2026-06-06
The Fl0w3r-Badge is a nice little gadget from the Chaos Communication Camp 2023. It consists of two flower shaped PCBs with a round RGB OLED-Display in the middle, provides several capacitive touch buttons, speakers, an audio input and output as well as the mandatory RGB LEDs.
At Gulaschprogrammiernacht 2026, Sarah [1] and I held a Talk about EMC testing methods where we used the results we got from several EMC measurements with the Fl0w3r.
If you are just looking for the slides from that Talk: Here you go.
Here are some hardware specs taken from the original docs:
So how to determine, what standards and limits should be applied to evaluate the EMC compliance of that nice little thing? Based on the usecase, this little Flow3r is likely classified as a multimedia device and with that information, we can take a brief look at CISPR 32 and CISPR 35 for limits and test procedures.
I set the limit for me not to do anything really harmful to this device - since this particular one is just borrowed from a friend of mine, I will omit all tests that will probably do permanent damage to the hardware. That includes specifically ESD, Burst and Surge. So I narrow all tests down to conducted emmission, conducted immunity, radiated emmission and immunity as well as some fun with near field probes.
While the criterea as well as the limits for all emission tests are defined in the mentioned standards, it is more difficult to determine when a Equipment under Test (EUT) fails a susceptibility test.
Some behaviour of our EUT might be easier to classify as ‘failed’. For example if the EUT turns off and never goes on again or it goes nuts while under test and starts to kill the crew.
But is our little Fl0w3r failing a susceptibility test, when the display just flickers from time to time?
Well, this depends on the device and its purpose. While it is negligible for our Fl0w3r if its display flickers, the same behaviour would considered inadmissible when it comes to the test of medical equipment.
To start with something easy to handle, I had a look at the conducted emmission characteristics of this cute little thingy. From Sarah I got a nice little USB-C Power Delivery LISN. Using this, I investigated a few configurations, starting with the Flow3r Badge beeing charged by a common USB-C wall charger. The control of the measurement process was realized with the help of a Python Measurement software [2], developed at TU Dresden’s Chair of Electromagnetic Theory and Compatibility.
Turns out, the Fl0w3r’s conducted emission from 9 kHz to 30 MHz remain under the limit.
Since the LISN’s frequency range extends up to 200 MHz it might be a good idea to have a look at this extended frequency range, despite the fact, that the limits for conducted emission are just defined in a frequency range up to 30 MHz.
With the help of a Rhode & Schwarz ZVL-6 Spectrum Analyzer, I was able to get a spectrum of the conducted emission in a reasonable frequency range from 9 kHz to 1 GHz.
Since the limit line of CISPR 32 is just defined in a frequency range up to 30 MHz, there is no acutal limit for conducted emission above.
While we were at conducted measurements, we can also investigate conducted immunity of our little Fl0w3r. This is done in a frequency range from 9 kHz to 80 Mhz - since our measurement setup is not enabled for a higher frequency range.
Initially this setup needs to be calibrated. To achieve this, a defined metal rod is placed inside the clamp. One end of this calibration rod is terminated with a 150 Ohm resistor to ground, while the end located at the EUT side of the clamp is connected with an 100 Ohm resistor in series to a powermeter included in our measurement equipment. With this setup our measurement software is capable to calibrate the coupled field strength to a desired value over the chosen frequency range.
When putting the Fl0w3r insulated 10 cm above the grounded tabletop, it withstands the coupled 10 V/m field strength over the whole frequency range. If it is just insulated from the groundplane by a piece of paper, the field coupled into the Fl0w3r is significantly increased to the initially calibrated field strength which leads eventually in a bootlooping Fl0w3r when the testing frequency is between 30 and 40 MHz.
Radiated emission measurements are usually performed at an Open Area Test Site (OATS) or an Anechoic Chamber. Suitable places for OATS are limited and the particular one at TU Dresden was completely unsuitable - One does not simply build an OATS right in the middle of an university campus next to the high voltage laboratory and with hundreds of students walking with their mobile phones next to your measurement setup. Since Anechoic Chambers are space consuming and expensive, we have to use a more compact test environment - A GTEM Cell.
A GTEM Cell consists of a pyramidical housing with a piece of sheetmetal, called septum, placed inside. The septum is electrically terminated with a resistor network while waves at higher frequencies are terminated by the RF-Absorbers placed at the rear wall. Since this is a special form of a TEM Cell wave propagation inside the working volume is approximately equivalent to wave propagation in free space. (When deficiencies such as higher modes leading to longitudinal field components are neglected.)
With that, a GTEM Cell is an ideal compact and cost effective environment for radiated measurements from DC to several GHz.
