This is a bit specific system related question about simultaneous data-acquisition for several transducers. Therefore hope you don’t mind for the long intro and the question. It would be too implicit if I wouldn’t write in detail.

In an indoors environment I have been encountering issues due to using a single ended system. Basically 14 voltage inputs from various transducers were coupled to a single-ended input data-acquisition board. The transducers are all DC like outputs and here some examples data for humidity, temperature, load cell amplifier ect. Some of the transducers have been used for many years and calibrated each year at calibration institutes.

So the system was such that the transducers are single ended and the DAQ was also single ended. Since the system was single ended, all transducer grounds were grounded together and this was causing impedance coupling and ground loops. Since the cable length are around 15 meters.

Now the system will be replaced such that this new data-acquisition board will be used with these isolated differential inputs input modules. So this new board will have isolated voltage input modules which will provide such main benefits: differential inputs, channel to channel isolation, low-pass filtering. So this seems fine for the input/daq side.

For clarity, below at Figure 1 I tried to draw the single-ended system used currently, and at Figure 2 the planned differential ended input system. In the diagrams below Trans represents a voltage transducer, coax cable the typical 50 Ohm coaxial BNC cable, and m depicts an isolated differential input module.

enter image description here

Here is my question:

My confusion here is that the transducer outputs(as you see in the datasheets) are not balanced and I don’t know how big are their output impedance. And since they are calibrated and almost all the transducers used are single ended outputs I’m stuck with the issue of not-entirely balanced system for this specific system.

If I use single-ended to diff-ended converters at the transducer outputs this might balance the system. But besides being costly to do it for each channel, my fear is to introduce offset or electronics noise by using such extra converters in the signal chain. And I checked such 0-10V transducers in the market such as temperature, humidity, barometric pressure transducers or load cell amplifiers in the market and they are almost always single ended output. It means their outputs are not balanced and the output impedance is most of time not given.

So I have three options:

  1. Not using any single-ended to diff-ended converters

  2. Using converters for each channel where I have no idea if I would introduce other problems

  3. Throwing out all the transducers and try to find new balanced output transducers which would be extremely difficult for each new device.

How should I approach this problem and do I really worry a lot about this? Does anybody have practical experience with such scenario?

Edit regarding a comment:

Here is my concern:

enter image description here

Will the module prevent the issue due to the source output impedance above?


***Will the isolation by the modules cause the input signal floating with respect to AGND of the DAQ? I thought in that case it is equivalent to buffering.

  • \$\begingroup\$ How about option 4: Pretend your single ended sources are differential, and connect them to your input modules as is. The input modules are galvanically isolated - no ground loops possible. \$\endgroup\$
    – JRE
    Commented Jan 7, 2019 at 12:06
  • \$\begingroup\$ @JRE Thank you for your comment. Please see my edit regarding your comment. I tried to express my concern more explicitly. What do you think? \$\endgroup\$
    – user1245
    Commented Jan 7, 2019 at 12:36
  • \$\begingroup\$ The input modules have an impedance of 500k (minimum.) So, your sensors set the line impedance. \$\endgroup\$
    – JRE
    Commented Jan 7, 2019 at 12:40
  • \$\begingroup\$ And, I don't know enough about the subject to make authorative statements - which is why I only made a suggestion in a comment. \$\endgroup\$
    – JRE
    Commented Jan 7, 2019 at 12:41
  • \$\begingroup\$ Bring a 117AC line cord near the cables from your sensors, so the electric field of the HOT wire gets coupled into your imbalanced sensor wires. Then examine the digitized code-spread, with varying distances between the power cord and the sensor wires. If you see the code-spread changing, then you have high-output-impedance sensors. \$\endgroup\$ Commented Jan 7, 2019 at 18:48

1 Answer 1


The following applies to signal transmission where the operating frequencies of the signal have a wavelength that is much, much greater than the cable length.

A "balanced" output means that an output presents an impedance from each terminal to "earth" that is equal. This means that any EMI hitting the interconnecting cable spreads interference equally to both conductors. It then follows that a differential amplifier can eradicate this "common mode" noise interference by subtraction. But, of course, the input differential amplifier must have a balanced input impedance too; however, given that input impedances are usually very much higher than output impedances, this is less of a problem.

A "differential" output brings to the party reduced emitted interference from the cable providing that the system is balanced (impedance balanced). Because the signal amplitude is also double what an equivalent single-ended output has, there is usually a 2:1 improvement in signal-to-noise ratio.

So, in your final diagram, ideally you would have a 100 ohm resistor duplicated in the negative lead feeding the differential input. That will impedance-balance the cable and it means that any external EMI hitting the cable is spread equally on both conductors.

See this Q&A for further details. This Q&A is also relevant.

  • \$\begingroup\$ I understand but all these type of transducers are singe ended (and unbalanced). None of them used here or manufactured in the market are balanced. If I had access to an example system would be great but I dont have such opportunity to see what they use. So I'm stuck with deciding should I convert each input to diff ended or just consider the transducer output impedance negligible. You mention adding 100 Ohm in negative lead but these transducers I linked none of their output impedance given. Im just guessing they are around 100 Ohm just a guess. \$\endgroup\$
    – user1245
    Commented Jan 7, 2019 at 13:16
  • \$\begingroup\$ Then you need to find out what your device is or measure it or add a controlled impedance in both wires close to the transducer that effectively overrides the transducer's own impedance. \$\endgroup\$
    – Andy aka
    Commented Jan 7, 2019 at 13:22
  • \$\begingroup\$ I see like adding 1k to each line(signal and negative lead) ? Makes sense as long as they dont introduce offset error due to voltage divider effect. These isolation modules have 100k input impedance. Maybe I should just leave with it and neglect the output imbalance. My main concern was 50Hz ground loops in SE system. Im really disappointed why manufacturers dont produce balanced transducers btw. \$\endgroup\$
    – user1245
    Commented Jan 7, 2019 at 13:28
  • \$\begingroup\$ You mean differential transducers not balanced transducers yes? \$\endgroup\$
    – Andy aka
    Commented Jan 7, 2019 at 13:41
  • \$\begingroup\$ There are such solutions analog.com/en/products/lt6350.html#product-overview which would also provide impedance balancing for each line but they require very clean power supply and in my case it will be so hard to install each time for every different transducer. Im talking about tens of transducers coupled to 14 channels different times. It is really pain to deal with that. Maybe the resistor method is better. If Im lucky maybe I dont even need these. \$\endgroup\$
    – user1245
    Commented Jan 7, 2019 at 13:47

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.