1=Output on 0=Output off
101010101010 (Clock = 6x output Frequency)
S R (SR latch one signals)
111111000000 (Phase one)
S R (SR latch two signals)
001111110000 (Phase two)
S R (SR latch three signals)
000011111100 (Phase three)
(Reset counter at this point then repeat)
Looking at that diagram; there appears to be 20 pulses per phase, 40 transitions, so 120 transitions in total. That would fit into a 120 byte EEPROM table and it's possibly less if transitions on multiple phases occur at the same time. A "delay this long then set outputs as this" byte pair could allow for a very simple program to generate what is required though one would have to calculate the data table.The better motor inverters generate SPWM waveforms, as shown in the attached image below. This would require a very large lookup table if using a Picaxe, which BTW was the original poster's request.
I thought the timing I posted could be very easily achieved with a PicAxe. The way I had to build mine was a very hard way of doing it, and I wish I'd had a PicAxe at the time. I hoped to inspire Hareeshu to decode the timing into a For..Next loop with a few nested If..Then statements that would produce the same outputs.One only needs to Google "three phase inverter waveforms", for dozens of circuits to appear. The simplest uses a shift register and a couple of inverter gates.
All of these circuits however, are "180 degree conduction" pulses. They work perfectly well but have a significant amount of harmonics.
The most worrisome is the 5th, as it generates torque in the opposite direction of the motor's rotation. This will cause significant inefficiencies and motor overheating.
The better motor inverters generate SPWM waveforms, as shown in the attached image below. This would require a very large lookup table if using a Picaxe, which BTW was the original poster's request.
View attachment 17979
Sure;@Fernando
Can you post the ASC file or link to the sim graphic?
Thanks
Goey
I'm not sure the PicAxe could manage the calculations required? It seems that "cos( 1*p1s )-cos( 1*p1e )+...+cos( 1*p7s )-cos( 1*p7e ) = ampl*pi/4" needs to be repeated for seven pulses, per quadrant, per cycle, per phase. Even with the maths co-processor, that would take some time.For sine generation - search "magic sine" at www.tinaja.com...