Wanted: inverter plans or schematics

[Tyler883]

Hi everyone,

Does anyone have plans or a schematic for a 12V to 120V inverter? I'll take anything you got, including ideas and suggestions.

Of most interest, would be plans or schematics to something like the Trace microsine 100Watt grid tie inverter....but I know the saying....beggars cant be choosers.

Thanks to everyone in advanced.

Tyler

Btw, I'm new to these forums, and would like to compliment everyone on their sincere interest in helping each other. I've seen other forums degenerate into a place of petty insults and arguements, and I hope this never happens here.

Please check my diary in a few minutes, and I'll have a breif introduction about myself

[jacquesm]

I've taken apart a trace 4Kw unit that got hit by lightning, the innards are pretty impressive. There are three pic chips that run the inverter, one for the user interface, one for the powerstage and one for all the peripheral stuff.

The thing uses three stacked transformers that have their primaries switched on or off as needed to build a pseudo sinewave. Because of the size of the transformers their filtering capabilities are quite good, and a fairly small choke at the output takes care of removing the switching noise.

The trace unit really shines in the grid connect portion, power can flow both ways on the grid connection, no idea how they pull that off.

The output has a single loop current sensor around the AC out wire that tells the machine what kind of output current it currently is producing, and it shuts down when it sees too much flow.

It also has a built in battery charger.

Cloning this thing would be pretty difficult, better by your fets in bulk !

If you'd just concentrate on the inverter portion, forget about the grid interconnect, the battery charger and the fancy uP driven user interface it's doable, but still pretty tough. The transformers are custom wound.

There also is a pretty hefty heatsink and a fan that forces air through the fets/heatsink assembly.

Trace has a patent on the switched/stacked transformer setup.

They got bought out (by xantrex), the original engineers have started a new company called 'outback', which is reportedly going to produce a modular inverter.

The microsine is really neat, but I have never found one in real life, around here we only use the 'fat' ones.

Wonder if you could use microsines to 'stack' as a cheap alternative to a larger inverter (but then again, the 4Kw unit is about $1 Can. per watt, which is not too bad for a sinewave inverter with all the goodies they throw in)

[Tyler883]

You are right. we can break the grid tie into several blocks.

How about 1) the inverter 2) the microcontroller that controlls it 3) the user interface 4) the battery charger 5) any others?

1. the inverter - this is the core to the design, it needs to include all the nessecary control curcuits and sensors so that a microcontroller interface with it.
   
2. the microcontroller - I not sure what kind of controller would be appropriate for the needs of an inverter. However, I am impressed with the ATMEGA128 because it it can be easily programmed by any computer, free software, a serial cable(with almost no parts required for the final connection to the microcontroller). The Satelite TV 'testers' use it as the core of one of their smartcard emulator solutions.
   
3. the user interface. beyond being able to reprogram it. we would want to define the other requirements to determine what we need here. Things like a current meter showing direction and amperes could be done as local instrumentation first, then later the microcontroller could add the bells and wistles like data collection.  
   
4. the battery charger - are readily available and reasonably low cost, and I would be tempted to buy a solution, then copy it if no patents were being violated. Here's my ignorance shining through....are there special requirements for the charger, like the chargers that electric cars have?
   
   

I have worked on products that use a single chip solution for establishing a charger profile that includes the different phases of charging a battery. I think that I could track down the data sheets on these and see if there is 12V and/or 24V version available?

[jacquesm]

5) optional: grid tie
6) safety, you don't want the thing to go 'island' when the grid tie connection does not carry power, csa and UL require that you disconnect and power down (ridiculous, but that's their requirement) to protect linemen from getting zapped by your inverter backfeeding the grid. Same on the output side, probably some kind of ground fault protection would be good.
7) a stacking option for 240 V single phase operation (not that hard, just a few comm lines to keep things synched up and to ensure simultaneous shutdown)
8 input for a solar panel bank
9 input for a windmill
10 output for a diversion load

That way you don't need all these extra boxes on the outside, again they could be
'modules' that connect to the main inverter but in an orderly fashion.

These would be in concert with the battery charger that would mean that there is only ONE battery charger, instead of the 4 controllers that I'm looking at in my system right now. (One solar charge controller, two battery chargers (one in each inverter) and one for the windmill, quite messy). It would also require only one battery temperature sensor instead of 4.

Even if right now you'd not design the machine to be grid tie you probably should be prepared to retrofit it.

