Protect PCB from close proximity radio transmission

Thread Starter

BeachCircuit

Joined Mar 26, 2021
5
I've got a circuit that has been plagued by RFI issues for years and I am going through the fifth version of my PCB design and I want to eliminate RFI issues once and for all.

The circuit and PCB is used for transmitting rocket telemetry and GPS data on amateur rockets that travel 10K feet to 150K feet high. I've got a PCB that is currently using a Teensy 4.1 MCU (Arm7), a UBLOX GPS, multiple I2C sensors (barometer, gyro, accelerometer), and it has over a dozen wires that connect to the board used to fire pyro separation, control cameras, sense events, etc. I also have a 1 watt LoRa radio in the 430 Mhz range transmitting using a 1/4 wave dipole antenna. Unfortunately, the radio and antenna need to be within inches of the PCB, as there is no practice place to relocate the antenna or radio. In the smallest footprint, the PCB and radio "avionics bay" are self-contained in a 3" round by 14" long tube.

My issue is that the one watt UHF packet transmissions often interfere with the 400hz I2C bus or the Teensy SPI bus used for flash memory logging. In my last round of design I isolated power for the radio and MCU, I used a star topology for my ground/returns, ferrite beads on all the IO lines, y-caps on power, and a third filled-in layer on the PCB tied to ground.

For my next version, I plan to put a complete RFI shield around the top and the bottom of the PCB components using the WE ShielDIY with the SMT clips on both sides. I also plan on connecting all IO to the shield with 470pf caps, along with 50 ohm resistors in series on each IO line.

Having a relatively high power UHF transmission so close to the board is really unfortunate, but there is no way to design around it. Let me know if you have any other suggestions, especially for filtering out external 430 Mhz noise.

Thanks!
 

LowQCab

Joined Nov 6, 2012
2,170
Use a 4-Turn Helical, Circularly-Polarized, Antenna for Transmitting and Receiving.
The Circular, Solid-Aluminum "Reflector" or "Ground-Plane",
( which doesn't work the same way that a "Vertical-Radiator" Antenna Ground-Plane works ),
almost eliminates any RF radiation from the "back-side" of the Antenna.
A Helical-Antenna with more than ~4-turns may make the Antenna too directional.
A 4-Turn Helical can simply be aimed strait up,
and will cover approx. a ~30-degree "Cone" with a very strong signal both ways.

You really don't need all that much RF-Power when You use
properly matched, extremely directional Antennas.
For your application, ~250mw should be way more than enough RF-Power, ( on either end ).

"Circular-Polarized-Antennas" generally don't play nice with other "Linear-Polarized-Antennas",
this tends to block-out outside interference from other RF-Sources,
which virtually always have Linear-Polarization,
whether they have vertical, or horizontal polarization, is almost irrelevant.

http://jcoppens.com/ant/helix/index.en.php
.
.
.
 

Thread Starter

BeachCircuit

Joined Mar 26, 2021
5
Hi @LowQCab , I have tried over a dozen different antenna designs, but my limitation is the amount of space I have combined with a low frequency requiring a larger antenna. The helical antenna would take up almost all of the 14" avionics bay and wouldn't leave room for the flight computer, radio, cameras, additional altimeters, etc. I've also tried less power (e.g., 250mw). You are correct that is all you need when the vehicle is at 100K feet and oriented correctly with good antennas on both sides, but rocketry is a hot mess when it comes to predictability. We have rockets that come in ballistic and land 5+ miles away and at 250mw you would lose them on the horizon, even with a good Yagi antenna. Orientation is also problematic with vehicles pointing in all different directions on descent. So, we've found that one watt LoRa, along with a very good Yagi is the most reliable. Sometimes we also put an antenna up 100' high at the launch site connected to a LoRa repeater and use a mesh configuration to better capture the horizon and "location in the dirt" transmissions. The repeater passes to our hand-held trackers.

@nsaspook , I am planning on using something similar called ShielDIY to create a Faraday case over the PCB. If it doesn't work I'll check out the one you suggested. Thanks!
 

MisterBill2

Joined Jan 23, 2018
12,399
I suggest trying an "on the ground" test of the transmitter to verify that it is the signal from the output and not from some intermediate stage that is the problem. Or maybe you have already verified that. One other possibility could be a power level shift, low power during ascent when the data system must work, and then shift to higher power when that portion of the flight is completed. (That concept might not apply, though.) Other that that, a simple trailing wire antenna could move the RF farther from the circuits that are affected. The wire could be very thin, perhaps #30 stainless, or possibly even like the army uses, or did use, for wire-guided rockets. Just a few feet would provide a lot of reduction.
 

Lo_volt

Joined Apr 3, 2014
268
It will help to twist a ground with each of the signals on your SPI and I2C buses. Keep in mind that you want the return path for a given signal (usually ground) to closely follow the signal. This is why it is recommended to twist power and ground or signal and ground on cabled connections. It's ok to add grounds on short cables such as yours to allow for twisting signals with grounds.

To explain, think about what is happening when a signal follows one path and the return current path follows a completely different path. This is relevant for both PC board traces as well as wired connections. If the signal and return follow two different paths, they have an open area between them where electromagnetic fields can flow. Any modulated EM field that does flow through this area will now induce current flow on the signal and it's return path. A strong enough field may induce enough current to affect and even disrupt the intentional signal. To guard against this, as stated above, the signal and its return must have the minimum possible space between them.

Additionally, twisting adds another level of protection. It's easy to do with cables, but not so with PC boards. EM fields are directional. When you twist a signal with it's return, not only have you reduced the area between the two significantly, but you've also rotated the opening where EM fields can influence your signals. At a small point along the twist the opening is in in the proper orientation but on either side, the orientation is 90 degrees out of proper and so will barely affect the signals. This greatly reduces the effect of external EM fields.

Bottom line: For cabled connections run a ground with each of your SPI or I2C signals and twist the signal and ground together. For PC board connections, make sure that your ground plane is continuous beneath a given signal trace so that the return path for your signal can closely follow the signal itself.
 
Top