diy 7 level inverter

[joestue]

So I'm intending to go ahead and build one of these things for a variety of reasons.

A friend of mine is going to try and program a PIC to output the correct truth table to operate the H bridges. Only thing I'm worried about is the exact timing, any dc bias caused by a tenth of a percent difference in duty cycle will cause the transformer to saturate and blow up.
The truth table for the 7 level inverter is rather simple and only 4 I/O lines are required with proper H bridge drivers, a 15 level inverter isn't much more complicated but setting the correct intervals to cancel the harmonics is a lot more complicated.

Of course, the transformers have to be custom wound, so that's the downside.


If we decide to put it all the parts on one board I could probably get the parts count down to 20$ or less, in case anyone wants one.
due to the low switching speed twisted pair or computer ribbon cable would be good enough to hook up all the mosfets to the H bridge drivers.

The only part i'm sketchy on is the buck/boost transformer on the output, the relays could be set up to adjust the output voltage from the incoming battery voltage, rather than the actual output of the inverter.
Way I figure it the efficiency can be as high as the transformers in practicality. mosfets are cheap these days, and the low switching speed means you can parallel any number of them.

[dnix71]

Is there no crystal driven analog circuit that would resonate a sine at 60 Hz? Or would that require coupling transformers to be stable?

[boB]

Looks basically like the Trace SW inverter topology.

boB

[Larsmartinxt]

Why not just use one transformer and feed it with a pwm modulated signal to create a "cine"?

[joestue]

I've been thinking about this more and I think the way to go is the pwm topology, but use multiple paralleled H bridges with separate LC filters before you get to the big 60Hz transformer.
Then as the current load increases, enable more of the H bridges.
That way the switching losses are decreased, and at no load only one out of say 10 sets of H bridges are working, so all you're left with is the transformer losses and one tenth of the static switching losses.
Each H bridge needs about 6$ worth of gate drivers and logic per kilowatt of switching power, so that's pretty cheap, and in practicality each H bridge could handle a kilowatt of power for about 10$ in mosfets.

one of the issues with PWM is the energy stored in the inductor, you need a high frequency inductor capable of absorbing all of that 20-50Khz harmonic power without losing much power.

another issue with PWM is making sure there is no dc offset voltage on the outputs of each half bridge. either that or use a half bridge topology and just eat the reactive current in the rail splitting caps.

things that make you go ... hmmm....

i suppose with the ratings on these mosfets, suppose you were to design a half bridge type drive, where the ac voltage at full load on the rail splitting capacitors was 5% of the rail voltage.
you then over build the mosfets such that they can take 20 times the design rating.
thus you would have inherent short circuit protection, at the price of exploding filter caps if the over current circuit failed.

another problem with PWM topologies
the current drawn from the dc filter capacitors is discontinuous.
if you have multiple h bridges and each is pwm'd, they each have to be separately filtered if you want to interleave the phases to reduce the rms ripple current drawn from the dc filter caps.
when you're dealing with 50 amps of ripple current at 50Khz this is non trivial.
for multi level inverter, the current ripple drawn is the difference between the steps.
so the more steps, the more the current drawn from the battery looks like 120hz, rather than 50Khz triangle waves in the case of pwm.

[oztules]

Joe,
I just did a thing here with circuits of and changes to 6kw pure sine inverter... low frequency
http://www.anotherpower.com/board/index.php/topic,780.msg7495.html#msg7495

May be useful in some way (even transformer selection perhaps)

..........oztules

[joestue]

where the heck did you find grid tie inverters for 75$ each?
with 15 kilogram toroid cores in them?!
lol, looks good man.
i'm surprised you got the idle current down so low.

[oztules]

A company here went belly up, and they sold heaps of these things at next to nothing.
even singly (not by the pallet that i did) they are still dirt cheap.

http://www.ebay.com.au/itm/Brand-New-1-5KW-PV-Grid-Solar-Inverters-Inspire-Solar-/261104169835?pt=AU_Solar&hash=item3ccb05736b

it was insane really.

It does show that using toroids, you can get the idle current down very low compared to the leaky originals.


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

[oztules]

The  gross weight is 38kg for the inspire grid tie unit...... you get your monies worth for sure with these.... the tranny is much heavier than 15kg with all the wire on it.

That inverter I converted weighs 58 kgs... heavy duty units for $760 including shipping to this remote island....

www.ebay.com.au/itm/NEW-Gen-Power-48v-240v-18000W-6000W-Pure-Sine-Inverter-Battery-Charger-Generator-/130765900315?pt=AU_Boat_Parts_Accessories&hash=item1e72414a1b

But with 5kwh idle power for the day, makes it a bit poor for economy, but that seems to be easily addressed.

Are you using three transformers to cut the idle current, or is there some other reason that three is better than one.

From tests here, one seems perfectly good if you use a toroid. The wave is perfect, and the stability is fine. Apart from the original idle current they seem a perfectly good simple way to make a powerful inverter.... and I like the fact that it acts as both a UPS ( from genny or grid pass through 30A 240v), and standalone. Seems they had boats in mind, as they call grid... shore power.

