The term feed forward has multiple meanings. In this post, it is an amplifier design principle that can help reduce waveform distortion problems.
In an earlier post, the expression “feed forward” was discussed with respect to a control function used in some switch-mode power supplies. However, the term has other meanings as well. We are now going to look at an amplifier design principle that is also called “feed forward,” which can help to reduce waveform distortion problems.
In Figure 1, we have an amplifier (yellow) that accepts an input signal and boosts that signal, but is somehow flawed in that it creates a distorted output waveshape (violet). Such a non-ideal amplifier might be pushing some kind of power limit or it might be working at extremely high frequencies, but for the sake of this discussion, we will mimic distortion-generating processes and leave the real world causes to the imagination.
Figure 1 This diagram shows a flawed amplifier creating a distorted output waveshape, and the feed forward distortion correction.
We can reduce the distortion effect by taking the distorted E2 signal, scaling it back down by the reciprocal of the yellow amplifier’s gain factor “A,” and doing a signal subtraction with the original input to yield E4. Multiplying that E4 by the same “A” yields E5, which contains all of the distortion products of E2 but with none of the intended signal itself.
Taking the subtraction of E2 – E5 cancels out the distortion products leaving E6 (green), which is an undistorted and amplified version of the input signal Ein. Whatever distortion the yellow amplifier produces, extracting the distortion products by themselves and feeding that and only that forward to the output subtraction point cleans up the distortion at the final output.
To further illustrate the concept, the example in Figure 2 works exactly the same way but for a very different yellow amplifier distortion waveshape at E2.
Figure 2 This example has a different distortion but the same correction.
The above waveshapes came out of some GW-BASIC programming, but the distortion cancellation principle can also be demonstrated in SPICE (Figure 3).
Figure 3 Here the distortion cancellation principle is demonstrated in SPICE.
The amplifier stage using U1 in Figure 3 has two diodes and two extra resistors placed in its feedback path to deliberately induce some waveform distortion. While ordinarily we wouldn’t create a distortion condition on purpose, for the sake of this demonstration we have done it. Then, with U2 configured as a unity gain differential amplifier, distortion correction is achieved.
Just to further belabor the point, if we change the U1 distortion in some other arbitrary manner, the distortion correction effect is achieved anyway (Figure 4).
Figure 4 This example has a different distortion than Figure 3, but the same correction.
We should note that this distortion correction principle has been previously described. In this Keysight literature, Feed Forward Amplifier, we find the following image:
Figure 5 This schematic illustrates another way to obtain distortion cancellation.
At these frequencies, three couplers, one summing network, trimmable gains, and delay line phase shifters are used to obtain distortion cancellation. However, the distortion cancellation principle is the same as we have reviewed.
Pretty spiffy stuff, isn’t it?
John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).