| Electronics > Beginners |
| Gate driver with overlap |
| (1/1) |
| mark86:
I'm looking for examples of a H bridge gate driver circuit with switching overlap (not sure if this is what it would be called?) ie. both switch pairs would briefly be on at the polarity change. The H bridge would be supplied by a current source so short circuit current is not an issue as it would be if it was supplied by a voltage source. In particular this would be used to drive MOSFETs (or possibly IGBTs) to convert a DC TIG welder into an AC TIG welder. The H bridge will be driven by a microcontroller at a maximum freqency of 250Hz. Because of this I could use the microcontroller to manage the overlap however I want to remove the risk of having both switch pairs turned on for longer than intended eg. in the case of a software bug or failure. Super fast switching is not really important in this case and limiting the switching speed might help to minimise HF noise? |
| MagicSmoker:
--- Quote from: mark86 on July 01, 2019, 12:20:58 pm ---I'm looking for examples of a H bridge gate driver circuit with switching overlap (not sure if this is what it would be called?) ie. both switch pairs would briefly be on at the polarity change. ... In particular this would be used to drive MOSFETs (or possibly IGBTs) to convert a DC TIG welder into an AC TIG welder. ... Super fast switching is not really important in this case and limiting the switching speed might help to minimise HF noise? --- End quote --- There's all sorts of ways to do this ranging from inconvenient to clumsy to downright difficult, but one simple solution is to use NPT* IGBTs for the bridge and drive them with complementary signals from a flip-flop or the like. You are basically relying on the slow turn-off of the IGBTs to get a little bit of overlap. Using a flip-flop divides the clock frequency in half and always outputs a 50% duty cycle square wave, though, and in welding you often want to vary the proportion of positive to negative current ("AC balance"). If you want to control AC balance then you will need to use one of the more complicated schemes or code up your own PWM timer in a MCU, but beware that most MCUs with built-in PWM timer peripherals automatically generate deadtime in the complementary outputs, hence why I said you might have to roll your own. * - NPT = Non-Punch Through, as compared to PT, or, you guessed it, Punch-Through. If an IGBT is not explicitly described as NPT then look for descriptions like short-circuit tolerant, rugged or intended for motor drives. |
| Siwastaja:
--- Quote from: MagicSmoker on July 03, 2019, 11:23:01 am ---but beware that most MCUs with built-in PWM timer peripherals automatically generate deadtime in the complementary outputs, hence why I said you might have to roll your own. --- End quote --- Some (STM32F334's HRTIM comes to mind) have the possibility to set a negative deadtime (i.e., overlap) on their deadtime control units. |
| Ian.M:
Look at the MCU's complimentary PWM with deadtime and consider what happens if you invert the outputs, so On swaps with Off and deadtime becomes overlap time. Many MCUs let you select PWM output polarity, so you may not even need external inverters. To handle software failure, you need to detect both outputs active (start of overlap) in hardware, and use it to trigger a monostable, reset when the overlap ends. If the monostable times out before the overlap ends, you've got a fault, and what to do then depends on what's safest for your circuit. It could be as simple as resetting the MCU. |
| mark86:
Thanks everyone. This has given me some good ideas. STM32's are my MCU of choice but I wasn't planning to use one with a high resolution timer. It looks like inverting the PWM outputs and using the dead-time insertion to create the crossover will be perfect. I'm thinking I'll just feed the two outputs into an AND gate then to a low pass RC filter to generate a hardware shutdown signal in a case where both outputs stay on. |
| Navigation |
| Message Index |