1
\$\begingroup\$

General question

When measuring the weak amplitute low-frequency (DC) current signal (in 0.1 - 1 uA) transmitted through FFC cable and connectors to a PCB with transimpedance amplifier (TIA) and ADC, if the total cable length is fixed, would having more FFC connectors introduce more noise to the signal?

Detailed description

I am designing a sensor array that utilize many photodiodes (reverse-biased) to measure the light intensity from different angles. The signal is weak (0.1-1 uA) and of low frequency (DC). Currently, each photodiode board is connected to the signal acquisition PCB with TIA and ADC circuits through FFC cable and connectors.

Now, to modularize the design and minimize the footprint, I want to merge all FFC cables from each photodiode board to an intermediate PCB and use a larger FFC cable to feed the bundled signals to the signal acquisition PCB.

In the scenario of weak voltage signal, I think more connectors means higher impedance and possible loss of accuracy of the analog voltage signal. In my case, would more connectors cause signal integrity issue for the analog current signal? Or is there any factor I should include in my design when choosing the connectors?

Thank you,

\$\endgroup\$
11
  • \$\begingroup\$ It's bad practice to feed raw photodiode signals over wire, and when you must, you use a coax. You really want a TIA preamp next to each photodiode. Muxing raw photodiode signals also has problems but some do it with careful design, but combined with raw signals wire, you are not going to like the results. Set up a circuit and with a wire and record the result, then direct mount the photodiode and look at the result. \$\endgroup\$
    – DKNguyen
    Commented Apr 28, 2023 at 14:53
  • \$\begingroup\$ Long paths between photodiode and TIA add capacitance. You can compensate for capacitance at the TIA's feedback resistor, especially if your interest is in low-frequency (DC). But if path lengths between each diode and TIA vary significantly, frequency response may also vary from diode-to-diode. Are your FFC cables shielded? If so, capacitance is likely higher. Are both diode leads routed by cable, or only one (the other being DC biased)? \$\endgroup\$
    – glen_geek
    Commented Apr 28, 2023 at 14:57
  • \$\begingroup\$ @DKNguyen I thought the COAX is for transmitting the high-frequency signals? Would the low-frequency signal also require COAX? I forgot to mention that the FFC cable's length was 200mm, and I actually got pretty satisfying result when putting the photodiode and TIA on different boards and connect them with the FFC. I am not sure if introducing more connectors would lead to worse signal quality. \$\endgroup\$
    – Yuanyi Wu
    Commented Apr 28, 2023 at 15:06
  • \$\begingroup\$ @YuanyiWu Coax is for noise sensitivity in this case. If you've already tested are are satisfied just add more connectors and test again. In my opinion, if you already have a board for each photodiode then you already have a carrier for a small preamp. \$\endgroup\$
    – DKNguyen
    Commented Apr 28, 2023 at 15:06
  • \$\begingroup\$ @glen_geek The FFC cable is not shield. Also the cable lengths from each photodiode to the intermediate PCB do not vary too much so I assume the capacitance should be more or less the same. Both diode leads are routed by the FFC cable. \$\endgroup\$
    – Yuanyi Wu
    Commented Apr 28, 2023 at 15:10

1 Answer 1

Reset to default
1
\$\begingroup\$

When measuring the weak amplitute low-frequency (DC) current signal (in 0.1 - 1 uA) transmitted through FFC cable and connectors to a PCB with transimpedance amplifier (TIA) and ADC, if the total cable length is fixed, would having more FFC connectors introduce more noise to the signal?

I have built cables for nV applications that were ~14in long FFC so there are similar things to consider.

It's not really the number of connectors, but how the grounds are connected. The most important thing is to consider ground loops and how the shielding is done on the FFC (FFC outer layers can function as a shield).

These are the things I would consider:

  1. Shielding - Can you get magnetic or electric fields in the environment that would generate 0.1uA in the FFC, if so then you'll need a shield for electric fields. For magnetic fields, the best thing to do is consider the loop area of the PD sensor conductors to the PD and minimise that. If there is chassis shielding FFC shielding might not be as important.

  2. Leakage, most low level TIA have voltage guards that can be extended out to the FFC. The material thickness of the FFC (usually kapton which has a high impedance volume resistivity of ~1.0 x 10^17 Ω•cm ) You can estimate the resistance and include it in spice models to see the effects.

  3. Grounding and shielding. The shields should not be tied together if they create loops in the cable. Grounds should also not be tied together if they create large loops as magnetic fields could couple currents onto the shields, on long runs of FFC these currents can couple into the PD signals if they are running next to the PD signals through mutual inductance (similar to a transformer).

  4. Connectors, If you require guarding signals, I would probably select a connector that you could run the guard signals and mostly surround the PD signals. There are also ways to directly solder FFC with guards to a PCB.

\$\endgroup\$
2
  • \$\begingroup\$ For extremely low currents, you may want to consider a driven guard instead of or in addition to a ground shield, too. That effectively eliminates both leakage and cable capacitance from the equation (not entirely, since we're talking about traces on flat-flex, not well-shielded triaxial cable, but still significantly). The driven shield should be uncoated (i.e. no soldermask) on the PCB and flat-flex if at all possible. \$\endgroup\$
    – Hearth
    Commented Apr 28, 2023 at 18:23
  • \$\begingroup\$ With a kapton cable you would have a four-layer cable with a shield or guard completely surrounding the signal on internal layer \$\endgroup\$
    – Voltage Spike ♦
    Commented Apr 28, 2023 at 18:48

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.