Can FET replace SPST switch for beam scanning? And FET as resistive mixer question.

Scanning patch array antennas become more popular: WiFi, cellular, sensors, etc. all want to receive or transmit signal in certain direction to reduce interference, and able to alter this direction dynamically. I reviewed few papers and datasheets on commercial products, and have many questions and doubts about trends in this area.

Different products require frequencies from 2.4GHz (WiFi) to 9-10 (Satellite) ~70GHz (Sensors).

What can be used to do beam-forming (must be cheap!):
1.Phase shifters. On of the way to form beam in certain direction is to alter phase of individual sub-arrays. This requires expensive phase shifters. Although it can be realized using varactor phase shifters at 2.4GHz or even 5GHz. For higher frequencies it requires expensive RF ICs. One IC for each sub-array.

2. Frequency-scanning arrays. Require too much bandwidth, or too log phase delay lines (losses).

3. Sub-arrays with different phase shifts, which are switched. There are switch IC available up to 5GHz. Higher frequencies - too expensive. But this way looks like the cheapest one.
At higher frequencies, is it feasible to replace SPST switches with FETs and get acceptable performance at least for some applications?

4. Unsure if this idea is correct: What if we make receiving patch array, where each sub-array have separate mixer? Then do phase shifting in time domain on IF low-frequency signal using DSP. After time domain signal from each sub-array is phase shifted, they are all summed up digitally. Could we achieve receiving antenna beam forming using this approach?
For example, we have two sub-arrays (A) and (B). If transmitter is closer to first sub-array (A), then signal at sub-array (B) will have some phase-delay (few GHz signal).
Output of mixer connected to sub-array (B) will have this phase-delay too relative low frequency of MHz. Now if pretty fast ADC is used, waveform from each sub-array can be phase-shifted in time-domain (somehow), then combined. Is it a digital beamforming? It seems the cheapest method (ADC price x sub-array count + DSP price). Only passive (for receive)

Using FET for oscillator, mixer, switch at the same time theoretically can reduce costs (BOM, buying only one component for mass production - higher amounts, lower price). Although can complicate design of mixer and switch.

>>>
And now FET as resistive mixer question: why resistive mixers are so rarely used? What is the best way to apply RF and LO to resistive mixer? Some authors apply them together, some to gate and drain, sometimes LO is at gate, sometimes at drain. Maybe if LO is applied to gate, biasing requirements are lower, but RF can't be too small. IF RF is applied at gate with good biasing, then LO applied at drain can provide stronger IF output (maybe).

Can we use FET resistive mixer for upconversion? How it compared to Schottky diode in upconversion tasks?

GaAs FETs can be used for RF switches in a frequency range, from near DC up to about 20 GHz.
FET resistive (passive) mixers are widely used for frequencies up to about 2 GHz. They have good port-to-port isolation and can be used also for up-conversion. At higher frequencies their insertion loss become to high, when compare to Schottcky diode mixers.
Resistive FET mixers generally have better linearity than diode mixers, but requires higher LO drive level.
The passive (resistive) and active FET mixers are very similar. The main difference is how the devices are biased. In the active mode the FETs are biased nominally in saturation and have DC current. In the passive mixers, they are biased near the threshold.

If fet resistive mixer insertion loss is too high, can it be compensated by using higher LO power?

Increasing the LO power helps to get lower insertion, but as expected, this increase in drive power has a limit.
In most cases the LO power cannot be higher than about +15dBm, depending by mixer topology.

Which port of FET must be driven to get up conversion effectively? Using only one FET device. I guess IF at gate, and LO at drain? RF output at drain too? Or maybe LO and IF at gate? I am using passive fet mixer

I have read material on FET resistive mixers and made simplest possible FET resistive mixer using classical configuration:
Grounded source FET, LO is applied to gate pin, gate pin is biased approximately at -1v, RF is applied at drain, RF pin is grounded through 80 Ohm resistor. I put the resistor because direct grounding through low-pass filter also shorts IF output. I tried also to use 1 KOhm resistor, but only noise is increased and some voltage +0.2v across the drain resistor (Vds=+0.2v), although no positive voltages are applied to schematic. In case of Rd=80 Ohm, Vds=+0.05v. Here is an image:

In all material i read IF is relatively high. But in my case it is audible frequencies.
I do not understand, is it right to add drain resistor in this case?

Your additive FET mixer configuration is unusual.
Try using the Gate for RF input, Source for LO input, and Drain for IF output.
In additive single transistor mixer, the active device should be biased for linear amplification.

I am concerned about feeding weak RF into gate, but i will try. I thought that with weak RF it would be difficult to achieve switching, or negative bias will be too sensitive. So LO is fed to gate. I found similar passive mixer configuration in many sources, patents and one very old RF book. Also i reviewed this material:
HP's "Application Note G005: Active GaAs FET Mixers Using the ATF-10136, ATF-13736, and ATF-13484"
http://www.hp.woodshot.com/hprfhelp/4_downld/lit/xrlit/ang005.pdf
It has many important information, but mixer is active. Very low current is used to ensure stability. So they use common source configuration, RF is fed to gate, LO to drain (not only). Currently i am very satisfied with the mixer, but want to ensure I got most of it.

Additive single transistor mixers they are not switching mixers, as other types of mixers. This is the reason that I mentioned that the transistor should be biased for linear configuration.

Regarding the well known HP app note, there are mentioned few input/output configurations for RF, LO, and IF.
But as you see, none of them is the mixer configuration that you posted above (with LO input to the gate, and RF and IF at the drain).

But that appnote is about active FET mixer. In most works LO applied to gate in passive FET mixers. But here is pretty interesting info:
https://www.miteq.com/docs/MITEQ_Mixer_Question_Answers.pdf

Q11: Are there passive modes of operation for the MESFET mixer other than LO on gates and
RF/IF on the drain source?
A11: It is possible to get very low conversion loss (-3 to 0 dB) by applying LO between the drain and source and
RF to the gate with IF output at the drain. In this mode of operation, the LO periodically powers the FET into the
active amplifier region and one obtains normal amplifier gain less the Fourier LO switching coefficient (approximately
-6 dB). The input IP3, burnout and noise figure for this mode of operation are all considerably lower than drain
source mixing. The lower limit of noise figure for this mode of operation is 3 dB because of the image response.

Click to expand...

LO is applied on passive FET mixers only in case of QUAD FET passive mixers, and not in single device FET mixers.
In fact LO is not recommended to be applied to gate (or base in BJT case) in any single transistor mixer type.