Pre- and De-Emphasis (RIAA, FM, CD, Tape)

The treble is boosted before the medium and cut by the mirror image afterwards: the signal arrives unchanged, while hiss and crackle become quieter. With the RIAA curve of the vinyl record, FM, CD emphasis and tape equalisation – also to listen to.

Explanation Tasks

What is happening?

What emphasis does

Many transmission paths add their disturbances mainly at high frequencies: hiss on a radio link, crackle in the groove of a vinyl record. Music usually has little energy there. The trick: before the medium the treble is boosted (pre-emphasis), afterwards it is cut by exactly the mirror image (de-emphasis). The signal arrives unchanged, while the disturbance becomes quieter during the cut.

In the animation the “Input u(t)” (blue; t stands for time) arrives from the left and passes the box “Recording EQ”. In the grey middle part it lies on the medium in teal, and the disturbance is added there. The box “Playback EQ” undoes the boost, and the “Output y(t)” (orange) appears on the right. The small curves in the boxes and the dashed curves in the frequency response show the two EQs, the solid curve shows the whole chain. With “on” and “off” you bypass each EQ separately.

RIAA: bass and treble of the vinyl record

The vinyl record uses emphasis at both ends of the spectrum. In the treble it is about the disturbances: hiss and crackle of the groove lie mainly there. When cutting, the treble is boosted, by 13.6 dB at 10 kHz.

In the bass it is about space. At the same loudness the cutting stylus moves at the same velocity; the excursion of the groove then grows inversely with frequency. Low notes would need wide grooves and cost playing time. That is why they are cut when the record is cut, by 19.3 dB at 20 Hz. The reference point is 1 kHz: there both EQs amplify by 0 dB. The playback equalisation is built into the phono preamp, which is therefore needed with every turntable with a magnetic cartridge.

Three time constants

The RIAA curve is defined by three time constants τ (tau): 3180 µs, 318 µs and 75 µs. Each belongs to a corner frequency f = 1/(2π·τ), i.e. 50.05 Hz, 500.5 Hz and 2122 Hz. The readouts show both, and the corner frequencies are marked dashed in the frequency response.

The playback curve rises towards the bass below 500 Hz and levels off again below 50 Hz; above 2122 Hz it falls at 6 dB per octave. At 20 kHz it is −19.6 dB. A time constant describes a simple network of a resistor and a capacitor, τ = R·C. Every phono equaliser is built from such networks. Mechanically it would correspond to a spring with a damper and no mass – unlike the resonators in the bell and shelving filter modules.

Mirror image with a limit

The recording EQ is the inverse of the playback EQ. For RIAA and FM it would keep rising without end above the highest corner frequency. No amplifier can do that, and the cutting stylus would be destroyed by ultrasound. The model therefore limits the boost with an additional pole at 3.18 µs, i.e. about 50 kHz. In cutting amplifiers it is called the “Neumann pole”, after the maker of the cutting lathes. The whole chain loses only about 0.6 dB at 20 kHz because of it.

In 1976 an IEC standard added a fourth time constant of 7950 µs to the playback curve: a gentle high-pass below 20 Hz against rumble. It is not used everywhere and is missing from this model.

How much noise do you save?

The readout “Noise through playback EQ” states how white noise on the medium changes through the playback EQ, measured between 20 Hz and 20 kHz. White noise has the same power in every hertz, so most of it lies at high frequencies. For RIAA it is −3.7 dB, for FM −6.5 dB (50 µs) or −8.1 dB (75 µs), for the CD −5.3 dB. RIAA saves less because the playback EQ boosts the bass and with it the noise there.

Try it: switch on the sound and switch both EQs on and off in turn. The signal sounds the same, the hiss changes. With crackle you hear that the clicks become duller with emphasis.

FM and CD

In FM broadcasting the transmitter boosts the treble with a single time constant: 50 µs in Europe, 75 µs in North and South America and South Korea. At 10 kHz that is 10.2 or 13.5 dB. The price is headroom: treble-rich music would overmodulate the transmitter, so transmitters limit the treble. If transmitter and receiver do not match, it sounds duller or brighter.

The CD has a rarely used emphasis with 50 and 15 µs. It is a first-order shelf: above about 10.6 kHz the boost levels off at 10.5 dB. A flag on the CD tells the player to cut. If it is missing, for example after copying to a file, the recording sounds too bright.

Tape: the other way round

For tape the standard defines the magnetic flux on the tape: it is constant up to a transition frequency and falls above it at 6 dB per octave. That is what the recording EQ in the model shows. Tape can only store high frequencies weakly, so in real machines the record amplifier boosts the treble until the flux meets the standard. The playback head differentiates the flux and the playback amplifier integrates it again; the two cancel and are left out of the model.

So playback has to boost the treble again, and that makes the tape hiss louder: by 10.7 dB with NAB, by only 8.2 dB with IEC at 15 ips (35 µs), by 16.5 dB with the type I cassette (120 µs). NAB additionally cuts the bass on the tape with 3180 µs. An IEC tape on an NAB machine therefore sounds about 8 dB too bassy at 20 Hz. So that the bass boost on playback does not go to infinity, the model limits it below 5 Hz.

What the model simplifies

Note on time scale: as in the other filter modules, the animation runs in slow motion; the factor is shown with the readouts. At 250 Hz the readout “Slow motion” shows 1 : 500. The sound, on the other hand, runs in real time. It computes the EQs digitally at your device’s sample rate; at 48 kHz they deviate from the curve by at most 0.5 dB up to 5 kHz, and by up to 1.5 dB at 10 kHz. A limiter protects against overly loud peaks, for example when only the recording EQ is running.

