Teensy 4.1 UHF RF spray

A

aberoev

New member
I am building a device that needs to be mounted very close to UHF microphone receivers. Unfortunately I've found the Teensy 4.1 sprays a bit of RF in the UHF spectrum, enough to interfere with the audio when WL mic signals are nearing the end of reception range. I haven't had this problem with similar Pro Micro based projects

I've tried lining my enclosure with copper tape and grounding the Teensy to that, also tried different USB cables and multiple ferrite cores but it doesn't help much/enough. I've also tried different clock speeds, and it seems to not affect this RFI. Sandwiching a thick piece of aluminum foil between the UHF receiver and the Teensy project does help, but isn't practical because the project is enclosed in a very small custom 3d printed case.

The amount of RF noise coming from the Teensy seems to be the same, whether it's soldered to my PCB project (couple of encoders) or standalone/just the board. The project is battery powered

Is there anything else you think I should try to solve this problem?

Thanks very much in advance
 
It might help to know the kind of work the Teensy does.
RFI is proportional to switching times and lengths of connecting cables, and current switched. Shorter is better. Slower is better. Lower current is better. Usually.
I am probably wrong, but I seem to remember there was me menion of changing slew rate of I/O pins. Worth looking for.
Applying ferrite cores can help, I'd suggest binocular cores which have a very high AL value for just a few turns. Clip ons and 1 turn might not be enough. Also you really should do all leads including power and ground.
That said, it's impossible to pin down without more information.
Good luck.
 
vk3txd said:
It might help to know the kind of work the Teensy does.
...
Click to expand...

Is the Teensy just in close proximity - or physically connected to the UHF receiver units?
Battery powered - not actively USB connected?
Is the battery voltage stepped to run the Teensy? Using?
 
Thank you for reading this, to answer your questions:

- The Teensy is connected with a PCB to some rotary encoders for acting as a midi controller for a audio field recorder

- The Teensy is in close proximity, not connected to the UHF receiver. The whole rig is battery powered from the same battery. Battery voltage is stepped down from 12 v via the field recorder No difference when changing or isolating power sources.

- The RFI doesn't change much whether a Teensy is connected to my PCB or not, or even programmed. There are no wires dangling about, see photo https://www.dropbox.com/s/x6vwn25ol1luby1/IMG_1552.HEIC?dl=0

- Changing the clock speed from 600 Mhz to 528 or even 24 MHz does seem to lower the RF spray by a very light amount

- Obviously mic transmitters are generally going to overpower the noise floor by a lot, but my goal was to eliminate this RF noise completely to allow maximum range on the UHF setup

Thank you very much for any ideas
 
Just for the sake of testing, can you try powering Teensy from 3 AA batteries? And remove the field recorder and any other gear. The idea is to confirm as much as possible that the RF noise really is coming from Teensy, rather than something else like the step-down converter from 12V.

Especially if you're seeing the problem at nearly the same level when Teensy runs at only 24 MHz, there's a strong chance something else is emitting the noise. Switching power supplies in particular can be very noisy.
 
shielding materials

Copper will make an electrostatic shield at all frequencies and a magnetic shield at higher frequencies. For an electromagnetic shield at lower frequencies you need a different metal, something ferrous. For a wide range of frequencies a common shielding material is Mu Metal: https://www.magnetic-shield.com/products/mumetal-brand-products/ and it is available even on Amazon (assuming that's really what you get). Tech data https://www.magnetic-shield.com/mumetal-technical-data/ and they have other alloys for higher intensity fields. MuMetal will provide shielding of both electric and magnetic fields. Here's a discussion on common shielding material tradeoffs: https://leadertechinc.com/blog/the-...ng-metals-and-what-you-should-know-about-them

On a past product in a custom molded plastic case, the case was spraying with a high-nickel shield. You can get something like that in a spray can ($60!): https://www.elexp.com/products/25841-340gsuper-shield-conductive-coat

If you have a wideband radio receiver that can give you some idea of the specific frequencies you are trying to block.

