How does PWM work? 2

[cccelcj]

Forgive my ignorance, but can anyone give a simplified explanation of how PWM works? I understand this is the fundamental concept behind most VFDs and that the duty cycle of the pulses at a certain frequency is proportional to maximum motor speed. Is this partally correct?

 

[itsmoked]

PWM is just that: Pulse Width Modulation. The main point of PWM is to create some average voltage via turning some maximum voltage completely on and off rapidly with the required duty cycle. The primary reason to do this is to avoid wasting energy in some semiconductor by dropping the undesired voltage in it(linear region) which also burns up the semiconductor device.

I believe most VFDs use a selected "carrier" frequency which is used as the PWM frequency. 3-10kHz

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[bklauba]

The chain of events / circuit path function is something like this:

The incoming AC (if DC is the input, skip this stage) is converted to DC. It then is filtered, sliced and diced, until a very robust DC buss, with highly variable voltage, and very short pulses is available (hence Pulse width modulation).

The selected waveform, perhaps sine, perhaps a profile of Lincoln lying on his back, from chin to forehead, whatever, is formed by varying the contribution of the DC buss pulses.

Thus, in something like a power conditioning system (ala Phase Technologies)a messy, droopy, noisy AC electrical service signal can be changed into a very clean sine wave. For this kind of product, the output is also sent thru a transformer, which allows establishment of a neutral, and also filters out much of the high frequency (frequency of pulsing) waveform.

A tad flowery explanation, but this seems to work for most.

BK

 

[alexit]

Is operating like light switch, longer left on the more the motor sees...

 

[cswilson]

To elaborate a little on what itsmoked said:

There are two frequencies of concern in a VFD, and the PWM frequency is not the "Variable Frequency". The PWM frequency is a fixed frequency substantially higher than the maximum variable fundamental frequency into the phases. This higher frequency can be thought of as a carrier frequency, as in radio AM and FM station frequencies, that are higher than the fundamental audio frequencies they "carry".

The most important term in the PWM abbreviation is "modulation". The purpose of modulation is to provide some fraction of the (relatively) fixed supply voltage to a phase at a given moment. One strategy is linear modulation, where for example, to provide 50% of the supply voltage, you turn the power transistor half-on, dropping half of the supply voltage in the transistor, providing the other half to the phase. As itsmoked alludes, the problem with this method is the very high power dissipation in the drive that is produced. (These are the "Class A" audio amplifiers that you see in audiophile stereo stores, with the huge heatsinks that are almost as big as the electronics box.) Still, there are linear modulated drives for motors, usually when the switching noise of a PWM power stage cannot be tolerated.

As itsmoked points out, PWM is popular because it keeps the power transistors in low-loss states for the vast majority of the time. In PWM, if you want to provide 50% of the supply voltage to the phase, you turn the transistor full on for 50% of the cycle, and full off for the other 50%. You count on the low-pass filtering of the motor's electrical inductance and the system's physical inertia to time-average the effect of this command pattern.

Now let's look at the "fundamental" frequency this modulation is carrying. In a typical VFD, this will vary from a few Hz up to maybe a few hundred Hz. The magnitude of these AC waveforms (and therefore the maximum PWM duty cycle) will be approximately proportional to the frequency (this is called the volts/Hz ratio). Remember that the PWM duty cycle for each phase will vary sinusoidally over the phase cycle.

Because a motor is always a generator, it will produce a back-voltage waveform on its phases of magnitude proportional to the speed. To produce any torque, the supply voltage from the drive must be greater than the back EMF voltage to be able to force current into the phases. Even at no load, the drive must be producing phase voltage waveforms of magnitude (and therefore of peak PWM duty cycle) proportional to the speed.

Curt Wilson
Delta Tau Data Systems

 

[waross]

With a carrier frequency of 4000 hz. there will be a pulse starting every 0.25 milliseconds. The width of the pulse can be from 0 miliseconds to 0.25 milliseconds. If you divide a 60 cycle sin wave into 0.25 milisecond slices, the width of each pulse will correcpond to the instantaneous voltage at that point in the sin wave. At a point in the sin wave where the voltage is 50% the pulse duration will be 50% of 0.25 milliseconds.
The pulses start as regular as clock work, they stop at the correct instant to produce the required instantaneous voltage. respectfully

 

[ozmosis]

cccelcj
whilst talking about the switching frequency (also known as pulse frequency or carrier frequency) then a general rule of thumb to remember is:
The higher the switching frequency, the lower the audible noise at the motor.
The higher the switching frequency, the harder the VSD/VFD has to 'work' and so it will generate increased heat within the power transistors (and most drives will require you to de-rate for high switch.freq).
The higher the switching freq, the more RFI (radio frequency interference is produced.
So, whilst a high switch.frequency is desirable, it needs to be balanced with the downside of increasing it.