And heres the basic AC side.

Attached Thumbnails

 

 

And here is my actual 900 watt GTI with the high resolution watt meter. Set up for 10x watt reading!
Thats 5 kwh if your wondering, and thats just from wind last night!
The analog control board is under the two big blue capacitors. The AC conneting SSR is just behind the left IGBT half bridge and the DC Connection SCR is behind the right side IGBT half bridge. The power factor capacitor and line filter are behind and under the power transformer. The little black capacitor in the center is the 100 uf filter cap on the H-bridge DC side.
The big blue capacitors are not hooked up. They are glued in but dont serve a purpose any more.
Yea, the pair of 200 amp 1200 volt IGBT's are way overkill but I got them cheep and figured I would use them. ($10 each on eBay) Toshiba MG200Q2YS42 if your courious.
This unit is dual range capable. 22 - 90 volt full range or 22 to 45 volt low range as its set right now. It can handle 35 amps at 45 volts for about 2 minutes (1500 Watts) or 90 volts at 18 amps before it overheats. So I have plenty of head room for those high gusts. The top efficiency is around 80 - 85%
turn on at 22 volts gives me about 15 watts feedback, line watts not input watts. input is more like 45 watts at that level.
I built it into an old 24 volt 25 amp power supply case if your wondering what it used to be.

Attached Thumbnails

     

 

Heres the Basic Voltage Sensing Deadband Switch.

Attached Thumbnails

 

 

I think i found what I need I have 6 of them there 30 amp 1000volt half H-bridges QM30DV-2H I played with it last night I stepped my 12 volts up to 240
but it was still a square wave

Attached Thumbnails

 

__________________
Burt
http://mybe80be.com
http://stores.shop.ebay.com/mybe80be__W0QQ_armrsZ1

 

Super!
Being current controled if you are using a transformer drive system for the time being you may need to add a few diodes in series with the base input to increase the turn on voltage point. And dont for get the current limiting resistors on the base input! The specs on the QM30DV do show a base to emiter voltage of 3.5 so you already may be good to go if you are using a 10 -12 volt AC control transformer.

If you use the 6 together (3 on each leg) I would a expect it could run at 15-20 amps Ave input current and probibly a fair amount more!
And possibly up to 500 volts DC input!
(If you have the right size capacitor on the H-bridge. Plus the proper line filtering circuit of course! And obviously you are using a 240 Volt AC transformer not a 12 volt!
That just theoretical, Actual real life often proves different.)

 

thanks to all for the great info...esp TCMTECH..and yes I to have seen PHD disease..

I have ideas and questions..

1) Why not switch AC in a random way so as to "spread" harmonics in frequency domain?? Variations of this trick are done to reduce EMI from very high speed clocks, etc.

2) Why not make every medium size PV panel with its own inverter module?? Does %eff suffer?

3) If every panel had an grid tie could each one also have a low cost wireless interface ( blue tooth? ) that reports that panels health to site PC??

4) For truely large installations A GPS receiver for panel position also?? Single chip receivers are now very chep..

5) How is the line phase sampled?? What about unwanted coupling of GT to phase sample of input?

6) GPS receivers can produce VERY , VERY accurate 1 pulse per second time that is same every where. Can this be used in some way to make GT better? A related question is phase movement of GRID phase to 1PPS phase..hmmm I will measure and post.

-Dan

 

I am not sure what everyone elses thoughts are on size an number of GTI's to use.
But for me I know cost per kw does go down rapidly with size increase And there is a slight efficiency gain too , but you also spend more up front for making one big unit.

Given that you are probibly working on a sytem that is build as you can afford it I would go multiple small units instead. Spread the cost over a longer time scale and you also have that advantage of if one goes down you still have others working.

I wish I still had my old low frequency power line spectrometer. I could post an actual picture of what my GTI looks like on the line side at full tilt.
And when standard power draw devices are on most of them have more odd harmonics coming from then most people would believe. The power in your house is not as clean as you may think.
Put a drop of tap water under a miocroscope. See all the little things you did not know were there? Your house power is like that too! If you start looking at your home power up close like that all of those with electrical OCD would never turn on another device ever again! They just might be letting more harmonic bugs roam free on the power sytem, OH NO!!!!
Use good proven ac line filter design for the power range your working with and leave the miroscopic harmonic bugs alone! They are there and they will always be there its just part of an AC system.
I use a real time voltage sampling method for phase monitoring. That simple cotrol transformer driving the H-bridge is a very reliable and simple way to make it work. just make sure you have a wide enough deadband for the line side (power ransformer) wave form to change from its positive half to negative half. (zero crossing region)
As I have stated before starting and stopping the H-bridge conduction cycles at around 20-30% of peak voltage works very well. And there is no complicated phase formulas to work with!
I found that region works well by trial and error. Burned up a few IGBTS finding out that the area of a sine wave close to the zero point is useless to try to dump power on. Vicious high amps to turn on and off and little usable counter emf to work against! harmonics generator anyone!?!
Plus the transformer and power factor correction need a dead zone for them to resonate properly. That way the line side winding is still working with a sine wave even if the switch side is dumping square wave power.
And when you are sampling the line voltage for phase reference do it on the line side of the filters not the GTI's power transformer side! far far less chance of starting a runaway resonance/feedback loop!
Believe me, a odd frequency power resonance or ossilation does not go well with anything in the house! Flourecent lights grumble and chatter, the refrigerator compressor makes an terrible sound, digital clocks loose track of time and then the harmonics police will come looking for you!

