Well, I have then a solution to do this with two 16 bit timers. I know of dead band because I learned by the hard way: I made a driver for 30A 12V DC motors, and the control was a little bit crazy at the begining. Ok, seems that resonant power supply is the only one that uses that topology, as explained in your link above. I never saw the proposed phase-shifted control schema for this application. This simplifies the activation/deactivation for transistors to only two signals (one for each pair), and a variable dead band.įor SMPS that drive transformers or even resonant mode powe supplies, again the schema Q1-Q4 and Q2-Q3 is used. That means that you don't need to turn on Q3 fon any reason, and that you can also activate/deactivate Q1 at same time than Q4 (and Q2 like Q3). When Q1 is deactivated and Q3 remain active, and Q2 became active, then the cicle starts again, but reversed. Once the current is stopped, it still remains activated Q3 and Q1, no more current will flow through coil, so this is like having nothing active. If you have a free wheeling diode, then only turning off Q4 (and also Q1 if you want), current flows also from left to right, but through D3 (diode in parallel with Q3), and D2 (supposed diode in parallel with Q2) if you also turn off Q1. When you turn off Q4 and turn on Q3, Q3 is reverse polarized, meaning that current try to flow from emitter to collector (in an NPN transistor!!). If we switch on Q1 and Q4 a current begins to flow trhoug the coil from left to right. Of course, brake is accomplished mantaining only one transistor on (and then some free wheeling diode also). This is a quite usual configuration to regulate DC motors speed with reverse direction/brake. That means that some free wheeling diodes are on. The difference in phase-shifted is at the deadh band, meaning that the time that ALL transistors are off is variable. Yes, I know full bridge topology, but always the control was done activating pairs Q1-Q4 and Q2-Q3, so signal each pair is the same (level shifting appart). I plan to use hardware deadband generator so that I can realize it with the existing number of capture and compare registers. fast PWM with xor gates but there is no Tclock/2 prescalar option to synchronize phase correct (512 ticks) channel with fast PWM (256 ticks) channel. because of the unsymetrical nature of the phase shifted 3. channel with phase correct but then i can not synchronize the 3.

The suggestion of Megausbfreak: The phase correct PWM mode is symetrical so I can not set a compare match for the 3. Unfortunatelly Mega128 has only 7 because OCR2 and OCR1C uses the same pin and cannot be used seperatly. Xor Method that rhilton suggests is a good idea but There are a total of 4 seperate outputs signals all having different compare and capture points so I would need 8 compare and capture registers.

Phase shift controller setup