Originally published by Paul Stenning in Electronics - The Maplin Magazine, July 1996
The dynamic range of an audio signal is the difference between the quietest discernible part of the signal and the loudest undistorted part of the signal. It is normally expressed in decibels (dB).
With Compact Disc (CD) the possible dynamic range is often quoted at around 110dB. In practice this sort of range is rarely used on commercial recordings, but the recording medium no longer imposes a restriction to the usable range.
The dynamic range available with cassette recording depends on the quality of the tape and the equipment. The limit is usually imposed by tape hiss at the lower end and saturation at the higher end. Noise reduction systems such those designed and licensed by Dolby offer a useful increase in the dynamic range achievable, often at the expense of some higher frequency signals at lower levels.
With a cheap ferric cassette and no noise reduction system, the dynamic range can be 20dB or possibly even lower. At the other end of the scale, a high quality metal tape in a hi-fi cassette deck with Dolby S noise reduction can achieve 70dB. A typical domestic system with Dolby B will give a dynamic range of perhaps 50dB.
The range available with vinyl records depends primarily on the spacing between the grooves. With wider spacing it is possible to achieve a higher recording level because the groove width itself can be greater. The lower limit is dictated by surface noise, which is dependent on the quality of the vinyl and pressing equipment used.
This explains the popularity of 12" singles in nightclubs - the wide groove spacing allows a higher recording level to be achieved and hence an increased dynamic range, up to about 60dB. The sound quality obtained from such records on a good quality playing system is often not far removed from that on CD.
On the other hand, albums where the record company have attempted to cram as much music as possible onto each side, have a very limited dynamic range, sometimes as low as 10dB. This is sometimes referred to as "groove jamming" and was common on the compilation albums issued by companies like K-Tel and Ronco in the seventies and early eighties.
Dynamic Range Processor
The Dynamic Range Processor presented here allows the dynamic range of a signal to be increased (expansion) or decreased (compression) by a variable amount up to 3:1. The unit was developed for use when recording music from various sources.
For example, when recording CD's onto cassette for use in a car it is helpful to use some compression so that the quieter sections are not drowned out by the engine noise. Also it is useful to employ some expansion when recording tracks from poor quality records or pre-recorded cassettes. Suggestions for using this unit and processing music from various sources are given later.
Obviously a system such as this could never be truthfully described as Hi-Fi, since any form of audio effect unit, by definition, "distorts" the original signal. However the unit has been subjected to extended listening tests and the design has optimised to produce the best possible results on a wide range of music.
The circuit is based on the SSM2120 dynamic range processor IC. This is available from various suppliers including Maplin and RS, and costs about £10. The IC contains two level detectors and two voltage controlled amplifiers, making it ideal for processing stereo signals. The circuit diagram for the two channels is virtually identical, so this discussion will concentrate on the left channel.
The level detection circuits contain a wide dynamic full-wave rectifier, logging circuit and a unipolar drive amplifier. These circuits will accurately detect the input signal level over a 100dB range from 30nA to 3mA peak-to-peak.
Referring to the block diagram of the level detector, the REC-IN input is an AC virtual ground. When applying signals a DC blocking capacitor (C-IN) is used since REC-IN has a DC potential of about 2.1V above ground. The value of the input resistor is set to give a +/- 1.5mA peak signal. For +/- 15V operation this corresponds to 10K.
The full wave rectifier is followed by a logging diode whose pair transistor has a fixed collector current set by R-REF. For 15V operation R-REF is 1M5. The signal on the LOG-AV output is the log of the average of the absolute value of the input current. The value of C-AV affects the attack and decay times of the circuit.
The LOG-AV signal is buffered by a unipolar amplifier stage. RL provides an emitter load for the output transistor while the 39K/1K feedback resistors give a gain of 40.
