I am implementing a voltage source ac-dc converter (230 V AC to 400 V DC) whose switching frequency is 2 kHz using IGBTs. Can anyone suggest the general time delay in practice that is provided between the upper and lower switches of any leg?
For IGBT converters and inverters, the usual delay time (or dead time) provided is between 2 to 4usec. More the time delay, the safer you are from accidental overlap of conduction leading to shoot-through fault in a limb, but more is the waveform distortion and vice versa.
You can take a look in the datasheet how long does the transistor take to close (turn on), and multiply this time by two, with the new transistor this can be shorter than 2uS, this time consider you are using the same gate voltage and gate resistor than indicated in the datasheet.
Hi, The name of this time delay is "Blanking Time" Choosing it depends on the Harmonic generated due to this time and also the increase and decrease in voltage when the current changes polarity. For 2kHz, 2 micro second blanking time is adequate.
Dead time constraints have to consider far more than just IGBT turn-on and turn-off times.
Any driver you could make use of will be characterized by it's delay time - usually deterministic - and the more important parameter "delay deviation between any two parts".
In optoelectronics, the delay deviation may be in a range of up to 400ns as long as the parts are brand new.
Due to aging, the delay deviation and the delay increase. In case you plan a design to last for 20 years - a usual assumption in PV/Traction/Wind-Application - you would have to account for this as well.
In case you change to inductive couplers like Infineon 1ED-Series, the delay deviation is down to a range of
Thank you sir for sharing your experience on the delay time analysis. I will also try to go with 1 uSec delay and if my results are coming adequate, then at least I can save 50% time for turning ON of the switch.
I think that the manufacturers of the bridge drivers of the high and the low side know this problem the best. In IR2184 the dead time was 500 ns, in other types it was adjustable from 400 ns to 5000 ns. We have used from 500 ns (using IR2184) to 1000 ns. Rather 500 ns. The dead time has some negative influence on the THD of the inverter output voltage and lowers a bit its value (this phenomenon is described in the literature).
The transistor dead time in an oscillator varies with the bus voltage. It varies from 1.5% to 20%. It is advisable to keep the dead time between 1 - 10%.
If the distortion introduced by a conservative dead time matters, then you can compensate it. See enclosed file. Unfortunately only available in Spanish.
That’s another topic but totally true, dead time introduces a small distortion (as larger the dead time, larger the distortion), there are not a lot of information about it, here there is a paper in English that mention it shortly during the introduction.
I forgot to add this: If you use a Z network (2 inductors and 2 crossed capacitors) you will get rid of the problem of dead time being too short. At the same time your inverter will be more robust as the short circuit protection will have plenty of time to turn off the IGBTs.
Take a look at this application note. I found it very useful. If you look @ datasheets of IGBTs, you'll find that the rise and fall times are very small compared to the delay times. The delay time are more significant in determining your deadtime.
@Morteza Kazemi , Thnx for your answer. you are exactly right. the time should be decided from data sheet and it must be higher than the turnoff-time of the switch.
I am working on H bridge inverter but unfortunately, i am confused about dead time.Any one tell me that how can i calculate dead time of mosfet for proper switching. i am waiting for ur answers thanks
It is recommended to select a dead band higher than the turn off time of the MOSFET that is being used for your circuit and it can be available from the data sheet.