A valuable tool for testing loudspeakers.
TONE BURST testing is a technique which is rapidly gaining acceptance in a wide variety of applications. Typical applications are in testing of hydrophones, signal-to-noise in telephone channels, reverberation chamber testing and in the determination of peak distortion in loudspeakers. With loudspeakers, tone burst testing has the further advantage that the speakers may be tested with their maximum peak power level whilst keeping the average sound output level low enough to not annoy the neighbors -- a considerable advantage indeed.
Some time ago our audio consultants, Louis Challis and Associates, asked us to build them a tone-burst generator and the resulting instrument has been used by them ever since wi th much success. The tone-burst test has been mentioned in several speaker reviews and, as a result, many people have asked for constructional details of this instrument.
-----------
MEASURED PERFORMANCE
TONE BURST GENERATOR.
On Time Cycles.
Off Time Cycles
Frequency Response 3 Hz - 300 kHz
Distortion 3 V input at 1 kHz Input Level Maximum Nominal range Input Impedance Output Noise Voltage with no input
Power Supply Current 2,4,8 or 16 2,4,8 or 16
+0 -3 dB
<0.02% 3 V RMS 100 mV to 1 V 47 k
<25 pV 4 mA
-------------
Fig. 1. Circuit diagram.
POWER RAILS OF IC2, IC3, AND IC4 NOT SHOWN PIN 16 OF IC2 IS + 6V PIN 8 OF IC2 IS - 6V PIN 14 OF IC3 AND 4 IS + 6V PIN 7 OF IC3 AND 4 IS - 6V PIN 7 AND 15 OF IC2 ARE RESET PINS AND -6V
Fig. 4. How to add a potentiometer to the generator for burst-on- tone operation. That is the generator gives a continuous tone level with tone bursts of higher amplitude at intervals.
TO PIN 9 ON IC4 47k LIN TO PINS 2,3 4 P. OF IC4 REMOVE LINK TO OV TO OV.
--------------
PARTS LIST - ETI 124
R1 Resistor 47 kr., 'MW 5% R2 " 47 IL 'MW 5% R3 " 47 k 'MW 5% R4 " 47 k 'MW 5% R5 47 k 'MW 5% R6 120 iMW 5% R7 120 I/4W 5%
C1 Capacitor 1 UF 25V electro
C2 ' 220 pF ceramic
C3 " 220 pF ceramic
C4 " 33 pF ceramic
C5 " 0.047 i/F polyester
C6 " 0.047 µF polyester
C7 " 0.047 l/F polyester IC1 Integrated Circuit LM 301A IC2 " 4520 ( CMOS) IC3 " 4001 ( CMOS) IC4 " " 4016 ( CMOS) IC5 " " LM301A SW1 Switch 1 pole 4 position rotary SW2 Switch 1 pole 4 position rotary SW3 Switch DPDT Toggle with centre off SW4
Switch DPDT Toggle PC Board ETI 124 8 AA size batteries 2 4- way battery holders and clips
Plastic case
Escutcheon 3 single RCA sockets 2 knobs
--------
4016 4066
QUAD ANALOGUE SWITCH. ON IF CONTROL IS HIGH ON RESISTANCE 4016 TYP 2800 4066 TYP 800
Fig. 2. Component overlay and interconnection diagram.
Note that there are six links on the board, including two under IC4, which should be installed first.
Fig. 3. Pin connections of the ICs used in the generator.
----------
HOW IT WORKS- ETI 124
The input signal is squared by comparator IC1 such that the output of the comparator will be high if the input is above +6 mV, and low if the input signal is below -6 mV. Resistors R2 and R3 provide the necessary positive feedback to cause the IC to act as a compatator. The output of the comparator is connected to both clock lines of IC2.
If the enable line is high these counters ( IC2) will toggle at the input frequency.
IC3/3 and IC3/4 form an RS flip flop where the output must be in either a high or a low state, that is the flip flop has only two stable states. If the output of IC3/3 is high IC2/1 is allowed to clock and, after tho number of input pulses selected by SW1 have been counted, the output from SW1 goes low. This low is coupled to the flip flop by C2 toggling the flip flop, disabling IC2/1 and enabling IC3/2. After the number of cycles, as selected by SW2, have been counted the flip flop is again toggled. IC3/1 and IC3/2 are used to square up the pulses generated by C2 and C3 respectively.
The input signal is also coupled to the output buffer, IC5, by the analogue switch 1C4/1. When this switch is closed (control signal high) the output of the buffer will be the same as the input. When switch IC4/1 is open IC4/2 will be closed and the output will be held at zero. Since these switches are controlled by the flip flop the output will be the required tone burst.
A trigger output is taken from the flip flop to synchronize an oscilloscope if required. A second output is also available from pins 4/i 1 of IC4 which is the reverse of the main output.
Switch SW3 forces the flip flop into either of its two possible states thus allowing continuous tone or no output to be selected as required. In the center position the normal tone burst is obtained.
-------------------
Fig. 5 (a) Top trace - the input tone burst of five cycles. (original
design). (b) Bottom trace - the response of a low-cost speaker at 1 kHz.
Note the reduced amplitude of the first half cycle and that ringing has
added another cycle at the end of the burst. The room reflection can
be seen on the trace after the burst.
Fig. 6. Printed circuit board for the Tone Burst Generator Full size.
142 x 47mm.
Fig. 7. Front panel artwork
DESIGN FEATURES
A tone burst must always be an integral number of cycles. If the burst is switched on or off part way through a cycle then undesirable transients will be produced that will mask the test results. Thus the burst must start and end exactly at the zero-crossing point of the sine wave in the burst.
