IMD Test Online — Distortion Check

Intermodulation distortion (IMD) is what makes overloaded gear sound harsh rather than warm: entirely new frequencies that were never in the music. Push two pure tones through any nonlinearity and the output gains sum and difference components, spaced at exact multiples of the original pair. This page plays the two standardized twin-tone signals, SMPTE and CCIF, and computes exactly where every distortion product must fall, so you can learn to recognize IMD spectra before pointing a measurement microphone at anything.

How to use it

1. Start quiet. Both tests are line-level signals in dBFS; the 60 Hz tone in SMPTE has real amplitude and will rumble small speakers.
2. Play the SMPTE pair. A 60 Hz stress tone and a 7 kHz probe tone sound together at a 4:1 amplitude ratio (the bass sits 12 dB above the probe). Press Stop at any time; it stays available while playing.
3. Read the sideband list. The panel shows the exact frequencies where 2nd- and 3rd-order products must appear around the 7 kHz probe: 6,880, 6,940, 7,060 and 7,120 Hz.
4. Play the CCIF pair. Two equal-amplitude tones at 19 kHz and 20 kHz, summed to a -4.4 dBFS peak. The panel lists their products at 18 kHz, 1 kHz, and 21 kHz.
5. Cross-check visually. Open the spectrum analyzer in a second tab while a pair plays. If your output path is clean you will see two probes and nothing between them; any rising sideband is your chain distorting.

The twin-tone standards and their math

A nonlinearity described by a power series y = a1·x + a2·x² + a3·x³ turns an input of two tones f1 and f2 into products at every combination m·f1 ± n·f2, where the order is m + n. Second-order terms create the difference tone f2 − f1 and the sum f2 + f1. Third-order terms create 2f1 − f2 and 2f2 − f1, and these are the dangerous ones: they land right next to the originals. Each 1 dB of increased input level raises 3rd-order products by 3 dB, so IMD near the overload point climbs steeply.

Two classical standards dominate audio measurement:

Standard	Tone 1	Tone 2	Amplitude ratio	Products to expect
SMPTE RP 120	60 Hz	7 kHz	4:1 (bass 12 dB up)	Sidebands at 6,880 / 6,940 / 7,060 / 7,120 Hz
CCIF (ITU difference)	19 kHz	20 kHz	1:1 equal	18,000 Hz, 1,000 Hz, 21,000 Hz
DIN variant	250 Hz	8 kHz	standardized pair	7,750 / 8,250 Hz region

SMPTE tests low-frequency power behavior: the big 60 Hz excursion drags the amplifier through its weakest region, and any gain change during that swing amplitude-modulates the 7 kHz probe, which is exactly what sidebands at ±60 and ±120 Hz mean physically. The classic bench setup sets the 60 Hz tone to +4 dBu and mixes the 7 kHz probe 12 dB lower. CCIF inverts the stress: both tones are small-signal and high, where tweeters, crossovers, and cheap converters run out of linearity, and the 1 kHz difference product falls squarely into the ear’s most sensitive region, so a fraction of a percent is easy to hear. The DIN method occupies the middle ground with a 250 Hz + 8 kHz pair.

Why IMD is harder to forgive than THD

Harmonic distortion of a single tone is musically related to the source: an octave or a fifth above it, partially maskable, and in even orders often pleasant. IMD products have no such relationship. A sideband 60 Hz above a 7 kHz tone beats against it, and the 1 kHz CCIF difference tone is an entirely new pitch that appears only when your chain distorts. Reference texts on sound reproduction consistently report that the same percentage of IMD is judged more annoying to listeners than the same percentage of harmonic distortion. This is also why guitar players run amps into clipping on purpose: power chords (root + fifth) deliberately generate 2f1 − f2 subharmonic content, which is musical IMD used as an effect.

The 3rd-order geometry explains the spacing rules. With CCIF tones 1 kHz apart, the upper product 2f2 − f1 = 2 × 20,000 − 19,000 = 21,000 Hz sits 1 kHz above the top tone, and the lower product 2f1 − f2 = 18,000 Hz sits 1 kHz below. The products march away from the stimulus pair three times faster than the tones themselves, which is what “3 dB per dB” looks like on a spectrum.

Accuracy, limits and troubleshooting

• This page computes products; it cannot measure your device. Without a calibrated input the readouts show where distortion must appear if your chain has any, not how much is there. Treat anything you hear as a hint, not a number.
• The 19/20 kHz pair is inaudible to most adults and may be removed entirely by your output path: Bluetooth codecs like SBC and AAC impose low-pass limits below 21 kHz, and some speakers’ crossovers roll off hard above 18 kHz. Absence of sound proves nothing; a visible 1 kHz difference tone on the analyzer proves the opposite.
• Ultrasonic energy at level heats tweeters. Voice coils that never see 20 kHz in music dissipate real power here. Keep the volume conservative and run short passes.
• If one tone sounds louder than the other, your volume stepped mid-playback or the OS applied per-channel effects; reset output levels and retry at a fixed position.
• 60 Hz rumble on laptop speakers is rolloff, not distortion. Consumer bass drivers cannot reproduce 60 Hz at amplitude; run the SMPTE pair on headphones or a speaker system that reaches it.

Related tools

• THD Test — the single-tone harmonic counterpart, with a verifiable self-test.
• Tone Generator — build custom twin-tone pairs beyond the standard frequencies.
• Spectrum Analyzer — watch sidebands grow around the probe tones in real time.
• Waveform Generator — compare intentional harmonic stacks with accidental IMD products.