0
\$\begingroup\$

I am trying to use a current source for applying current pulses (0-20mA and 10-20ms width) to a resistive load inside an MRI scanner room. Pulses are applied when an optic trigger is detected by the MCU. The MCU also controls the selection of the source/sink channels of the multiplexer. A diagram is given below.

The RF signal of the MR system appears at the output of the optic to voltage converter, and thus MCU receives false triggers. This issue was solved by disabling the interrupt of the MCU for a while after the first trigger pulse is detected.

The RF noise appears also at the address selection lines and at the current channels. I tried using Faraday cage by putting the entire circuit inside an aluminum box, using coaxial cables for current channels and grounding their shielding at the near endings. However, RF signal still observed.

Any suggestions to eliminate this RF signal? Do LP filters will help here? if yes, what are your recommendations for the filter design?

Notes: 1- Current channels are passing to the MRI room through a hole in the wall.

2- The output of the optic2voltage circuit is (0-3V).

3- All the circuit components are fed by a single power supply (+15/-15v).

enter image description here

A view of the detected RF signal between the terminals of a current channel is shown below

enter image description here

| improve this question | | | | |
\$\endgroup\$
  • 2
    \$\begingroup\$ You are aware that the "M" in MRI stands for "magnetic", right? What are you doing to minimize the effects of induction on your current-loop wiring? \$\endgroup\$ – Dave Tweed Sep 5 '16 at 15:14
  • 1
    \$\begingroup\$ Going through holes in a shield, without further information, is doomed too failure. You need to do a thorough audit of where all the signal currents flow, and where their respective ground currents are. Coax cables help. Grounding the outer of each coax as it passes through shields is usually necessary. Conceptually, you want to put a Faraday shield round each signal element. Physically, that shield will usually be a box at the end, a connector, a coax, another connector, etc, etc. You need to be quite thorough in your analysis of the physical circuit. \$\endgroup\$ – Neil_UK Sep 5 '16 at 15:53
  • 1
    \$\begingroup\$ What's up with that "doctored" photograph anyway? What exactly are you trying to show us? \$\endgroup\$ – Dave Tweed Sep 5 '16 at 21:40
  • \$\begingroup\$ @DaveTweed, I think that using shielded coax cables minimizes the indcution. right? The photo is taken from the oscilloscope, showing the RF signal and its peak-peak amplitude (4v pp). \$\endgroup\$ – n.na Sep 5 '16 at 22:27
  • 1
    \$\begingroup\$ OK, I'm baffled. How can you possibly "image current density distribution" with an MRI without the intense magnetic field having a strong effect on the current itself? Sounds like the wrong technology for the job. \$\endgroup\$ – Dave Tweed Sep 5 '16 at 23:31
1
\$\begingroup\$

What you're trying to deal with right now is generally called "electromagnetic immunity" of a device. Your device has to operate is a very harsh RF environment and still maintain a level of performance that it showed when operated in lab conditions. A typical quick-fix attempt would be to add ferrite beads to your cables just outside and/or just inside the shielded box. Consider them for power and signal cables alike (minus the optical fiber).

However, given the very strong field produced by the MRI, additional interference mitigation may be necessary. Be sure that each chip has sufficient power supply bypass capacitance to avoid conducting RF from the supply to the chip's I/O. Your thought of using low-pass filters on the signals is a good idea, but remember that the filtering will reduce the signal slew rate and add propagation delays. There are ferrite chip inductors specifically intended for suppressing high-frequency noise. You can even find common-mode choke chip inductors that would work for your outputs. Since your outputs are current sources, the additional DC resistance shouldn't be a big problem.

You haven't indicated what construction method your device uses. My advice is geared towards PCB designs. The stray wiring of some styles of prototyping could make managing electromagnetic immunity much more difficult.

| improve this answer | | | | |
\$\endgroup\$
  • \$\begingroup\$ Thank you for the valuable information. Do I need to put LPF for both source and sink lines? or it is enough for source lines only? Regarding the device construction, it is a PCB (MCU+Current source). I only added the MUX with stray wiring to enable using four electrodes rather than two. However, this switching works fine only in lab conditions. I thought maybe this is because the RF noise and thus I am trying to reduce it. \$\endgroup\$ – n.na Sep 6 '16 at 13:07
  • \$\begingroup\$ The stray-wire connected multiplexer may exacerbate any current EM immunity problems you're having. For the internal (shielded-side) filtering, you should put them on all of the connections, preferably symmetrically. Your goal is to block the RF energy as close to the case of the device as possible, preventing it from interfering with your other circuits. \$\endgroup\$ – user2943160 Sep 6 '16 at 22:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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