Bluetooth the the GTI sytems. That would be interesting if you could get each to supply a real time output of whats going on. Data log the whole system. I bet a university or company doing alternative energy research might pay a bit to get real life system log info.

 

Here is something interesting to think about. calculate the length of a 50 or 60 hz wave. Lay that over the distance of your continent you live on. Now factor in that power plants are dumping onto it at all points along it and power is being pulled off of it. Then factor in phase drift caused by the load changing locations durring a 24 hour period. And the supply point does not have a fixed origion. Every power plant and every load along that wave length is floating not static. The master wave ties everything together but it does not have a fixed point of origion or end.

Interesting! No true single master point to reference things to on a time vs wave point scale.
Perhaps thats why clocking does not so work well. The phase peak will always change depending on where you are on it and where the power is coming from and going to at that instant. It is only relative to your position not fixed.
Washington state is some degrees off from florida if you do the math. But who is the master reference? Is washington running faster than Florida or are they running behind Florida? Who is front and who is back?
See the wave relitivity problem?
You have a multiterawatt standing wave but yet there is no true ends that it is bouncing between!

 

Hi All;

I am into GPS so I know a bit about that and see it getting into everything..

Yes now I understand the deadband , that is by looking at line phase when you are not "ON" you get it "unperturbed" grid phase. Clever.

I had another idea on the distributed approach, that is a GT in each PV panel.

My understanding is that filtering and magnetics of size are involved with "true" ~sinwwave approach??

Is it possible that with a distributed approach the resulting inverter output is a sum of approximate modified sinewaves?? ( is the GT zero crossing not now a integrated sum of zero crossing time instants whose mean is ~1/60 sec, but average period set precisley by crystal reference)

If the start stop time is also modified in a noise like fashion as described above could it be that resultant sum has better sinwave properties with smaller magnetics ( lower cost?)

By using blue tooth is it possible to coordinate the sum in fashion that improves %eff also??

Just for giggles one of the first GPS displined clocks , design by HP, was used to record Time of Arrival of power line fualts at distant locations..then by propigation of wave a distyance can be calculated, location of fualt found.

-Dan

 

Hey TCMTECH;

I have seen "inexpensive" software that takes audio input from PC and does FFT on it, plots spectrum.. could that be a low cost power line spectrum analyzer??

You might have to modulate a carrier at ~ 5 to 10 khz to get into bandwidth of this instrument, but once cal'ed it should be fairly honest interms of spectral lines ...the ine I saw had sub hz resolution...think it went from low hz to 20KHz..

-Dan

pS excuse typos..Fault is not FUALT.

 

There rated at 30 amp 1000volt how you get "15-20 amps" 3 of them where controlling a 30amp 460volt motor they are made to handle 2.7kw . So if they can handle 2.7 kw at 1000 volts looks like they could easy pass 75amps at 12volts maybe even a 100amps and not get hot seeing I have heat sink that is 20 pounds to go with them?

__________________
Burt
http://mybe80be.com
http://stores.shop.ebay.com/mybe80be__W0QQ_armrsZ1

 

DanGPS first.
neat way to use GPS. I think i understand the priciple you are talking about. its sort of similar to those line thumpers that echo a wave off a break in a electrical line. but much bigger and longer distances can be tested. or am i way off?

Second Part : what?
Actualy I sort of get it but i need a better expaination. It still seams like a complicated line noise maker just waiting to happen. But again I could be completely missunderstanding the concept your trying to discribe.

Third part: never thought of using an audio spectrum program that way!
If it works I am putting it on my CD in the Dumpster Diver tips section. And I will credit you at this site for it!

