250 watt grid tie inverter build

[tcmtech]

The biggest issue I have come to see is that the purists assume the national power grid systems are mathematically perfect devices which of course they are not.

What little line noise that one home may make does not go very far considering that in a typical urban residential area it is unlikely that the home level service power comes from more than a block away before meeting up with its first step down transformer that brought it in from a substation some distance away.

That fist transformer is the first inductive filter that noise sees and from there there are numerous other utility transformers circuit breakers power factor correction capacitor banks surge/lightning suppressors and a fair amount of line resistance as well which all dampens that line noise even further.

Point is if the grid can take a direct lightning strike to a main feed line a few miles from anyones home and that home often times sees little more than a power blink a few watts of odd harmonics of locally generated electrical noise is probably not going to be noticed outside that persons own house!

Thats my theory anyway!

 

[johansen]

The biggest issue I have come to see is that the purists assume the national power grid systems are mathematically perfect devices which of course they are not.
Point is if the grid can take a direct lightning strike to a main feed line a few miles from anyones home and that home often times sees little more than a power blink a few watts of odd harmonics of locally generated electrical noise is probably not going to be noticed outside that persons own house!

Thats my theory anyway!

I'm not sure on why the industry has such tight standards.
But the fact of the matter is the power flow is controlled to better than 1 part in 300, or the frequency will start drifting more than they can get away with.
whether or not that was the motivation for making the UL1741 standard so difficult, i don't know. I would like to think there isn't a conspiracy.

As we have it though, if you want a legal grid tie then you upload to the grid through a separate meter and get paid the industry rate which is often one third of the residential rate. I think the harsh reality is most of the grid tied folks are running solar and the grid doesn't need more power during the middle of the day, it needs it at 5 pm, or whatever the peak is for that day.

This is the primary reason why the industry is so excited about being one day able to suck power out of your plugin hybrid and only pay you the difference. (never mind the fact that the battery depreciates at around 25 cents per kilowatt hour, and that's being really generous.)

on the harmonics, they really are the devil when you need your half gigawatt distribution transformers to have better than 99.5% efficiency.

 

[Fendel]

The "senoidal" tension curves our greds seams to be made of micro coils, they are all plenty of picks to the inner side.
So, with a lot of grid tie inverters and assyncronous generators (simple motors) the picks are outsides, and can compensate some of the firts order harmonics.

 

[selimy]

Hi guys,

am new to this forum. Pretty much read most of the inputs regarding gti's here on this site. Tried a few ideas, all worked as expected. Wonder if there are any circuit ideas out there to diy-build an H5 or HERIC gti like the ones built by German companies SMA/SUNWAYS. They apparently have 96-99% efficiency. It looks simple as the H bridge is extended by one or two IGBTs, but how to drive them is another story. Appreciate ideas, thoughts or references. Rgds

 

[johansen]

the biggest issue is controlling the power flow, not the efficiency.

i can build 98% efficient buck regulators with just regular irfz44 mosfets hard switched at 60Khz, and it wouldn't be hard to build two of them, modulate them opposing each other from 20-80% or 30%-70% as the dc link voltage changes, and then you take the output of the two buck converters and feed it into a 98% efficient toroidal transformer and you get a sine wave at 96% efficiency.

trying to get that last 2% out is simply not worth the money imo.

 

[johansen]

so i think i've figured out how to make a GTI from a line commutated H bridge (use IGBT's; SCR's won't work for this, if they stay on its bad news)

but you set up the H bridge so that the switches do not turn on until the line gets to +10 or -10 volts. --this satisfies the intent of the UL1741 standard, which requires cycle by cycle checks to see if the grid is present.

now the output of this h bridge is going to hard rectify the line. if you connect a capacitor across the H bridge, you will in a sense have made a capacitive load on the ac line, using a dc capacitor.

so this capacitor has to be essentially negligible at 60HZ. so we use a 10uF cap.

and then we hook up a boost converter or a flyback converter to this capacitor.
if we use a boost converter, we have to keep the H bridge from turning on until the ac line voltage is higher than the dc voltage going into the boost converter, otherwise there is nothing to limit the current. (so, 1 diode drop in the boost converter, 2-4 volts drop in the h bridge, 12-16 volts going into the boost converter, and we set the h bridge to commutate at +10 or -10 volts.