In the original unit, designed for Louis Challis, preset times can be independently selected for the on and off periods of the burst with the exception that the burst time is automatically modified to give an integral number of cycles. The preselected on/off ratio, however, is independent of the burst frequency.
To give the required control range, six switched ranges as well as a variable control are provided for both the on and off periods. Other features of the original unit are the ability . o start at any point in the cycle as well as the zero crossing point, a phase- inverting switch to select either the positive or the negative half cycle first and an OFF LEVEL control to set a base tone level which is modified when the tone burst occurs. In addition the dc level of the output can be set and a switch is provided to select burst, pure tone or off as required.
When it came to redesigning the unit as a project we decided that many of the features offered by the original design were unnecessary for the user concerned only with testing speakers.
Hence the unit has been redesigned in a greatly simplified form.
Instead of using monostables to generate variable on/off times we now divide the input with a counter to obtain times that remain in the same ratio regardless of input frequency. We settled for the ability to select 2, 4, 8 and 16 cycles for the duration of either period, as this compromise greatly simplifies the circuitry. We still have the switch to select tone, tone burst or off, but the OFF LEVEL control has been deleted. The latter control may quite easily be added, however, as shown in Fig. 4. The output dc level control and the starting- point phase change have also been deleted.
Since we only need half of a CMOS 4016 IC, to give the required output, the other half may be used to give an inverse output if required, that is, the reverse output is on when the other is off and vice versa. This output is not buffered or brought out to the front panel. If it is intended to load this output with less than 47 k it is recommended that a 4066 IC be used instead which will handle loads down to 10 k. For loads of lower impedance than this, a buffer such as is on the normal output should be used.
CONSTRUCTION
As with any project construction is greatly simplified if a printed circuit board is used. However the layout of the unit is not critical and any other suitable method, such as Veroboard or Matrix board may be used if desired.
We strongly recommend that sockets be used for the CMOS ICs, especially if a printed circuit board is not used, as these devices are quite easily damaged when soldering. The use of IC sockets also facilitates later servicing. Also remember that, unlike TTL, all unused inputs of CMOS must be connected to either the positive or negative supply rail.
The plastic box that we used measured 160 x 95 x 50 mm and is very convenient in that the printed circuit may be held in position by sliding it down behind two of the pillars to which the front panel is screwed. The front- panel overlay on the prototype was made from Scotchcal but, as the amount of lettering required is quite small, this may readily be done directly on the panel by hand or with Letraset.
Shielding of the internal wiring is not required providing that the unit is kept away from strong 50 Hz fields. If operation in the vicinity of strong fields cannot be avoided then the unit should be mounted in a diecast box.
USING THE UNIT
The testing of loudspeakers is very difficult indeed and much effort is still being spent to find test methods which will not only give an accurate understanding of the relative effectiveness of the design, but which will be easy to reproduce.
One of the main problems with speaker testing is that the speaker cannot easily be isolated from its environment. For example, reflections from the walls of a room modify the response, seen by a microphone, no matter where the microphone is placed in the room. If one could eliminate reflections then the situation would be improved considerably, and hence the use of anechoic (echo free) chambers for testing speakers. But such chambers are very expensive to build and consequently not readily accessible to the amateur.
A further problem is in assessing the transient power handling capability of the speaker. Speakers will handle far greater peak transient power than is indicated by their RMS power rating.
This is a very important attribute of loudspeakers in handling musical transients. Any attempt to assess this with a sinewave signal may result in the destruction of the speaker due to thermal failure - apart from also being extremely noisy.
---------- How the unit is assembled.
The use of a tone-burst generator minimizes both these problems. How this is achieved is better understood by examination of Fig. 5. This shows on the upper trace a five cycle 1000 Hz burst that is fed to a loudspeaker.
The second trace shows the same burst as picked up by a microphone in front of the speaker. We notice that the burst has been changed by the speaker and an examination of these changes can tell us a lot about the speaker. For example we notice that the first half cycle has not reached full amplitude and this indicates that the speaker would have some difficulty in reproducing high frequency transients.
Next we notice that instead of five cycles there are now at least five and a half. This could mean one of two things. Either there is a speaker/room resonance or, the speaker itself is continuing to vibrate after the original excitation has ceased. Which is it? We can determine this by changing the position of the speaker to see if any change occurs in the shape of the burst, if not it is caused by the speaker itself, and if it does then it is a speaker/room resonance. A speaker that lengthens the burst unduly will sound muddy in that region. Of course the speaker must be examined over its whole range to gain a thorough assessment of performance.
It is of course possible to eliminate room reflections simply by performing the tests outside. However unless one lives in a very quiet area, background noise will introduce problems - and your neighbors are unlikely to appreciate the noise that you will generate.
By varying the off period we can also select a ratio where the room reflection, the oscillation seen after the cessation of the burst, does not interfere with the first few cycles of the burst and the response versus frequency of the speaker may then be assessed from the amplitude of the first half cycles that are stable in amplitude. Thus it is possible to gain an appreciation of the frequency response, transient performance and quality in terms of ringing of a speaker by careful use of the tone-burst technique.
The transient power handling capability of a speaker may be assessed by selecting a fairly long off to on ratio for the burst and by feeding the burst to the speaker via a high- power amplifier. If for example an off to on ratio of 8:1 is used then the peak power will be eight times the average power. Thus the speaker may safely be driven to a peak level where a predetermined amount of distortion occurs. Take care that the amplifier is capable of providing the peak power required.
Of course a tone-burst generator may be used for a wide range of testing. We have mainly concentrated in this article on its application to the testing of loudspeakers.
The circuitry of the tone- burst generator may easily be modified for us P as a 'silent switch' for A/B speaker testing. The method of doing this is shown on pages 106 and 107.
--------------