Comms Board Walkthrough

• custom tranciever
  • a lot of this is from the recommended schematic sheet, most of the values are too
  • rf circuitry is pulled from the manufacturer datasheets
  • vna testing: input and output ports. to decouple the rf system and isolate it they added these 0ohm resistors, two of them overlap so you can solder either one or the other (horizontal or vertical) this helps to test
  • UFL connector→SMA→VNA
  • You can make more a loop between test points
  • SPI interface on the left: one of them is an interrupt, regular GPIOS
  • decoupling capacitors also included, crystal oscillator
  • RF testpoints are
• Testing this? spectrum analyzer, can use with VNA
  • this would be transmisison trsting
  • work with shaker for understanding rf side
  • can observe impedence with the VNA
  • they did SPI testing
  • VNA is a 2 port measurement device-input and output-it sends waves and shows outputs
  • main purpose is to just transmit and recieve a signal-fw writes over spi and it sends out a transmitted signal and then recieves and demodulates
  • 16db output power max- we want 31- so we amplify signal with frontend
  • chip handles modulation-shift keying wave, it handles all of that
  • half duplex-cant transmit and recieve at once
  • 0734120114, on wifi cards - large pad footprint
  • adapter for UFL to SMA → most testing has this type
  • rf docs: Custom Transceiver Design
  • gpio on bottom, rf pins on right (filters)- PA is power amplifier, TRX_SW→ most done in fw
  • PA-RF out, digital signals from microcontroller→ ic→ analog→ comes out through PA
  • other way around→ antenna recieves RF data, piped through same path goes to PA and LNA, chip only selects one to take as the input. only transmit or only recieve sections. LNA-low noise amplifier.
  • for rf ICs u can connect different power supplies (different decoupling caps on the left is because of this). all power pins to the same supply
  • 3v3a - 3v3 analog - quieter power rail
  • oscillator: tuned resonant tank, generates sin wave at a frequency, allows to keep track of the passage of time
  • most passives out of PA are for filtering
  • wide band antenna may pick up other frequencies - this could influence our signal and it basically gets rid of everything you don’t need (harmonics analogy). need to clean signals for RF
  • so reciever should recieve the same signal regardless - caps and inductors fiter
• frontend:
  • interface between transciever and waves
  • this does all of the amplification and some filtering as well
  • they had a LDO to step down their 5v voltage to 4.2v - dissapates in heat. couldve used buck but this was a small step. IC doesnt support high pulse?
  • dont use buck for RF because of the ripple - can affect signals and analog components. LDO doesnt cause this problem - no switching components here
  • TR/EN/BYP pins: gpio pins, it can either transmit or recieve - we also need to tell the frontend so thats what this does. BYP-bypass recieve mode, shuts down and does no amplification
  • truth tables and diagrams in the datasheet are super helpful
  • dont use bypass as a low power mode,be careful cascading amplifiers
  • cc1120 chip is the tranciever
  • 50 ohm trace impedence: may have documentation. everything has characteristic impedence-cap/ind. wrong impedence will cause signals to reflect back and lose strength-need right power output
  • parasitics = non idealities of a component. lump element model. 0.5c electrical speed. the period and distance that travels becomes more important, the lump element model doesnt work in high speed signals. a distributed model assumes small distributed components along wire lengths. Back to the 50 ohm trace: not literally 50 from a multimeter, the 50 comes from if you send a signal along this trace - before it gets to the other end it doesnt know whats there. As it travels it sees the distributed peripherals and essentially RLC filters. If you feed a signal through one section of this youll getan impedence. usually res is much smaller than cap and ind. LC filter slows down signal progression though. multimeter would see 50 ohms in the first nanosecond of measurement→when the signal has not reached the end
  • at high frequency, ur essentially always at that middle state
  • not lumped elements, signal doesnt see all the elements all at once, just at a point, so thats where the impedence of a transmission line come from, its what the high end of a signal sees
  • saw 1 and saw 2 pins represent a saw filter: trims of frequences higher and lower than some value, has a center. 433Mhz is our target frequency-centers around this. this is not exact with decimals (licensing requirement??) they picked something for now. close to ideal. how does the +/- setting of thid work - numbers are in the datasheet. these can be tuned
  • end of last summer amateur radio association walked them though process (might have it on notion), 433 is an amateur radio band→ will need to look into this
  • dk if you’d use the same without selfie sat. other bands in the range might be useful, shouldnt have to scrap frontend [follow up about this, and how selfie sat is changing things]
• VNA: vector network analyzer. main purpose is to measure or quanitfy cps of a system. RFFE = RF frontend. datasheet explains parameters.GAIN: input, how much stronger is it, these are s parameters (reflection etc.), you can also observe how systems vehave and what happens to the signal you send through - you get a frequency vs intensity plot. → used purely for testing, in pairing with a spectrum analyzer
• if chip is in transmit mode, and tx in antenna mode, youll see gain of the chip that matches the datasheet
• chip has filters inside, so u may see 0 or negative gain with some frequencies.attenuating - large negative
• VNA through saw filter - passive - no gain - 0db. vna also lets u see small attenuators too - no complete efficiency. there is a passthrough frequency

