• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Grid Tie Inverter Schematic 2.0


Most Helpful Member
Thread starter #1
Since many of you have been patiently waiting for a new and more complete GTI design I figured I would give you some new schematics for a slightly more advanced system to think about for a while. :)

This is in a way a continuation of the first Grid Tie Inverter Schematic thread from last year so I will not be going over everything in fine detail immediately regarding the operation and function of the exact circuits. Nor am I going to spend much time regarding the defense and argument of what is legal, illegal, safe, or unsafe in any country, location, or personal application relating to experience or inexperience by any individual. To that end here is my general liability statement pertaining to the issues the arm chair safety Nazis seem to continually have so much trouble dealing with. :p

This GTI build up Thread is intended for conceptual purposes and does implement live line power level voltages in some parts of the circuits which if mishandled can injure and or kill. Should you decide to build and use such a device based on this or other GTI schematics I or anyone else has provided freely on the internet you yourself are liable for your accidents, injures, property damage, and other unforeseen happenings that may negatively affect, injure or possibly kill yourself or others due to your choice to attempt to build and use such a device.
By building such device as these schematics pertain to in original or modified form you may be breaking laws and could be subject to possible fines, jail, and or prison time for it.
Should you choose to build such a device as this and intend to use it for saving energy in your home or dwelling you may very well be breaking the law. Beware by building and using such a system you will be deemed a pirate cogeneration operator which is considered illegal in many countries!

Basically this means you have to be proficient enough at understanding schematics and electrical and electronics principles to be able to safely build and fine tune a design or circuit such as this and adapt it to your specific needs required for your location and get away with using it as well. What you do with information you find on line and how you apply or misuse it is basically not my problem. ;)

Once again this is a “conceptual circuit” lacking the full safety interlocks and other features that a full legal system would have and use. However when built properly this circuit is a fully functional GTI system based on an exact circuit design I have been using successfully for years now in multiple GTI’s of greatly varying power handling capacities I have built over the years. These as drawn schematics however do not show the various indicator LED’s in place that normally indicate when specific components are on or off on the real as built GTI units I use them in.

The first schematic, ‘Power handling circuit.’ is the high current higher voltage power handling system which is based on a standard H bridge design. In this application the IGBT’s are driven from actual driver IC’s and not a small control transformers as was shown in the old thread. By using the driver IC’s a more precise zero crossing point timing and switching device efficiency is possible which also improves the overall performance and control of the system. This also gives the greater advantages of being able to manipulate the waveforms in different ways relating to additional current and voltage control feedback circuits that can be used to protect and limit the power that may be applied to the GTI power handling systems as well. As you will note there are multiple lines, [To (A)] through [To (K)], running to different control circuits not shown in the ‘Power handling circuit’ schematic. They relate to the ‘Control circuit A’ schematic and an optional ‘Control circuit B’ schematic which specifically handles the higher level safety interlock functions. Plus there is a ‘Control circuit C’ schematic which adds full PWM waveform shaping and further current control capabilities plus external interface functions as well.

The ‘Power handling circuit’ and the ‘Control circuit A’ are the basic systems that will allow for full operation, with limited system protection, of this type of GTI system. The other two control circuits are the extra support systems needed for additional features but are not essential to the basic system operation and will not likely be covered any time soon. However if built properly this more basic design does allow for them to be added on at any time by simple plug and play additions.

Unfortunately I am not sure how quickly I will be able to get to doing the more in depth descriptive write ups on this GTI design being I have been hired to drive truck in the oil fields and work long hours. For me this is a low priority hobby project of mine and nothing else. You will get what I give when I have time to give it. this may be a better schematic and system design than the first one I gave out last year but unfortunately you will still have to think a bit more if you choose to use it.

If your electronics skills are good you will be able to work out what everything does and why it is needed. This circuit is highly adaptable and can be used with any input voltage and current within the working ranges of the components chosen, It can adapted to work on a low voltage low current system that can work with as little as a few tens of watts from a 12 volt based system up to that multi kilowatt capacity system that run at inputs of up to around 400 volts and over 100 amps. The adaptability comes from the design only needing a few changes in the sizing of a few specific components. There are only a few capacitors, resistors, diodes, switching devices, plus the power transformer and its related line side post filtering circuit that determine what voltage and current capacity this circuit can effectively work on.