As a stand alone battery system, it can also handle a grid tie inverter ( that solar inspire for example) on the output, and then  let the grid tie charge the batteries, and drive the loads, while it just sits there and controls things.... best of all worlds I think. ( needs a dump load system on the batteries to control excess from the grid tie solar if the batteries are fully charged though... other wise the voltage rises above spec, and the inverter turns off, as does the grid tie for that matter)


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

[oztules]

Joe,
I have been rewinding a motor or 2 lately, and it strikes me that a decent motor core could be used as a toroid (press it out of the motor casing). The teeth.... would they give a leakage path, or can they be ignored.

A quick and dirty test here says that for 50hz, I need about 4.25sqinches of laminate (bottom of the teeth to the outer round of the stator core) to get 1 turn/volt. ( 5" high by about .85" thick... outer to the inside depth/ bottom  of the tooth)

If the leakage is not so bad, it may give you a simpler path (and dirt cheap) to get a decent motor core and wind that for a single 2-3kw transformer output, rather than bother with thet triple core  topology you re looking at......

Is there any reason to go three transformers, given that I have now proven the toroid as a suitable low magnetising current alternative.

I don't need one of these things now, but curiosity is piqued.

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

[joestue]

I've tried winding toroids from motor cores, they suck for the most part, unless you have a rather large core that doesn't have any bolt holes through it, and you have to cut the teeth out, but that's not impossible with a milling machine.

turns out with two transformers you can get 9 levels, which is good enough for me i suppose, i don't know yet, a harmonic trap wouldn't be that big for the 9th harmonic, but it does suck up power i suppose.
also, if the source can be floated, then you only need one transformer rated for 1/3rd the total KVA, at 2-3 times the frequency.

My interest is building my own purpose built inverter.
what i'm probably going to do is when i get my solar panels made (if this happens lol) i'll run them all in series, then boost or buck the voltage to 256 volts, run this into the big H bridge, and the smaller one runs into a 3:1 transformer which is then connected in series to the output of the other H bridge.
this gets me 240vac at whatever KVA the solar panels can produce, and to split that only requires a transformer for half the difference in kva between the two lines

the solar panels would be floating at +/- 256 volts.
the switching elements would be IGBT's backed up with mosfets, so it would be more or less bomb proof and short circuit proof.

[dgd]

A company here went belly up, and they sold heaps of these things at next to nothing.
even singly (not by the pallet that i did) they are still dirt cheap.

http://www.ebay.com.au/itm/Brand-New-1-5KW-PV-Grid-Solar-Inverters-Inspire-Solar-/261104169835?pt=AU_Solar&hash=item3ccb05736b

it was insane really.

It still is insane, thanks for the link I bought two of these A$99 each

Dgd

[joestue]

This project is officially killed, it is simply not worth my time.

unless someone wants to program a uproc for me

[oztules]

Dgd,
These grid tie inverters have some odd foibles.
They under report the power they make, sometimes by as much as 15% or more. If they claim to be doing say 1300 watts , they are in fact doing 1400 watts or more.
I have checked these with 3 different watt meters, and they all agree.... the clincher is that when the solar drops off below 100 watts or so,it reports wild figures for the input current and the input wattage... they go negative ( actually say things like 300000 amps etc, ie gone below zero and around the clock again.)
They still function normally, and report the output figures as positive, but under report... worse as the power goes up.
They under report the power inputs as well, and the watt hours... whoever calibrated the current  adc did not do a good job  of it.

They island properly and do everything else very well, but be warned, design the system so that the panels will not over power the current into them, as they will report a fault, and go off line, rather than just raising the voltage, and shedding the excess power.... or simply even just turning off and trying to start again, they can stay turned off instead until the next day if they experience over current from the panels.... thats the only annoying thing if you have been overzealous in your panel selection.... simply software silliness from some idiot programmer.

Nicely built, and have worked flawlessly but for those observations...... cloud lensing is not your friend if you have gone skinny on the margins.


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


Other than that, they seem fine

[dgd]

Oztules,

Thanks for this info. I have not yet set one up with PVs to try it out but I will do so soon.
I agree they do seem to be well built.  Excellent value for $99

dgd

[jack11]

I wonder if the programmable DDS (direct digital synthesis) chips people use in communications to synthesize various complex waveforms, can somehow be used to generate stable and programmable sine wave reference in an inverter.

The benefits would be: cheap, fully programmable frequency and phase, amplitude resolution probably exceeding the quality of grid waveform, ability to synchronize the output phase of multiple inverters for stacking, ability to quickly synchronize with the grid waveform, .............

Form this point on this small-signal reference DDS waveform would need to be "amplified" to the desired AC power levels. I wouldn't know how, since I know little about the internal inverter design.

[joestue]

A common PLL can be hooked up to an LC filter and get a clean 57-63 HZ sine wave. use the sine to control the pwm duty cycle, vary the power by changing the bus voltage. below a few kilowatts the power control system doesn't have to be very sensitive, just keep the leakage reactance of the transformer high.

but if you want 99% power factor and 99% efficiency, you've got to control the power flow to within .5% or you can't meet those specs.
also, if the grid waveform is sufficiently distorted, to pump a sine wave into the grid, you'll actually be pushing a lot of harmonics into the grid, trying to make the grid into a sine wave, this will increase losses in your inverter.