Digital Signal Processing: Codes for digital magnetic recording

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting

Departments | Features | ADs | Equipment | Music/Recordings | History

The binary data representing an audio signal can be recorded on tape (or disc) in two ways: either directly, or after frequency modulation. When frequency modulation is used, say, in helical-scan recorders, data can be modulated as they are, usually in a non return-to-zero format (see below). If they are recorded directly, however, say, in stationary-head recorders and compact disc, they have to be transformed to some new code to obtain a recording signal which matches as well as possible the properties of the recording channel.

This code should have a format which allows the highest bit density permitted by the limiting characteristics of the recording channel (frequency response, dropout rate, etc.) to be obtained. Also, its DC content should be eliminated, as magnetic recorders cannot reproduce DC.

Coding of binary data in order to comply with the demands of the recording channel is often referred to as channel coding.

Non-return to zero (NRZ) This code is one of the oldest and best known of all channel codes. Basically, a logic 1 is represented by positive magnetization, and a logic 0 by negative magnetization.A succession of the same logic levels, though, presents no change in the signal, so that there may be a significant low-frequency content, which is undesirable for stationary-head recording.

In helical-scan recording techniques, on the other hand, the data are FM-converted before being recorded, so this property is less important. NRZ is commonly used in such formats as PCM-1600 and the EIAJ format PCM-100 and PCM-10 recorders. Several variations of NRZ also exist for various applications.

Bi-phase Similar to NRZ, but extra transitions are added at the beginning of every data bit interval. As a result, DC content is eliminated and synchronization becomes easier, but the density of signal transitions increases.

This code (and its variants) is also known as Manchester code, and is used in the Video 8 PCM recording format, where bits are modulated as a 2.9 MHz signal for a logic 0 and as 5.8 MHz for a logic 1.

Modified frequency modulation (MFM)

Also called Miller code or delay modulation. Ones are coded with transitions in the middle of the bit cell, isolated zeros are ignored, and between pairs of zeros a transition is inserted. It requires almost the same low bandwidth as NRZ, but has a reduced DC content. The logic needed for decoding is more complicated.

A variation is the so-called modified-modified frequency modulation (M2 FM).

Three-position modulation ( 3PM)

This is a code which permits very high packing densities, but which requires rather complicated hardware. In principle, 3PM code is obtained by dividing the original NRZ data into blocks of three; each block is then converted to a 6-bit 3PM code, which is designed to optimize the maximum and minimum run lengths.

In this way, the minimum possible time between transitions is twice the original (NRZ) clock period, whereas the maximum is six times the original.

For detection, on the other hand, a clock frequency twice that of the original signal is needed, consequently reducing the jitter margin of the system. This clock is normally recovered from the data itself, which have a high harmonic content around the clock frequency.

High-density modulation-1 (HDM-1)

This is a variation upon the 3PM system. The density ratio is the same as 3PM, but clock recovery is easier and the required hard ware simpler. It is proposed by Sony for stationary-head recording.

Eight-to-fourteen modulation (EFM)

This code is used for the compact disc digital audio system. The principle is again similar to 3PM, but each block of eight data bits is converted into 14 channel bits, to which three extra bits are added for merging (synchronization) and low-frequency suppression. In this way, a good compromise is obtained between clock accuracy (and possible detection errors), minimum DC cur rent (in disc systems, low frequencies in the signal give noise in the servo systems), and hardware complexity. Also, this modulation system is very suitable for combination with the error-correction system used in the same format.

EFM+ Based upon the knowledge gained on EFM used in CD players, a new and better performing modulating system EFM+ has been developed for SACD and DVD-Audio. The advantage of EFM+ is the more efficient use of the memory. Instead of eight-to-fourteen modulation, where three merging bits are additionally required for synchronization and low-frequency suppression, making it an eight-to-seventeen modulation in reality, EFM+ is modulating the 8-bit to a 16-bit word including low-frequency suppression.

Keeping in mind that to avoid reading problems by the optical system, a minimum three and maximum 11 subsequent zeros are allowed, 351 words of 16 bits meet the specifications. Only 256 words are required to convert the 8-bit word, leaving us about 95 additional or spare words. Depending on the DC value of the digital signal, different codes from these 'spare' words are used to minimize this DC component. The gain of memory is due to the fact that instead of the conversion from eight-to-seventeen (14 + 3 merging bits), only 16 bits are used for EFM+.

Prev. | Next

Top of Page   All Related Articles    Home

Updated: Monday, 2019-08-26 20:57 PST