2 Kw Home GTI Pow&#101.png 2 Kw Home GTI  Con.png 2 Kw Home GTI  Con.png
Last edited:


Most Helpful Member
Thread starter #2
Power handling circuit.

In order to adapt this circuit to work with different voltages and amp loads there are only a few basic changes that need to be made to specific components.

The main power transformer needs to be matched the continuous and peak loads that will be produced by the input source. The way this GTI design works will allow most good quality transformers to run at working wattage levels equal to their VA ratings provided that the C2 and C8 values are close to correct. C2 and C8 are part of the two power factor correction and waveform shaping LC tank circuits that change the highly variable input waveforms into the acceptable and reasonably clean sine wave based power that goes back onto the utility lines.

Using the IR2113 driver IC’s to control the switching devices also greatly reduces the switching spikes and associated harmonic ringing effects that the old direct control transformer driven switching device design used. This is how this circuit can get by with using much less capacitance on C2. Now around .2 uf per input amp is needed for a general starting point. Depending on the transformer, switching device operating characteristics, and the waveform shaping methods used this ratio could still need to be raised or lowered though.

R1 and C3 are optional and are mostly just for additional spike and potential HF ring suppression. Should they be used they will need their values adjusted if a high frequency PWM waveform shaping and control method is implemented.

C1 is the primary filter capacitor and can be any common larger value electrolytic. In this schematic I have it shown as a 2500 uf value however depending on the size, stability, and peak power capacity of what ever input source is being used this value may need to be much larger.
This schematic has it sized for a input source that has a good peak power reserve and stable supply capacity up to around 40 - 50 amps. If the power source is of a less stable design or comes from a location that has a long supply line between it and the GTI the relating input voltage drop from line resistance could require that this capacitor may need to be far bigger. Possibly even 10X what is shown here for operation at a stable 40 - 50 amp load condition if the line losses or source pull down are bad enough.
Regardless of the power levels being used over sizing this capacitor will never hurt anything regarding the GTI’s overall performance and will likely pick up some effective returnable power just from smoothing out any supply line losses in many cases.

C2 is part of one of the power transformers LC tanks plus works as the primary switching spike suppressor/snubber for the H-bridge system. Electrolytic capacitors will work Okay for this if their working voltage is rated many times higher that the systems actual peak waveform voltage. Preferably larger value Poly or Mylar capacitors or AC power factor capacitors work best due to their having superior ESR numbers over electrolytic capacitors. The ratio of capacitance to input amps is not exact due to the variation in the quality and characteristics of the power transformer and the switching devices or how they are driven as well. A good high quality transformer will have far less switching spike energy and ringing harmonic noise than a low grade one. Having the switching devices solidly turned on and off with minimal transition times helps reduce these effects also.
In this circuit a 10 uf poly capacitor is used and has no problems cleaning up the switching spikes while running in the 40 -50 amp peak ranges occasionally encountered in normal real life day to day operation.

Q1 – Q4 are the primary switching devices. IGBT’s or Mosfets are most efficient and are directly drivable from the IR2113 IC’s. The same rules apply as in the old basic design in regards to the working voltage and current capacities that these devices need to be rate for. A minimum of 1.5X working voltage headroom and a 4X working current capacity are highly recommended. Being the IR2113 IC’s have a 600 volt working limit they are not recommended for use in a system that may see greater than 400 volts on the H-bridge circuit.
Being this circuit was designed for a GTI I designed to have a working capacity up to around 50 amps I used rather large 200 amp rated IGBT devices. Also the IR2113 IC’s do have high output current capacity so they are capable of driving multiple individual IGBT’s or Mosfets paralleled together while still working at higher frequency ranges.

D1 is the primary power handling and feedback blocking diode so it must be rated for the same working parameters that the Q1 – Q4 devices are rated for as well. The 1.5X minimum working voltage overhead and 4X working current rating applies here as well. If a high frequency PWM waveform shaping system is being used this diode must also be capable of working in that frequency range as well.

C4 and C5 on the drive IC’s are presently sized for use in a low frequency application where long on times with moderate gate capacitances and leakage effects may be possible. For other working frequencies or driven devices refer to the IC’s specs sheets to correctly size them to the specific application.

