I've been handling field test equipment orders and deployments for about 4 years now. In that time, I've personally made (and documented) more than a dozen significant mistakes, totaling roughly $12,000 in wasted budget and lost time. One of the most frustrating was a full day I lost in September 2023 because I couldn't get my Anritsu MS2090A to connect properly during a 5G NR field validation.
It wasn't a hardware failure. It wasn't a firmware bug. It was me. I'd skipped a step on my mental checklist that I'd used a hundred times before. The result? A $3,200 piece of gear sitting idle while I chased phantom problems.
Since then, I've created a rigid 6-point checklist for any field test setup, especially when using a new instrument or after a firmware update. This isn't a full user manual—it's the specific stuff I've messed up, so you don't have to.
This is for when you're setting up an Anritsu MS2090A (or similar handheld spectrum analyzer) for an out-of-the-box test, after a firmware update, or when you're getting weird readings that don't match the simulation. It's the 'first 10 minutes' checklist that saves the next 10 hours.
There are 6 steps. Don't skip 'em.
This sounds insultingly basic. I know. But the physical connection is where I lost my day. I was using a brand-new test cable that looked fine. No kinks, no bent pins. But the SMA connector wasn't fully torqued on the MS2090A. It was hand-tight, but in a hurry, 'hand-tight' is often 'barely touching.'
The check: Use a torque wrench for the first connection. It sounds overkill for a field test, but a loose connector creates micro-arcs and intermittent loss that look like a signal drop issue. On my MS2090A, that looked like a sudden -6 dBm drop at 3.5 GHz. I spent hours blaming the antenna. It was just the damn connector.
Also, visually inspect the center pin on the analyzer's port. I've seen debris from a cable cap get lodged in there. It only takes one tiny speck of metal to throw off a calibration.
This is a specific gotcha on the MS2090A. It's a beast, but it eats battery. If you're running it on battery power, the unit automatically dims the screen and can throttle the receiver after 15 minutes of high-bandwidth scanning to preserve power. I didn't know this until I read the deep-dive manual post-failure.
The check: Before you start critical tests, confirm the battery is above 80% and you've set the power profile to 'Performance' mode, not 'Power Save'. You can find this in the Utilities menu. Otherwise, you might be measuring the analyzer's thermal noise floor instead of your actual signal. That'll mess up your PIM test results fast. This was accurate as of firmware version 6.02. The UI might have shifted since then, so verify current settings in the manual.
Here's the big one. The MS2090A has a built-in InstaCal feature. I assumed it was always up-to-date. It's not. When you turn the unit on, it might still be using the calibration coefficients from the previous session or from the factory firmware version.
The check: Go into 'Calibration' and check the 'Cal Date' and 'Cal Temperature'. If the unit is >5°C from the last cal's temperature (which happens when you pull it from a cold truck into a hot rooftop), it drifts. You must initiate a full InstaCal once the unit has been powered on for 15 minutes and stabilized. The 'surface illusion' is that the unit is ready to go. The reality is that temperature-induced drift in the IF chain can cause 1-2 dB of error at 6 GHz.
From the outside, it looks like you just power it on and go. The reality is the first 15 minutes are 'warm-up time' for the oscillator to stabilize. I learned this the hard way after a $890 redo because my readings were consistently 1.5 dB low.
Auto-mode is tempting. But for field work, the auto-reference level algorithm on the MS2090A (and really any analyzer) will jump around if there's intermittent interference. You'll see the trace bounce up and down as the unit adjusts its gain.
The check: Start with a manual reference level at 0 dBm. Set the input attenuation to 10 dB (or 20 dB if you suspect high-power signals). This prevents the pre-amp from getting overloaded by a burst of signal while you're scanning a wide band. It's a simple fix that prevents the 'noise floor floating' headache. Then, adjust down as needed once you see the actual signal level. Doing this saved me from a false 'No Signal' alarm on a 5G site.
This isn't a technical step. It's a mental one. When you see a weird peak or a signal that doesn't match the planning file, do not immediately start tweaking settings. I've done this. You change the RBW, you adjust the span, you switch the detector mode—and suddenly you've lost the original signal and you're chasing a ghost that was just an artifact of your own adjustments.
The check: Freeze the trace. Save the screenshot to the internal memory (the MS2090A makes this easy with a dedicated button). Take a photo with your phone of the cable connections. Then, go back to Step 1 and verify the physical setup. Nine times out of ten, the weird signal is a reflection from an unterminated port or a loose adapter. In Q1 2024, I created a simple 'Frozen Trace First' rule for my team. It’s prevented at least three false-site-alarm calls.
Another personal mistake. You finish the test. You're hot, you're tired, you pack up the unit. The next tech turns it on, sees your last settings, and starts testing on a different band with your previous calibration. They might not check. It's tempting to think you can just shut it down. But the MS2090A boots into the last-used state.
The check: Before you power off, reset the unit to its default startup state. In the 'File' menu, there's a 'Reset to Factory Defaults' (not a full factory reset, just the measurement settings). This clears the calibration state and sets it to a neutral condition. Then, label the battery as 'Used' so the next person doesn't start a critical drive test on a half-charged unit. It's a small thing that prevents big misunderstandings.
This is a note I always include. The Anritsu MS2090A gets regular firmware updates (version 6.05 was a big one for 5G NR support). But after an update, the unit always requires a new full calibration. The InstaCal coefficients are stored per firmware version. I missed this once and my PIM test results were completely unreliable for a week before I realized the upgrade had reset the calibration database.
Bottom line: treat a firmware update like a new instrument. Run the full 6-step checklist again. It’ll save you the embarrassment of a re-test.