so how do we modulate the boost converter?
well, since its a low voltage design we have to move the current sense from a resistor between source and ground, to using the on resistance of the mosfet as the current sense. this adds a few more dollars but increases the efficiency significantly.
and we operate the boost converter in DCM, using a 1$ L6562 configured the same as it normally is.
with one exception: the voltage sense resistor network to make the current track the grid's sine wave, we put a 12 volt zenar in series with it, this forces the L6562 to have 12 volts of deadspace across the zero crossings.

and instead of measuring the output voltage of the boost converter, which is normally accomplished by dividing the typically 400 volt output down to 2.5 volts, and feeding this into the error amp, what we do is we take the current flowing into the H bridge (we can use a negative referenced current sense resistor) and we take the voltage across this resistor, lowpass filter it at about say 6 Hz cut off, amplify it 10 fold, (so we don't have to burn up 2.5 volts in the current sense resistor)
and we feed that into the error amp.


obviously a boost converter is not practical because we have to have as much deadspace around the zerocrossings as we do voltage going into the boost converter.
so a flyback is obviously more appropriate.

in real life though, what we probably want to do is either use a buck-boost converter, or a buck converter, and run the GTI from a much higher input voltage.

FYI: this design can only add reactive power to the grid, it cannot receive reactive power, so if the grid is lacking in third and 5th harmonics, it will add 3 and 5 to the grid, but if the grid has a surplus of those harmonics, there is nothing in this gti design to "pull the grid" back to a sine wave.

i have a crude schematic drawn up but its not in front of me right now.

I *think* it is possible to use a DCM boost PFC chip to run a buck converter in DCM and have it feed a line commutated H bridge... but it would be better to use an average current mode control scheme.

 

[Barbe]

Hi...i follow your diagram is not working mate...

 

[Barbe]

Sorry guys..its actually working...i used battery...12vdc but my issue is my output 117v only my primary on 10v...any suggestion

 

[Barbe]

 

[tcmtech]

Input DC voltage might be a bit low for the power transformer you are using or the zero crossing deadband too wide due to not having enough voltage coming from the gate drive transformers.

How are you measuring the voltage and are you able to get any actual feedback to the grid when it connected?

 

[Barbe]

Im measuring using voltmeter....mate, also im using 100uf/400vdc caps am i right do ineed to change it to ac? Sorry but im not familiar with the flow..though i have background in electronics..i have very little knowledge on this one...if you dont mind could you please explain to me regarding the feedback to the grid how can i assure that is actually working? But your design is perfect...but i think youre corect my input dc is not enought(12dcv/4amps battery i used for testing). Thank you again in advance for your help...

 

[Barbe]

All IGBT is in perfect temperature....at the moment im using normal diode...not scr...GBPC5010 i use only the two pin..and my powr transformer primary is 12vac 0-12. Its actually 12-0-12 but when i first used it is not working so i deside to make it 12volts and i parallel the primary...and its working...i get 176-177ac secondary and my primary is 10ac measuring by voltmeter...and the four control transfo have the same secondary 10ac...

 

[Barbe]

 

[Barbe]

Here is my actual build of your design...thank you for this in advance for making this in public youre actually helping others...

 

[tcmtech]

If your power transformer is a 12 VAC unit running it in a grid tie application you would need around 18 volts DC at the input before you start seeing the unit loading up.

For a 12 volt transformer the peak voltage would be around (12 x 1.414) ~17 volts. Then add in a your switching device forward drops and blocking diode forward drop and the small loses with running the transformer in a step up mode and likely you wont see full output into the grid until around 18 - 20 VDC input.

 

[Barbe]

Ah okay then..thanks for the info..seems i need to increase my dc input..i will let you know then later...thanks mate

 

[Barbe]

Hi mate i found this in the net..apreciated if you can have a look..if this is possibly working or need some modefication? **broken link removed**

 

[tcmtech]

I have better things to do that review other peoples designs.

If they come from another forum the overall tone of the thread tends to show whether the design was workable or not.

 

[Barbe]

Yeah mate..is that workable or not...seem you look more prof...than student

 

[tcmtech]

I just throw mine together with bits n' pieces of old salvaged whatevers I have laying around.