C3 is the primary power factor correction capacitor and also the primary and final line side waveform clean up device for the GTI system. In this circuit the power transformer is roughly a 2 KVA unit out of a large UPS unit which worked best with a 35 uf capacitor for the LC tank circuit. This capacitors value will be different for every size, type of transformer, line voltage, and line frequency used and will have to be sized accordingly.

L1 and L2 represent an off the shelf line filter device like what is common to most electrical and electronics appliances and devices. It is optional but still highly recommended. Simply use a filter unit that is rated for the line amps and voltage you are working with. If you are not familiar with what one looks like they are most commonly a silver or grey box that has the two power lines going in one side and then coming out the other. They are most often directly attached to the where the primary power goes into an appliance or home electronics device.

TR1 is a Solid State Relay whose input is represented by the opto isolator SSR in the ‘Control circuit A’ schematic. Size the SSR device accordingly to your GTI’s peak line voltage and current handling requirements. In some locations it may be necessary to use a SSR for each line as well so design accordingly.

From here you should be able to design and build a workable circuit capable of being tuned to the proper working characteristics of the components you have chosen. Ideally an oscilloscope is needed for analyzing the specific waveforms related to the fine tuning of each section of the system but basic trial and error will work if the rough numbers I have given are followed. Unfortunately due to the vast variations in the quality and type of the components that may be used there is no guarantee that the numbers I have given will be exactly right the first time. ;)
Last edited:


Most Helpful Member
Thread starter #3
Control circuit A

This is the primary control system that handles the basic synchronization and line connection interface for the GTI unit. Its purpose is to take the two halves of each sine wave and convert them into properly timed signals that can control the switching device driver IC’s. It also determines when the GTI should connect and disconnect from the line as the input power levels change.

TX1 is a standard step down transformer with a 24 VAC center tapped secondary winding. It supplies the power for the control circuits and the base timing reference points required for generating the proper zero crossing dead bands needed during each half of the switching cycles. Being the control circuits are a small electrical load a transformer of only several VA is needed. If built properly the control circuits themselves will only use around 1 - 2 watts peak.

The zero crossing dead bands are controlled by the dual op amp IC U2. Its purpose is to compare the voltages between each side of the transformers secondary to the reference voltage produced by R4 VR1 D5 voltage divider circuit. As the sine wave, as sampled by D2 or D4 and trimmed by ZD1 and ZD2, rises on one side of the center tap it creates a positive voltage on the + input of the corresponding op amp. That voltage is compared to the voltage of the – input and once the + input is higher than the – input the op amps output goes to a high logic level output.
By using this comparison method the zero crossing band gaps can be effectively varied to limit the inrush and post shut off currents the power switching devices need to handle during each half of the switching cycle. This simple comparison method makes it possible for true real time control over the switching devices during each half of every cycle. If one half of one cycle goes missing for any reason or simply falls below the comparators switching threshold its timing reference is ignored and no switching device actions are initiated for that half of the cycle.
The [To (K)] input through D7 also makes it possible for optional biasing of the zero crossing dead bands by using an external input voltage to raise the comparators – base reference voltage to any point higher than the minimum level set by VR1. This can achieve a basic degree of cycle position and width adjustment that can be used for an effective input power and switching device current limiting system. Also by being able to take the – reference voltage to a point above the highest point of the + input reference signals the op amps outputs can be held in a low output state regardless of the + inputs timing signals thus preventing any timing logic from being sent to the driver IC’s should it be necessary.

U4 is a common LM431 precision voltage reference IC that is use with the U3 (555 timer IC) to control the GTI’s line connection logic. R5 VR2 R6 creates a voltage divider circuit that is referenced to the DC input of the GTI unit. Depending on how much standby power is required for the power switching devices, primary power transformer, and the control circuits there will be a minimum input voltage where the GTI can not effectively return power back onto the lines. Using VR2 to set this minimum operating voltage point is necessary to prevent the GTI from taking more power from the lines than what the control circuits need when the input power source is to low.
When U4 is activated it discharges C3 through R8 and D6 and sets the 555 timer to produce a high output which turns on the primary power SSR and connects the system to the lines to send power back. When the input voltage drops below the set point of U4 it turns off and allows C3 to charge through R7 and VR3 until it reaches the threshold limit of the 555 timer where it then turns its output off and disconnects the GTI from the lines putting it into standby mode.
The standby time delay is necessary so that the GTI is not constantly cycling on and off line if the input power source has regular dips that go just below the minimum power return limit of the GTI. These dips may be simply from shadows crossing solar panels or brief dips in the wind when using a wind generator system. The time delay can be set from 100’s of milliseconds to tens of seconds. By changing the value of C3 that range can be made longer or shorter.

