Until yesterday I would have bet everything that a passive mixer, driven by 25% duty cycle gets rid of the 3rd, 7th, 9th etc harmonic; while the 50% version has all odd-order harmonics. I swear I have read this in about every single paper in that area so that I started assuming it "granted common knowledge" without ever actually verifying it.
Now in my simulations this is not the case and indeed, verifying a sinusoid wave, sampled 4x per period with zero-order hold all all odd order harmonics, same as the 50% version ("rectangular wave").
Can anyone help to find out what could I mixing up here?
Thanks!
PS: In case my question is unclear or it is not entirely clear what I mean I am happy to provide simple MATLAB scripts comparing the spectra between 50% and 25% mixer
I'm guessing you're just using a normal two-path mixer and reducing the applied duty cycle from 50% to 25%, which is wrong. The gains come when you add more switching paths, like N=4 and D=25%, or N=8 and D=12.5%, and so on.
the only thing i ever heard was that driving the LO port of a mixer with a square wave instead of a sine wave MIGHT improve some of the spurious outputs. Marki microwave has some mixers that deliberately operate in this mode for spurious output reasons
I think you refer to the influence of the duty-cycle on a non-linear device, as a frequency multiplier (and not a mixer).
This document from Charles Wenzel explain you what is about with harmonics generated vs duty-cycle (Figure 2):
There is no reason to use other than 50% LO duty-cycle in an RF mixer. Any other duty-cycle might reduce the output harmonics, but for sure will degrade mixer performances (all of them).