A mixer takes two signals and mixes them together. The resulting output is usually both frequencies, plus their sum and their difference. For example, if you feed a 5 MHz signal and a 20 MHz signal, you’d get outputs at 5 MHz, 15 MHz, 20 MHz, and 25 MHz. In a balanced mixer, the original frequencies cancel out, although not all mixers do that or, at least, don’t do it perfectly. [W1GV] has a video that explains the design of a mixer with a dual gate MOSFET, that you can see below.
The dual gate MOSFET is nearly ideal for this application with two separate gates that have effectively infinite input impedance. [Stan] takes you through the basic circuit and explains the operation in whiteboard fashion.
Oddly, you think of a mixer as adding two signals, but it really multiplies them if you think about the trigonometry involved. Consider a sine wave:
f=A*sin(ω*t+φ)
In this formula, A is the amplitude,ω is the radian frequency (that is 2 times pi times the frequency in Hertz) and φ is the phase (in radians). Of course, t is time. Now, consider two frequencies:
f1=A1*sin(ω1*t+φ1) f2=A2*sin(ω2*t+φ2)
Keep in mind a cosine wave is just a sine wave with 90 degrees added to the phase. If you simply add these two together you would get f1+f2, but to get the frequencies to add we really need ω1+ω2 and ω1-ω2 to appear in the output. Remember that sin(a)sin(b)=1/2[cos(a-b)-cos(a+b)] — or if you can’t remember, look it up. So multiplying two sine waves results in two phase shifted sine waves (cosine waves) of the sum and difference frequency. This is why the symbol for a mixer has an X in it and not a plus sign.
Mixers are a fundamental part of all but the simplest transmitters and receivers. Many times, you’ll see people use an IC like the NE602 or NE612 that incorporate an oscillator and a mixer together. There are other chips out there, too.
Ah! the NE602! 1980s cognates of things more modern; e.g. nrf24l01; but without all the fab digitry. Alas, poor lsi mixers; I know them, Horatio.
The NE602 has been superseded by the SA612. Still in production and readily available.
I do not see the relation between the ne602 and the nrf24l01…. but that may just me…
There is none in fact, except being both RF devices.
Nitpick: one could also use a NE602 to make audio effects, but clearly it’s primary use is as RF mixer and local oscillator.
I had (mistakenly) thought that dual gate MOSFETs went the way of the Dodo years ago, but I just noticed that Digikey has stock of parts from both NXP (BFR1105, BF992) and Toshiba (3SK292, 3SK291). This is good news for fans of retro receiver topologies.
Sorry, that should be BF1105R, with the reverse pinout w.r.t. BF1105.
Dual gate mosfets can be emulated by connecting two jfets like that: upper jfet source with lower jfet drain, all remaining pins becoming D, S, G1 and G2.
https://i.stack.imgur.com/6Nyhg.gif
Very interesting, this is the 1st time I heard that.
Be sure to match those two FETs!
There’s an irony hear. In the early seventies someone wrote about a cascode preamp using two jfets, and he dismissed dual gate MOSFETs for various reasons, but including their sensitivity to static. The early ones were not gate protected, so they came with wire shorting out the leads and endless detail bout protecting them before they were in circuit.
Then later the author recanted, protective diodes had ben added to the gates of MOSFETs, but then he added “dual gate MOSFETs are really a cascode arrangement”. Nobody gave it much thought, they were available and relatively cheap. Then they became hard to get, and in recent times things like your schematic have appeared to replace them, and it certainly looks like a cascode amplifier.
Micel
I hear another Jenny List “Transistor Circuits” being typed up!
Nitpicking: “w is the radian frequency” should be “ω is the radian frequency”. ω just looks similar to a w, but is a totally different letter.
Oh, and “f=A1*sin(ω1*t+φ1)” is missing a “1”, and “p is the phase” is as wrong as “w is the radian frequency”.
Arggh. I always type them straight and then go back in WordPress to fix them up and just missed. Sorry — fixed.
Also, · and × are much nicer multiplication operators than *.
This circuit seems like it would be ideal for a direct conversion receiver (heterodyne), considering the low coupling between two gates.
There’s a classic article in QST I think in 1968, where “direct conversion” was used. I don’t know if the term had been used before that, but it became common as a result. It actually used four hot carrier diodes in a mixer, which were fairly new at the time.
But then the dual gate MOSFET was common in direct conversion receivers, the MOSFET available about 1968. But there were issues, so endless articles bout improving the mixer. Double balanced didn’t seem to help, though eventually in the simple ones the NE602 double balanced mixer IC took over.
