RF Amplifier for Cubesat plasma thruster

[Weston]

I recently wrote up one of my research projects on a class E inverter for a cubesat plasma thruster for a RF design contest NXP is running, some people here might find it interesting:

http://superlab.stanford.edu/rfchallenge.html

The inverter is 40W at 40.68MHz and similar to a lot of the class E tesla coil stuff. Working on this reminds me of how way back in the day there was a competition on 4HV for who could make the highest frequency tesla coil, perhaps I should start working on that again.

[Da_Stier]

Very interesting to read for sure.

The circuit is kind of similar to a lot of ouput stages in Q-Switch power supplies.
Thee run at around 32MHz most of the time.

Great work. 

[klugesmith]

Great work there, Weston.

Please tell us when the vacuum chamber testing happens.

Is there Cubesat-oriented literature about managing the temperature of electronic subsystems in space?
I guess the extreme cases are full sunlight and full shade.
Aside from temperature, how does vacuum make things different for circuits with hundreds or thousands of volts inside, or sensor circuits where nano- and pico-amp leakage currents could be problematic?

What about Cubesat telemetry and remote control?  Are there optional or mandatory packages for that stuff, including the ground side?  Does a 3U box have enough room for a heat-resistant re-entry pod with micro-SD card inside? 

[Weston]

Space rated electronics is something I am still figuring out and I have not found much reading material on yet. It's a hard vacuum so there are no electrical leakage / ionization issues. It mainly involves lots of thermal de-rating, which is one advantage of this amplifier with a to-220 power transistor. A previous version used a SMD GaN part which had thermal issues even before vacuum chamber testing. You also have to worry about components with electrolytes, like batteries or capacitors.

Everything on the PCB can be heavily de-rated except for the gate driver IC, which I am a bit worried about. Its a little WLCSP part and there is no version in a larger package. The part dissipates ~ 0.2W. I can try to add more thermal vias around it if I need more thermal dissipation in vacuum. Worst case it has a built in thermal shutoff which should prevent lasting damage if it overheats.

Cubesats are large enough where you can manage to communicate effectively, space is not very far away, everything just moves really fast sideways. I don't work on that side of the project, but there is some sort of ground station network run by the academic community. 

[klugesmith]

Thank you!

... space is not very far away, everything just moves really fast sideways.

Not trying to hijack your thread, Weston.  The really fast thing reminds me of the Huygens probe to Titan, launched with a comm-link design bug that almost precipitated a total disaster.  Story is well told here:
http://www.thespacereview.com/article/306/1

Short story:  Huygens depended on radio link to mother ship Cassini from the time they separated, though Titan atmospheric entry, descent, and landing.  Range rate, and changes in range rate, of many km/s.

The radio receiver designers correctly planned for the Doppler shifts in carrier frequency. 
But getting the data bit stream out of noisy demodulated signal depended on having a very exact baud rate.
And, uh, Doppler affects baud rates and carrier frequencies by the same ratio.
If astronomers saw a red-shifted pulsar in the sky, the redshift factor would apply to its apparent rotation rate.

[klugesmith]

Back to nominal Subject, Weston's RF amplifier.

After thermal vacuum, does Cubesat testing include random and swept-frequency vibration on a shake table?

Be a shame if the rocket ride happened to make that toroidal coil resonate, mechanically, enough to break a connection.  In my ancient and limited experience, most of the strength demands on spaceframes (?) and subsystems are just to survive the ground-to-orbit trip.

And speaking of coils, those blue-and-white SMD inductors sure are pretty next to the TO-220 part.  The L1-C3 and L2-C2 branches are electrically in parallel, and simple circuit theory says the L-C order doesn't matter.  Is one branch swapped on purpose to reduce coupling between the inductors, or in consideration of some other parasitic circuit effect?  It makes the schematic diagram typography easier. :-)

[Weston]

Good observation on the inductors, they are indeed placed like that to reduce the coupling. This is an early electrical prototype, mainly to evaluate the performance of the MRF101AN in class E and the other component selection. Vacuum testing of this version for the team in Australia to test it driving the actual load of the plasma thruster and to get some early results on the thermals.

The finished system is going to be more mechanically robust. It is also going to have some more control and monitoring circuity, and possibly its own energy storage.

Interesting to hear about a failure due to doppler shift of the baud rate, I am surprised it had to be so precise.

I have heard that a common issue with cubesats coming from the academic community is a failure to achieve a stable orientation / to de-spin. Hopefully that will not happen with this project!

[shrad]

For widening your vision on this kind of oscillator/amplifier combo, you may want to document on HF soldering of plastics which uses the same frequency range and topologies, having to feed large electrodes and be reliable under non ideal conditions (manual operation of soldering machines, with or without coupling between electrodes, able to handle harsh condition of industrial use, etc...)

I have several units at home that I got for the HF mosfets principally, and it rang the bell in the back of my head...

[Weston]

As an update, my project won third place! https://www.nxp.com/products/rf/rf-power/homebrew-rf-design-challenge-2019:Homebrew-RF-Design-Challenge-PG

I have heard of dielectric heating of plastics, it seems pretty interesting but I have not found a good use for it. It might be an interesting project though. A way to make bending acrylic easier?

For that you would probably want a linear design, it will better withstand a load mismatch and you can get more power per transistor without exceeding any ratings. The refrence design in the MRF101AN datasheet gets a ~82% efficiency, which is pretty good. In space applications it can be hard to dissipate power so I really tried to optimize efficiency for my design.