Another option would be to be able to use one inverter to produce 240V 50Hz for Europe.

The battery charger in the trace will do 60 amps or so on a single 120V line, I modified my generator before I got the second inverter to work in 120V only mode for a more even load on the head, but with a stacked pair it would be better to switch 240 from the generator to battery voltage (less line loss, generator is usually situated as far away from the living quarters as possible).

Power flow through an inverter is really complicated, from what I've been able to glean from the non-burnt wires the inverter basically 'pools' all it's energy at the battery terminals, the various units are interconnected at that point. flow from the grid is optional, as is the external generator hookup. I can draw you a functional block diagram if you want, but that will take me some time and it will have to be from memory, but I'm quite sure I still remember how it all fit together. (that's the point at which I realised that this was way over my head,
so I bought one and a year later another one).

There is a 'bypass' relay that powers the output directly from the generator if generator power is present, the inverter synchs up with the generator so that in stacked mode (2 inverters) and charging on only one leg the output stays in phase.

If you draw more power than the generator can provide it will switch back to inverter mode, this is one of the really neat, but totally unadvertised features of the sw series.

I don't have any experience with the interconnect, but I know a guy that installs these for a living and he can tell you all about them (he's an authorized xantrex dealer). He went through the approval procedure and has a few intertie systems running in the region.

The only atmel project I have ever done was really simple (stepper motor controller), no idea how suitable they would be for something like this, but in my view if you have seen one microcontroller you have seen them all. I realise that they all have their 'following', but the arguments tend to be religious in nature, in the end every controller can do almost every control job. The days of the 6502 are long gone

From what I have seen you will need a minimum of 16 digital lines and a bunch of analogue ones (which can be multiplexed if neccesary). A 10 Mhz controller should be able to do the job easily. One thing about the pic series is that they are dirt cheap (I have a couple of tubes full of them here that I bought at $4 a piece, 16c871's), don't know about other types / brands.

If you want I can sign up for the firmware, I have nothing to do anyway As I said I don't know the atmel too well but it's a fairly easy chip to program, easier than the 'pic', which has a pretty restrictive assembler.

[Tyler883]

Hi Jacque,

This is great stuff. And I would like to preserve the main content. If you don't mind, I'll volunteer to maintian a diary on this.

"5) optional grid tie "

Yes, I would agree that this should be high on the priority list. I'm sure there is more than one way to make an inverter, but I'd like to find a way that works well in a grid tie application. I think it would be a more time efficient approach because it could help prevent a second spin on the inverter design. Also, it occured to me that I might be misreading your comment about 'optional', if so....yes, I think the grid tie features should be optional.

"6) safety....CSA, UL"

A safe design would be more appealing to different types of users. Personally, I don't care about CSA regulations but I do want a safe inverter, and I think that the inverter should be 'CSA capable'.

Where I work, we produce a high cost, low volume device that is CSA certified one batch at a time. The CSA inspector comes in and tests each individual device and approves them. They charge a fixed rate per visit( so far each visit has always been around a half day) of a few hundred dollars.

...more to come .....

regards

Tyler

[Tyler883]

Hi Jacque,

You've mentioned some stuff that I didn't consider, and I think my weakness is that I have no first hand experience with inverters, and I am certainly not up to speed on the available features/modes that different models offer. For example, when you mentioned inverter mode vs generator mode, I havent yet considered a generator because my personal options for alternate energy need to reflect that I live in a city. Calgary, BTW. There's alot more to this than I thought, but I am very interested, I promise to do alot more reading and learning.

Also,

w.r.t. the choice of microcontroller: I hear you on your comments about microcontrollers, I too think that most would work well. Perhaps, a way to make a good selection would be to sit on this for awhile, and see what specific needs develop(probably not an issue) and see who wants to do the programming( this might be the only real thing that matters).

My assembler experience is a bit dated. I've only worked with 6502 and 68xx instruction sets. However, I have experience loading files into an Atmega microcontroller( wink wink) and have lots of experience troubleshooting and testing microcontroller based products. Someday I plan on doing some emmbeded programming. The inverter might be a good project, but I've assumed that there are very talented people out there that would like to get involved, and that I could offer assistance in the areas that I do well at already.

...I've got a family due this evening,  cheers!

Tyler

[jacquesm]

Hi Tyler,

I can mail you one of my xantrex manuals (I have two of them anyway) so that you get an idea of what you're in for, there's quite a bit to these machines.