As in a real recording, the input signal is limited to 20 kHz. A square or sawtooth therefore rings slightly before and after every jump (Gibbs phenomenon), at the input as at the output. Without this limit the recording EQ would turn every jump into an arbitrarily high needle. The disturbances are simplified: the hiss is white, and the clicks are short rectangular pulses at random intervals. Real record crackle, tape hiss and FM noise have spectra of their own. All EQs are ideally linear. Real FM transmitters, cutting amplifiers and tape machines also limit the level, and tapes lose treble in a way that the standard only sets as a target.

Tasks

Practice questions

Questions from studio, stage and mastering – to think about, discuss and check with the module.

1. Why is the bass cut when a vinyl record is cut?

Show answer

At the same loudness the cutting stylus moves at the same velocity. The excursion of the groove is then inversely proportional to frequency: at 20 Hz it would be 50 times as large as at 1 kHz. The groove would become so wide that it touches its neighbours, and hardly any music would fit on a side. The RIAA curve cuts 20 Hz by 19.3 dB, so the excursion becomes about nine times smaller. The phono preamp boosts the bass again.
2. A CD with emphasis is played on a player that ignores the emphasis flag. What do you hear?

Show answer

The treble stays boosted: by 4.5 dB at 5 kHz, by 7.6 dB at 10 kHz and by up to 10.5 dB at the very top. Bass and midrange stay unchanged. The recording sounds clearly too bright and harsh. Emphasis CDs mostly date from the 1980s; when they are copied to files, the flag is often lost.
3. An FM receiver from North America (75 µs) receives a European station (50 µs). How does it sound?

Show answer

The receiver cuts more than the transmitter boosted. At 1 kHz almost nothing is missing (−0.5 dB), at 5 kHz it is −2.8 dB and at 10 kHz −3.5 dB. The programme sounds a little duller, but it also hisses a little less. The other way round, a European receiver would sound too bright with an American station.

Tasks

Predict first, then try it out, then open the solution. “Load starting point” sets the matching initial values.

1. Reading the RIAA curve

A 20 Hz sine runs through the RIAA chain. Read from the readouts how much the recording and playback EQs amplify. Repeat this at 1 kHz and 10 kHz. What does “Total at f” show each time?

Load starting point

Show solution

At 20 Hz the recording EQ cuts by 19.3 dB and the playback EQ boosts by 19.3 dB. At 1 kHz both are 0 dB – the reference point of the standard. At 10 kHz the recording boosts by 13.6 dB and the playback cuts by 13.7 dB. The total stays at about 0 dB each time. The small difference at 10 kHz comes from the limiting pole at 50 kHz in the recording EQ.
2. Forgotten equalisation

The playback EQ is switched off – like a turntable connected to a line input without a phono preamp. How does the 250 Hz sawtooth change in the oscilloscope? How does it sound? Switch on the sound with the speaker button.

Load starting point

Show solution

The output now carries the recording curve: the fundamental at 250 Hz loses 6.7 dB, the 10th harmonic (2.5 kHz) gains 3.7 dB, the 40th (10 kHz) as much as 13.6 dB. In the oscilloscope the edges of the sawtooth become sharp and emphasised; it sounds thin and harsh. A real turntable would also be far too quiet, because a magnetic cartridge only delivers a few millivolts.
3. Noise with and without emphasis

Both EQs are on, and there is hiss at −30 dB in the groove. Look at the output and read “Noise through playback EQ”. Then switch both EQs off. What changes in the signal, and what in the noise? Try the crackle too.

Load starting point

Show solution

The signal stays the same, because both EQs together are flat. The noise gets louder without emphasis: the playback EQ lowers white noise between 20 Hz and 20 kHz by about 3.7 dB. With sound you mainly hear that the hiss is brighter without emphasis. The clicks become rounder and duller through the playback EQ, because their high-frequency parts are cut.
4. FM: 50 or 75 µs?

A 10 kHz sine is transmitted over FM with 50 µs (Europe). Read how much the transmitter boosts and how much the receiver lowers the noise. Then switch to 75 µs (North America).

Load starting point

Show solution

With 50 µs the transmitter boosts 10 kHz by 10.2 dB and the noise drops by 6.5 dB. With 75 µs it is 13.5 dB of boost and 8.1 dB less noise. The price: treble-rich music needs 3.3 dB more headroom at the transmitter, otherwise it is overmodulated. That is why transmitters limit the treble, and with 75 µs they have to do so more.
5. Tape is different

Choose “Tape NAB”. Compare the dashed curves in the frequency response with those of RIAA. What does “Noise through playback EQ” show? Switch to “Tape IEC 15 ips” and “Cassette type I”.

Load starting point

Show solution

Tape works the other way round: the flux on the tape falls towards the treble, and playback raises it again. This makes the tape hiss louder, by 10.7 dB with NAB. With IEC at 15 ips (35 µs) it is only 8.2 dB, with the cassette (120 µs) 16.5 dB. The shorter the time constant, the less treble boost on playback and the less hiss. That is why the cassette needed noise reduction such as Dolby.

Settings

Sound

Level calibration (dB SPL)

MIDI keyboard

Pre- and De-Emphasis (RIAA, FM, CD, Tape)