Warpverter

[Warpspeed]

Looks like inverter 1 + 2 is sine wave enough to power dumb loads like motors and heaters.
Can you post links describing your inverter project from the start?
View attachment 171101

Yes that is true.
Trace and Xantrex manufacture inverters that use the same Warpverter principle and they only use three inverters, and that works perfectly well.
My ROM that contains all the gate drive data has eight bits, so adding a fourth inverter is not that difficult. The fourth inverter is also very small.

But if you are manufacturing inverters, the extra parts increase the cost, and something like 3% distortion with three inverters is still pretty good.
The fourth inverter is really overkill, but I decided to do it anyway.

 

[CoolWill]

I've now built several inverters. But when I first got started, one of the avenues I researched was the Warpverter. I am still very impressed by the ingenuity of the design. Ultimately I went the standard PWM route due to well-worn simplicity and not desiring to wind 4 transformers! It took a while to understand the theory of the Warpverter and although I haven't built one, it is an edifying exercise.

 

[AntronX]

Its also fully bi directional. Really nasty highly reactive loads can bounce out of phase current right back onto the dc bus.

So batteries can be charged from the grid via this inverter by slightly delaying waveform generation phase and voltage? How good the power factor be?

 

[cs1234]

I want a Warpverter, just to say I have one. Unfortunately, just looking at @Warpspeed 's well written posts makes my eyes glaze over and I am forced to recognize my inadequacy.

 

[Warpspeed]

So batteries can be charged from the grid via this inverter by slightly delaying waveform generation phase and voltage? How good the power factor be?

No. This is not a grid tie inverter, there is no way to synchronise a Warpverter to the grid.
Its intended for off grid applications only.

However when running off grid, you can couple up a grid tie inverter to a Warpverter to back charge a battery through the Warpverter.
The Warpverter then becomes the grid, as far as the grid tie inverter goes.

The only disadvantage is that you could overcharge the battery doing that, as the grid tie inverter has no way of knowing when the battery is full.

 

[AntronX]

Ok so it's a control algorithm problem to solve. I was wondering if warpverter power stage topology is capable of operation in that mode.

 

[CoolWill]

Ok so it's a control algorithm problem to solve. I was wondering if warpverter power stage topology is capable of operation in that mode.

A separate system could monitor battery charge and operate a contactor to disconnect the grid tie inverter AC.

 

[Warpspeed]

I've now built several inverters. But when I first got started, one of the avenues I researched was the Warpverter. I am still very impressed by the ingenuity of the design. Ultimately I went the standard PWM route due to well-worn simplicity and not desiring to wind 4 transformers! It took a while to understand the theory of the Warpverter and although I haven't built one, it is an edifying exercise.

I have been playing around with these ideas for over forty years. Its all now finally come together, and its very satisfying seeing others successfully build this type of inverter. Unless you are aiming for about 5Kw or more, pwm is a much simpler and more sensible approach.

Its at the higher power levels, that makes home brew pwm very difficult.
It can be done, and has been done, but its not so easy getting the reliability.

Winding four transformers is a huge undertaking, and not to be taken lightly.
But if serious power is your goal, a Warpverter is likely to cause you much less long term grief.
Its a lot of work, but you only have to do it once !

 

[Warpspeed]

Ok so it's a control algorithm problem to solve. I was wondering if warpverter power stage topology is capable of operation in that mode.

Yes.
PWM or Warpverter is just a means of turning dc into ac. Two very different types of power stages.

Its how the process is controlled, in other words a software algorithm problem or what generates and synchronizes the inverter to the grid.

 

[Hedges]

A separate system could monitor battery charge and operate a contactor to disconnect the grid tie inverter AC.

Just shift frequency of timebase up and down to control frequency-watts GT PV and regulate battery voltage and charge rate.

I know this is all a rather difficult concept to get your head around to begin with, but please try to bear with me....

Clear as mud for some, I'm sure. But I follow it completely.
Considering I've dealt with transformers a bit, also transistor amplifiers.

Check out my threads on transformers and 3-phase. The issues of saturation are quite relevant.
Also, not detailed much here, is my measurement of BH curve and building "Chan" model simulations. Different core materials and annealing relate to the different curves.

 

[Hedges]

Yes, that is it EXACTLY.

Single phase?

Should be able to put 2 secondary windings on each transformer. That will give 120/240V split-phase. Full wattage available as surge for 240V or for either 120V phase. Continuous if secondary windings of suitable gauge.

I can see why it doesn't work parallel to another source like grid. No adjustable voltage or current. Best you could do is phase shift between charge/discharge to regulate average current, but seems quite problematic.

Does AC Vpeak and Vrms vary as battery voltage varies? Or do you change timing and number of steps to compensate?

 

[toms]

I’m glad this has it’s own thread, and that there are a few here that appreciate the design. I have followed this idea for a long time across various forums, and it has often been misunderstood or dismissed.

I feel this is a good reflection of the quality of the membership base of this forum.

 

[Warpspeed]

How the voltage regulation works.

