PA Signal Processor (Part 1)
A professional quality stereo audio control system for group and theatre PA use featuring:
• Audio Limiter
• LED Peak Program Meters
• 3-way Active Crossover
• Balanced Line Output
Parts Cost Guide £90 less case
Nowadays it is usual for a rock group to perform on stage with a high power sound system termed the 'PA'. This two part article describes designs for an audio limiter, a peak program meter and a three band active crossover to help exploit the full capabilities of small and medium sized PA rigs. Although the prototypes were built as stand alone pieces of equipment containing all these functions, there is no reason why the individual designs cannot be installed in existing equipment thereby providing a cheaper form of upgrade.
Many smaller PA rigs used for Pub and Club work expand as and when money is available and as a result are often not as efficiently organised as they would have been had the entire sum of money been available at one time. Often, the quest for higher powers will have resulted in several amplifiers and speakers being purchased and run directly in parallel. The next step to change this form of system into a more orderly arrangement is often avoided because of an assumption of great cost (probably true) and a lack of suitable parts or designs. The E&MM Signal Processor is intended to be added to such a system and will generally give a considerable improvement in sound quality, and possibly surprisingly, an increase in apparent volume level. The processor consists of three functional blocks: limiter, peak program meter and active crossover together with a power supply. The function of each block will be covered in some detail before the actual design and construction are described.
Figure 1 shows the block diagram of the stereo prototypes which were built into a 19" rack case. It is intended that this case should be used next to the mixer (or even disco deck), although there is no reason why it should not be used stage side, or why the various functions should not be split up, since each function is self contained. The high impedance input of -14dB sensitivity is connected directly to a level control and thence to a x1 buffer to drive the limiter. The limiter can be disabled as required. The output from the limiter is available directly (PA Slave Output) to drive additional PA systems or tape recorders, or via a three band crossover with individual level controls on each band. A peak program meter (PPM) uses a convenient output from the limiter (although it also may be used on its own) and displays the audio level as a moving dot on a line of light emitting diodes. Two extra switches are included, a mono switch which is very necessary if the signal source should fail in one channel, and a mute switch which simply switches off all the sound outputs without affecting the meter — useful for setting up levels.
The limiter and PPM will be covered in this part, the active crossover, a suitable power supply and an optional balanced line driver in the second and final part next month.
In the circuit diagrams which follow only the right hand channel is drawn. The components for this channel all have a fixed offset numbering starting at 100, whereas those in the left channel start at 200. Some components do not have an offset, these are common to neither channel and are all to do with the power supply or decoupling.
During construction, particularly if the complete unit is to be built, it is suggested that each board (or set for the PPM) is tested before final installation because it is much easier to debug on a bench. Testing procedures for each board are given after the constructional description, although final calibration is left until the end of the second article.
It is strongly suggested that only new and branded components are used in the construction of this equipment since it must be reliable even after a few years on the road.
A Limiter is inserted into the audio path to prevent overload of the PA system caused by sudden peaks in level (e.g. shouting down a microphone). In this way a limiter performs no action upon the audio signal, until the level of that signal reaches a pre-determined threshold. Once the threshold is exceeded the limiter progressively attenuates the signal keeping the output constant at the threshold level. The response of an ideal limiter is shown in Figure 2.
A limiter should not be confused with a compressor, although the action is very similar. A compressor reduces the dynamic range of the signal over the entire range and does not have a defined threshold like the limiter. As a result the effect of a compressor is much less obvious, but generally compressed music is much less exciting to listen to because of the lack of dynamic range. Compression is nearly always used when a record is cut, since the dynamic range of a record is low. As stated, in the PA facility the limiter is used more for protection and does not act on the music if the system is properly set up. Thus a limiter is more suited to the requirements here.
A limiter must be capable of reducing the level of music quickly, so as to prevent the overload; 10ms is usually considered a suitable time. The decay time (time to return the level to its original level) must be reasonably long to help avoid sudden shifts in music level. It is important to note that the attack time should not be less than 10ms because the bass response will be affected.
The effect of a limiter is audible since the music level will dip (the effect known as 'breathing'), but the overall effect is far less objectionable than the distortion generated by overload.
A block diagram for a limiter is shown in Figure 3. The output from the voltage controlled attenuator is fed to a precision rectifier, and thence to an attack/decay time constant. The voltage output of the time constant reduces the signal level if the threshold is exceeded by increasing the attenuation. This system is a feedback control path, essential because the VCA chosen does not have a linear voltage/attenuation characteristic.
In the PA signal processor a stereo limiter is required and the best way to do this is open to some debate. A compound limiter with a common rectifier/time constant is apparently the best way since any gain reduction will occur on both channels, thus preserving the stereo image. However, such a system will not detect overload in one channel if the other is lightly driven because of the combined nature of the control path. The logical development is to provide a peak detector for each channel and develop a common control voltage from this. When additional circuits are added to allow differences in the VCA module to be cancelled out, the circuit becomes very cumbersome. Accordingly a pair of identical but separate limiters were used. This proved to perform as well as the complex limiter in actual use. Obviously, the image problem still exists, but this was not very noticeable; again it should be remembered that the limiter will not operate during normal performance.
Figure 4 shows the circuit of the right hand channel of the limiter. The VCA employs a common and cheap 2N3819 FET. Experiments were done with a new generation VCA integrated circuit (the B&B audio 1537) but this was found to generate immense distortion under severe overload, most probably due to cross-modulation in the input stage. Admittedly, this limiter will distort at the same level, but the distortion is much less objectionable and may be regarded as 'soft'.