Without the full ‘Control circuit B’ in place the connections [To (J)], [To (K)], and [To (I)] (from the power handling circuit schematic) are all tied to [To (E)] and grounded to an off state.


Most Helpful Member
Thread starter #4
Basic GTI theory of operation.

A grid tie inverter is not a stand alone inverter like the common 12 volt DC to 120 VAC devices most people are familiar with. A GTI uses an entirely different approach to how it produces its AC side output power relative to its DC side input power. The more common misconception is that a GTI works on the same principals as a common stand alone inverter and can thusly provide an AC output even when not connected to an AC line. Or that if the AC line goes dead and the GTI is being supplied adequate power to its DC input side it will continue to keep feeding AC power back and create “islanding” or other dangerous power feed back conditions. This is a false belief and should be considered generally a wrong way of thinking as to how a GTI works. By design a GTI system is not capable of working as a stand alone inverter which generally prevents it from ever being able to work in a stable free running stand alone condition while at the same time back feeding power to any type of load on the AC line side of the unit.

The primary reason a GTI can not work as a stand alone inverter is simply due to how it works in regards to creating the AC voltage and current it superimposes back onto the AC power line itself. Each half of the AC wave form is both referenced to and created in phase with each relative half of the AC sine wave it is being superimposed onto. This means that a GTI has no internal master oscillator or clock that determines the 60 Hz frequency or correct phase switching event timing of the output devices. The GTI power output is entirely dependent on the AC line it is being fed back into. Without that AC line voltage and current present the unit itself has no timing or phase reference points to work with and thus no stable power can ever be put out on its own let alone support any type of load itself. Simply by design alone a GTI can not work without the AC power source being present for it to feed back into. Without it it will quickly destabilize and will tend to stall into a naturally off line state regardless of how much input power is available or how small the AC line load may be.

There are three basic factors that are in play regarding the operation of a GTI which by their nature are what make a GTI different from a common stand alone inverter device and thus prevents it from ever working as a stand alone inverter device when being used in realistic operating conditions.

1. In a common stand alone inverter unit there is a master clock or oscillator that tells the output devices to switch at a constant frequency. A GTI has no master clock or oscillator that governs when the output switching devices are turned on or off. Without being connected to a stable AC line a GTI has no means of producing any type of stable AC output. Without that master clock the GTI control and power circuits always want to go to an off state without the master reference points being supplied by the AC line itself.

2. In a common stand alone inverter the primary output power is derived from a constant voltage DC rail that gets converted into an alternating current by the output H bridge switching devices. A GTI has no constant DC rail voltage rather only the highly variable voltage that is supplied by the DC input power source which can vary from zero to slightly higher than the peak of the AC line voltage on the switching side of the main power transformer. A GTI is a variable voltage source inverter not a constant voltage source inverter so it cannot support any type of loads by itself.

3. In a common stand alone inverter the input voltage must remain within a specific range and be able to supply sufficient current for it to carry any load on its output. A GTI does not require a stable input voltage from its supply side power source but instead works with the highly variable input power from sources that have limited or variable voltage and current capacities. A GTI is designed to work with input sources that can have wide voltage and current swings and by design only use what power is readily available from that source. The output side of a GTI can not work as a stable power source due to its input source not being constant or stable and it has no provisions for it to stabilize its own output to compensate for the variability's of the input power source as well.

This is basically why a GTI cannot create an unsafe feedback condition during a power outage in realistic working conditions. By its design a GTI it is not capable of working in a stand alone mode condition.
Last edited:


Most Helpful Member
Thread starter #5
Zero crossing dead band.

The purpose of the zero crossing dead bands is to create a safe turn on and turn off point for the switching devices to operate between during each relative half of the sine wave and also to provide a sufficient dead band zone between each half of the sine wave as it reverses its relative polarity during each half of each cycle. It also has the added function of hiding the slight phase lag effect of the control transformer where all timing reference points are taken from and any other time lag created by the switching control systems as well.

