Various aspects of home brew inverters

Author Message
iannez
Newbie

Joined: 05/07/2019
Location: Italy
Posts: 23
Posted: 11:36am 27 Jan 2021      

Goodmorning everyone!

thanks Poida for illustrating the logic you constantly implement.
you use the powers of two as truncators to then keep the 8bit in the calculations.
as a bitshift >> 8 (2 ^ 8) which the cpu interprets quickly.
and the PID as a filter of a certain order via software.

i like this type of mod, something slowly I understand

I actually experienced the possibility of increasing the number of NPWM p
oints to give a better shape to the output wave (not that at 20khz it is bad!)
unfortunately I don't have a distortion meter or a better oscilloscope that allows me to visually check.
the rise in frequency is the consequence.

I calculated a whole series of useful frequencies that gave out
integers that can be used.
there may be problems with the PLL but that's another story

PPWM: 200 | FREQ: 80000 | NPWM: 800
PPWM: 250 | FREQ: 64000 | NPWM: 640
PPWM: 256 | FREQ: 62500 | NPWM: 625
PPWM: 320 | FREQ: 50000 | NPWM: 500
PPWM: 400 | FREQ: 40000 | NPWM: 400
PPWM: 500 | FREQ: 32000 | NPWM: 320
PPWM: 625 | FREQ: 25600 | NPWM: 256
PPWM: 640 | FREQ: 25000 | NPWM: 250
PPWM: 800 | FREQ: 20000 | NPWM: 200
PPWM: 1000 | FREQ: 16000 | NPWM: 160
PPWM: 1203 | FREQ: 13300 | NPWM: 133
PPWM: 1250 | FREQ: 12800 | NPWM: 128
PPWM: 1280 | FREQ: 12500 | NPWM: 125
PPWM: 1600 | FREQ: 10000 | NPWM: 100
PPWM: 2000 | FREQ: 8000 | NPWM: 80

i found the limits using timer1 mode14 with values

PPWM: 200 | FREQ: 80000 | NPWM: 800

because it goes beyond the memory ability, that would be 1600 per fullwave.

I haven't tried with the PROGMEM but I think the limit is the same.

and below

PPWM: 1000 | FREQ: 16000 | NPWM: 160

it might not make sense because the frequency would become audible
on an H-bridge and the number of NPWM points would be too low in my opinion.

I remember that my application is for now different from the inverter one.
I say for now why the Chinese inveter taken on aliexpress must arrive: D

now i just need to have a super imposed sine wave
in phase and with the same zerocross.
adjusting amplitude should decrease the overall output voltage or increase it.
almost like putting two secondaries of a transformer in series.
Transformer AVRs work like this through relays that switch on various secondaries.

so...

I took a while to understand how I could reset the distance between the two zerocross of the waves.
so automatically have superimposed wave with concident zero cross,
without using the _pcorrect variable that I introduced and which refers to the analog pin.

i thought if i used another ISR interrupt (INT1_vect) on pin3
on which then to enter with our filtered rc wave
(of course a square would be better but it should go anyway)
in the ISR I could calculate an additional pcint.


ISR (INT1_vect)
{
  pcint1 = pcount + (v1low == 1? NPWM: 0) - NPWM; // get position in 50Hz waveform, -200 to 200 counts
                                                                                  // pcint is position in output waveform when this interrupt occurs
}


the difference between the two pcint of the two ISRs should give the distance
in pcount between the two zerocross.


_pcorrect = pcint1 - pcint;


if the logic is correct you put this value in the final calculation of the phase_error.

what do you think about it?

thank you soon, A.
Edited 2021-01-27 21:43 by iannez