Electrolytics have a maximum current rating called ripple current. They also have a maximum reverse voltage rating that is usually quite low (1.5V typ). In addition, they have a maximum temperature rise, also quite low (10C typ). Don't confuse this with temperature rating which is environmental.

I think you're exceeding all of these quite a bit. I looked up a 1000uF @ 200Vdc aluminum electrolytic. Ripple current was rated at 2.6 Amps at low frequencies.

I think you're fortunate in not having blown one up yet. At a minimum, their lifetime would be drastically reduced.

Here's the right way to use them for AC in series-parallel:

The balancing resistors are necessary because the caps are not usually matched very well. 10k-20k should be okay. The diodes should have a PRV rating at least equal to 2x your max output RMS voltage and should be able to handle the line current.

Generally, if you can feel the caps getting warm, you have a problem.

Hope the wind doesn't cause too much damage on you. It's pretty quiet here at the minute

Firstly well done.

Secondly, you are not using this board correctly.

Please write down the exact specification and anything else you can give us for motor cap specs.

Members of this board live in just about every country in the world and we therefore have access to all sorts of dumpsters, shops, closing down electronic companies to name a few.

If I can find some of what you want I would gladly post them to you gratis.

I think we all owe you something for your contributions.

Regards,

Try putting the capacitors in parrallel, each reverse biased from each other, with the diodes in series with them as blocking diodes. This will yeild twice the capacitance since you now have them in parallel vs the series configuration you are working at this time. It will actually be a dual half wave structure, since they are opposed and each will work with half the cycle.

Another idea is to create a bridge configuration, just like rectifiers, but with the electrolytics, and add the blocking diodes and stabilizer resistors accordingly. You might also consider bleeder resistors on the caps, if you get spikes from the half wave structure that this configuration creates.

I'd be intrested in waveforms from a scope as the current is raised with voltage and frequency. There  may be a resonant point in an RC reactance which is developed by the load.

This entire concept for heating is highly plausable, and will always produce output in proportion to the greater the wind that normally blows the heat out of the house.

Is this right?  I thought it would be volts X amps /1.73.

It is the line voltage you are measuring isn't it?

Could the elements be cut to a length that allowed the prop to "spin" free up to whatever rpm without capacitors?

Just curious

thanks to you all who have put much thought into what i am doing. it seems that i have opened a can of worms, but a good can of worms it is. i don't think there has been much work done in this area and it certainly seems to be a good discussion ground with many good opinions and ideas for this combination of gen/caps/heating.

well none of the capacitors have blown up yet :)

i took many measurments today as the wind was still very good but not as severe as yesterday.

just to review some info in case some people do not have it all.
the gen is a converted 7.5hp motor. it is a 12 pole unit that produces 240 volt AC at 600 rpm. it is currently connected 1 star with line to line resistance of 2 ohm.

the element heating bank consists of 6 elements. originaly, 3 elements were connected delta and the other 3 elements were connected delta and paralelled with the other delta. rating is 18 kw at 208 volt 3 phase.

i was running 1100 MFD per phase  with 3 elements connected delta, but capacitors were getting hot.

during yesterdays wind storm, i had 1650MFD per phase with 6 elements connected 2 star. the capacitors now ran only slightly warm to the touch. the mill did seem to experience a slightly harder time starting up from zero rpm.

3. elements connected 1 delta 8.6 ohm
6. elements connected 2 delta in paralell 4.6 ohm
3. elements connected 1 star  25 ohm
6. elements connected 2 star in paralell 13.5 ohm

the caps are:
MEPCO/ELECTRA

3110. DC112T200BMA1
1100. MFD 200WVDC
362. -8237          I googled, but could not come up with a spec sheet for these caps

#1 the AC voltage across the output terminals of a pair of caps back to back was 25% of the incomming voltage of the gen. at 75 volt and more from the gen, the terminal voltage of the caps dropped to 20% of incomming gen voltage.

#2 the current measured between the caps and heating elements was approx 1/2 of that measured on the incomming gen lines.

#3 at approx 100 volt incomming from the gen, the input and output current to the elements becomes approx the same.

