Technical obstacle – communication design at 2.4GHz – done.

Well, under Vancouver’s rain, range tests was successful – sprayed by shower antennas and prototype boards showed stable reception over 33th street up to distance 1200m. In the middle of the tests shower became pouring rain, and at the end (in 15min) equipment become totally wet (like submerged in water). Etalon BT recipient was able to pickup signal

Cubesat prototype

on distances of 10m at the same time(!) LNA was able to get 0dBm signal on distance 190-160m. Thanks to the weather (loss of a signal), the last question, “how to design communication equipment for 2.4GHz?” was passed. No problem in PCB, no problem with LNA, impedance matching was The Problem.

Actual fix was: the connection cable (for prototype only) from the LNA’s PCB to the antenna was not a coaxial cable with 50om impedance, but “microphone” shielded cable with an impedance of 33om. When that was accounted (together with physical dimensions of all connectors) and series:16pF->shunt:1.2pF->sr:5pF->sr:3.3nH was changed to sr: 2.7nH->shnt:1.0pF->sr:22pF->sr:3.9nH rain become a sunshine in the end of the tunnel.

If anyone is still reading this, then the following is an "Complete idiot's guide. Amplifier, 2.4Ghz." (case when 2.4GHz capable measuring equipment, network analyzers, oscilloscopes ant etc. pricy, totaling 20,000$ min, average== 40,000$ are unaffordable):

a) Trust impedance’s calculation – needs to try different calculators to confirm (in spreadsheet) different formulas, same input -> calculations -> same output==result. The best calculator conformed is by Dan McMahill: http://wcalc.softpedia.com/  or download from different page http://publisher.brothersoft.com/dan-mcmahill.html No needs to be afraid to take integrals by Runge-Kutta formulas to confirm calculations (just kidding! – some knowledge in math is OK). People already spend a lot of time and study to get proper approximation’s formulas, they knows what they are doing, needs to use already accumulated knowledge. For impedance matching the best is Smith’s software (free version does allow using less than 6 components).

b) Calibrate capacitance meter – needs to have at hands different SMD capacitors with known values, calculation of a delta from readings takes 10 min. Needs to use the same meter’s pins for measurement.

c) Better to use SMD components - to solder the best is an infrared oven with a profile temperature’s control (todays it is on E-bay), to rework the best is a hot-air station, good tweezers is mandatory. Needs to calibrate the oven – usually it requires 10mm stand inside the tray, needs to place the electronic thermometer on the stand, insert the tray into an oven (check a max measuring range for the thermometer) and by lowering or elevating the stand, needs to get the best match for profile on LCD screen and actual temperature readings, it is taking a time (couple of hours), but oven’s calibration is mandatory. Soldering paste - for earth is Pb40%Sn60%, - for space In60%Pb40%. Soldering paste is preferable, but indium paste sometime unaffordable, alloy InPb is expensive – better to mix by weight Pb with In (lead from 3WET store == fishing equipment, and indium from e-bay). If soldering can be done by the alloy only (not by the paste) thanit is better to use a hot-air instead of an infrared oven. If soldering will done by an soldering iron than amount of the alloy can be controlled. . . (designers! Pls! do not laugh – it is my way).

d) Measure impedance of any piece of a wire pretending to be named as the coaxial cable – (on a picture connection btw PCB and the antenna actually done by a microphone wire – all designers was laughing looking on it). To measure an impedance cut piece of cable 20cm, 40 cm, 60cm, and measure the capacitance btw the central wire and the shield. Open a calculator for a "coaxial transmission line" and enter physical parameter of a coaxial cable. One parameter left unknown, it is Er (dielectric constants for insulator), change it to match the capacitance per mm(or m or ft) with already measured for the small wire. Confirm for a different wire’s size – take the medium for Er values. Then the value for the impedance in calculator will be what are you looking for.

Cubesat prototype

e) In PCB calculations needs to use “coplanar waveguide with ground plane” instead of a microstrip - do not be confused with the name “microstrip” – the microstrip requires empty space btw components on PCB – usually it is unaffordable. Difference in the microstrip and the “co-we-wi-gr-pl” is 25%. Difference also in == for microstrip it require precise thickness, for waveguide it less dependable.

f) In any PCB/wire/coaxial calculation needs to use as much as possible details (connectors are separate elements, soldering pads are separate elements). No need to burry yourself under lot of details, but more details brings better result, anyway all small details will be accounted, better to do it at the beginning than at the end.

g) Do not trust what component’s supplier send,(it is not their fault – mistakes are possible during packing 100 small, same looking pieces of paper, with 0.5mmx1mm components embedded in it) measurement needs to be done before use, capacitors can be different values, resistors - different precisions, inductors- varies, instead for a capacitor you can get the resistor, that can burn active components and etc. No needs to be a paranoid, but basic component‘s checks better be done.

h) Needs to measure PCB parameters, manufacturer's spec needs to be verified – order PCB with big "testing" microstrip(or “co-we-wi-gr-pl”) and measure physical dimensions of dielectric. Measure physical dimension with PCB with plates on top and bottom and measure capacitance of a tested element. Use calculator to get range of Er – error in 1pF will give a range of a dielectric constant (Er) for a PCB.

i) Better to use two layers PCB instead of 4 (actually that is preference for ExpressPCB – for different PCB manufacturer that can be a different case) variation of Er and thickness of dielectric brings less variation for impedance for “coplanar waveguide with ground plate” in case of 2 layers board then for 4 layers.

Thanks Gregory for impedance calculation explanation. Thanks Boris for rifle sight (scope pointing), Thanks Shura for explanation of interference on transmitting antenna/diagram. Thanks Luda & Serg for catching cold under Vancouver’s winter rain.

Cubesat prototype

Now it is a time for cubesat/ground station 1WT transmitter (with possibility boost power to 10 WT), and time to assemble everything.

Share this Page

Become a Part of the Future! Sign Up for Our Newsletter: