I want to measure the voltage drop across a shunt, but the ADC performing the measurement will be far from the actual shunt - perhaps five meters.

I'll need to measure voltage drops across the shunt to about 0.1 mV, so stray signals of that magnitude could be a problem.

Can I just run two pieces of 22 AWG wire from the shunt to the ADC, or would something like twisted lair help cancel out noise, and ensure the wire runs are the same length (the latter may not matter since the current flow is minimal - only enough to satisfy the ADC cap).

The environment is a camper van - so it is fairly noisy as it has various DC and AC wiring running all over the place.

  • 1
    \$\begingroup\$ Why is it not possible to locate the ADC at the shunt? That's the advantage of digital signals: put the ADC near the shunt, minimizing noise, and run the digital signals (with high noise immunity) wherever they need to go. \$\endgroup\$ – uint128_t Feb 6 '16 at 7:15
  • 4
    \$\begingroup\$ You are right to be worried about noise at that signal level. You are doing Ah monitoring, right? It should be OK to low-pass filter the signal. Put the filter near the ADC. You can use multiple stages to maximize the attenuation. \$\endgroup\$ – mkeith Feb 6 '16 at 8:22
  • 1
    \$\begingroup\$ But to address the actual question - yes, by all means use twisted pair rather than separate wires. In some cases your noise and power-line pickup may still be objectionable, and you may need shielded twisted pair. Or even coax. Or even doubly-shielded or triply-shielded coax. \$\endgroup\$ – WhatRoughBeast Feb 6 '16 at 16:33
  • \$\begingroup\$ local instrumentation amplifier --> ADC --> RS485 --> twisted pair --> remote intelligence unit \$\endgroup\$ – JonRB Feb 8 '16 at 22:46
  • \$\begingroup\$ Can you draw power from the remote circuit? If yes, you might make a little capacitance multiplier circuit with a simple transistor and connect a piece of coax to the emitter and make the shield ground. Then at the ADC end you might make an integrator circuit with an op amp with resistors to scale the voltage to match the range of the ADC. \$\endgroup\$ – squarewav Feb 8 '16 at 22:52

Common mode voltage (DC and noise on whatever it turns out to be) is likely to be your biggest problem. A shunt is a very low impedance source and is easily filtered if you don't need fast response. You don't need a twisted pair in that case, though it won't hurt.

If you filter the shunt voltage well at the ADC end and use a differential amplifier with a bipolar supply (such as an instrumentation amplifier) to amplify the voltage up into the ADC range you should be okay.

If you're planning on going directly into an ADC with an on-board PGA and unipolar supply, I think you're going to have serious problems.

  • \$\begingroup\$ I don't need a fast response - I would be sampling only a few times per second, likely. I'm planning on a using a differential amplifier such as the ADS1115. \$\endgroup\$ – BeeOnRope Feb 9 '16 at 16:34
  • \$\begingroup\$ With a bipolar supply (eg. +/-5V), I hope. \$\endgroup\$ – Spehro Pefhany Feb 9 '16 at 17:09
  • \$\begingroup\$ No, it will only have a unipolar supply - 0 to 5V. I don't follow how that matters for the ADC? \$\endgroup\$ – BeeOnRope Feb 9 '16 at 17:25
  • \$\begingroup\$ It means that your amplfier's common-mode voltage range can't extend much below ground at the point of the amplifier. Since currents flow through an automotive chassis causing differences in potential, this is, in my opinion, not a good plan. I am assuming your shunt is connected to the chassis at some point near the battery (and far from the amplifier). It might work out.. or it might not. \$\endgroup\$ – Spehro Pefhany Feb 9 '16 at 17:32
  • \$\begingroup\$ The ground the amplifier will use is directly wired from the battery negative, and not through the chassis. That same ground is also one input to the differential amplifier, with the other behind the other side of the shunt. So I don't think the chassis is involved here at all. \$\endgroup\$ – BeeOnRope Feb 9 '16 at 17:37

I would make sure you use shielded coax with a known characteristic impedance. Besides helping with noise, if you have any clock running on your circuit, you'll need to impedance match your circuit or the reflection will force you to run it really slow. (On the order of KHz) Even if you don't use a clock, the interference from the reflected signal would mess with your readings, especially since it's an analog signal.

  • \$\begingroup\$ Shielded coax seems like overkill here. I should have been clearer in my original question that the signals I want to measure (current flow from a battery) only varies slowly with time - i.e., sampling a few times a second is enough. No clocks are involved. \$\endgroup\$ – BeeOnRope Feb 9 '16 at 16:31
  • \$\begingroup\$ @BeeOnRope oh yeah, if it's just a battery, then the line will settle. I guess the noise would be your only concern then and plenty of suggestions have been given there. Good luck! \$\endgroup\$ – JosephQ Feb 9 '16 at 17:07

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.