This dead band area can be accurately monitored or be manipulated by using simple voltage comparator circuits that monitor each half of the wave and provide a solid high or low logic output that relates to when the AC sine wave voltage goes above or below a reference point determined by the reference voltage applied to the comparators negative inputs. By comparing the two voltages a basic PWM method is produced that can be used for multiple control functions of the power handling circuits such as input current limiting, switching device peak current limiting, and power transfer control.
By moving the reference points up or down in relation to each half of the AC sine wave the dead band time and on state time of the switching devices can be varied one way or another. By turning the switching devices on at a point that would be low on the sine wave and close to the zero cross point a wide on state time is created during each half of the wave. This will allow for a low GTI input power source voltage to be picked up and fed back into the AC lines. However when a much higher input voltage and current is available this can create a problem for the switching devices now needing to switch on and off at very high currents which can damage them due to a gross over current switching conditions.

The reason for this is can be thought of in a way as variation of basic ohms law.
If the GTI power transformer was a 250 VA rated unit designed to have a 10 VAC nominal rating on the low side windings, H Bridge side, its peak voltage would be approximately (10 x 1.414 = 14.14) volts. Its winding resistance may be around .1 ohms so basic ohms law would dictate that if a 12 volt input power source was applied to the transformer the current it could pass would be around (12 / .1 = 120) amps. 250 VA can be equated to around 250 watts so ideally the transformers input current should be around 21 amps on average.
The easiest way to limit the current is to simply reduce the effective voltage across the winding by either reducing the input voltage or by increasing the counter EMF of the winding to make the 12 volts appear to be less than it really is. If the zero cross dead band switching point is raised to a level of 6 volts then the transformer is producing a 6 volt counter EMF against the 12 volt input crating a 6 volt differential in the transformers windings instead of the full 12 volt differential. At 6 volts the switching devices now will only see half of the original 120 amps and will need to switch a peak of 60 amps instead of 120 amps.
The 60 amp peak may sound like it is still considerably above the 21 amp working capacity of the 250 VA transformer but it only represents a fraction of the total half of the sine wave. From the initial zero cross point up to 6 volts there is no current flowing through the switching devices. Also from 12 volts to the peak of the wave at 14.14 volts and back down to 12 volts there is another zero current flow time.
The actual switching device conduction current will start at 60 amps at the 6 volt point and then rapidly drops back to zero as it approaches the 12 volt EMF equilibrium point. Above that 12 volt input voltage point the sine wave peaks and then begins to drop down to the 12 volt point where once again the switching devices will start to conduct again until the 6 volt point is reached where they now are turned off while conducting 60 amps once more.

So what does all that mean? Basically the two times during each half cycle where the switching devices are conducting represent about 1/3 of the effective wave and averages the total current out to around 20 amps. Because of these peak Vs average amp ratios it requires the H Bridge switching devices to have at least a 4:1 current ratio over the estimated average current that the GTI will be subjected to.
In this case a 250 watt GTI working on 12 volts would have a 21 amp average draw but the switching devices may be subjected to regular current switching peaks 3+ times higher than that.

Although this is not an exact mathematically or electrically correct description or explanation it still gives a simple but rather accurate and realistic ratio of switching currents to input voltages and a general representation of what is going on with the switching devices during the zero cross dead bands and what they are subjected to during the corresponding switching points as referenced to each half of the sine wave.
Once the actual inductance of the transformer and other circuit characteristics are factored in this peak switching current number drops further and a realistic zero cross switching point of around 20% - 30 % of the peak voltage is typically a safe starting point to work from.

With additional current sensing and control circuits to limit the peak switching currents seen by the switching devices this zero cross dead band can work at lower relative percentage levels but once below a certain point the effective switching amps will become very high even though the overall power transfer to the line side of the GTI unit will become less and less. This has the effect of placing a practical lower limit on the working voltages and currents relating to the switching points zero crossing dead band.
Last edited:
Hi tcmtech, I have got to say that this circuit is awesome and I can't wait to build it up and give it a go.
Just a quick question of curiosity, why have you used 100R gate drive resistors? Why not something like 10R to reduce switching losses? I'm assuming that this has been chosen as a balance of dv/dt versus switching losses?


Most Helpful Member
Thread starter #7
Well sort of. Ideally the value for those resistors will depend on what the ratings and design of the switching devices are. If you have IGBT's or Mosfets that can work at high switching speeds by all means use lower values.