Referring back to the main circuit diagram, the relation to the components shown on the block diagram can be clearly seen. U2:B is the level detector section if the SSM2120. The value selected for C-AV (C2) is 2.2uF, which was found by experimentation to give the best overall performance on a wide range of music. The level detector was found to respond to lower frequencies rather more than higher frequencies, so the input resistor (R17) was shunted by a high-pass filter (C7 and R18) which begins to take effect above about 1.5KHz. The values of these components were also established by experimentation and listening tests.
The level detector output passes to the dynamic range control (RV1) via a filter circuit (R6 and C3). This filter smooths the sharp attack edges of the signal to reduce the audible effects when a high level of expansion is used.
SW1, and SW3 on the right channel, are contained in one double pole component. When the switch is open, each channel operates independently, and when it is closed the two channels are controlled together. D1 and D2 ensure that the higher level detector output at that instant reaches the base of Q1. The transistor is configured as an emitter follower, and it's base-emitter voltage drop compensates for the voltage drop in the diodes (to within 100mV). The final control signal is buffered by U3:A.
Audio Signal Path
The audio signal into the unit is buffered by U1:A. This is preceded by a DC blocking capacitor (C1) to remove any DC offset on the input. The value of C1 together with R1 is set to give a low frequency roll-off, with a -3dB point at about 30Hz. Allowing lower frequency signals through at a high level can cause an unpleasant pumping effect on signals with a high bass content.
The input to the voltage controlled amplifier (VCA) section of the SSM2120 (U2:A) is a virtual earth. The audio voltage signal is converted to a current by R15, while R11 and C4 ensure stability. The current output from U2:A is converted back to a voltage signal by U1:B. C5 ensures stability, while C6 is a DC blocking component.
The current into the CFT input of U2:B is adjusted by RV2 for minimum distortion. This operates by trimming out internal voltage offsets, and the setting method is described later.
The VCA has two control inputs. Increasing the voltage on the + control input causes an increase in gain, while increasing the voltage on the - control input decreases the gain. SW2 (ganged with SW4 on right channel) selects the appropriate input depending upon whether expansion or compression is required. On the prototype a three position rotary switch was used with the centre position connected to neither input to give a "flat" setting. Both control inputs must be connected to ground by resistors not exceeding 200R. R16 sets the amount of compression or expansion obtained with RV1 at maximum, the value of 2K2 giving about 3:1 which in practice is ample.
In a fixed range system R16 would be connected directly back to the output of the level detector (pin 3 of U2:B), the value being selected to give the desired level of compression or expansion. The additional components are necessary to allow a smooth manual adjustment of the compression or expansion level.
For clarity the power connections to U2 are shown separately on the power supply circuit diagram. R37 is a biasing component for the VCA sections, and sets the output current.
The circuit requires a supply of +/-15V at about 50mA. This is derived from a 15V transformer with the usual rectification and smoothing components. The voltage is regulated by a pair of standard three-pin regulators (U4 and U5). Although the 78L15 and 79L15 TO92 cased 100mA devices would be adequate, it was felt that standard TO220 cased 1A devices would be better able to dissipate 600mW of heat. Indeed the power supply could power two circuits if separate compression and expansion arrangements were required for a record/playback system. No heatsinking is required.
The whole circuit with the exception of the transformer and switches is constructed on a single sided PCB. The dual track pot (RV1/RV3) should have PCB mounting pins otherwise it will need to be mounted with short pieces of stiff wire.
Due to the cost, a socket is recommended for U2. If you do not have a 22 pin socket, use a 14 pin socket and an 8 pin socket next to each other. Do not fit U2 into the socket until the power supply has been checked. The completed PCB is then fitted into a suitable case and wired up as shown. A single DPDT toggle switch is used for SW1 and SW3, and a single 4P3W rotary switch is used for SW2 and SW4.
Ensure the mains connections are adequately insulated - if necessary add some insulation tape while carrying out these tests. With U2 removed, connect the unit to the mains and switch on. Connect the negative lead of a meter to a suitable ground point such as the wire form J15. With the positive meter lead connected to pin 4 of U1 the reading should be 15V +/- 0.5V, and on pin 11 there should be -15V +/- 0.5V.