left side

• ANT pin: another RF network like the CC1120. antenna is here. this is a pi filter. →read up on pi and t filters though
• cap datasheet: +/-2%. NP0 code: each section describes one specific parameter. number is dielectric - look this up. when you select decoupling caps theres x7r, x5r etc: if you use those for rf itll be bad - theyre cheap and have hgih capacity for size but they vary a lot in capacitance.
• important to have caps that are 2% and compare NP0 Gx0 etc. and look at the voltage rating too. adding dc bias can drop 80% of the actual capacitance.

Chip has configurable gain - 31db. csdc req- cant output more than 1 watt of power. db vs dbm documentation. nts: need to read requirements.

rf frontend can amplify after 26/31 dbm? cc1120 outputs 11dbm. VAPC pin needs to be tested

DAC is final comp, but RF frontend board doesnt have it. can be hooked up to VNA, connect to pot and vary it, see the results and then convey it to figurmware.

feed in 1/0 and it varies the output voltage, fw talks to it over i2c

vapc is a pin on the board, changing it changes gain of the power amplifier, for testing u can use a pot, DAC is for the final competition board

DAC might be connected to 5v? or 3v3? important to double check this

Layout

• 4 layer board is important for RF, this one is 6 layers though- recc’d by prof.Shaker
• trace widths are super important
• all the vias: they made them test points, stitching was too close and they got confused and didnt make them vias.RF traces can interfere,this is their way of preventing interference between traces
• i2c lines may have them too, RF lines can interfere with digital lines and vice versa.
• for next time: remove thermal relief on stitching vias. outer one just keeps the pores contained
• via stitching for ground plane depends on frequency - regular every such distance. empty space should be stitches as well. calculators online will recommend spacing for you
• UFL connectors: dont route between ground pads - they recommended not routing between these pads.google rf considerations
• keep them short

Special design decisions

What went well?

What didn’t go well/tasks for fixing it?

• need to correct stitching
• all the functional components are present
• still worth testing, good vs great
• backup DAC sourcing might be nice. hits max at 1.9v on VAPC,dac only outputs a max of 1.98 v - not the best margin
• backup RF projects for learning experience might be nice if RF dries out
• backup transciever project wasnt finished - some projects off the shelf can help with this. we can still test this module - would need fw work

Improvement ideas

What needs to be purchased to complete the boards?

• VNA - this ones important, spectrum analyzer - even if its nano. hard to get your hands on RF testing equipment
• cc1120 stock as backup might be nice
• nano VNA could be improved with attenuators

How much documentation already exists within our team about this design?

Where are the datasheets and research materials used located?

• ask kaveet if necessary

Anything else about the boards we should know?

Link budget - ask aryan. antenna theory

deployable antenna - antenna is a patch antenna right now, deployable is needed for a launchable satellite. neurosat one - dream antenna. shouldnt affect anything on the hardware end