For me I typically have large multi hundred amp rated IGBT's on my H bridges and their naturally slower turn on and turn off times have negligible power losses associated with them being they are switching at the 60 Hz line frequency. At 60 Hz the dv/dt and related losses dont add up to much. However if a higher system voltage or a higher frequency PWM waveform shaping method is used then the potential switching losses will become far more of an issue.

Originally I used the 100 ohm resistors simply because the big IGBT's I use have such large gate capacitance values that they caused odd problems in the control circuit power that would cause odd switching ringing or other control circuit interference issues if the snubber capacitors on the control circuit IC's where marginal or the physical layout of control circuit was less than ideal.
It was basically just a quick fix that sort of just ended up staying in the base schematic design more or less even though I dont necessarily use 100 ohm resistors in the actual GTI builds myself.

None of the circuits values or designs I have here are set in stone so by all means if something needs to be changed to suit your particular components or circuit requirements do so.
Last edited:
I see what you mean, so once I have selected the switching devices and tuned the on/off switching times I should further tune the z-cross dead band.

Any advice on tuning C8???


Most Helpful Member
Thread starter #9
Depending on what size and quality of transformer and line voltage you are using that value is also highly variable. The more direct and simple way does require a bit of trial and error to find the closest realistic value.
Basically what C8 does is balance the transformers power factor at near 1 and work as a snubber or damper for any harmonics or ringing effects that may occur if the switching devices are not being driven properly or are being driven hard at a full load. Being the input side of the transformer is seeing a square wave from the H bridge the output side LC tank effect helps further clean that up and round off the square wave to produce a reasonable sine wave shape that is acceptable and will not cause harmonics problems on the line side of things.

The best way to tune the LC tank is by reading the idle current of the transformer and C8 combined and then adding more capacitance to C8 until the line current reaches its lowest level. You are looking for the point where the LC circuit goes from being an inductive load to a near resistive load but not all the way to being a capacitive load on the line when the system is idle.

Due to the variable nature of the GTI circuits as they feed power back at different rates means that just getting close to a neutral power factor when the GTI idle is going to be close enough.
Ok so this what I have done so far....

I have built the circuit using a 100VA toroidal (240V-36V). However I have not yet inlcuded the UVLO part of the circuit at this stage.

Tuning the z-cross dead band didn't work to well using an LM358 op-amp, the slew rate is to excessive. I basically had to replace the comparator reference with a 500k trim pot divider to get the half-cycles to not overlap.

I used a 70V DC power supply to act as the DC source. I managed to get 3A out of it. I'll get some waveforms of this shortly. I'm assuming (or hoping) that most of this power was being delivered back into mains, although I won't know for sure until I get my hands on a power analyzer (I have one lined up for next week). I'll also be able to tune C8 further for a close to unity power factor once I have that piece of equipment.

Now everything seemed great, and I was on top of the world! UNTIL I disconnected the GTI from mains and it kept running!!!! Now I haven't spent much time yet looking into this, however I'm guessing that this could be avoided by increasing the z-cross deadband??? Any advice???? I have some waveforms of the primary vs secondary voltages which I'll post shortly as well.


Most Helpful Member
Thread starter #13
At 100 Va on 240 volts you probably could do without the C8 power factor correction capacitor all together.

As far as LM358's are concerned at 50 - 60 Hz they should have more than adequate slew rates to work with a direct line frequency when set up in a comparator mode. I am just guessing but it may be that they may need pull down resistors on the + inputs to get them to turn off faster. Some op amp IC's can work directly off of diode input sources but others hold enough internal capacitance that they wont turn off fast enough without the inputs being pulled down to the common. A 10K to 100K pull down on each of the + inputs should cure the slow response time issue.

Without knowing more about your specific components it may be possible that the small power transformer and control transformer are close enough in size that they can create their own feedback loop and having a slow response time in the control circuits could conceivably make the circuit as a whole self oscillate at a low enough frequency that I can keep itself running at near line frequency.

Normally on a full GTI system there is also the over and under line frequency shut down circuit and over and under line voltage circuits that would catch any self oscillations or over or under line frequency drift or voltage changes and shut down the switching devices.
But since thats not part of this circuit and is not shown in this design, Part of "Control Circuit B", it may be possible to get a self oscillation to keep the system running while disconnected from the lines. If you placed any sort of load greater than the output capacity of the unit on the output as it was being disconnected its likely that it will loose the self oscillation and just stall out from a overload condition.
I am just a guessing at this point.

