I would like to ask for advice on shielding the following setup.

I have a measurement target sitting inside a Faraday cage, connected through a shielded cable to a measurement circuit. The cable shield would be connected to the Faraday cage. The circuit mainly includes a microcontroller, AD converters and a single op-amp inverting transimpedance amplifier. The circuit sits inside a metallic box connected to the cable shield.

As the instrumentation needs to be battery powered (3.3V), the transimpedance amplifier is referred to the mid-level potential of 1.65V (wrt battery negative terminal) to allow measurement of current flow in both directions. This in turn means that at zero current flow, the measurement sample is at 1.65V wrt the negative terminal of the battery powering the measurement circuit. The analog signal bandwidth to be recorded is on the order of 10Hz-500Hz.

The device would occasionally have to be connected to a desktop computer through a USB interface. Then, the USB ground has to be connected to the negative terminal of the battery, 0V for the microcontroller and ADCs, -1.65V regarding zero transimpedance output.

I have received conflicting advice on how best to shield this circuit, one idea being to connect the shield to 1.65V, coupling the shield to the zero level of the transimpedance amplifier or alternatively, connecting the shield to the battery negative terminal, zero for the digital components.

In either case, a further question is whether the USB cable shield can or should be connected to the metal enclosure of the measurement circuit.

I have read a number of other posts on shielding, mainly as this, this and this, however I have not been able to piece these together for my case. Any help pointing me in the right direction would be much appreciated.


1 Answer 1


My advice: galvanically isolate it from the computer and USB.

While you haven't really described your problem, the fact you want a faraday cage implies you need it to be quiet, and anything connected to a computer never is.

You can isolate the digital interface quite easily.

If you are using I2C, then this is available galvanic isolated. For very low noise (in wafer probing RF testers), it has been used with a fibre connection

You will see that is isolates the serial (uart) interface to the USB (FT232). The coupler IC's are very easy to use. The only vice is they have capacitance across the barrier, and they also use UHF/Microwave RF signals. Neither are so much of an issue at your frequencies.

The TOSLINK fibre connection is great for very low noise: no metal enters the faraday shield, no cables come close, No RF is used, there is no capacitance. (and they are cheap and easy to get).

It is easy to feed uart over toslink - just hook the modules to your FTDI usb-serial board. The only drawback, is using a two wire serial (RX/TX only), so if you need handshaking, it has to be xon/xoff

[TOSLINK is the plastic optic fibre used by common audio equipment]

  • \$\begingroup\$ Galavnic isolation seems like a good idea I haven't thought of, thank you. We are currently using a USB to UART converter as part of the circuit. What would you think of using an USB-USB isolator chip on board, such as this? In this case the usb on the computer side would not be connected to the device shield, if I understand correctly. Do you have any tips regarding which voltage level to connect the shield to (analog zero or battery zero)? \$\endgroup\$
    – rob
    Apr 24, 2018 at 11:56
  • \$\begingroup\$ You can connect the USB shield, to mains earth if you like. Laptops etc often float at Vmains/2. Connecting the faraday cage to mains earth is optional. What you have to do is NOT have a path for current to flow from the computer, through the cage, to somewhere else. Galvanic isolation breaks that. If you are going to be doing serious low noise work quite a bit, I would go for TOSLINK fibre. It is perfectly easy and cheap, and you can have absolute confidence there is no sneaky capacitive copling or contact between cables/shields etc. \$\endgroup\$
    – Henry Crun
    Apr 24, 2018 at 22:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.