Poidas "Inverter Software Control" Topic.


Author Message
Warpspeed
Guru

Joined: 09/08/2007
Location: Australia
Posts: 3267
Posted: 08:19pm 02 Dec 2019      

Mike,

Stability is a non issue because nothing is fed back. It simply can never oscillate because there is no feedback loop. The correction may be slightly imperfect giving less than optimum regulation, but it will be totally and unconditionally stable.

This voltage feedforward system is well tried and proven in my own Warpverter which has been powering my home, with zero problems for one full year. It has never missed a beat.

Andrew has developed his own Warpverter software using a Nano, whereas my own Warpverter is totally hardware. Both my own hardware derived feedforward, and Andrews software derived feedforward work in exactly the same way, and our two very different circuit boards are directly interchangeable with identical plugs and connectors.

I have run both boards in my inverter, and Andrew has run both in his inverter.
These two boards can be swapped over in less than a minute.
The basic inverter functionality is identical in both, although Andrews board being a Nano has some extra added features, my more basic hardware board lacks.

The way mine works, is that the incoming dc voltage is measured 25 times per second with a dual slope analog to digital converter. As suggested previously, the upper half of the measurement range goes from 00 at half scale to FF at full scale.

That selects one of 256 different lookup tables, that produce all of the eight different mosfet gate driver waveforms to the inverter.
As the input voltage goes up and down, different lookup tables provide the most appropriate waveforms to drive the transformer primaries.

This generates a low distortion sine wave when all four transformer secondaries are added together in series. There is absolutely no feedback, nothing is connected to the secondary except the 230v load.  Andrews Nano has identical inverter functionality, but his is all done with magic and number crunching, instead of the five volt discrete logic I use.

Exactly the same feed forward control concept would work just as well with pwm as with a stepped sine wave.

The current sensing idea is something quite new, I have only thought of in the last few days and have not yet tried the idea out in hardware.

It will be extremely easy to implement, and I have ordered a suitable 100 amp Hall sensor which is not going to arrive here for another three weeks at the earliest.

The concept behind this, is that at the moment the +2.5v voltage reference for my analog to digital converter comes from a TL431 fixed voltage regulator.

I can use a single supply 100 amp Hall sensor that runs off +5v, to measure the inverter dc input current.  The Hall sensor (HSTS21) has an output of +2.5v with zero measured current, and of course the output voltage goes up and down depending on measured current direction and flow.

I can connect a potentiometer between the fixed +2.5v from the TL431, and the variable +2.5v output of the Hall sensor.  The wiper of the pot then provided my +2.5v dc reference voltage to the A/D converter.

If the wiper is at the TL431 end, the voltage reference will be a constant +2.5v with zero correction of the inverter output voltage from input current variations.

If the wiper is at the Hall sensor end of the potentiometer, the +2.5v reference to the A/D converter will vary with the inverter input current, in both magnitude AND DIRECTION.

There is a subtle point about this. The A/D converter is ratiometric, so that the measured parameter (inverter input voltage correction) and the A/D reference (inverter input current correction) multiply, not simply add or subtract from each other.

I should be able to tweak that pot so that the inverter output voltage can be corrected to compensate for the resistive losses in the switching bridge, and transformer windings that causes the inverter output voltage to droop under heavy load.

The interesting thing about that, is that the inverter output voltage normally falls off with increasing load as expected.  
But when reverse power is being fed from a grid tie inverter back into the battery, the inverter output voltage will rise by a similar amount due to the same resistive losses in the bridge and the transformer.

When back feeding power, the Hall sensor will correct for the voltage rise, just as it corrects for the voltage droop under normal inverter load.
Should be possible to find an optimum position where the inverter output impedance can be set to zero with perfect voltage regulation in either direction.

Its a fairly simple idea but difficult to put into a very few clear concise words.

If any of this is not clear, I will try to clarify 
Edited 2019-12-03 09:17 by Warpspeed
Cheers, Tony.