Signal Generator Calibration in Singapore: What Is Verified and Why Accuracy Matters

What is verified during signal generator calibration

Signal generators span a wide range of instruments, from basic function generators (sine, square, triangle, 1 Hz to 1 MHz) to high-frequency RF synthesisers (100 kHz to 40 GHz) and arbitrary waveform generators (AWGs). The specific parameters calibrated depend on the instrument type, but the core verification covers the following:

The calibration scope is confirmed against the instrument's specification sheet. For each parameter, the as-found reading is recorded before any adjustment, the reference value is stated, and the deviation and uncertainty are calculated. If the instrument is adjusted, as-left readings are taken after adjustment. The certificate states which parameters are within specification and which (if any) require adjustment or are outside the manufacturer's stated tolerance.

Why signal generators drift. The physics of frequency and amplitude error

Crystal oscillator aging

Most signal generators derive their frequency reference from a crystal oscillator. Either a temperature-compensated crystal oscillator (TCXO) or an oven-controlled crystal oscillator (OCXO). Crystals age: the physical resonant frequency of the quartz element drifts slightly over time as molecular stress in the crystal slowly relaxes after manufacture. A TCXO typically drifts 1–5 ppm per year; a well-disciplined OCXO may drift less than 0.1 ppm per year. For a 1 GHz signal, 1 ppm corresponds to a frequency error of 1 kHz. Whether that matters depends on the application, for general-purpose testing, 1 kHz at 1 GHz is negligible; for narrowband communications testing, it may cause measurement failures.

Attenuator degradation

The output amplitude of a signal generator is set by a combination of a variable attenuator and a fixed-gain amplifier. Mechanical step attenuators use precision resistive networks with relay contacts. Over thousands of switching cycles, the contact resistance of the relay switches drifts, introducing errors at specific attenuation settings. This is why amplitude accuracy should be checked at multiple output power settings. A generator that reads correctly at -10 dBm may be 0.5 dB off at -40 dBm due to a specific attenuator step that has degraded.

Output amplifier gain drift

The output amplifier that drives the signal to the connector has a gain that varies with temperature and age. At frequencies above 1 GHz, amplifier flatness (the consistency of gain across the frequency range), also degrades. This produces amplitude errors that are frequency-dependent: the generator may be accurate at 100 MHz but read 0.3 dB high at 3 GHz because the amplifier's gain has risen at higher frequencies.

Applications that require calibrated signal generators in Singapore

Electronics manufacturing and production test

Production test lines for RF circuits (mobile phone modules, Wi-Fi chipsets, Bluetooth devices, and power amplifiers), use signal generators as stimulus sources. A production test that relies on a signal generator whose amplitude is 0.5 dB off will systematically accept devices that are marginal on gain, or reject devices that are actually within specification. In either case, the product quality and the customer relationship suffer. Annual calibration with stated amplitude uncertainty catches this drift before it causes systematic test failures.

Telecommunications infrastructure testing

Singapore's telecommunications infrastructure (base stations, fibre networks, data centre links), requires periodic testing with calibrated test equipment. Engineers testing receiver sensitivity, transmitter output power, or cable loss use signal generators as reference sources. If the generator's output level is not calibrated, the test result cannot be cited as a measurement. Only as an observation with unknown accuracy.

Aerospace and defence at Seletar and Changi

Avionics test (testing NAV, COM, IFF, and radar systems on aircraft), requires signal generators calibrated to traceable standards under CAAS Part 145 and AS9100. Avionics frequency and power tolerances are tight; an uncalibrated signal generator used to verify a VOR or ILS receiver may pass or fail the aircraft on incorrect evidence. SAC-SINGLAS accredited calibration provides the documented traceability that aviation regulators require.

Calibration interval and in-house vs outsourced calibration

The standard calibration interval for a signal generator is 12 months. This is the interval recommended by most manufacturers (Keysight, R&S, Anritsu) and accepted by most quality and compliance auditors. In-house calibration of signal generators is practical only for organisations with calibrated reference instruments. A frequency counter, a calibrated power meter with appropriate sensors, and for RF generators, a spectrum analyser. The reference instruments themselves require calibration at a higher level of accuracy, typically by a SAC-SINGLAS accredited lab. Most organisations find it more cost-effective to outsource signal generator calibration to an accredited lab rather than invest in and maintain the reference instrument infrastructure required to do it in-house.