Calgary eh ? Not exactly next door, but at least within the same country. Only a 3,000 km car ride then...

I cut my teeth on the 6502 (kim, bbc) and 6809's (coco), then 68K (atari ST), then ibm pc's and SGI machines (indy's and o2) (but I never did any assembler on the SGI's).
Some Z80, pic and atmel stuff as well as a laser game on a 8051 (which really sucks).

The last few years I have been busy developing web stuff mostly, I'm sort of getting away from that because I've had enough of it.

The control loop for the cpu that switches the coils on and off is the most time critical thing in the whole machine, you'd probably have to count that out cycle by cycle or all the clocks in your house will be off (unless they're battery operated ., other than that it's mostly doable by polling a few lines. I think that's the reason xantrex split their control module up into several cpu's, like that the cpu that controls the coils has absolutely no other duties and it can never run out of it's tight time critical loop. The inputs to that cpu then become 'produce power', 'sychronize with this pulse' (from an integrator that samples line or generator voltage) and 'shutdown'. The outputs woudl be on the order of 'currently producing power', 'synchronized', 'overload' and so on. The UI cpu then takes these and reports them, the misc. cpu switches relays and determines the housekeeping functions to be done based on these, it also performs a whole bunch of realtime measurements (temperatures, voltages, currents and so on). I can ask my friend if I can have that damaged unit again (I think he still has it, afaik it's unreparable, it certainly looks like that) and shoot some pics of the innards. The whole thing is neatly split up in several boards (fet board, cpu board, mains io (which has a bunch of current sensors and relays on it too), user interface).

A big decision is the input voltage, xantrex make their line of 'sw' inverters in 12,24 and 48 volts, the bigger ones only in 24 and 48. I don't know how much impact
that has on the circuitry, but I can imagine that at 12 V and any appreciable output you're looking at some pretty hefty currents. I know many people use 12 V systems but personally I wouldn't want to go near one, way too many losses. Of course half the world will disagree with that... 12 V inverters are pretty cheap these days ('modified' sw or even square wave), no point in trying to create something you can build yourself at 5 times the price that it takes some chinese manufacturer to make it, ship it across the world and still turn a profit, and give you a 1 year warranty to boot. Some of these things are sold for less than what I would pay for the parts - in bulk. Pretty depressing. But the 'high end' seems to be wide open, nobody has afaik produced an diy version of any inverter > 2000W with all the trimmings. (or even without), certainly not a multi-step sinewave approximation unit. It's a pretty big project, that's for sure.

One other thing, I'm ok with looking at other peoples stuff but I would not want to 'clone' it, also there might be a patent on the switched coil design (someone told me that there is, but that's only hearsay). Beter to look at the functionality and then design the whole thing from scratch. Even if it is patented, as long as you are not 'producing' a device you're in the clear. I'll see if I have some time this weekend to do a patent search. Since xantrex designed their sw line a few interesting components have arrived on the scene, most notably igbt's, I wonder if they would give you an advantage versus a fet based design.

Would be cool to do the whole thing modular, add power as you need it in 1Kw blocks or so... I can dream can't I ?

Enjoy your weekend.

 Jacques

[Tyler883]

Hi Jacques,

Thank you for offering the xantrex manual, however,I just checked their website, and they have operator manuals, data sheets, and application notes available in .pdf format. This will be a good place for me to start.

I'm building a fence today. I've got to rent an auger. I'll have more time this evening.

regards

Tyler

[Tyler883]

To anyone that might be reading this thread,

I've started some related work in my diary. Also, we will likely continue to use this thread and possibly create new ones, too.

The birth of the project looks official. Jacque and I have both both agreed to pitch in, and we would welcome like others to help out.

Well, I've got to run. I got the fence posts in yesterday, so the worst is over.

thanks

Tyler

[RC in FL]

I have one also that got zapped by lightning.  It worked but thought there was no AC1 power so stayed on inverting mode.

Problem was the voltage sense monitoring transformer on the power connection board. It is the small transformers that basically take the 120 vac down to 13 vac which is actually measured by the control system.  

The primary side of the transformers are very fine wire and easily zapped by a lightning surge.

Would be a good idea to at least put a ZNR suppressor soldered on back of PC board on primary side of these transformers.