The transformers are wound with primaries to suit the lowest incoming dc voltage.
At the lowest design voltage, the transformers are switched to produce the full 81 step voltage range and nominal ac output voltage.
Because the transformer ratios are fixed, increasing the dc voltage would produce a corresponding increase in ac output voltage.
My own Warpverter works between 90v and 180v incoming dc voltage. Yours might be any 2:1 dc voltage range.

The waveforms shown below are with only three inverters running which produces 27 steps.
With fewer steps its much easier to see the changes that occur.
The fourth inverter produces such a smooth sine wave its very difficult to see anything happening.

At minimum dc input voltage of 90v you can see the full 27 steps.
If the input voltage is increased to 95v, notice the top step right at the peak has become slightly narrower.
In fact all of the steps are made slightly narrower, but its most noticeable right at the peak.
This overall change in the waveform corrects for the 5v dc increase, producing the same nominal rms ac output voltage.

If we keep slowly increasing the incoming dc voltage, eventually the top step becomes narrower and narrower, finally disappearing, and we have fewer steps peak to peak. This provides very fine output voltage control, and we can have a regulated output voltage over a very wide 2:1 change in dc input voltage.

The change in waveshape is produced by jumping between lookup tables in ROM.
The jump always occurs right at a zero crossing, so large output voltage corrections can be made without causing any discontinuity in the sine wave. There are 256 different lookup tables that cover the 2:1 dc input voltage range.

What determines which lookup table to use is a measurement of the incoming dc voltage. That is measured 25 times each second, and the ac output voltage is fully corrected every second mains cycle right at the zero crossing. This system does not require any voltage feedback from the output of the inverter. Its very fast acting compared to feedback, and it can never become unstable in the way feedback sometimes can.

The only disadvantage is that due to slight voltage droop with increasing load in the transformers, and voltage drop in the IGBTs, the ac output voltage does fall slightly with increasing load. In my own inverter its about a ten volt drop at 5Kw of load, compared to zero load.
The grid does that anyway, and its never been a problem here.

That can be corrected by measuring the DC input current to the inverter and using that to modify which lookup table is selected beyond what the incoming voltage measurement dictates. I have tested the idea and it works as expected, but have not bothered to build another control board to incorporate it into my own inverter. Other Warpverter builders are using the current measuring modification and are very pleased with it.

This voltage regulation all works backwards when back charging a Warpverter from a grid tie inverter. It still regulates the ac voltage, preventing it from rising when the grid tie inverter is back feeding.

Here are some waveforms showing the effect of different dc input voltages on the ac output with only three inverters running.
The oscilloscope is set for 100v per division.
Ac output voltage is held pretty constant even though the vertical height of each step increases in proportion with the increase in dc input voltage.

 

[CoolWill]

Just shift frequency of timebase up and down to control frequency-watts GT PV and regulate battery voltage and charge rate.

The heart of the Warpverter is a done deal. I'm sure that could be added, but as an off-the-shelf unit, it might be easier as a separate control scheme.

 

[AntronX]

There are 256 different lookup tables that cover the 2:1 dc input voltage range.

How many Kbytes total for entire ROM ?

 

[Warpspeed]

Each lookup table is 1K.
256K total

 

[Ellcon123]

No. This is not a grid tie inverter, there is no way to synchronise a Warpverter to the grid.
Its intended for off grid applications only.

However when running off grid, you can couple up a grid tie inverter to a Warpverter to back charge a battery through the Warpverter.
The Warpverter then becomes the grid, as far as the grid tie inverter goes.

The only disadvantage is that you could overcharge the battery doing that, as the grid tie inverter has no way of knowing when the battery is full.

Re: back charging. I have a LF off grid inverter with a cheap fleabay grid tie inverter on the output. This plugs into a smart plug and I use Home Assistant to turn it off at a set battery voltage and then on again at a lower voltage. Been working great for some months now.

 

[AntronX]

Very cool design. Essentially a hardware based state machine.

 

[Warpspeed]

Very cool design. Essentially a hardware based state machine.

I started out using microcontrollers, but quickly realized that the only way to make the whole process fast enough was with lookup tables.
Real time number crunching is just not practical. I went through using dual port rams, all kinds of things.
Finally went to pure hardware, its quite simple really.
Just cycle through 1K of ROM at 50Hz (or 60Hz) and latch that onto the output.
The bit patterns stored in ROM are direct on/off gate drive data. Eight latched output bits switch eight half bridges through opto isolators, there is no data manipulation required.
Use a very accurate and stable dual slope 12 bit analog to digital converter to select one of the 256 lookup tables.

The dual slope integrating A/D averages out any ripple or noise on the battery, which can be considerable if for example there are solar controllers or other stuff connected to the battery.

 

[AntronX]

Do you store data bits for full sine wave cycle or just 1/4 of it and flip/invert in hardware? I would like to look at detailed implementation of the EEPROM. Using parallel data output of ADC and driving address select bits of EEPROM is very clever. I would probably just have an 8 bit microcontroller in the middle of it all running assembly code.