.65A

i am a little stuck on where to go next ?

do i need a larger core

[willib]

oops almost forgot

[Nothing40]

Hum,Try increasing the frequency..maybe the 40-80khz range?

Usually when I see cores that size,they're in a SMPS for a PC,or something..Usually good for a couple hundred watts or so..

[Ungrounded Lightning Rod]

Fortunately, torque is proportional to current.  So you can measure your current rather than measuring torque directly.  (RPM and volatge are related similarly, but measuring the generated voltage is complicated by the voltage drop from the current through the wiring resistance.)

However, since this is a charging application, output power is proportional to charging current.  So measure THAT (averaging across the cycle).

Easy way to track max power point is to vary the pulsewidth by a small amount (a couple percent) at a rate (say one Hz) slower than the mill's reaction, and measure the delivered power at the narrow and wide settings.  Then:

 - If the wide setting is producing more power, widen them both slightly

 - If the narrow setting is producing more power, narrow them both slightly

 - If they're about even, leave 'em be.

And repeat.  That makes a feedback loop that walks you right to the peak of the power curve and then follows it around as the wind varies.

[Darren73]

willb,

I believe your wire is too thick for the frequency, the switchmode power supply design book I have quotes that for a 37% copper utilisation factor the wire would be 0.038", this is the max wire size you should consider, the optimum would be smaller than this, you can wind many in hand for this to achieve the current capacity required, but a platted winding (also known as litz) is the best option.

regards

Darren

[willib]

by increasing the input voltage to 12V , i was able to vastly improve things

this may get tricky ( as far as control )

but i am now up to 23W output for 27W input. 84% efficient

you can hear the coil screamming , still it dosnt get hot ( a good sign)

i turned it up too high(once), and did not look at the input current

input current was 8.11A at 11.88V output was .92A @ 30.58V

the Fet was hot but did not explode or anything and the coils were just warm , so i'm satified , for now.

the 84 percent data :

same two coil inductor as before 12V battery input , the output battery consists of one 12V sla and two 6V sla batteries

so they are not as strong as the power i'm pumping into them hence the 30V on the output..before the test they were 25.93V total

input 2.26A @ 12.15V output 0.77A @ 30.1V

as for control ..? some more thoughts , just typing out loud here:

i think there should be another few lines of code , because if it is programmed to get max current (output), it may try to overdo the input current like i accidently did.

assuming the alt can/could supply that much

as the rpm change the input voltage will rise and fall , as the voltage rises and falls, the efficiency changes because the boost has to work harder , so to speak , to boost a smaller voltage  , maybe tracking the input current (to the boost converter )is what we need to do ?

im going to have to think some more on this

[willib]

i woul if i could but the pic16F88 maximum PWM frequency is ~19khz

[willib]

i had the two coils that i paralleled on hand , so that is what i used , was your quote for 19Khz ?

[Darren73]

the quote was for 19kHz, the size is per wire of the wires in hand, the number in hand only affects your max current, rule of thumb is max 1000A per square inch of copper, hope this helps.

regards

Darren

[willib]

interesting proposal , i'm gonna have to think a lot more on this

thanks everyone for their input

[RP]

I couldn't tell from the picture, is that a laminate core or powdered ferrite?  If it's laminate then I'd reduce the frequency to about 7khz if you can stand the noise.

[Opera House]

I think you would do much better looking at the input voltage rather than the frequency to control the PWM.  This insures that you get some power at low wind speeds and adding a slope function allows the mill to turn at a higher speed in higher winds for more power.  Just monitoring frequency will not take into account resistance in the system.  

[jimovonz]

"assuming the alt can/could supply that much" - If you try to "overdo the input current" (inertia issue aside) your blades simply become less efficient (stall even) as there is not enough energy available in the wind. This leads to a reduction in current into the battery. If your using current into the battery as your control variabe (I can't see why you'd use anything else) then you'd adjust your duty cycle accordingly.

[jimovonz]

What do we want to achieve? We want to maximise current into the battery. In all cases more current into the battery is good, less current into the batery is bad irrespective of the voltage (or frequency) at any point in the system. Is it not then logical to use current as the control variabe in your MPPT? While it may be possible gain improvements using a system based on frequency or voltage, such a system would be hard wired to give the best output for a specific system state, would be a pain to 'tune in' initially and most likely would suffer when anything in the system changes (such as ageing of the batteries, system loading, state of charge, blade tip errosion etc). Using current would give you a universally applicable controller that would would give you the best output regardless of these other variables.

[willib]

Hi Jim

ive been programming PICs again

i just got the 16F877A to correctly read an analog input , and output PWM accordingly , i was using the 16F88 before .

I'm programming them in pascal , assembly is tough , but pascal at least i'm used to it.

If ANYONE wants to play with Varible PWM , email me and i'll get the number of people interested .

I can program them for ya'all

we have to decide on which chip to use , but first i need the number of people interested.

just send me an email and say "I'm interested" !,or reply to my new diary ,

i"m not looking to make money on this , I'll start a new diary later on and ask the same question.

the stuff you can do with varible PWM is amazing !

it really is a lot of fun!