How signal generator calibration is actually performed

Calibration of a signal generator verifies three independent parameters, each requiring a different reference instrument. Frequency accuracy is verified using a calibrated frequency counter or, for higher precision, a GPS-disciplined frequency reference, comparing the generator's actual output frequency against the counter's reading across multiple frequency points spanning the generator's full range. Amplitude accuracy is verified using a calibrated power meter with an appropriate power sensor for the frequency and power level under test, measuring the generator's actual output power at each of several amplitude settings across its range, since, as noted above, amplitude error is not consistent across all output levels. Where the generator supports modulation (AM, FM, or digital modulation formats), modulation accuracy is verified using a vector signal analyser or modulation-capable spectrum analyser, confirming the generator's modulated output matches its programmed modulation depth, deviation, or symbol characteristics within the manufacturer's specified tolerance. All three measurements are taken as-found before any adjustment, then as-left if the generator supports internal calibration adjustment, with both data sets recorded on the certificate.

Phase noise and spectral purity: a less obvious but real accuracy factor

Beyond frequency and amplitude accuracy, a signal generator's phase noise, random short-term fluctuations in the output's phase, close to the carrier frequency, affects how usable the generator is for certain test applications even when its frequency and amplitude are both perfectly within tolerance. High phase noise can cause a receiver under test to show poorer sensitivity than it actually has, because the generator's own instability is being measured alongside the receiver's genuine performance, or can produce misleading results in adjacent-channel power measurements common in telecommunications testing. Standard annual calibration typically verifies frequency and amplitude accuracy but not always full phase noise characterisation, since phase noise measurement requires more specialised equipment (a phase noise analyser or a sufficiently capable spectrum analyser with phase noise measurement capability) and is usually only performed where the specific application, precision telecommunications testing, radar system characterisation, is sensitive enough to require it. If your application depends on low phase noise, confirm with your calibration provider whether it is included in the standard scope or needs to be requested separately.

External frequency reference locking for higher accuracy

Many higher-end signal generators support locking to an external 10 MHz frequency reference rather than relying on their own internal crystal oscillator, and using this feature correctly can materially improve frequency accuracy for applications that need it. Locking the generator to a calibrated, more stable external reference, an OCXO frequency standard, a GPS-disciplined oscillator, or another instrument's reference output, effectively transfers the external reference's superior stability to the generator's output frequency, bypassing the accuracy limitation of the generator's own internal crystal entirely. This is a common technique in telecommunications and aerospace test setups where multiple instruments share a single high-stability reference distributed across a test rack, ensuring all instruments in the chain are frequency-locked to the same source rather than each drifting independently against its own internal reference. When a signal generator is calibrated while locked to an external reference, this should be noted on the certificate, since the stated frequency uncertainty reflects the reference actually used during calibration, not the instrument's own internal oscillator specification.

Harmonic and spurious output: a fourth parameter beyond the basics

Beyond frequency, amplitude, and modulation accuracy, a signal generator's output is never perfectly pure; it contains harmonics (energy at integer multiples of the set frequency) and spurious signals (unwanted energy at frequencies unrelated to the set frequency, often originating from internal mixing products or power supply switching noise) at levels well below the fundamental but not at zero. For most general-purpose testing this is a secondary consideration, but for applications testing a receiver's selectivity or adjacent-channel performance, or for EMC pre-compliance work where the generator itself is meant to represent a clean, known interference source, harmonic and spurious content can materially affect the validity of the test result if it is not characterised and accounted for. A generator's datasheet typically states maximum harmonic and spurious levels relative to the fundamental (in dBc, decibels relative to the carrier), and while routine annual calibration does not always include a full spurious-output sweep across every frequency, confirming these specifications are still being met is worth requesting specifically for applications where a genuinely clean signal, not just an accurate frequency and amplitude, is the actual requirement.