You may also run into a overall efficiency limit using a small power transformer being that as transformers get smaller in Va ratings their overall efficiency drops off fast. A large 2 KVA transformer may be over 95% efficient at full load but a small 100 VA may only be around 70% efficient at best.
Even without the over and under line frequency and voltage controls in a real life situation when the power goes out there is enough load on the electrical systems to easily pull the GTI down and stall it out. Just unplugging it from a wall could apparently still allow for enough self oscillation to keep it self running if the circuit conditions where just right.

To be honest just unplugging one from the wall while its running at load is not something I have normally ever done on purpose. Being I use my wind generators as a power source the sudden loss of loading on them would cause a rapid input voltage spike that could damage things. I think the same could hold true with solar panels as well. being loaded the output voltage would be kept down but when unloaded their output voltage could easily rise well above the GTI's input limits and cause damage.

In a proper GTI system setup there has to be a way to dump the input power to an alternate load or tell the input power source to shut down if the unit losses its line connection or has a problem itself. Thats what the additional safety interlocks and control circuitry in the "control circuit B" does. Without it at this point this circuit is to dumb to know when or how to protect itself.
I added the pull-downs on the + inputs of the op-amp and it did solve the problem with the half-cycles overlapping giving me much more control in adjusting the dead-band. Thanks!

The 100VA transformer is just a start, its the largest I could find amongst all my bits and pieces. I'm planning on getting a 2kVA custom wound toroidal transformer as I can't find one available anywhere! Before I do so, could you please give me some advice on the relationship between the turns ratio and the input/output voltages? Lets say for arguements sake that the primary voltage varies between 50V-100V, and the line is at 230Vac. Should I have to transfomer wound for 50V/230V?

Regarding the self-oscillation, an MCU would definetely be the way I'll go to check for line frequency, voltages etc. And I'll try out your theory of the transformers being to close in size once I have my 2kVA wound.


Most Helpful Member
Thread starter #15
The best way I can explain the input voltage to RMS AC voltage is that the peak input at full power tends to be roughly around 10% over you peak voltage of the transformer. If you have a 50 VAC rated transformer your peak voltage will be about 1.414 times that which is 70.7 volts. given that a transformer has a little bit of internal losses you will probably see around 78 volts at the roughly 2 Kw point.
However by changing the width of the zero crossing dead band you can get your input voltage to go even higher by reducing the time the switching devices are on for each half of the cycle. A very short duration on time will let the input voltage go far higher than the roughly 78 volt point. But the down side is the peak currents that the switching devices may have to handle during that shorter on time could be far more than the present 4x over current design can take so a higher current rated switching device may be nessisary if the PWM durations are short and the peak currents are high.
The actual currents will still be limited by the inductive and resistive effects of the transformers windings but without knowing what they are its impossible to guess at.

As far as using toroid transformers you will likely get a added overload capacity but I have no idea as to how much. I typically have always used old E & I core type transformers from old power supplies, battery chargers, and industrial equipment take off's. Most of them rapidly loose their efficiency as the frequency goes up from 60 Hz so they always tended to want to stall out if disconnected. This naturaly poor high frequency response would cause the output voltage to drop off which would in turn cause the control transformer to keep reducing the effective conduction times until the GTI would just stall. Your toroid transformer may be more efficient at higher frequencies so it can keep a self oscillation going at a level great enough to keep the control transformer up and running at a high enough level to keep the control circuits alive. Small transformers tend to be able to work at higher frequency's better than bigger transformers can.

The other problem of your GTI being able stay active by free running is that the switching devices could be getting subjected to considerable ringing effects and voltage spikes without the line power present to dump the power to. I am still guessing at it though.
There is a good chance that if you hooked an O-Scope up to your primary side you probably wont like what you see when the power line is disconnected. The few times I ever lost a line connection while a GTI is powered up caused the switching devices to pop a few seconds later. That was the main reason why the automatic shut down and load dump circuitry was added for real life unattended operation.

