Are playback only cassette tape noise reduction systems like DNR offer better subjective performance than Dolby’s noise reduction systems?
By: Ringo Bones
Playback only noise reduction systems for use in cassette tape playback and reducing hiss when listening to FM radio broadcasts became widely available during the early 1980s. One of the most popular of these was the one based on the LM1894 IC used on what is referred to as a dynamic noise reducer or DNR. DNR units based on these types of ICs were praised for providing better subjective noise reduction results – sound quality wise – when compared to the then ubiquitous Dolby B noise reduction system used in entry level hi-fi cassette tape decks. But first here’s a discussion on how the DNR works.
According to its National semiconductor datasheet, the LM1894 is a 14-pin dual inline package stereo noise reduction IC for use with audio playback systems. This dynamic noise reduction – or DNR – system is non-complementary, meaning it does not require encoded source material. A noise reduction circuit using this IC is compatible with virtually all prerecorded tapes and FM broadcasts. Psychoacoustic masking and an adaptive bandwidth scheme allow the LM1894 IC based DNR to achieve 10dB of noise reduction. A DNR circuit based on the LM1894 IC can save circuit board space and cost because of the few additional passive components required. Described as a non complementary and single-ended noise reduction system that is compatible with all prerecorded tapes and FM broadcasts, has a 10dB CCIR / ARM weighed effective noise reduction, has a 4.5 volts to 18 volts power supply range that’s compatible for installation in automotive applications and a 1 volt RMS input overload.
The dynamic noise reduction system is a low pass filter that has a variable bandwidth of 1,000 Hz to 30,000 Hz, depending on music spectrum. The DNR system operates on three principles of psychoacoustics. The first of which is that white noise can mask pure tones. The total noise energy required to mask a pure tone must equal the energy of the tone itself. Within certain limits, the wider the bandwidth of the masking noise about the tone, the lower the noise amplitude need be. As long as the total energy of the noise is equal to or greater than the energy of the tone, the tone will be inaudible. This principle may be turned around; when music is present, it is capable of masking the noise in the same bandwidth. Next, the ear cannot detect distortion lasting less than 1 millisecond. On a transient basis, if distortion occurs that is less than 1 millisecond in duration, the ear acts as an integrator and is unable to detect it. Because of this, signals of sufficient energy to mask the noise allows the LM1894 DNR IC to open its operating bandwidth to 90-percent of its value in less than 1milliseconds. Reducing the bandwidth to within 10-percent of its minimum value is done in about 60-milliseconds: long enough to allow the ambience of the music to pass through, but not so long as to allow the noise floor to become audible. And lastly, reducing the audio bandwidth reduces the audibility of the noise. Audibility of noise is dependent on the noise spectrum or how the noise energy is distributed with frequency. Depending on the tape and the recorder equalization, tape noise spectrum may be slightly rolled off with frequency. Depending on the tape and the recorder equalization, tape noise spectrum may be slightly rolled off with frequency on a per octave basis. The ear’s sensitivity, on the other hand, greatly increases between 2,000 Hz and 10,000 Hz. Noise in this region is extremely audible. The DNR system’s low pass filters makes the noise less audible by filtering steady-state hiss in this frequency. The low frequency spectrum of the music will not appreciably open the DNR’s bandwidth, thus hiss in the 2,000 Hz to 10,000 Hz part of the audio spectrum is not heard.
The LM1894 IC is often used in conjunction with the 18-pin dual inline package LM3915 LED bar-graph driver array driver IC as a bandwidth indicator. The resulting set-up cosmetically resembles the dbx 563X Hiss Reducer that was sold during the late 1980s and early 1990s. Sadly, by around 1986, the LM1894 DNR IC was becoming hard to find in most DIY electronic stores – especially if you’ll going to buy just one IC instead of “a minimum quantity of 1,000 chips” as often advertized in popular electronics magazines at the time. While building my own DNR unit back in 1987, I was forced to replicate the transconductance amplifier portion of the LM1894 IC using discrete FETs.
In use, playback only single-ended DNR type noise reduction systems provide subjectively better sound quality results in comparison to Dolby noise reduction systems like Dolby B and Dolby C because most DNRs operate their cutoff frequency in the in the 1,000 Hz to 2,000 Hz part of the audio system, while Dolby B operates in the 500 Hz zone, thus making DNRs sound less “dull” than an over-aggressive Dolby B. But in actual use, Dolby encoded prerecorded tapes – in either Dolby B or Dolby C encoding – tends to perform well in a typical DNR unit in comparison to tapes recorded without any Dolby noise reduction because a Dolby encoded cassette tape has a substantial treble boost than ones recorded without Dolby.
