Radio Design 101 - RF Mixers, Part 2 of Episode 5 

This was such a fun class to teach, and one that students really enjoyed, that I am hoping the videos provide some of the experience that the college classroom affords. Thanks for the comment !

Thanks. Good point. I probably should use the term non-linear circuit rather than switching mixers. The BJT mixer (timestamp 6:50) may not fully switch on and off - but it does have non-linearity and transition softly between 'on' and 'off' during the LO cycle. So the diagrams are definitely idealized to a degree. I actually never liked the commonly used formula f_IF = N f_RF + M f_LO because the N value is practice is mostly limited to N = 1. Unless one overdrives the radio front-end - but that does occur in situations (as noted in the Epilogue episodes in this searies). And as shown there, we can get intermods too... As you noted - it's definitely more complicated ! (but that's something we try to avoid in good receivers of course)

Thanks for the question. In the videos, I'm almost always talking about "signal" flow. In this case, the signals (the RF and the local oscillator signals) are AC sinewaves, so they can pass through the capacitors without any problem, while the capacitors 'block' any DC 'bias'. As you noted, capacitors do block DC. But they are also used for "AC Coupling" - letting the AC portion get through. To go a little deeper into this, think of the voltage and any node in the circuit as AC+DC. For example, at the input of an amplifier, there is a "DC bias" required to get the transistor turned on. But the input signal is not just a flat DC line. It has the signal "riding" on it. Some examples are shown in the amplifier design discussions of Episode 3. Time 10: 14 to 11: 00 shows one example here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-UUlqW-vSq9M.html Hope that helps.

For most mixers, the RF and LO ports have to be within the RF range of frequencies (MHz to GHz depending on the model). But... The IF port typically works down to DC, so it includes audio frequencies. In our classes we routinely mix down to audio range at the IF output when building Doppler radars. That said, if you mean can you use audio frequencies at other ports - the answer is - maybe - it depends on the mixer type. Many have transformer inputs, so they won't pass audio. But something like the NE/SA602 (or 612) IC could be AC coupled with a sufficiently large DC blocking capacitors (e.g. 10 to 100 uF) on the RF and LO ports. (Of course they're still switching-mixers. If you want a pure multiplication, then there are analog multiplier ICs, or you could just digitize two signals with 2 ADC channels and multiply them in the microprocessor I suppose :-) Hope that helps.

@@MegawattKS i mean homemade mixer like using 4148 diode and Ferrite As transformer or , using dual gate MOSFET mixer for low frequency in khz ?

@@omsingharjit I think that would be possible. However, keep in mind that the mixer is a switching design, so you will get out many, many, frequencies in addition to the sum and difference frequencies. The frequencies (n*f_LO +/- m*f_rf, with m and n signed integers) will be in the audio range but the waveform and sound that come out will be pretty complicated (and 'messy'). That's true for the 602/612 devices too. I guess it depends on the goal/application. Do you plan to have a narrow bandpass filter to select the desired 'product' ?

Excellent ! The good news is that the high-frequency content typically doesn't get much further down the receiver chain since the IF amp is typically lowpass or bandpass :-)

I'm starting to work on the next (and final?) video. The good news is that it will leverage the content already covered for the IF limiting amp (with some tweeks), and the demod uses the SA602/612 which we're already familiar with. (In the class we used chips like the LM3189 or SA604 for IF amp and demod, but I didn't have any of those, so sticking with the basics. Those chips use a Gilbert Cell internally for their demod, so the techniques are basically the same...)

@@MegawattKS i am wondering whats the signal strength that would get a station receivable.....my SDR(USRP) had been unable to detect stations, my place gets poor signal strength. Is -55 dBm after LNA a signal strength that could translate into audio? (i could always test with my own SDR station, if commercial stations reception end up not working).

@@cholan2100 Yes. that's actually a reasonably healthy signal :-) A good FM broadcast band receiver should work down to within about 20 dB of the thermal noise floor (kTB). Here, B is the channel bandwidth, so 200 kHz, giving a noise floor of about -121 dBm if everything is perfect in the receiver (including low noise figure in LNA). So if all goes well, you should be able to get signals as low as -100 dBm at the antenna. In the class final exam demos, we set the threshold sensitivity spec at -110 dBm and the RF input power needed for 30 dB audio signal to noise as -100 dBm. Again - these are signal levels at the antenna, so if you have 20 dB in the LNA, you've got -75 dBm at the antenna and a good 25 dB margin :-)

@@MegawattKS looks like electrical light fixtures and other electronic devices are dumping so much noise in the band. To my surprise even USRP SDR(despite being in a metal enclosure) on its own is radiating(unintentionally even when its not programmed to radiate anything). So not surprised its having trouble receiving receiving the FM stations. I had a powercut by chance in my place, all of sudden my receiver started receiving stations.

Hi. Interesting problem. It sounds like maybe the LO drive is not strong enough to cause the needed multiplication to happen. For a design like that at 11:30 (or more properly at 12:30 since that includes the output filter you mention), the LO signal at the base needs to be at least 100 to 200 mV peak (with the emitter AC grounded) to force the transistor to turn on and off over the course of the LO cycle. That 'chops' the RF signal, meaning its _multiplied_ by a squarewave (which contains sin(w_LOt)) at the LO frequency. If the waveform looks like the LO, AM modulated at the beat frequency, then it sounds like the result is just the addition of the signals, with no multiplication happening. I.e. it's just something like 2 sin [½ (w_RF + w_LO) t] cos [½ (w_RF - w_LO)] = sin (w_LO t) + sin (w_RF t). So its just the addition of the LO and RF sinewaves, with no new frequencies in its spectrum. What are the frequencies and amplitudes being used for testing? And is the circuit something like that at 12:30 (with a good emitter bypass)?

@@MegawattKS thank you very much!

I checked some other ICs above 500Mhz. I found they are called up converters or down converters. No one is simply called mixers. Do you know why they can only do up or down conversion instead of bidirectional conversion like 602/612? What has been improved to trade off that functional loss?

@@sullivanzheng9586 High frequency devices are difficult to use; they oscillate without extreme care. Up converters help this by blunting the input some and down converters help this by blunting the output some. Even so, they are VERY finicky. This is also why RF amplifiers have low gain.

@@byronwatkins2565 As I search on mouser and LCSC and I found a few "RF gain block" ICs such as TQP3M9009 (50M-4GHz Gain is 27dB@50Mhz, 0.8dB NF) or BGA2869 (DC-2.2GHz, Gain 31dB@50Mhz, 3dB NF). It looks like their gains are not that low? And I also find their bandwidth are crazily wide and NF are very low. And they have 50ohm input and output impedance so save me lots of time design and fine tune impedance matching circuit. I wondered if they are good to use as LNA after attenna for the FM receivers I am building? I found very very few people use these ICs when they build their own radio. Am I doing something wrong here? I also tested several samples of these RF gain block ICs, the S21 parameters are very very impressive and noise is almost negligible, so if I reason from first principles, these ICs can replace those FET/BJT version of LNA amps, can't they? (As long as their working frequency covers low freq, like all the way to DC)

@@sullivanzheng9586 Be very careful to maintain 50 Ohm traces, to bypass power well, and to limit the bandwidth. Also, be aware that the broad dynamic range of FM station received power makes receiver design tricky: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-pRsXqeU9Vtw.html I don't know why others don't use these for lower frequency, but I am sure that the cost and the tendency to oscillate are considerations.

@@byronwatkins2565 My guess about few people using broadband LNA is that interference may be a big trouble. Just like Epilogue of this video series says, strong AM signal may overshadow/blind our target band (FM). JFET LNA usually have some band selection "for free".