I've never understood why in bridge output circuits like MOSFET H-bridge in sine wave inverters or DC motor controls, or like the IGBT 3-phase output of a VFD, both high side switches and low side switches are both PWM'd from the same signal. Driving high side FETs is harder than low side, so why PWM them? Why not just leave them fully on for the entire period of the fundamental wave and just PWM the low side switch? And if you're going to do that, why even use a FET as the side switch? Why not use a SCR? Wouldn't that be cheaper and simpler? Especially if you're talking huge power like an EV motor controller, which is what I have in mind as I ponder this. Transistors rated 500+ amps are not cheap, unless you're talking about SCRs.

 

I've never understood why in bridge output circuits like MOSFET H-bridge in sine wave inverters or DC motor controls, or like the IGBT 3-phase output of a VFD, both high side switches and low side switches are both PWM'd from the same signal. Driving high side FETs is harder than low side, so why PWM them? Why not just leave them fully on for the entire period of the fundamental wave and just PWM the low side switch? And if you're going to do that, why even use a FET as the side switch? Why not use a SCR? Wouldn't that be cheaper and simpler? Especially if you're talking huge power like an EV motor controller, which is what I have in mind as I ponder this. Transistors rated 500+ amps are not cheap, unless you're talking about SCRs.

I thought most good motor controllers only switched high side when there is a direction change. I've always coded them that way. P-channel (even PNP BJTs are more difficult to switch, slower to switch and code becomes much cleaner (easier to monitor) and heat management is much easier (much lower switching losses).

 

I've never understood why in bridge output circuits like MOSFET H-bridge in sine wave inverters or DC motor controls, or like the IGBT 3-phase output of a VFD, both high side switches and low side switches are both PWM'd from the same signal. Driving high side FETs is harder than low side, so why PWM them? Why not just leave them fully on for the entire period of the fundamental wave and just PWM the low side switch?

Low side switching is possible only when upper switch is open, otherwise there is a short circuit, but it's possible to switch other leg and it seems to be used on some sine wave inverter projects. Typically high side is PWM controlled on those what I have seen (on internet, not real) but the principle is the same. It works with resistor load but reactive load can cause problems.

When high side switch is off, it's possible to PWM low side in order to get wanted average output voltage. In the sketched image of one leg of three phase inverter, PWM of S2 has no effect to output voltage which is not good for PWM control. Current doesn't commutate to D1 at all in shown situation.

 

What you are describing is SPWM With Bipolar Switching for sine wave generation in modern circuits. SCR inverter technology is old school technology used before power MOSFETs and other high power switching devices were available.
Page 383, 394.
http://www.introni.it/pdf/GE - SCR manual 1972.pdf

https://www.tntech.edu/engineering/pdf/cesr/ojo/asuri/Chapter2.pdf

Typically in a motor controller or inverter that uses PWM synthesis (Sinusoidal-Pulse Width Modulation) to generate the fundamental drive wave are using half-bridges per leg with poly-phase generation and full-bridge for single phase. The PWM frequency is much higher (10 KHz or higher) than the fundamental wave and we need to maintain precise timing of ON and OFF times for commutation. Your typical SCR can't do that without some help. How do you propose to turn off the SCRs at 10 KHz?

 

What you are describing is SPWM With Bipolar Switching for sine wave generation in modern circuits. SCR inverter technology is old school technology used before power MOSFETs and other high power switching devices were available.
Page 383, 394.
http://www.introni.it/pdf/GE - SCR manual 1972.pdf

https://www.tntech.edu/engineering/pdf/cesr/ojo/asuri/Chapter2.pdf

Typically in a motor controller or inverter that uses PWM synthesis (Sinusoidal-Pulse Width Modulation) to generate the fundamental drive wave are using half-bridges per leg with poly-phase generation and full-bridge for single phase. The PWM frequency is much higher (10 KHz or higher) than the fundamental wave and we need to maintain precise timing of ON and OFF times for commutation. Your typical SCR can't do that without some help. How do you propose to turn off the SCRs at 10 KHz?

My assertion is that you shouldn't need to PWM the high side switches, and my question is "why am I wrong?"

Please take a little walk with me in my head and let me know where I'm stumbling...

let's say we have a DC motor we want to control direction of. We can make a crude h-bridge out of relays/contactors/solenoids; whatever we call them, electromechanical switches.

Now we can run the motor full voltage forward or full voltage reverse. That's cool, but not so elegant. It would be better if we had variable speed. The simplest way, since we already have an electromechanical h-bridge in place, is to insert a single transistor, between the bottom of the H-bridge and ground. We do that and now we can choose direction by steering current with the electromechanical h-bridge and we can control speed by PWMing that transistor.

But let's say we didn't already have an h-bridge made of electromechanical switches. Let's say we only have 2 electromechanical switches, and we also have 2 transistors. We could construct an h-bridge out of this; use the electromechanical switches left and right at the top of the h-bridge, and use the transistors at the bottom left and right. Now to go one way, you energize the top left relay and PWM the bottom right. Or vise versa to go the other way.

