The current is already rectified by the diode bridge, you don't need a triac, an SCR will work, and has lower losses...
...
Hi, yeah I knew that of course.
I'm using a TRIAC as it can be gated on with a negative current pulse, SCRs require a positive current pulse which would require a high side pulse driver, a lot of extra circuit complexity. Obviously I have kept the voltage and current sensing low-side to suit the PIC ADC. Also I have more options in my junkbox of high current TRIACs, I have not bought SCRs in 30 years (or ever?) and only have real old salvaged SCRs.
The TRIAC and SCR have a very similar forward voltage so there's little difference in losses, and actually the TRIAC is the coolest part of the whole device with the transformer, bridge react and current sense resistor all running at higher dissipation. (I'll post a photo of the new heatsinking soon).
...My two yen would be to use integral sinewaves, (your second method), it will produce far less distortion and EMI.
Thanks for the suggestion.
That still gives me an issue with the 4A current limiting. At the moment in the initial "bulk current" phase of charging it limits the current to 4A as the primary regulation, and can be half an hour before the battery gets to 14.5v to go to the second stage "bulk voltage" charging. Using the phase angle control I'm measuring the current at 16 points per half cycle, which gives a good average current per half cycle and the regulation process can adjust the phase angle for the next halfcycle. That gives a fast reliable closed loop current control fixed at 4A average.
At 4A the phase angle is over 75% so using the pulse skipping system instead it would give 3 or 4 pulses On and 1 pulse OFF, so I would need to average current over 10 or 20 pulses to get some reasonably accurate current average. That's really going to slow down the current control loop which will be a problem.
To 4pyros; (re the noise from phase angle switching) actually it is surprisingly good. The soft transformer type and huge inductance means the current pulses are quite rounded on the front edge.
To ChrisP58; Thanks for bringing up flux DC imbalance in the transformer. I don't really like the idea of doing pairs of halfcycles as it will give worse regulation, but it should be easy enough in software to keep a running count of + and - halfcycles and ensure there is no average imbalance and short term imbalance is kept small. From testing so far it looks quite random, in that it will run for a short time with alternate + cycles then jump to alternate - cycles after a fraction of a second.
At this point I'm leaning towards a phase angle control for the bulk charge, then switch to pulse skipping for the 13.8v float charge for the last few hours as that does not need the current limiting.
I did notice yesterday the pulse skipping at low duty cycles (low average output current) did have a hotter transformer than the earlier test using the phase angle switching at the same low average output current. Something to think about?