My PIC produces 100Khz PWM signal and drive half bridge mosfets via IR2111 mosfet driver. The circuit is works well but when duty is between %80-%99 the output voltage doesn't go up. But when the duty cycle is %100 i see full voltage. Everything is normal with 60Khz PWM signal in all duty cycles.
In other words:
100Khz PWM
------------
%0-%80 : 0-15V
%80-%99: 15V
%100 : 17V
60Khz PWM
------------
%0-%100 : 0-17V
My PIC is 18F4620 which has 20Mhz Xtal. I put a lot of 100nF near the all ICs.
What is my problem?
The 100 kHz PWM may be too fast for the other components in your circuit. The turnoff comes before the output can rise to the full value. When the PWM is off then the output can rise to the full value. Trying to do this without an oscilliscope is like trying to box with one hand tied behind your back. You should at least rent one or borrow one. How did you select and qualify the magnetics without being able to see what's going on? I'm amazed that you would tackle such a project without the proper tools. Would you try to frame a house without a square, a saw, and a nail gun?
You didn't state what Mosfet's you are using, just the driver. If you can name that we might have a clue, a schematic would help, or at the very least a picture or two of how you're building this. The drivers on delay is only 130ns and off is only 65ns worst case so whatever is causing the problems it's past the FET driver, likley the FET itself or some parasitics in the way.
Are you aware that whatever value you place in PR2 will define the maximum PWM value. Once the PWM value is equal to PR2*4, the output will be 100%. This is the same as when CCPR1L = PR2.
It probably has something to do with the typical 1400ns of deadtime (700+700) built into the IR2111 which is there to prevent FET shoot though currents. 100Khz has a period of 10:mu:S and 1400ns is 14% of that.
Yes, because the IR2111 will wait apx 700ns after turning Q1 off before turning Q2 on.The IR2111 will also wait 700ns after turning Q2 off before turning Q1 on. Add this to the 850/150ns Ton/Toff times respectively, and you have a problem with low duty cycles at 100Khz. The relatively long switching times also mean that the slow drain/source body diode of Q2 will come into play when Q1 turns off and the inductor kicks back. A schottky diode in parallel with Q2 may help with this but you will still have the deadtime issue.
Oh Kchriste, Thank you for detailed information. I have just read LT1158 datasheet. I think it is better. Which one is better in this case? I think LT1158 has shorter deadtime if I didn't read wrong.
The LT1158 datasheet doesn't specify a deadtime. They state that the deadtime is dynamically adjusted based on the charge remaining on the FETs gate when the LT1158 tries to turn it off (They call it "Adaptive Non-Overlap Gate Drives"). Their typical application circuit does mention a PWM frequency of 0-100Khz so it should be OK, but I would have preferred to have seen a higher safety margin.