How to Select the Right Analog Multiplexer for High‑Precision Sensor Arrays

When you’re trying to read a dozen tiny thermocouples or a cluster of strain gauges, the little chip that decides which signal gets to the ADC can make or break your whole project. A bad choice can add noise, limit speed, or even damage the sensors you spent weeks calibrating. That’s why picking the right analog multiplexer matters more than ever in today’s push for higher precision.

Know Your Signal Landscape

What’s the voltage range?

Most sensor arrays work with signals that sit in the millivolt to low‑volt range. If the multiplexer’s on‑resistance (R_on) is too high, you’ll lose a chunk of that tiny voltage before it even reaches the ADC. Look for a part with R_on well below the source impedance of your sensors—typically under 100 Ω for high‑precision work.

How fast do you need to switch?

If you’re sampling at a few hertz, any old 1 kHz part will do. But for fast imaging arrays or vibration monitoring, you may need a device that can settle in a few microseconds. Check the “settling time” spec: it tells you how long the output takes to stabilize after a channel change. A rule of thumb is to pick a part whose settling time is at least ten times faster than your sampling period.

What about leakage current?

Leakage current is the tiny amount of current that leaks through the switch even when it’s off. For high‑impedance sensors like piezoelectric accelerometers, even a few picoamps can skew the reading. Look for a multiplexer with leakage in the femtoamp range if your sensors are that sensitive.

Match the Architecture to Your Design

Single‑ended vs. differential

Single‑ended multiplexers route one signal line per channel. They’re simple and cheap, but they pick up more common‑mode noise. Differential multiplexers switch both the positive and negative sides of a signal pair, keeping the noise common to both lines and canceling it out at the ADC. If your sensor array is prone to EMI—think industrial environments—go differential.

Number of channels

It’s tempting to buy the biggest chip you can find and leave room for future expansion. But larger multiplexers often have higher on‑resistance and slower switching because the internal switches share the same control bus. If you only need eight channels, an 8‑to‑1 part will usually give you better performance than a 32‑to‑1 part.

Power budget

Battery‑powered data loggers can’t afford to waste milliwatts on a hungry multiplexer. Some parts have “low‑power” modes that shut down unused switches. Check the supply current spec and see if the chip can run from the same voltage rail as your ADC—this reduces the need for extra regulators.

Practical Tips from the Bench

When I was building a 16‑channel temperature scanner for a greenhouse, I started with a 32‑channel part because I thought I might add more sensors later. The first test run showed a 20 % drop in signal amplitude. The culprit? The on‑resistance of the larger chip was about 250 Ω, which, combined with the high source impedance of the thermocouples, formed a voltage divider that ate most of the signal.

Switching to a smaller 16‑channel device with 30 Ω R_on restored the full voltage swing. The lesson? Bigger isn’t always better—pick the smallest part that meets your channel count.

Another time, I tried a cheap single‑ended multiplexer on a high‑speed photodiode array. The data looked jittery, and the ADC was complaining about “over‑range” errors. The problem turned out to be the settling time: the chip needed 5 µs to settle, but I was switching at 200 kHz, giving it only 5 µs total per sample. The solution was a faster differential part with a 0.5 µs settling time, which cleaned up the waveform instantly.

How to Test Before You Commit

Breadboard a prototype – Hook up a single channel of the multiplexer to a known sensor and measure the voltage drop across the switch. This gives you a real‑world view of R_on and leakage.
Measure settling – Use an oscilloscope to watch the output as you toggle channels. Count how long it takes to reach within 1 % of the final value.
Check crosstalk – Apply a signal to one channel and look at an adjacent idle channel. Any noticeable voltage is crosstalk, which can corrupt neighboring sensor readings.

If the prototype passes these three simple checks, you’re probably good to go.

Bottom Line

Choosing the right analog multiplexer for a high‑precision sensor array is a balancing act between resistance, speed, leakage, and architecture. Start by listing the exact needs of your sensors—voltage range, impedance, and required sampling rate. Then narrow the field by matching those needs to the specs: low R_on, fast settling, low leakage, and the appropriate single‑ended or differential layout. Finally, verify the choice on a small test board before committing to a full design.

At Analog Switchboard we’ve seen too many projects stumble over a cheap part that looks good on paper but falls apart in the field. Take the time to do a quick bench test, and you’ll save yourself hours of debugging later.