spwm design and application through power mosfets

Thread Starter

qwertyqwq

Joined Jul 11, 2019
20
hi everyone !
im designind a inverter which low power at beginning.
the generating spwm way is comparing two type of signal(sine and triangle) . the main signal is 50Hz sine wave which generated by icl8038 pulse generator. the carrier signal is 22kHz triangle wave which also generated by icl8038 pretty old but good ic.
the compared spwm is at first zero and going to opamp's Vcc voltage and at least going back to zero . all those times the spwm signal never goes back to zero. because of that situation , icannot use any logic ic's and also mosfet drivers. the problem exactly beggining here !!!
there is a pic of oscilloscope down !
1) what should i do ?
2) should i decrease carrier waveform frequency ?
3) should i use something different way to create spwm signal ?
4) is my spwm signal is good? (at image of oscilloscope, the spwm signal is clamped and in normal it has negative and positive alternans)
5) which way is more suitable for inverter design comparing two type of signal or using a microcontroller ?
the comparison maded by TI's opamp LM324.
IMG_4034123123123.png
 

cmartinez

Joined Jan 17, 2007
8,218
hi everyone !
im designind a inverter which low power at beginning.
the generating spwm way is comparing two type of signal(sine and triangle) . the main signal is 50Hz sine wave which generated by icl8038 pulse generator. the carrier signal is 22kHz triangle wave which also generated by icl8038 pretty old but good ic.
the compared spwm is at first zero and going to opamp's Vcc voltage and at least going back to zero . all those times the spwm signal never goes back to zero. because of that situation , icannot use any logic ic's and also mosfet drivers. the problem exactly beggining here !!!
there is a pic of oscilloscope down !
1) what should i do ?
2) should i decrease carrier waveform frequency ?
3) should i use something different way to create spwm signal ?
4) is my spwm signal is good? (at image of oscilloscope, the spwm signal is clamped and in normal it has negative and positive alternans)
5) which way is more suitable for inverter design comparing two type of signal or using a microcontroller ?
the comparison maded by TI's opamp LM324.
View attachment 182031
It would be of great help if you first posted a schematic of your circuit.
 
Last edited:

RPLaJeunesse

Joined Jul 29, 2018
252
Driving a PWM MOSFET requires gate rise and fall times in the <50 nanoseconds range, preferably in the single nanoseconds. Think of that as your goal. At 22kHz, a period of about 45 microseconds, the LM324 bandwidth and slew rate is way too slow to use as your comparator. Look for a proper comparator that will pass a 22kHz square wave without much distortion. I'd say anything with more than 100ns prop delay is too slow. And use lots of good RF capable bypassing, such a comparator would be made with transistors good to 100MHz plus.
 

Thread Starter

qwertyqwq

Joined Jul 11, 2019
20
Driving a PWM MOSFET requires gate rise and fall times in the <50 nanoseconds range, preferably in the single nanoseconds. Think of that as your goal. At 22kHz, a period of about 45 microseconds, the LM324 bandwidth and slew rate is way too slow to use as your comparator. Look for a proper comparator that will pass a 22kHz square wave without much distortion. I'd say anything with more than 100ns prop delay is too slow. And use lots of good RF capable bypassing, such a comparator would be made with transistors good to 100MHz plus.
thanks for advice .
you were right, i totaly miss that issue. how silly i am !!!!
i have changed comparator from LM324 to TL074(low noise j-fet input opamp) and the spwm result is pretty good than according to the former.
It would be of great help if you first posted a schematic of your circuit.
here is my circuit diagram .

1.png
now i have another problems ;
1) when i plug (in real , not simulation) the center tapped transformer's Vcc [ TR1(S2) ] the spwm signals periods are changing and its decreasing about 2 times of main freq(main1=50Hz and after plug in main=23Hz ). why this happening ??
2) the center tapped transformer is 12-0-12 / 220V 6W. the problem is voltage of secondary winding. everytime i have only peak to peak 2.6V on my oscilloscope screen and nothing more ! im thinkig of the reason is my center tapped transformer because of the low power's . and another thing you should now , the power circuitry needs 0.5 amps (i measured it ). so why this happening ?
thanks in advance ;)
 

RPLaJeunesse

Joined Jul 29, 2018
252
OK, I can see a few things that are potentially problematic.
0) Look at when the PWM signal switches from U2A to U2B. I think you will find its at 50% duty when this happens, it should be at/near zero, like in http://www.frequencyinverters.org/image/frequency-inverter-pwm.jpg You have a serious modulator design problem.
1) If TR1 is 50Hz rated it has the wrong core for this use, it needs to be rated for operation from 22kHz to maybe 400KHz. Laminated iron cores are only good for low frequencies, and a modulated 22KHz wave has significant energy content past 1MHz at low duty cycles.
2) I see nothing to capture/clamp the energy in the leakage inductance of TR1. Expect Q1 and Q2 to be destroyed eventually.
3) Using 1k gate resistors is really high for what would normally be fast switching. With 1k expect lots of dissipation in Q1 and Q2. For a similar device (IR2110) the gate resistors shown in an-978 are 3r3.
4) Using the IR2101 is illogical, you don't need the (expensive) high-side voltage rating. In fact you shouldn't even be using the high side outputs, just a low side driver is adequate. But the logic sense may be wrong and you will need and inverter on each U2 output.
5) Power circuits like these demand careful layout and tight bypass caps. Even a first pass prototype needs a PCB with a great ground plane. Or look into the "Manhattan" construction technique. You show no bypass caps above. Each IC power pin needs at least one 100nF part to ground placed tight nearby.
6) You show relatively large polar caps for C1 and C2. These likely don't have adequate frequency response for the intended use, and are 10 times the size of the ceramic caps of an-978.
7) D1 and D2 need to be fast diodes, and 1N4001s aren't. That will kill the gate drive rise time, increasing Q1 and Q2 dissipation. Use fast parts at a much lower current rating. Even 1N4148s might work OK.
8) Your TL074 has bipolar supplies, I doubt U2 does. U2 will not be happy with its inputs driven below ground. Either clamp U1 at ground (not easy) or provide a resistor network (likely 2-3 resistors) that does the level shifting and limiting.
9) Running the TL074 open loop as a comparator causes it to go into saturation. It slows down coming out of saturation. Not a good choice for this application. Look at real comparators for the PWM part.
10) Just what are you using for U2? does it have the Schmitt trigger inputs it needs?

With all the above you will need some time to address everything, good luck!
 
Top