The source of the 2N3819 FET is held about 4 volts off ground by the resistor capacitor combination. Now, as the FET control voltage (on the gate) starts to rise above the pinch off level the FET resistance starts to fall, thus attenuating the signal by virtue of R103. Since the FET resistance is determined by the gate/source voltage, it is obvious that by increasing the source voltage the effect of a given gate voltage is reduced. The source voltage is determined by the preset position, and hence the preset serves to set the limiting threshold. It is well known that the distortion produced by an FET can be much reduced by superimposing half the drain voltage on the gate voltage — C102, R104.
The output from the VCA is fed to the x5 buffer to increase the signal level to 0dB, which in turn feeds the following stages and the precision rectifier, built around IC101. The operation of such a rectifier is slightly tricky and it will be more fully understood if it is realised that the positive and negative signal paths are completely different. If the input is negative, then D102 and R112 conduct creating a virtual earth at pin 6 of IC101. This effectively 'earths' the end of R116. The negative signal is also fed to pin 2 of IC101 and thus emerges as a positive signal, of the same amplitude. A positive signal, however, causes D101 and R113 to conduct, hence the input end of R116 goes negative. The positive signal is also routed by R114 and R115 and the resultant summation gives one 'unit' of negative signal since R116 is half R115 plus R114. This negative 'unit' is inverted by IC101a to again give a positive output. Thus both negative and positive inputs give positive outputs with the gain determined by R117. The output from this rectifier is smoothed by the time constant (D103, R120, C106 and R121) and fed to the gate of the FET. The limiter is disabled by shorting the time constant to ground, hence turning the FET hard off.
A bonus is available in that the precision rectifier output is always present and exactly follows the output signal level. As such, the output from the rectifier can be very conveniently used to drive the PPM, thus avoiding the need to provide another rectifier. The output level from the rectifier is reduced by R118 and R119 to a level suitable for the PPM.
The circuit diagram also shows the input buffer and the low pass filter. R102 is included to prevent damage to the IC when the mono switch is used. Note that the level input to the limiter should not exceed 200mV RMS and an attenuator must be used if higher levels are required.
All the components for the stereo limiter and the buffers are mounted on one circuit board (Figure 5). The board is designed so that all the switch functions are available at the front of the board, thus minimising wiring. Note that alternative packages for 2N3819s exist and not all have the same pin layout.
Testing will be more easily accomplished if a scope and signal generator are available, but a meter can be used with less accuracy. Connect the board to a current limited supply and switch on. Check that the power rails are correct on each IC and then connect an audio input. Disable the limiters by linking the appropriate pins on the PCB. Check that the output is present and five times larger than the input. Enable the limiter. Adjust the input level to be larger than 200mV and then alter the presets. At some point it should be noticed that the output starts to be attenuated. Repeat for both limiters. If nothing happens check the output of the rectifiers and the time constant, remembering that a standard meter will alter the time constant. The final setting up of this module will be covered in the last stages, once it is installed.
The peak program meter uses a string of 16 LEDs, of which the bottom 12 are green, followed by one orange and three red. The orange will be set to indicate an output of 0dB, and thus the reds will indicate degrees of overload.
In this PPM only one led is lit at a time rather than the usual bar graph display, where all the leds up to the level point are lit. This means that the colour change that occurs indicating an overload is much more obvious — particularly out of the corner of the eye when attention is focused on the stage. This type of display is termed a 'moving point' for obvious reasons. The PPM uses the output of the precision rectifier in the limiter section, but it is worth noting that the circuit will work with any audio signal, not necessarily rectified; in this case only the positive peaks in the audio will register.
The LED string is driven by a Siemens UAA170L bargraph driver (Figure 6). This was chosen in preference to the LM3914 series because it offered a 16 LED resolution rather than just 10 from the National chip. The integrated circuit is connected according to the application notes, where Vref min and Vref max refer to the minimum display voltage and maximum voltage respectively. R130 determines the LED current, which should not exceed 50mA.
A PPM should have a fast attack time and a slow decay time, and since both are different from those used in the limiter, a different and slightly more complex peak detector is used to achieve an attack time of 5ms and a decay time of 2 seconds. The detector employs a 311 comparator. When the voltage on pin 2 is lower than that on pin 3 the output of the comparator goes low, charging the capacitor by means of TR102 and the attack defining resistor R125. Obviously, when the converse is true the open collector output of the comparator is held high by R123 and nothing further happens. The preset RV126, reduces the proportion of the capacitor voltage to allow calibration of the complete unit.
Three printed circuit boards are used, of which two are identical driver circuits. It is suggested that a socket be used for the driver ICs. The third board houses both sets of LEDs (Figure 7). Assemble the driver boards, noting that the top board uses double-sided PCB pins for supply connections and insert the wire links into the display board, but do not insert the LEDs until the front panel is complete and final assembly is under way. Then, the LEDs should be fitted into the holes on the display board, but not soldered in. The board is then offered up to the front panel and bolted to the pillars. The LEDs are then guided into the relevant holes with a screwdriver, pressed into place and their wires soldered to the board and trimmed. Note that no LED holders were used in the prototype. The bottom driver board should now be bolted to a bracket, thence to the display board, and linked in with 8 pieces of bare wire. Test this board. The upper board may now be mounted in the same way and a ½" 6BA pillar mounted between the two boards to keep them steady. Three pieces of bare wire were used to connect the power to the bottom driver board, the wires being soldered to three double-sided PCB pins inserted into the top board.
Connect to a power supply and switch on. The bottom LED should light. Apply an audio signal and confirm that the display point moves. If nothing happens check the peak detector and if the display performs erratically a short between two drive lines or a broken drive line is nearly always the cause.
Feature by Chris Lare
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!