I also agree the Trace SW series are super units.  Only thing they are missing is a connection for battery current shunt sensor.  The battery charging / generator startup criteria would be much better with a battery current sense deriving a amp-hr useage over just the battery voltage criteria it is limited to.

[Chiron]

The large Trace inverters use line/load sensing to tell if the grid fails. The ones I worked with are "switched" at 8kHz. While the IGBTs for that phase are "off" the controler senses the line voltage, if the line voltage is either too high or low the controler pauses the turbine immediately and shuts down the inverter. Believe me, it's no fun when an interconnect opens up at full production and your uptower when she goes down. You wouldn't believe how much a 60Meter tower can sway

I've worked on larger Trace inverter/VSCF controllers on the Zond 750 turbines. These have even more complex controls and electronics since to maintain 60Hz on the stator the rotor hase to be "excited" with a pseudo-sine wave between 0 and 20Hz depending on rotor speed, in addition, above syncronous speed (1200RMP at the gen shaft) the controler draws current from the rotor. 250kW of he total 750kW of the turbine comes from the rotor at 1320RPM. While not "rated" officialy for that high an output I have seen a Z750 putting out 900kW for short periods during testing.

The Z750A/60 converters use 2 matricies with 6 IGBTs each, 2 matched per phase. There is a "line" matrix and a "rotor" matrix. Under syncronous speed the line matrix is a rectifier/charger for the shared capacitor bank. It's job at that point is to keep a constant voltage on the cap bank drawing from the grid, so the rotor matrix can act as an inverter to the rotor to keep the stator current in phase with the grid.

As the generator spins up faster than synchronous the matricies change roles, the rotor matrix draws current from the rotor windings and the line matrix inverts it and puts it back on the grid. In both operating modes the line matrix keeps a constant voltage on the capacitor bank.

The way they accomplish this gave me some ideas and I'm in the initial stages of designing a small turbine, about 250W, that I plan to tie to the grid with an inverter uptower to take the "wild" 3 phase AC and convert it to 240V/60Hz matched to the line frequency with an single phase version of the Z750 line matrix. 2 large capacitors charged to +and-200V and direct coupling to the 240V line through a switched mode inverter modulated at about 30kHz to keep heat losses to a minimum. I don't expect to turm my meter back with the unit but it's just a testbed/prototype for this and some other Ideas I've gotten from working on the utility sized ones.

[Perry]

Hi all,

Is there anyone still interested in this thread?

I'm presently peeling back the potting compound on one of my microsine inverters to see what makes it tick.

Hoping to gain a few ideas to help build my own unit.  

Cheers,  Perry

 

[Perry]

Ok,  keep an eye on my files.  

I'll add photos and list of components. Hopefully one day a schematic.

I could use help identifying some of the components. Any volunteers?

Cheers,  Perry

[Perry]

As you can see in my file section I'm moving forward slowly.

I'm down to a bare board now. I've started to draw out the schematic of the DC input section.

I need help trying to identify some of the surface mount components.  The first is a transistor marked 1C9I.  Any ideas?

Cheers, Perry

 

[Perry]

Ok, Started with the 24V DC input section.

Uploaded the beginings of a schematic.  I used a freeware program called kicad to draw it out so anyone could download it and view it.

I'm learning the software as I go and some of the symbols (Mosfet, etc) aren't exactly right but you'll get the idea. (I could spent some time and draw out the correct components when I get sometime, but prefer to move forward.)

I'm wondering if anyone is interested in my project? Feels pretty lonely around here.  

Post some comments....good or bad.  Cheers, Perry

[TomW]

Perry;

The smart thing to do here, rather than jump on another old thread, is post a Diary where you can post this in one place that makes sense.

Cheers.

TomW

[Perry]

Ok, as per TomW's suggestion I've moved this to my diary.

Perry

[TomW]

Perry;

I think you will find the Diary to get you more feedback. Plus you can keep it all together in an easy to follow location.

Cheers.

TomW

[Perry]

Seems to have caught a little more attention posted in my diary.

Took a little time away from my project to resize the pictures and set up the diary, but I'm sure it'll be worth it.

I wish I had more time to dedicate to this little poject. I look forward to moving forward on it when ever I get a couple free minutes and it's a tough choice to sit and document it instead of ripping and tearing.

Sunday is looking like a good day for me to make a few big jumps forward again. (Winter should favour this indoor project.)

Thanks again for the suggestion TomW.

Cheers, Perry