Calibrating vector and digitally modulated signal generators

Modern signal generators used in telecommunications, particularly those generating digitally modulated waveforms for 4G, 5G, or Wi-Fi standard testing, add a further layer of complexity beyond the analogue AM/FM modulation accuracy discussed earlier. These vector signal generators must accurately reproduce complex digital modulation formats (QPSK, various orders of QAM, OFDM waveforms), and their calibration extends to verifying Error Vector Magnitude (EVM) performance, a measure of how closely the generator's actual output constellation matches the ideal, mathematically defined constellation for the modulation format in use. A generator with elevated EVM will produce a signal that a receiver under test could interpret as its own performance limitation, when the actual source of the error sits in the generator rather than the device being tested. Calibrating this class of instrument requires a vector signal analyser capable of demodulating the same complex waveform formats, materially more specialised equipment than the frequency counter and power meter sufficient for a basic CW (continuous wave) signal generator, and is a distinct scope question worth confirming explicitly with your calibration provider if your work involves modern digital communication standards testing.

Reading a signal generator's calibration certificate correctly

A signal generator calibration certificate typically presents results as a table of frequency and amplitude test points with as-found deviation and stated uncertainty at each, and interpreting it correctly means checking the specific points relevant to your actual usage, not just confirming the certificate exists. If your application only ever uses the generator between 1 and 2 GHz, a deviation noted at 10 GHz on a wideband generator's certificate is largely irrelevant to your work, while a marginal result at 1.5 GHz deserves closer attention even if the overall certificate shows an aggregate pass. Similarly, if your work depends on amplitude accuracy at low output levels (testing receiver sensitivity near the noise floor, for instance), specifically check the as-found deviation at those low amplitude settings, since, as covered earlier, amplitude accuracy is not uniform across a generator's full output range and a certificate showing excellent accuracy at 0 dBm says little about performance at -60 dBm. Reading the certificate against your actual frequency and amplitude usage pattern, rather than treating a certificate as a single pass/fail verdict, is what turns the calibration record into genuinely useful engineering information rather than a filing-cabinet compliance artefact.

Signal generators in EMC pre-compliance and susceptibility testing

Beyond production test and telecommunications applications covered above, signal generators play a specific and demanding role in EMC (electromagnetic compatibility) susceptibility, or immunity, testing, where the generator, typically feeding a power amplifier and antenna, deliberately creates a defined RF interference field to confirm a product under test continues to function correctly in its presence. This application places particular weight on amplitude accuracy at the specific field strength the relevant EMC standard requires, since an under-calibrated generator producing a weaker field than intended could pass a product that would actually fail against the standard's genuine requirement, while an over-calibrated generator exposing the product to more interference than the standard specifies could cause an unnecessary and misleading failure. Singapore electronics manufacturers and EMC test houses performing susceptibility testing, whether for formal compliance or internal pre-compliance screening before sending a product to a full accredited test lab, depend on their signal generator's amplitude calibration being both accurate and current, since the entire test's validity rests on the field strength actually applied matching what the standard and the test report claim was applied.

Environmental storage and handling between calibration cycles

How a signal generator is stored and handled between calibration cycles genuinely affects how well it holds its calibrated accuracy through to the next due date, particularly for the frequency reference discussed earlier in this guide. Repeated large temperature swings, moving a generator between an air-conditioned lab and a warmer storeroom or vehicle, place additional thermal cycling stress on the internal crystal reference beyond what stable, controlled storage would produce, and can accelerate the gradual aging drift the physics section above described. Physical shock during transport, dropping a generator or subjecting it to a hard knock while being moved between test locations, can shift internal component alignment in ways that show up as a step change in calibration results at the next due date rather than the smooth, gradual drift a well-handled instrument typically exhibits. Neither factor changes the standard 12-month calibration interval on its own, but a generator with a documented history of rough handling or environmental extremes is a reasonable candidate for closer interim monitoring or a shortened interval, following the same as-found-data-driven logic applied to interval decisions throughout this guide.

Connector care on the generator's output port

The output connector itself, whether a Type-N, SMA, or other RF interface, deserves the same connector care discipline covered in our companion guide to RF calibration standards and methods: correct mating torque, periodic pin depth gauging, and protecting the connector from the physical damage that repeated, careless cable connection over years of bench use can introduce. A signal generator with a worn or damaged output connector can show amplitude deviation that has nothing to do with its internal circuitry and everything to do with a degraded mechanical interface, and this specific failure mode is sometimes missed during troubleshooting precisely because it is easy to assume amplitude drift always originates internally. Treating the output connector as a maintained, inspected component, not an inert fixture, protects both the accuracy the internal calibration certifies and the physical connector's own service life.