AMC1302: Need help to evaluate the schematic and layout for LCL filter current sensing with AMC1302

Hi Jiri,

Thanks for the prompt response.

1. You're right about the shunt resistor, I think I can increase the shunt to 0.4mOhm or 0.5mOhm total, or probably even better to use 4 terminal. Anyway, that will be done for the final layout here.

2. On the isolated power, yes we are aware of that... we already do it on the actual internal schematic.

3. Can you please help to evaluate the layout for 3 phase down below? due to safety isolation constraint, we must place the AMC farther from the shunt especially for the current sense on the right side.

Thanks,
Arief,

Hello Arief,

I was also thinking about suggesting a 4-wire shunt but quite frankly, I am not sure if you would benefit from it given the fact that there are three of them in parallel.

The layout looks reasonable to me. Still, please consider a 1uF cap in parallel with C20, C19, etc.

As we speak about longer traces - In this situation we also suggest placing a ferrite beads (402 size) in series with the inputs, eventually one for the ground:

I would start with a 0402 0-ohm resistor in such location and thus having the PCB prepared for ferrite beads in case you find you need them during testing.

Best regards, Jiri

Thanks Jiri,

Sure I will update by adding 1uF with C19 and C20,

Originally I was thinking of simply use ground plane between the AMC and the shunt res.
Anyway, why would you recommend to add ferrite in series an also at the ground connection? and I also started to think if it also means the ground plan should not be used or how to layout the ground (VSS1,VSS2,VSS3) connections?

After your next reply, I think I can close this question

Thanks

Well, I must admit that some recommendations come from the intuition or experimental results. In some AMCxxxx device datasheets or EVMs you may see ferrite beads. They often improve EMC behavior as the modulator sampling (the input of the AMC device) runs at high frequency, e.g. 20 MHz.

Additional benefit is that you somehow "weakly" couple the device to the source of fast transients. For example, in-phase shunt current sensing circuitry for motor drives jumps up and down on high voltage potential. These jumps are rich in harmonics. For this reason, ferrite beads damp some of the higher-frequency content.

The problem is that it is always difficult to predict the behavior and it typically takes some experiments with the system to get it right.

Some systems might require it, some not. 

Do the three high-voltage traces "jump" up and down with the HV potential?

If so, this area (green) will couple a lot of noise from the second "phase".

When I am thinking (aloud) now, I would rather see three small traces going there:

Best regards, Jiri