If this is OK, switch off and insert U2. Set presets RV2 and RV4 to the centre position. Switch on again and check the voltage at pins 1, 7, 8 and 14 of U1. These should all be between +0.5V and -0.5V. With RV1/RV3 set fully anti-clockwise the voltages on pins 1 and 7 of U3 should be within the same range. If these voltage checks are OK, the unit can be tested with an audio signal.
Connect the inputs of the unit to a suitable audio source such as a cassette deck, and connect the outputs to an amplifier and speakers.
Set SW2/SW4 (compression/expansion) to the centre "off" position, SW1/SW3 (combined/separate) to the open (separate) position and RV1/RV3 (range control) to minimum.
With the unit switched on you should be able to play the tape through the amplifier as usual, with no degradation in audio quality. If you have a high quality system you may notice a slight reduction in the bass response due to the input filter.
Now switch the unit to compression (SW2/SW4 anti-clockwise) and gradually turn the range control clockwise. The first thing you will probably notice is an apparent decrease in volume, but if you turn up the volume control on the amplifier it will not sound quite the same. It should sound thinner, probably with a reduction in the bass. Vocals will be less pronounced and will tend to merge into the music. Any tape noise or record surface will be more noticeable, particularly between tracks. The higher the range control is set, the more pronounced these effects will be. The centre off position of SW2/SW4 allows a quick comparison.
Set the range control back to minimum and select expansion. Increasing the range control should now have roughly the opposite effect to that described above. Any bass beat will be noticeably more pronounced, and a pumping effect may be noticeable if the control is set above about half way. Vocals should be more pronounced and noise between tracks should be reduced.
The effect of the separate/combined switch (SW1/SW3) is less apparent. For testing it is best to choose a track which has a pronounced difference between the left and right tracks. An early stereo Beatles recording is ideal, since the music is on one channel and the vocals are on the other. Select compression and turn the range up to about half way. Now set the amplifier balance control towards the channel that has the music. With SW1/SW3 set to "separate" the music level should be reasonably consistent, and with the switch set to "combined" the music level should reduce noticeably whenever vocals are present.
Sounds are very difficult to describe adequately in words, but the above descriptions should be clear enough for you to establish if your unit is working correctly.
If you have access to an audio signal generator and an oscilloscope, the VCA's can be trimmed for minimum distortion. Connect the signal generator to both inputs and the oscilloscope to one output. Set the signal generator to give a 10Hz sine wave (or the lowest frequency available if it does not go this low) with an amplitude of about 3V peak-to-peak (1V RMS). Set the unit to expansion mode and adjust the range control to about mid-way.
There will probably be some noticeable distortion on the waveform displayed on the 'scope, at the zero crossing points. Adjust the relevant preset to minimise this distortion. This will occur when the zero-crossing distortion is equal on the rising and falling slopes of the waveform. Repeat this for the other channel. The unit will not normally be operating at such a low frequency so the problem will not be significant in practice even though it may look quite bad on the 'scope.
If you do not have access to suitable test equipment, the presets should be left in the centre positions.
Use of this unit obviously depends on the original source material and what you are trying to achieve. I will give a few examples here.
My main use for the unit has been to attempt to improve poor quality recordings of music. Obviously there are limits to what can be achieved - to quote an old cliche "you can't make a silk purse out of a sows ear". If you use the range control in moderation you should be able to achieve useful improvements without adding too many undesirable effects.
Compilation albums suffering from groove jamming, and lower quality pre-recorded cassettes, can benefit from some expansion. This will tend to bring out the bass and vocals, and reduce the effects of noise. Normally turning the control up to about one third is sufficient, you are only trying to compensate for the compression that was used when the record or cassette was recorded.