If you know the radiating area of your board, you might be able to apply some PCB techniques such as running signals on inner layers with outer layers power, guard traces, etc. But if it is the Teensy module radiating, then your PCB can't fix that. If the culprit is traces running from Teensy you can try some slew rate limiting techniques and/or impedance matching and termination. A scope will show if you have high frequency components on signals leaving the Teensy and going elsewhere.

But wait - I just re-read your update: "The RFI doesn't change much whether a Teensy is connected to my PCB or not, or even programmed. " - is this with the Teensy powered and just sitting idle (whatever that really means if it has no code to execute), or not even powered? I'm not clear what the "Teensy not connected" means - installed? Powered? Something else? In your photo it looks soldered on so can't be simply unplugged. Can you not power the Teensy, and the RF is still there?

Let us know if/how you get this solved! Best regards...
 
I might have concluded things too fast and have not been clear, I apologize for that. English is not my mother tongue, so some things are hard for me to explain clearly. I've spent the evening evaluating a proper real time UHF (500-700 mhz) scan, and have found that:

- The field recorder USB power supply (which is also the midi host) is not introducing any significant increase in RF noise compared to powering the Teensy 4.1 with 3 AA batteries

- A brand new Teensy 4.1 out of the box, not yet programmed, powered via 3 AA batteries emits a very light amount of RF noise

- Introducing code @ 600 MHz clock speed makes no difference to the bare Teensy. Running code @ 24 MHz clock speed lowers that noise slightly

- Introducing my PCB increases the noise significantly compared to the "bare Teensy", lowering to 24 MHz lowers the noise slightly

Again, these RF noise levels are likely not going to be a problem for anyone, since the mic transmitters generally will overpower by a huge amount, I'd just love to learn how to make a project like this completely RF silent, as those extra inches of UHF reception range can make a difference for me.

I am going to look for this MuMetal material, I am not sure if I can source it (small country problems), perhaps it goes by other names?
 
aberoev said:
- Introducing my PCB increases the noise significantly compared to the "bare Teensy"
Click to expand...

Maybe if we could see the PCB and know enough about its design, better advice could become possible.

But without seeing anything specific, there are some general rule of thumb ideas about PCB design to lower RF noise. Here's a few I can think of right now. I'm sure others will have more...

An unbroken ground plane (no traces routed, no large gaps) can usually help. Typically this is only possible with 4+ layers. Ground fill / pour on 2 layer boards is rarely very helpful, and in some cases it can even make the problems worse if large sections of ground are only connected in limited ways which can allow them to act as a big antenna.

Resistors in series with high speed signals, like SPI clock, can reduce noise and give a better overall signal quality. Usually resistors between 33 to 220 ohms are used. The OctoWS2811 library page has some info about the need for these resistors when sending LED control signals down long wires. Scroll down to "Signal Quality".

Power lines can act act as an antenna. Decoupling capacitor very close to the high frequency loads helps, but often the problem is unavoidable because rapidly changing current has to flow through the wire, and any loop area between the power line and the ground return creates a magnetic field as the current changes. This problem mostly happens with 2 layer boards, especially if the power line has a serpentine routing which is common when the board has limited space for routing signals. Dedicating a plane inside a 4+ layer board greatly reduces this problem, because every trace with rapidly changing current has a ground return path directly underneath, minimizing the loop area which can create a magnetic field. If you already have a 2 layer design, there's really no easy way to try this quickly. But these days 4 layer PCBs are not very expensive. While it's more than 2 layers, it's still a *lot* cheaper than adding special metal shields or other post-design RF reduction methods.