A big shift happened when someone terminated a hot carrier diode mixer properly. I think it was Roy Llewellyn. The concept had been there previously, when diode mixers were used in superhets, but since direct conversion was for simple receivers, nobody thought about complicating them. That brought a wave of complicated direct conversion receivers, which often included phasing to get single signal reception.
As some have pointed out, dual gate MOSFETs have become less common, other devices providing the same end game. They made such a splash when they became available, but over time provide little that other devices don’t do.
Michael
Even before that (and darnit I don’t have the year but not too long after WWII) was the Costas Loop in IRE, which used direct conversion and I and Q channels to lock the LO on to what would have been the carrier in suppressed carrier DSB. I actually built one back then, and AM in stereo is possible using the phase shift SSB method to get each sideband separately – the “acoustic” image is that of the multipath propagation on long distance SW and is…cool.
Either QST or Radio AM handbook once had a circuit for this IIRC using a 1496 balanced mixer.
https://en.wikipedia.org/wiki/Costas_loop
This gives better reception on fading AM than other methods since carrier fading is meaningless and doesn’t cause distortion anymore.
I’m glad to see Hackaday doing these articles on the fundamentals we old farts had to learn – as it was nearly all there was. I have to say that knowing how things really work inside, down to the metal and even electron flow, has been a boon thoughout my career and life.
May I suggest adapting the opamp design tutorial and the phase locked loop one from say, the National Semi Apps handbook circa 1978?
Both were very enlightening and even life changing for this old engineer and got it across to the intuitive sense where all others had (and still do) fail.
I often say you don’t need to know how a car works to drive one but you can bet every driver in the Indy 500 knows exactly how they work. Of the best and worse jobs I ever had was taking apart microprocessors under a microscope to figure out why they failed for a “big semiconductor company” that has had a name change since then. Great job because it was high tech/sciency. Bad job because of politics and things. But even though I don’t often need to understand how an ion implantation can go wrong on a MOSFET inside a logic gate, it is surprisingly helpful from time to time.
You know, I had the TI Op Amp book back in the day and it was a model of obscurity with plenty of math equations. I finally worked out on my own (probably not the first, though) what I would later read in the Art of Electronics as “The Golden Rule of Op Amps.” Now Phase Locked Loops is a whole other thing, but probably my second favorite complex architecture to work with students about. The first being Analog to Digital conversion. So many different things you can do there that are interesting.
I liked the National opamp tutorial as it finally explained to me how *reduced* transconductance in the input pair could lead to better slew rate (I was designing audio amps on the side at the time, eventually improving on Meyer’s Tiger .01). Understanding pole splitting, and why a limit was then the transverse PNP was also good. It made it easy to understand why all the best opamps for audio (at the time) had FET front ends, and that fast audio amps with transistors had emitter degeneration resistors in the input pair (or quad).
Ion implantation was too expensive then to be a thing, AFAIK. We weren’t all that far from vacuum tubes then.
Now I run into people dealing with aggregations of charged particles thinking there’s some new magic when the tube guys worked out focus. bunching, and space charge effects quite awhile back…
Their PLL finally got it through my head about the tradeoffs between loop bandwidth (as opposed to main oscillator range), the ability to keep lock into the noise and capture and track range…nothing else did.
Above all, unlike the physics I now do – worked examples with units defined really helped me get past the jargon.
Physicists seem to have a nasty habit of switching units in mid-equation to avoid writing c or pi, for example, and if you actually try to MAKE something based on not knowing that…well, getting ludicrous numbers is a clue you need to go back and really understand – and not just throw around the symbols to impress the unwashed like so many do.
Not to thread-jack, but I found these online from TI (who I guess wound up buying almost everyone).
Worth spreading around in a place like this:
Opamp tutorial: http://www.ti.com/lit/an/snoa737/snoa737.pdf
And funny, the same mixer math as above is also in this PLL app note: http://www.ti.com/lit/an/snoa651/snoa651.pdf
This is a lot different than the RCA PLL phase detector in the 4046, which might more accurately be described as a frequency locked loop.
Sorry, without an edit button I’m not brave enough to try to make the links clickable.
Seekers who don’t know this stuff already are in for a treat! Magic reduced to sufficiently advanced tech, in other words.
Last time I searched for dual-gate MOSFETs, they were nearly unobtainable. Or $20 a pop.