Setting the expansion too high tends to cause a couple of noticeable effects. The most apparent will be a pumping sound on the bass - difficult to describe but easy to hear. A more subtle effect will be a fluttering variation in level, similar to that obtained from a poor quality tape suffering from drop-outs. If either of these effects are audible the dynamic range should be turned down so that they are not objectionable.
Expansion can also improve the sound from tapes recorded on systems with automatic level controls, although sometimes these level controls are so drastic that little can be done to eliminate their effects.
It is also worth trying some expansion on MW and LW radio stations, providing you have good reception without too much interference and noise. A vast amount of compression is used on AM broadcasting to ensure there is sufficient signal to mask the background noise. Most independent local radio stations broadcasting on FM also use compression in varying degrees, as well as limiting to cope with presenters who are not too proficient with the level controls. This explains why the BBC national stations often sound so much better on decent equipment.
If you have a recording that is distorted due to excessive recording level, a little compression can reduce the audible effects significantly. Compression also helps reduce the effects of drop-outs on cheaper cassettes, and can make bad scratches on records less pronounced.
Some live recordings - particularly those which are genuinely live and have not been processed in the studio - can sound echoy and hollow. A little compression will calm the reverberations and place the vocalist back with the instrumentalists where he belongs.
When CD was first launched there was a rush by record companies to release old material in this new format. Most of these were superb, but in some cases they really shouldn't have bothered. I have heard a couple of examples where the master tapes were clearly in poor condition with noticeable tape hiss and drop-outs. In some cases the recording company used an excessive amount of analogue expansion when transferring them to CD, giving audible pumping effects. Such poor CD's are not limited to the budget labels - indeed some of the worse examples come from major record companies who realy should know better. Poor quality CD's like this benefit from some compression to mask the problems, and reduce the dynamic range to the original intended level.
Compression is also useful when recording CD's for playing on a car cassette player. The advantage of some compression is that you won't have to turn up the volume to hear a quieter section over the engine and road noise, only to be deafened by a louder section. Obviously this is more of a problem with the lower quality equipment fitted by car manufacturers.
In most cases the separate/combined switch would be left in the "separate" position so that each channel operates independently. The "combined" position can be used when processing mono recordings or recordings where there is very little difference between the two channels. The "combined" setting is also useful when dealing with tape problems such as drop-outs and distortion which often affect one channel more than the other.
As I stated previously, compression and expansion should be used in moderation. Turning the control up too high will often cause unwanted effects to occur, which may be worse than the problem you are trying to reduce! The centre "off" position on the compression/expansion switch is useful for quick comparisons - you will often find that the unit is having more effect than you thought when you were twiddling with the control.
- SSM2120 Block Diagram
- Circuit Diagram Page 1
- Circuit Diagram Page 2
- Circuit Diagram Page 3
- PCB Artwork
- PCB Component Layout
- Interwiring Diagram
- Front Panel
|Resistors (all fixed resistors 0.25W 5% or better)|
|1||RV1/RV3||4K7 Lin dual pot (PCB mount)|
|2||RV2,RV4||47K vertical preset|
|100n box poly 0.2" pitch|
|4||C2,C8,C10,C16||2u2 63V radial|
|2||C4,C12||2n2 box poly 0.2" pitch|
|2||C6,C14||470n box poly 0.2" pitch|
|2||C7,C15||10n box poly 0.2" pitch|
|2||C17,C18||470u 35V radial|
|2||C19,C20||100u 25V radial|
|1||SW1/SW3||DPDT toggle switch|
|1||SW2/SW4||4 pole 3 way rotary switch|
|1||Transformer 15-0-15V 6VA|
|1||22 pin 0.3" width IC socket|
The word "Dolby" and the various Dolby products named in this article are trademarks of Dolby Laboratories Licensing Corporation.
This project, including all text, images and diagrams, is copyright 1991 - 2003 Paul Stenning. No part of this article may be reproduced in any form without prior written permission from Paul Stenning and WallyWare, inc. All details are believed to be accurate, but no liability can be accepted for any errors.