Ok cool, but now we want to upgrade from that DC motor to a single phase AC motor. We shouldn't really need to change anything. Let's say we only want to run 5Hz or something. The electromechanical relays should switch fast enough. We'll be PWMing the transistors (let's go with MOSFETs) at 10s of kHz, but the relays only need to switch at 5Hz, the fundamental frequency.

Ok that works (in my head at least) but we want to go faster, up to 60Hz say. The relays aren't going to cut it. So we replace them with SCRs. SCRs should have no problem switching the fundamental frequency of 60Hz, and getting the SCRs to turn off shouldn't be a problem because we cut the current with the low side switches.

Cool, now we have a hybrid SCR/MOSFET inverter! But the single phase motor doesn't work very well on variable speed, we need to upgrade to a 3ph motor. So we can add 1 more SCR and 1 more MOSFET. Now we have 3 half-bridges, each with a SCR on top and a MOSFET on bottom. We control them the same way as before, switching the high-side SCRs at the fundamental frequency and switching the low-side MOSFETs at the PWM frequency.

Sick! Now we have a hybrid SCR/MOSFET VFD! ...right?...
If not, why?

 

My assertion is that you shouldn't need to PWM the high side switches, and my question is "why am I wrong?"

Please take a little walk with me in my head and let me know where I'm stumbling...

let's say we have a DC motor we want to control direction of. We can make a crude h-bridge out of relays/contactors/solenoids; whatever we call them, electromechanical switches.

Now we can run the motor full voltage forward or full voltage reverse. That's cool, but not so elegant. It would be better if we had variable speed. The simplest way, since we already have an electromechanical h-bridge in place, is to insert a single transistor, between the bottom of the H-bridge and ground. We do that and now we can choose direction by steering current with the electromechanical h-bridge and we can control speed by PWMing that transistor.

But let's say we didn't already have an h-bridge made of electromechanical switches. Let's say we only have 2 electromechanical switches, and we also have 2 transistors. We could construct an h-bridge out of this; use the electromechanical switches left and right at the top of the h-bridge, and use the transistors at the bottom left and right. Now to go one way, you energize the top left relay and PWM the bottom right. Or vise versa to go the other way.

Ok cool, but now we want to upgrade from that DC motor to a single phase AC motor. We shouldn't really need to change anything. Let's say we only want to run 5Hz or something. The electromechanical relays should switch fast enough. We'll be PWMing the transistors (let's go with MOSFETs) at 10s of kHz, but the relays only need to switch at 5Hz, the fundamental frequency.

Ok that works (in my head at least) but we want to go faster, up to 60Hz say. The relays aren't going to cut it. So we replace them with SCRs. SCRs should have no problem switching the fundamental frequency of 60Hz, and getting the SCRs to turn off shouldn't be a problem because we cut the current with the low side switches.

Cool, now we have a hybrid SCR/MOSFET inverter! But the single phase motor doesn't work very well on variable speed, we need to upgrade to a 3ph motor. So we can add 1 more SCR and 1 more MOSFET. Now we have 3 half-bridges, each with a SCR on top and a MOSFET on bottom. We control them the same way as before, switching the high-side SCRs at the fundamental frequency and switching the low-side MOSFETs at the PWM frequency.

Sick! Now we have a hybrid SCR/MOSFET VFD! ...right?...
If not, why?

We basically do the same thing today with:
https://www.tntech.edu/engineering/pdf/cesr/ojo/asuri/Chapter2.pdf
Page 31:

2.3.3 SPWM With Modified Bipolar Switching Scheme (MBPWM)[14]

In the inverter employing the bipolar switching scheme, switches are operated in such a way that during the positive half of the modulation signal one of the top devices in one of the switching leg is kept on and the two other switching devices in the other leg are PWM operated, and during the negative half of the modulation signal one of the bottom switching device is kept on continuously while the other two switching devices in the other leg are PWM operated. The output voltage is determined by comparing the control signal and the triangular wave.

BJTs or power MOSFETs or IGBTs are just better switching devices than an SCR for this type of application. We switched from SCR's to MOSFET inverters.

 

We switched from SCR's to MOSFET inverters.

Right. We also switched from electric cars to ones with internal combustion engines, and look at us now, clambering back to "old" technology.

I was just thinking that maybe some money could be saved as I looked up the prices of semiconductors capable of 500+ amps.

We basically do the same thing today with:
https://www.tntech.edu/engineering/pdf/cesr/ojo/asuri/Chapter2.pdf
Page 31:

View attachment 265892View attachment 265893


BJTs or power MOSFETs or IGBTs are just better switching devices than an SCR for this type of application.

Thanks, I am glad to learn that I wasn't 100% off the mark, but I did not consider that in order to do what I described, would require the low side devices to also go through periods of fully ON while the top ones are PWM'd. In order to stay true to the concept I envisioned, the 3 phases of the motor would need to be broken apart and energized independently (3 separate h-bridges to 3 separate coils), which does not cheapen or simplify anything.