Keysight Multimeter, Used Keysight Optical Spectrum Analyzers & Other Test Gear: A Mistake-Based Guide

There isn't one 'right' way to source test equipment. If someone tells you there is, they are probably trying to sell you a very specific solution. I say this after eight years of handling instrument orders and after making, and documenting, fourteen significant mistakes that added up to roughly $27,000 in wasted budget. This article is the checklist I now use so that I don't repeat those errors.

When I first started in this role, I assumed the safest choice was always the newest, highest-spec instrument from a famous brand. That assumption cost me time, money, and a few awkward conversations. What I learned is a little more nuanced: the right buying decision depends on what you are going to measure, what uncertainty you can tolerate, and how much time you have to make the measurement.

Scenario A: You need repeatable electrical measurements, not spectral analysis

If your daily work is voltage, resistance, current, or low-frequency AC, a good digital multimeter will do more for you than a rack full of specialized boxes. A Keysight Technologies multimeter like the 34461A has a one-year basic DCV accuracy of about 0.0035 percent, according to Keysight's published specifications (keysight.com). But the spec is only part of the story. What I mean is, the real value shows up when you put several of these on a bench and they all agree at the fifth digit. That agreement is worth more than the individual accuracy number.

Personally, I would buy a used 34401A or 34461A with a valid calibration certificate before I would spend the same money on a brand-new meter with no calibration traceability. I started my career buying one calibrated Keysight Technologies multimeter at a time, on POs so small that some vendors laughed. The vendors who treated those small orders seriously are the ones I still call for five-figure orders years later. Small does not mean unimportant; it means potential.

Scenario B: Used Keysight optical spectrum analyzers—only if you need to see the spectrum

Used Keysight optical spectrum analyzers are a different animal. They are excellent for measuring wavelength-resolved properties like side-mode suppression, channel spacing, or optical signal-to-noise ratio. They are the wrong tool if you just need to know total optical power. That sounds obvious, but I made this exact mistake in September 2022. I bought a used Keysight OSA for close to $8,000 because I was worried about accuracy. The project actually needed a center wavelength confirmation and a power reading. A wavemeter, or a photodiode plus a precision multimeter, would have handled it for far less. The OSA sat unused for three months.

If you do need a used Keysight optical spectrum analyzer, buy one with a current calibration certificate, not one that 'works fine.' The detector response drifts. In Q1 2024, I almost bought a clean 86142B for $5,200. The calibration quote came back at $3,100 plus a six-week turnaround. That's the total cost of ownership you have to consider. I should add that this is not about blaming the seller; it's about knowing where the money goes after the purchase.

Scenario C: Water flow meters and AFM60A encoders—interface matters more than brand

Test equipment is not all electrical. Sometimes you are validating a water flow meter, or you are measuring motor feedback with a rotary encoder. A water flow meter seems simple, but your uncertainty model has to include linearity, repeatability, and the analog output wiring. If the data acquisition system has a ground loop, the meter will get blamed for a problem that is actually in the signal path.

The same thing happens with absolute encoders. A SICK AFM60A, for example, is a quality unit, but the exact model determines the output protocol. I once ordered an AFM60A assuming it had SSI output because an earlier model in the same family did. This one had PROFINET. I didn't catch it until the cable harness was built. That mistake cost $410 in custom cables and a one-week schedule slip. The lesson, in my experience, is to read the full part number and the output interface before you buy, not after.

Scenario D: How Agilent fittings for HPLC columns work

Another thing I had to learn the hard way was how Agilent fittings for HPLC columns work. I am an electrical engineer by training, so this was outside my comfort zone. Here is the short version: Agilent HPLC column fittings are compression fittings. The nut pushes a ferrule onto the column tubing, and the ferrule deforms to create the seal. That ferrule is mostly a single-use part. When you remake the connection, you should replace it. (Should mention: I reused a ferrule once and tightened it with a wrench, which made the leak worse.) The fitting itself is not the expensive part. The expensive part is the day of calibration data you lose when it leaks.

If you are working with legacy Agilent hardware, assume that any fitting you cannot verify is suspect. Replace the ferrule, confirm the torque, and run a leak test before you start a long measurement. This sounds basic, but it is exactly the kind of thing that disappears from a project checklist when someone is in a hurry.

How to tell which scenario you are in

There is no universal answer, but there is a useful set of questions.

1. Do I need wavelength-resolved data? If yes, a used Keysight optical spectrum analyzer might be justified. If no, measure intensity with a photodiode and a Keysight Technologies multimeter, or something equivalent.

2. Is this an electrical precision problem? Choose a matched set of digital multimeters and keep the same model family across your test rack. That consistency will reduce your comparison uncertainty.

3. Is this a flow or motion problem? Map the sensor output to your acquisition system before you order. Whether it is a water flow meter or an AFM60A encoder, the protocol and the wiring define how long the integration will take.

4. Am I using fluid fittings? Treat the ferrule as a single-use part until you have reason to trust it. Then replace it anyway if there is any doubt.

The real lesson, from my perspective, is that the best instrument is the one that answers your question with known uncertainty and at a cost you can justify. I still buy Keysight. I also buy cheap pressure sensors when a project only needs a go/no-go check. I would rather be the engineer who buys the right tool instead of the impressive one. And if a small order is what you have, find a supplier who treats that order with respect. I try to be that supplier now, because I remember how it felt when I was the small order.