Diodes used for mixing - why needed?

[zincshow]

Diodes used for "mixing" - why needed?

At the site http://www.st-andrews.ac.uk/~www_pa/...rt1/page1.html there is a very comprehensive explanation of a "mixer diode". My question: if the diode is always forward biassed to produce an output frequency of fs - fl (source minus local frequency), why is the diode even needed? Would not this same circuit without the diode produce the same strong "beat" at the source minus local frequency?

Thanks from someone pretty confused about this.

 

[0xDEADBEEF]

Yes without a diode you would get a beat, but the signal would still only contain the two frequencies but not the beat frequency, as you could see in a Fourier transform. If you apply a low pass filter still nothing would get through. Only to your eye it looks like the periodic wobble is a frequency. After rectification or using any non linear element you actually get components of the beat frequency into your signal.

 

[zincshow]

Thank you, as you see I am only on page 1 of the online seminar so I have a lot to learn. To follow-up, I visualize the frequency of alternating current as electrons moving one way, then moving back, then moving forward.. No bias on the diode makes sense to stop the backward motion of the electrons, but if you bias the diode as they do here, is it proper to visualize "more" electrons moving forward, then electrons not moving, then "more" electrons moving forward, then not moving...? Can I keep this visualization or do I have to abandon it and "shut up and calculate" as someone famous once said.

 

[0xDEADBEEF]

I didn't look at the course. If the diode is biased enough then the current will be very large during the positive cycle and slower but still positive during the negative cycle. If it is not biased a small current will even go backwards in the negative cycle but in the positive cycle the current is much larger.

 

[f95toli]

Thank you, as you see I am only on page 1 of the online seminar so I have a lot to learn. To follow-up, I visualize the frequency of alternating current as electrons moving one way, then moving back, then moving forward.. No bias on the diode makes sense to stop the backward motion of the electrons, but if you bias the diode as they do here, is it proper to visualize "more" electrons moving forward, then electrons not moving, then "more" electrons moving forward, then not moving...? Can I keep this visualization or do I have to abandon it and "shut up and calculate" as someone famous once said.

No, it is never "proper" to visulize current as moving electrons as this level. The "water in a pipe" analogy works well for basic concepts, but fails miserably for non-linear components such as diodes.

The reason why you use a diode is that it is a non-linear component, which is what you need to build a mixer. Any non-linear component can be used for mixing, not only diodes.

Your link seems to be broken, but the diode is presumably forward biased because that puts in a nice range where the current-voltage relationship is approximately quadriatic. Hence, the diode follows a "square-law" behaviour which is the "ideal" behaviour if you are building simple mixer.

 

[rbj]

a "mixer" in this context is a modulator and what you are trying to do is accomplish multiplying one signal with another. the way they do this is by adding (or "mixing") the two signals together and then applying that sum to a non-linear device that will have cross-products coming out. the diode is just a non-linear device.

 

[Enthalpy]

And if the non-linear function is expressed as I=a0+a1*V+a2*V²+a3*V³+...

then V² gives the desired non-linear effect, as it creates a product from the sum:
[cos(2pi*F1*t) + cos(2pi*F2*t)]² contains 2*cos(2pi*F1*t)*cos(2pi*F2*t)
which equals cos[2pi*(F1+F2)*t] + cos[2pi*(F1-F2)*t], the desired frequencies.

Hence what we call a "mixer" is more than an adder.

If V³ terms are important you also obtain 2*F1+-F2 and F1+-2*F2.

When nonlinearity isn't desired it's called "intermodulation", of second and third order, and characterized by an "intercept point".

Once I made a mixer with the third harmonic of the local oscillator, to save a frequency tripler after the quartz oscillator. I had an IC mixer where the LO entered a differential pair, so the strong drive made a square current there, with important third harmonic. Of course the preamplifier must be narrow band.

On the other hand, a good mixer should mix only with the local oscillator, not between various strong input signals, and then the V³ term is not desired. Dual-grid MOS are better than bipolar transistors for that, and a passive mixer with a full ring of 4 diodes even better.

The diode ring is subtle in that it multiplies its two inputs but, if the local oscillator is strong enough, behaves almost linearly with respect to the signal input. That's its superiority, which is possible only if signal and LO inputs are separated.

Other circuits obtain it from symmetrical design but are more complicated without a clear advantage.

Also fun: varicap diodes make mixers as well, and can amplify the signal despite being passive, by taking power from the local oscillator. They add little noise, and can be cooled to sub-chilly to be more silent.