Can I generate pure sine wave with just a clock and amplifier?

Question

I see pure sine wave DC to AC inverters cost a lot more than the modified sine wave ones due to added complexity.
Can't we just take a DC source, feed it into a quartz clock to create a sine wave, then amplify it up in order to create a pure sine wave? If so, why do they even bother creating these modified sine wave inverters as should a simple clock + amplifier be pretty cheap?

Answer 1

Yes, you could in theory start with a sine signal and amplify it to make a inverter. The result would be rather inefficient though. The amplifier has to work with lots of signals, whereas you know exactly the signal you want. The general amplifier approach ignores this and therefore doesn't avail itself of optimizations that would not be valid for amplifying general signals.

If you used a efficient class D amplifier, then the result might not be too bad. One way of looking at a sine wave inverter is as a class D amplifier, except that it also synthesizes the signal directly instead of having to faithfully follow some analog input signal.

One important optimization that the general amplifier approach is missing is that distortion can be much higher for a inverter. A few percent is a lot for a amplifier, but not much at all for a power cycle.

Also look at what amplifiers cost that can put out the voltage and power you want a inverter to put out.

Answer 2

I don't know how you think that feeding DC "into a quartz clock" will generate a sine wave but let's assume that it does.

^{simulate this circuit – Schematic created using CircuitLab}

Figure 1. H-bridge AC inverter.

Let's say we want to make a 1,200 W inverter running off a perfect 12 V battery with perfect transistors. The current required will be somewhere around 100 A.

Our sinewave signal gradually turns on Q1 and Q4 allowing current to flow from left to right through XFMR1 primary. We gradually turn them off again and start to turn on Q3 and Q2 to reverse the current.

Now consider what happens when the transistors are only partially on. Let's say that the current is at 50 A and about half the voltage is dropped across Q1 and Q4 - i.e., 6 V. We can see that the power dissipated in the transistors is 50 x 6 = 300 W. (P = VI). This is all lost as heat, is very wasteful of energy, will require a large heatsink, will heat up the room and may take the transistors outside of their safe operating area.

The alternative is to pulse-width modulate the transistors at full voltage with wider pulses as the sine increases. In this mode the transistors go from off (I = 0, so P = 0 in the transistor) to fully on (V = 0 [almost], so P = 0 [almost]). Since this is so much more efficient it is the method of choice.

Answer 3

You can turn a square wave into a reasonable sine wave by filtering .The filter LC Parts are large by modern standards .If you generate a multi stepped approximation to the sinewave that you want the filter size shrinks a lot .This is the basis of multilevel switching and it has been and is used and will be used .This multi level or stepped switching means a more complicated output stage but it gives low switching losses and a small filter .