Excellent questions so far! Thanks. You have pointed out a few errors in my schematics and theory of operations writeup's I will need to address in later work. After I came up with the original control circuit design I always kept to using similar parts and components and have never put much thought into the fine details of how the circuits worked or would work using different components other than what I normally have always used. Once I came up with a reliable basic control system some years ago I just stuck with that design ever since being it was so adaptable, easy to replicate and add more functions onto later. I see it has a few limitations that can cause unintended problems with using different components than what I always took for granted to just work.
Last edited:
I'm still a bit confused about the transformer design. You mentioned that the voltage would be 1.414 x the input voltage, I do not understand how that relates to the turns ratio? That is simply the peak acheived once the square wave is made sinusoidal, right? But where does the turns ratio come into play? Shouldn't the RMS output voltage be simply the DC bus voltage x the turns ratio?

What I really need to know is what my turns ratio should be. Given that my DC bus is spec'd for 50V-100V, and my line is at 230Vac, what should the rated voltage be for the other side of my transformer. My assumption is that the output of the transformer should be higher than mains so the current would flow into mains instead of out. So that means I would need to exceed the voltage rating on the primary to get the higher than rated output (even though the output would be clamped by mains)?

So what I'm trying to get at is, I know the secondary of my transformer has to be rated at 230Vac, which won't be exceeded because it is clamped by mains. And I know that my DC bus varies between 50V-100V. So do I design my primary to be rated at 100Vac, which wouldn't produce enough volts on the secondary at 50V DC bus level? Or do I design the primary for 50Vac, which would be exceeded substantially when the DC bus is at 100V?

Regarding the toroidal vs E-I core's, I have been speaking to a few manufacturers and they have all confirmed what you said about the efficiency along with toroidals being much quieter, but much more expensive, so E-I it is!
Sorry about the incompetence when it comes to the transformer design, that stuff is like black magic to me!

We all try to stick with the parts we are familiar with, it really removes alot of chance of error that way, so it isn't a bad thing. Anybody who tries building a power circuit up such as yours, and doesn't take some quick measurements before loading it up, well they deserve some smoking switches, that way they'll think twice before powering up the second time!


Most Helpful Member
Thread starter #17
I am not sure what you are going for exactly.
For a system that works at around 50 - 100 VDC input you would need a transformer with approximately a 64 volt secondary. (90/100) / 1.414 = 63.65 VAC RMS.

However by using the adjustable zero crossing dead band the input voltage can be tweaked higher from that base calculate limit point. For a 50 - 100 volt input I would probably start with a standard issue 48 VAC secondary type step down transformer and adjust the zero cross dead band to where it puts the full rated load on the power source at 100 volts. At 50 volts DC input you may only get a fraction of the peak power feeding back unless an automatic zero cross dead band adjustment was built into it so a lower transformer voltage could used of perhaps around 36 - 40 VAC RMS.
By introducing a little feedback from the input + side of DC power source to pull up the negative referance point voltage on the zero cross circuit op-amp comparators its possible to make the GTI have a more gradual rise on the peak input voltage and power point.

The problem with that is setting up a load curve that matches your power source. With solar panels its not really a big deal but with a wind power source that has the capacity to keep pushing the input voltage higher and higher as the zero cross dead band gets wider and creates smaller conduction times. At some point the input voltage will be too high and the resulting peak currents during the short duty cycles of each half of the wave will beyond the working capacity of the switching devices.

The narrowing of the duty cycle works to lower the average input amps or raise the input voltage but there is a limit to how far each can go. With a stable input power source that has a limited voltage capacity the shorter duty cycles work as an effective current and power limiting system but when the input power source does not have a limit to how high it voltage can reach it just makes the peak amps during the conduction times go higher and higher while the conduction times keep getting shorter and shorter from the resulting feedback loop.

To make that work a way of setting a minimum input load point and a point where the load on the input goes back to the exponential loading effect has to be implemented or a input voltage/peak switching current run away will happen.


Most Helpful Member
Thread starter #19
Everything you need to know is already provided in the write up. The schematics and critical part numbers are at the bottom of the first post. Just click on them to get the full screen views.

The rest of the design is up to you and the how and why of selecting parts are detailed in the write ups.
Control circuit B

When you get a moment can you post control circuit B. I would be interested to see what you have come up with.
I am just starting to gather all my parts. I think I have located the IGBT's for a very low price (free). So I will start designing my circuit.

Latest threads

EE World Online Articles