For measureing the radiated emission of a device under test, one has to measure it in at least three orientations, since the GTEM Cell can only detect electromagnetic waves moving towards the GTEM Cell’s apex. To have results obtained in a GTEM Cell comparable with results measured at an OATS or a Anechoic Chamber, a specific correlation algorithm is used. The factors into this corellation algorithm are beside the measurement results already mentioned, the geometry of the GTEM Cell and the position of the device under test in it.
In the obtained measurement results are shown, that there are several peaks above the limit. It is noticeable, that specificially at 80 MHZ, 400 MHz and around 1 GHz are several emissions that exeed the limit in CISPR 32 by a by a lot.
When connecting a power supply with a suitable USB-C cable to the Fl0w3r, the measured emissions are a plenty larger than in the battery powered case as the connected USB-C cable acts as an antenna.
As the Fl0w3r is already placed in our GTEM Cell, we can also examine its radiated immunity without much effort. To achieve this, we just need to put a Field Probe in place and switch our RF Switching Matrix to immunity testing using a Signal Generator and an RF-Amplifier. Our Python based measurement software [3] then levels our field strength at every frequency point to our desired field strength of 10 V/m in a frequency range from 30 MHz to 1 GHz. This is not yet compliant with a standard, but it has shown that this might be a suitable method to enhance speed and precision of susceptibility test procedures in GTEM Cells. There is more to follow up to this particular topic, I guess.
Turns out, the Fl0w3r is not susceptible in the described frequency range and field strength. To estimate the interference threshold we have to use a test method that is capable of higher field strengths in our desired frequency range.
A Reverberation Chamber is quite the opposite to an Anechoic Chamber, as there are no absorbers attached to the metallic walls. With this, it can be described as a Cavity Resonator where propagating electromagnetic waves are reflected at the walls. With constructive and destructive interference these electromagnetic waves will form field strength maxima and minima. Following this principle a Reverberation Chamber has a minimum working frequency which is in case of the one at TU Dresden’s Chair of Electromagnetic Theory and Compatibility around 200 MHz.
To shift the position of these interferences a Mode Stirrer is placed inside the chamber. When this Stirrer rotates it causes an other relection path of some waves propagating, leading to other interference patterns in the chamber’s working volume. In a well stirred reverberation chamber a statistically uniform field is desirable but achieving this is not as simple as it might be, since the mode stirrer is only capable to generate limited statilsticially independent field distributions.
The maximum field strength in a Reverberation Chamber can increase to several hundered Volts per metre. Thinking of a microwave oven is likely as they are working in a quite similiar way.
Putting our Fl0w3r inside the chamber might be a bad idea since high field strengths might cause actual damage to it, but I gave it a try. Since testing in a reverberation chamber is quite time consuming, I just had a look at some frequencies, where the Fl0w3r already had strong emissions. The following video shows the Fl0w3rs reaction to a increasing field strength at 400 MHz.
Surprisingly the Fl0w3r is capable to withstand fieldstrengths up to 200 V/m, which is extremly high for a device that is not constructed to work in an area of high field strengths.
Other frequencies I examined were around 1000 MHz where the Fl0w3r as well withstood about 200 V/m. Sometimes several effects like a self operating menu were observed.
When having a brief look at the circuit and the associated layout, one may wonder, why a ground plane is missing between the layout part which contains the actual circuit and the part used to mount the USB-C connector.
The circuit around the USB-C connector looks quite common, nothing special to see here.
Despite the overall layout of the Fl0w3r looks great, this particular part might have a significant influence to the overall emissions. The lack of a proper ground plane in this part forces the current on it’s way to the USB-C connector on a unfavorable path which eventually forms a small transmission loop antenna.
But this only explains the emissions increasing intensity when the Fl0w3r is charging. So where is the origin of the emissions?
Sarah had a closer look to the measured wired emissions with her Spectrum Analyzer.
One may notice that the marked peaks in the measured spectrum are evenly spaced with 80 MHz distance. Grepping for ‘80 MHz’ in the ESP32 Datasheet [4] yields just 4 results, making it easier to narrow down the source of the observed emissions. Beside the maximum SPI Clock, the SD/SDIO/MMC Host Controller works likewise at a clock of 80 MHz. Since the SPI Clock is distributed over the whole board, there enough opportunities to radiate the clock signal as well as its harmonics.
Maybe this documentation of these rather quick and dirty measurements may help to demystify some common EMC measurement procederes. Even tough most of the shown measurements were not fully compliant to EMC standards they may give an idea which measurements are common in EMC measurement.
In Conclusion the Fl0w3r now is fed of all those tortures and now happily wears its tinfoil hat.
[2] H. G. Krauthäuser, “Conducted Emission”