#4 there are 3 sets of 2 caps back to back per phase. each set seems to measure 1/3 of total current per phase to elements.

i have now removed the caps from service until the more experienced electromics people seem to agree on a proper way to use these caps for my purpose.  no point blowing up some caps if it is avoidable.

i have now resorted to my original way of controling the gen to heating elements with contactors and voltage relays.  it works well, but not as "slick" as with the capacitors.

the wattage on the higher end of the voltage scale seems to be identical if i am using the caps or the relay controls.

this has been quite a learning curve for myself and i am sure for quite a few of you. my electronic knowledge is limited so it takes me a while to absorb and understand all the explanations and suggestions from everyone.

again, thank you to everyone for all your thoughts and knowledge sharing. i hope between all of us that we can come up with a good simple way of using caps for heating purposes.

my method of calculating and matching up heating elements/gen/cpacitors/connections is based somewhat on my own knowledge and an awful lot of try this, try that, record results, analyze, try and get closer to ideal.

its been a lot of fun and enjoyment.

First of all, the idea is to have twice the resistance of the heater bank at cutin speed. That is, when it cuts in, the generator sees only half the load it sees at full speed. This is not a fixed rule, btw, but a recommendation or rule of thumb from someone to Zubbly in an earlier thread on the 7.5HP genny.

Since Z's heater bank was (originally) 25 ohm, we want the caps to exhibit a 'resistance' (in reality, a reactance, Xc) of 25 ohm as well, at the cut-in frequency of 200 RPM (20Hz) (my figures may be slightly off, I'm doing this all from memory a month after the facts...)

We are determing the capacity needed PER PHASE, so, we need to know the resistance of the heater per phase as well. In this case, R = 4.6 * 1.73 = 8 ohm.

Xc = 1 / (2* pi * f * C)

with f =20 Hz (200 RPM in Zubbly's case)
C = unknown capacity
Xc = 25 ohm (the reactance of the caps; 'must' be the same as resistance of the heater bank)

plugging in these values we get:

8 = 1/ (2 * 3.14 * 20 * C)

after a bit of calculations we get the result

C = 1000uF

This is for the situation in 2-delta (total resistance 4.6 ohm; resistance per phase 8 ohm). From memory, I recall that we ended up at 1500uF when we performed this calculation about a month ago. We may have used slightly different numbers, but you should get the idea.

We haven't taken into account here any line resistance; a lot of line resistance would lower the capacity requirement.

Also, this is NOT an exact calculation that yields a 'correct' answer; there's lots of room for variation and it will still work fine. This calculation gives a rough ballpark figure for how much capacity is needed. And, it's based on the assumption that at cut-in the reactance of the caps should be the same as the resistance of the heater. But, .5 times, 1.5 or 3 times may be other valid values... Who is to say what is the perfect amount ?

As stated, it gives a ballpark figure from where one can start experimenting, as Zubbly is doing. In the latest situation he now uses about 1600 uF per phase, and his resistance of the heaters is now (in star & paralleled) about 8 ohm per phase (13.5/1.73). (which is, incidentally, the same resistance as in the previous example)

Xc=1/(2*pi*f*C) = 1 / ( 2 * pi * 20 * 1600 E-6) = 5 ohm.

His reactance of the caps is in the new situation (with 1600uF & star/parallel heater) 5 ohm at cut-in speed, whilst his coil resistance (per phase) is 8 ohm. As a consequence, the generator will see a bit more load at cutin than in the original situation (1100uF). I understand from Zubbly's information that the genny doesn't speed up as well as it used to, at cut-in. The extra caps could explain this.

(I hope Zubbly is still reading this; he's not into this fancy math stuff anyway ;)  ) Again, the calculated results are just a ballpark figure. Everything stands with the initial assumption (how heavy should one load the genny at cut-in)

Questions? No questions.

Do you have any protection circuit in case the circuit goes open - ie heater burns out or capacitor burns out.  

Do you have a thermostat and a diversion load for the times when the room you are heating is too warm?

What is the resistance of one phase in your gennie?

Do you have any type of break or manual furling on your machine?

I used to work with this on single phase configurations at one time, but never tried it with three phase.

Utility companies use concfigurations with in Line Capacitors and inductors with the three phase lines and on substations to help balance lines and nodes in voltage on long grid runs. The R was the line which is miles of internal wire resistance. The frequency is fixed at 60 cycles, made it easy to offset standing waves and nodes where there were voltage peaks and dips over portions of the grid.