Adding a ferrite bead can also RF block the noise, especially useful on power supply lines. Where to add these varies by design. On Teensy 4.x, you can see 2 of them are in the circuit. One is at the power source, which is meant to keep noise possibly caused by the high frequency changes in Teensy's power consumption (which capacitors should mostly supply) from escaping back up the USB cable, which could act like an antenna if not a good quality shielded cable. The other is at the point where power is consumed by the DC-DC converter for the CPU. It steps 3.3V down to 1.15V or 1.25V depending on CPU speed. The ferrite bead is meant to keep the high frequency pulsing current consumed at the DC-DC input from impacting the 3.3V power which spreads to the rest of the PCB, and could be connected by wires to external circuitry in many applications. Where to put these ferrite beads requires some intuition and usually experimentation, but it's usually pretty simple to severe a power wire or PCB trace and solder a ferrite bead across the gap for a quick test. Don't forget to put a ceramic decoupling capacitor on the other size, if the design doesn't already have it by the place you cut.

Whether any of these apply to your circuit boards design, I can't know, but hopefully this generic info helps.
 
Last edited:
Thank you very much for your replies to this issue I am having, very helpful. However the "bare" Teensy still emits an amount of UFH noise that is a bit too much for this application (close proximity to UHF mic receivers). Could you tell me if the Teensy 4.0 would be any different? This midi controller project I am making is dependant on the Teensy capability of multiple midi cables, so I would really love to be able to use one for this project.
 
aberoev said:
...

- A brand new Teensy 4.1 out of the box, not yet programmed, powered via 3 AA batteries emits a very light amount of RF noise

- Introducing code @ 600 MHz clock speed makes no difference to the bare Teensy. Running code @ 24 MHz clock speed lowers that noise slightly

- Introducing my PCB increases the noise significantly compared to the "bare Teensy", lowering to 24 MHz lowers the noise slightly
...
Click to expand...

The out of the box Teens is running Blink code - assuming at default 600 MHz. Adding alternate code at 600 MHz not making any difference doesn't suggest the Teensy itself is the problem ... the next line says the PCB in use "increases the noise significantly" - so components on it or the metal pathways it provides are the creator/amplifier of the noise.

aberoev said:
Thank you very much for your replies to this issue I am having, very helpful. However the "bare" Teensy still emits an amount of UFH noise that is a bit too much for this application (close proximity to UHF mic receivers). Could you tell me if the Teensy 4.0 would be any different? This midi controller project I am making is dependant on the Teensy capability of multiple midi cables, so I would really love to be able to use one for this project.
Click to expand...

Is the "emits a very light amount of RF noise" an improvement using the 3 AA batteries and then unacceptable and making 'more noise' when powered otherwise (directly by USB or the power supply at hand)?

Given the commonality of the 1062 MCU (though smaller package) and support parts and layered design of the Teensy PCB it would seem a valid assumption when placed on the same PCB the same noise will appear. p#9 calls out a set of notes that could be behind the PCB's creation/amplification of the noise observed. It may be the T_4.0 part placement on the inch smaller board might change the resulting noise - it could be better or worse depending on what element is behind the emission.
 
For future reference you cannot expect to co-locate sensitive UHF receivers in the same box as a high frequency digital processor unless stringent RF design techniques are used. 600MHz for instance will probably necessitate LC filtering on all leads from the processor (after all 600MHz is slap-bang in the UHF band) - you should be using a Faraday cage at UHF, a bare plastic enclosure is not a viable option.

For layout of a motherboard an understanding of RF layout can avoid making accidental antennas from the PCB traces, especially by ensuring groundplane everywhere - 4 layer or more is really really useful for this.

One issue with small microcontroller modules like Teensy's is a lack of good grounding support - ideally every 2nd or 3rd pin would be ground to create a ground network that follows the signal paths down to a motherboard - otherwise the entire module can act as a patch antenna. One possible solution is to place the whole assembly in a Faraday cage with RF-absorbing card. Then the radiated energy get absorbed before it can leave through any leads/cables. Or you can place a grounded box over the module and solder it down to the motherboard at regular intervals along the perimeter, creating a Faraday cage just for the module.
 
You must log in or register to reply here.
Back
Top