November 11, 2026 fct test-station hardware-startup guide

Setting Up Your First FCT Station: A Hardware Startup Guide

You're three weeks from shipping your first 500 units. The PCBs are assembled, the firmware is flashed, and your CM says everything looks good. Then your co-founder asks the question nobody has answered yet: "How do we know these actually work?"

You start Googling. Test equipment vendors quote you $50,000 for an automated system with a six-week lead time. A test engineering consultant wants $15,000 just to write a spec. Your CM offers to "do some basic checks" but can't tell you exactly what they'll test or how. You're staring at a crate of 500 boards and wondering if you should just power each one up by hand.

Here's the thing most first-time founders don't realize: you don't need a $50K system to start. You need a $5K jig, a clear test plan, and the discipline to grow from there.

Start with the question, not the equipment

Before you buy a single pogo pin, write down what "working" means for your product. Not in general terms — in specific, measurable terms. If you're building a BLE sensor, "working" might mean: 3.3V rail within ±3%, accelerometer reads gravity on the Z-axis within ±0.1g, BLE advertises within 2 seconds of power-on, deep sleep current below 15 μA.

These are your pass/fail limits. They come from three places: your component datasheets (the accelerometer spec says ±0.06g typical), your system requirements (the battery needs to last 6 months, which means sleep current can't exceed 15 μA), and your field experience (you've seen BLE init take up to 1.8 seconds on your dev boards, so 2 seconds gives margin).

Write them all down. Every measurement, every limit, every unit. This document — your test specification — is more important than the fixture itself. A $2,000 jig running against a clear spec catches more defects than a $20,000 system running vague "power-on checks."

The $5K starting line

Your first FCT station has four pieces: a fixture, a power supply, a few instruments, and a test script.

The fixture is a bed-of-nails jig — a plate with spring-loaded pogo pins that press against test pads on your PCB. You designed those test pads into your board, right? (If not, that's your first lesson for Rev 2. For now, you can probe component pads or vias, though it's less reliable.) A fixture house machines the plate from acrylic or G10, drills holes at your test pad coordinates, and presses in pogo pins rated for 50,000+ insertion cycles. FCT-specific fixtures typically cost $2,000–$10,000 depending on board complexity (PCB&Assembly). Lead time: 2–3 weeks. The operator drops the board in, closes the clamp, and every test point makes contact simultaneously.

The power supply is a bench-top programmable DC supply — something like a Rigol DP832 or Siglent SPD3303X (Bester PCBA). You need programmable because your test script controls it: power on at 3.7V, measure quiescent current, ramp to 4.2V, check overvoltage protection. A good programmable supply costs $300–$500 and talks SCPI over USB or LAN.

The instruments start minimal. A digital multimeter for voltage and current measurements. A USB-to-serial adapter for firmware communication. A BLE dongle if you're testing wireless. If you're measuring analog signals, add a basic oscilloscope. You probably already own half of this from prototyping. Budget another $500–$1,500 for anything you're missing.

The test script ties it all together. A Python script running on a laptop is perfectly fine for your first station. Google's OpenHTF framework is purpose-built for this — open source, Python-native, designed for hardware test sequences (Nominal). Each test step is a function: power on, wait 500ms, measure rail voltage, compare against limits, log result. The script controls the instruments via SCPI or serial, runs the sequence, and outputs pass/fail with every measured value recorded. No LabVIEW license needed. No proprietary software lock-in. Your test code lives in git alongside your firmware.

Total cost for this starting setup: $3,000–$5,000 (FixturFab). Time to get running: 2–4 weeks including fixture fabrication. Test time per board: 2–5 minutes with an operator loading boards manually.

What your CM already has

Here's something founders often miss: many contract manufacturers already have test infrastructure. They have bench power supplies, multimeters, sometimes even custom fixtures from previous clients. The gap isn't equipment — it's specification. The CM's test technician is perfectly capable of running your test sequence. They just need you to tell them exactly what to test.

When you send your CM a clear test spec — every measurement, every limit, every step in order — you're not asking them to do something they've never done. You're giving them the instructions they've been waiting for. Most CMs appreciate this. It removes ambiguity, protects them from blame when issues arise, and makes their test station operators more efficient. A defined 90-second test sequence is easier to run than a vague instruction to "check if it works."

The challenge is getting the results back. Your CM runs the tests, but the data lives on a local PC at the factory. You get an email: "Batch 042, 498/500 passed." What failed? Why? What were the margins on the units that passed? Were any measurements drifting toward the limit? Without the raw data, you're flying blind.

Growing from manual to automated

Your first 500 units ship. Then 2,000. Then 10,000. At some point, manual fixture loading and a 3-minute test cycle become the bottleneck. That's when you invest in automation — not before.

The path looks like this. First, you shorten the test cycle. Your 3-minute test probably has 30 seconds of actual measurement and 2.5 minutes of waiting — waiting for the BLE stack to initialize, waiting for the sensor to stabilize, waiting for firmware to flash. Optimize the test firmware to initialize faster. Overlap measurements where possible. Pipeline firmware flashing during other tests. You'll get to 60–90 seconds without changing hardware.

Next, you add a second fixture. Two operators running two stations doubles throughput for $3,000 in additional fixture cost. This carries you to 10,000+ units per month.

Only when you're shipping 20,000+ units monthly does full automation start making sense: robotic board loading, conveyor integration, multi-up fixtures testing 4 boards simultaneously. Dedicated automated FCT platforms run $50,000–$200,000+ (PCB&Assembly). By the time you need it, you'll have the test data from thousands of units telling you exactly what to automate and where the bottlenecks are.

The data is the real product

Here's what separates a test station from a quality system: the data. Every board that passes through your fixture generates a record — serial number, every measurement value (not just pass/fail), limits applied, timestamp, operator ID, firmware version, station ID. This record is the board's birth certificate.

Six months from now, when a customer reports a failure, you'll query that serial number and see exactly what it measured at the factory. You'll check whether other boards from the same batch show similar margins. You'll trace the component lot and find that the accelerometers from that supplier were consistently reading 0.08g high — still within spec, but drifting. That insight only exists because you recorded the actual measured value, not just "PASS."

AN & Factory's test station integration captures this data automatically. You define your test spec on the platform, the station software runs at the CM, and every result syncs to the cloud in real time — signed and tamper-proof. Your CM doesn't need to email CSVs. You don't need to build a data pipeline. The test record follows the device from factory to customer.

Your first week

Here's what you can do this week, with no budget approval and no new equipment. Open a document. Write down every measurement that defines "working" for your product. Voltage rails, current draw, sensor readings, communication interfaces, firmware verification. Put a number and a tolerance on each one. That document is your test specification, and it's worth more than any fixture you'll ever buy.

The fixture, the instruments, the automation — those are just tools for executing the spec faster. The spec itself is the decision that your customers won't be your test department.

Need help setting up your first test station? AN & Factory provides test station software, signed test records, and fixture design guidance. Get in touch.

Sources

Have thoughts on this? Join the discussion on our forum.

← Back to blog