Be80be

True logic. But...
Heres the problem. when you switch them on into the transformer you are putting that 12 volts aginst the sine wave low down by the zero end. Your esentialy trying to dump 12 volts into a far less than 12 volt counter electromotive force.
If you can measure your transformer secondary resistance. Say its a good sized copper winding of say .05 ohms. Dump 12 volts across it you get 240 amps. 12/.05 = 240
put even 6 volts counter emf from higher in the sine wave and you are still pushing 6 volts back above that. 6/.05 = 120
Actual inductance will reduce that number more but as you can see thats how you get a high amp load at every half wave turn on but then have to turn off that high amp load at the end of each half wave.
Real life taught me that one. I kept popping IGBT's for no apparant reason until I got a hall effect type DC current sensor and hooked up my O scope to its output.
my average amps were only about 10 or so but points where the IGBT's were switched on and off were pushing 40 plus.
I know that PFFFT.... SNAP sound by heart! Semiconductor switch distruction!

With proper current control feed back then yes you could possibly push your 30 amp rated transistors at 20 amps plus per device. BUT it would still be risking a current overload and possible die burn up. You would have to check the actual manufactures specs to see what they are rated for disipation wise in this type of switching application. IF your in the safe zone your fine but get close to the edge and PFFFT!

Real world inverter circuits that I have studied and repaired have a typical 4 to one current rating and a two to one voltage rating over their actual running amps and running volts at the switching devices.
A good model plasma cutter with a 25 amp draw at 240 volts typicaly has the switching H-bridge circuit built to a 100 amp 800 volt or higher capacity per corner of each leg.

Some do go less than that but are more prone to burn outs and others go even higher and rarely burn out. Century Vs Hypertherm! Snap-on Vs anyone but century!

Snap-on, craftsman, marquette, solar, napa, some black and decker and alot of the generic welder and plasma cutters are made by century! Yep your high dollar Snap-on is a Century! Now arent you proud to know you spent 3 times the money as what a century with the same ratings costs to help fund a poor starving multi millionares rediculous hobbies!

 

You asked for a "clearer" explanantion of my idea on "randomizing" a modified sinewave..

Actually I see this as part of the sum of modified sinewaves from a distributed GT , say for 20 or so panels ..

If you modulate the starting/ending angle in a random fashion, but keep the overall period fixed via xtal time base that should spread the harmonic lines generated by modified sinewave? Pure harmonics can only be generated when all timing edges remain "fixed" ( my understanding)

Are you moving/modulating start/end angle of modified sinewave to achieve current control?? If so then I would expect a Harmonic line to have a width somehow proportional to rate and depth of modulation. In other words if you zoom in on a harmonic line , its not a line but is widened by your current control method?

If instead you add randomly ( by small angle deviation) on top of the current contol point you seek ,whose average value is zero then you have not modified average current "command" but you have further widened or spread any harmonic line??

I see this idea to work inconjunction with summing multiple panel based GT units in such a way that the sum of modified sinewaves has lower harmonic energy per "unit bandwidth" , and that due to random nature has lower probability that positive adding of modified sinewaves increases harmonic power levels?

If I decrease harmonics I should get a better sinewave, right? Is it possible such a method of an array of GT ( 1per panel) needs lower cost filtering(smaller magnetics) and acheives higher %eff ?

I will try and find the HP article on the power line fault detection using GPS clock and post.

-Dan

 

see attached for the app note from HP that describes ( as part of it) the motivation for HP developing that product, which was intertie fault detection over large distances.

Attached Files

HP_Smart_CLk.pdf (1.40 MB, 318 views)

 

Interesting!
I have not had to aproch the real world problem of multiple GTI's running high total power outputs yet. But I feel you do have a good and possibly very valid point. Plus what you were saying about spread out of harmonics seams reasonalbe too.
I do think that multi GTI setups could possiblly produce some harmonics amplification problems, maybe.
However I tend to aproch things from the simple (Complexity + time + money) Vs (practicality + reliability + user intelegence) Formula.
Complex, hard to build, and difficult to understand makes the potential user less likely to want to use it.
Plug it in, turn it on, walk away and never look at it again. That sells!
Also production issues are a real factor on a DIY anything. Can it be built and work reasonably well just from cheap common found parts I already understand?
Or is it potentialy difficult to build and parts are expensive and hard to work with?
Cheap, simple, reliable and functional always sells!

I will take a 60% top efficiency that is user freindly and reliable over 95% efficiency thats difficult to work with or on.
After all we are talking about using energy sources that only cost what it takes to capture them, right?
Do I spend $500 on a GTI that runs 95% efficient conversion from my 100 watt solar panel or do I spend $150 for 60% efficient GTI and just buy another 100 watt panel to make up the power loss difference with interest?
It purely comes down to how each person aproches the concept.

I am leaning hard on the simple, low buck reasonalble to build end with the help I am giving here. But I do also know there are more complex ways to make this work with some added gains and benifits. But right now I am still focusing primarily on simplicity and actual how it works theorys.

Start basic and go up. Each person will top out at where they are comfortable. Some will be running the basic two transformer unit design years from now, others will have moved on to far more complex and complicated designs that far outshine my own!