Keysight Bench Digital Multimeter + Function Generator: The Sensor Layer You're Probably Forgetting

Buy the Keysight Technologies bench digital multimeter and a function generator first—that's the right instinct. But if you stop there, you're going to miss the sensors that make the measurement actually useful. I've made that mistake, and it cost me roughly $4,300 in wasted hardware plus a two-week delay on a customer qualification test. The fix isn't a bigger budget. It's realizing that fork sensors, proximity sensors, and sometimes a FLIR thermal camera belong in the same initial setup, not in a later 'accessories' order.

I've been specifying test benches since 2017, first as a lab tech, later as the engineer our team calls when the 'simple' measurement keeps failing. I have personally documented 14 significant equipment ordering mistakes, and the biggest one was treating sensors as an afterthought. Here is what I learned—and what I do now to keep our team from repeating it.

Why the DMM Is Only as Good as the Sensor Layer

When I first started building benches, I assumed the DMM was the whole story. Buy a good meter, I thought, and everything else is just wiring. Three budget overruns later, I learned a meter gives you a reading—it doesn't tell you whether the signal you're measuring is real, conditioned, or being picked up from the wrong sensor.

From the outside, a benchtop multimeter plus a signal source looks like a complete setup. The reality is that sensors decide whether your measurement is meaningful. A mismatched fork sensor or proximity sensor can output a signal that looks perfectly fine on a display but is completely wrong for your fixture. That is not a calibration problem. That's a system design problem.

On the DMM side, I settled on a Keysight 34461A after watching it stabilize a 10 µV signal my earlier meter could not hold. Keysight's published basic DC voltage accuracy for this class of meter is around 0.0035%, which is the kind of number that doesn't matter until you are chasing a 20 µV drift in a reference design. It mattered the first week we tested low-side current on a prototype.

One more thing: buy the meter with data logging enabled from day one. The last thing you want is to discover you need logging and have to send the unit back for a firmware option upgrade. I have done exactly that, and I do not want you to.

The Function Generator Keysight Offers Is Not Optional

The function generator Keysight offers in its 33500B series became the other half of my bench. When I ordered my first one, I chose the cheapest model without thinking about output amplitude. My circuit needed 20 Vpp; the generator could only deliver 10 Vpp. So I bought an external amplifier, which introduced noise, which then made me question my DMM. The added amplifier worked. Not ideal, but workable. That chain of frustration was avoidable if I had spent five minutes on the spec sheet.

Why does this matter? Because if you are driving a fork sensor's emitter or testing a proximity sensor's response, you need a clean, controllable stimulus. 'Clean' is the key word. A function generator that produces a crudely quantized waveform can create false triggers and intermittent behavior. Keysight calls its waveform technology Trueform—practically, the sine wave does not look like a staircase. For sensor testing, that is more than marketing.

My rule now: write down the maximum voltage swing, the output impedance, and the frequency you actually need before you search for a part number. Do not browse models first. It saves the kind of mistake I made.

The Forgotten Middle: Fork Sensors

Fork sensors are one of those components that never make the 'main instrument' list, but they are everywhere in automated test fixtures. What is a fork sensor? At its simplest, it is a U-shaped photoelectric device with an emitter on one side and a receiver on the other. You place the object between the prongs. If the object interrupts or re-establishes the light beam, the sensor changes state. They are used for counting parts, checking presence, finding a datum edge, or detecting markers on a moving web.

My fork sensor mistake was embarrassingly simple. I ordered one with a 2 mm slot width because it looked compact and 'sensitive.' Our carrier rails were 5 mm thick. The sensor did not fit. The special cable option made it non-returnable. $280 wasted. So glad I checked the drawing before ordering the second lot—I almost bought the same part again.

If I had one sentence for you: fork sensors are simple, but the mechanical envelope, output type (PNP vs NPN), and response speed matter far more than the brand. The brand's job is consistency; your job is matching the geometry.

Proximity Sensors: The Least Glamorous, Most Critical Piece

Proximity sensors are the workhorses of any bench fixture that has moving parts. They answer one question: is something there? But there are two main families, and they are not interchangeable. Inductive proximity sensors detect metal only. Capacitive proximity sensors detect materials that change the local dielectric field—including plastic, wood, or even your hand.

I learned this the expensive way. I used a capacitive proximity sensor to detect metal pins inside a metal fixture. It kept seeing the fixture itself. Every false trigger made our DMM log a bad reading. An engineer walked over, swapped in an inductive sensor, and the problem disappeared. The cause was not the meter; it was the sensor choice. My initial assumption—'a sensor is a sensor'—was wrong.

Tip: when you spec a proximity sensor, write down the target material, the sensing distance, and the ambient temperature. Inductive sensors are simple and robust, but they do not work on non-metal targets. Capacitive sensors are more flexible, but they need careful adjustment because they respond to everything around them. That is the kind of nuance that no one puts on a marketing page.

What Is FLIR Thermal Camera? (And When You Actually Need One)

What is FLIR thermal camera? In one sentence: it is a camera that sees infrared radiation and turns temperature differences into an image you can look at. FLIR originally meant Forward-Looking Infrared, and the name stuck. For a test engineer, a thermal camera is a tool that lets you see which component on a PCB is actually heating up, instead of touching every capacitor with a finger and hoping.

I used to think thermal cameras were for building inspections and outdoor surveillance. Then a prototype started drawing 600 mA more than it should. I spent two days swapping ICs, checking solder joints, and questioning the power supply. A colleague brought a FLIR thermal camera to the bench. In five minutes, we saw the regulator that was running at 85°C. It was the wrong part, not a soldering issue. The meter and function generator did their jobs; they just could not tell us where the heat was.

Do you need a FLIR camera on day one? Not always. If you are working only with low-power analog circuits below 12 V, a thermal camera might be overkill. But the moment you see 'unexplained current draw' in your test log, the cost of a thermal camera is tiny compared to the engineering hours you will burn without one.

Boundary Conditions: Where I'd Push Back on 'Buy Everything'

Now the part where a lot of blog posts stop and I won't. Not every lab needs every one of these items on the first order. Here are the exceptions I have learned the hard way:

• Skip the fork sensor if your target is transparent and you haven't tested it with the actual material. Many fork sensors rely on beam interruption; a clear film may not block enough light. Test with a sample first.
• Do not buy a capacitive proximity sensor for a metal-only application. Buy inductive. It will save you the adjustment headaches I caused.
• A thermal camera is not a substitute for a current probe or a DMM. It finds hot spots, but it does not tell you why the current is high. Use it with, not instead of, your Keysight instruments.
• If your team works mainly on low-voltage digital logic at low current, you can probably postpone the FLIR camera until an overheating issue appears. When it does appear, you'll know within the first hour.

The old 'DMM plus signal source is enough' thinking comes from an era when bench instruments were the only tools available. That has changed. With sensors and thermal imaging, you can now see causes, not just effects. The fundamentals haven't changed—voltage, current, resistance still rule—but the execution has transformed. Adjust your checklist accordingly.

Note to self: this article is exactly the checklist I wish I had in 2017. If you are about to place a bench order, save it, and remember the $4,300 lesson. (And yes, the '10% for sensors' line in our budget now exists because of that lesson.)