The picture below is for a part of larger circuitry, but it's the part where EMC noise is coming from and causing the conducted emission test to fail.
Any hints on when to start to deal with this circuit to limit the noise?
7
-
\$\begingroup\$ What did you use to isolate the root cause of the disturbance to that part? The datasheet for some AD parts in this series do recommend a noise suppression capacitor from primary to secondary (Y-capacitor in some cases). Does the evaluation board for yours use one? \$\endgroup\$winny– winny2025-07-17 10:36:04 +00:00Commented Jul 17 at 10:36
-
\$\begingroup\$ When IC11 (ADUM1201) is taken out, the noise drops to acceptable levels. I have no evaluation board; I am testing the whole product for EMC in a conducted emission test setup. \$\endgroup\$Elektroneu– Elektroneu2025-07-17 10:39:48 +00:00Commented Jul 17 at 10:39
-
\$\begingroup\$ Good. Same series, different device: analog.com/media/en/technical-documentation/data-sheets/… Flip to page 17. Two Y-caps in series for EMI. \$\endgroup\$winny– winny2025-07-17 10:59:32 +00:00Commented Jul 17 at 10:59
-
\$\begingroup\$ Don't they mean C2 and C3 (not C4) across the isolation boundary? \$\endgroup\$Elektroneu– Elektroneu2025-07-17 11:05:28 +00:00Commented Jul 17 at 11:05
-
\$\begingroup\$ Good catch. Indeed. \$\endgroup\$winny– winny2025-07-17 11:07:49 +00:00Commented Jul 17 at 11:07
|
Show 2 more comments
1 Answer
\$\begingroup\$
\$\endgroup\$
1
Any hints on when to start to deal with this circuit to limit the noise?
Just after the common-mode chokes, you'll need shunt capacitors, for the common-mode currents, to the chassis on each wire, because the chokes won't attenuate them completely.
In conducted-mode EMC (150kHz to 30MHz) or radiated-mode (30MHz to 1GHz or more), the impedance of the shunt capacitors must be fairly low, otherwise the common-mode currents will flow along another path, the one with the lowest impedance. But here, your design works against you. The capacities of your TVS diodes are so low that they will have too high an impedance and the return path will not be efficient.
The equivalent capacitance of your return path is (8x22nF) in series with 12pF (TVS), i.e. approximately 12pF between signal and ground. This means that for a disturbance at 30MHz, the impedance is around 442 Ω.
You can place 10nF capacitors, for example, in parallel with the TVS diodes. This value is a compromise between signal completeness (filtering effect) and the low impedance of the disturbance return path.
-
\$\begingroup\$ Thank you. This is really a great point. \$\endgroup\$Elektroneu– Elektroneu2025-07-18 04:45:57 +00:00Commented Jul 18 at 4:45