GSM, DECT and ISDN all have different masks for the audio response. Of these DECT has the most relaxed in terms of the distance between the upper and lower limits. TETRA terminals should have a response which ensures that the handsets 'sound right' to the user at each end of the call without any major peaks or dips in the response curve. Therefore the response curve in the table below has been selected for handset, headset and handsfree operation.
It should be noted that the microphone capsule is particularly sensitive to air pockets in the front of it, these should be eliminated by the manufacturing process or at least made in a repeatable manner. A small gap of 0.5mm in the front of the capsule can result in a 15dB peak in the response at around 3KHz. In general sealed-back earpiece transducers are much less sensitive to mechanical variation than microphones.
Initial measurements should be made at the Linear PCM point within the terminal using pure tones for handset and headset units and artificial speech for handsfree.
An end to end test should later be performed using artificial speech to ensure the validity of the previous measurements.
Initial measurements should be made at the Linear PCM point within the terminal at one twelfth-octave intervals for frequencies in the range 100Hz to 8KHz inclusive using a pure tone at a sound pressure level of -4.7dBPa. See [P.57, Table 1] for a list of frequencies.
For sending tests the Mouth Reference Point shall be in accordance with [P.64] and an artificial mouth shall be used [P.51].
For receiving tests a digital signal generator shall be used to generate a signal equivalent to pure tone of -16dBm0 [P.64]. This will need to be scaled appropriately if it is applied at the Linear PCM point within the terminal. The sound pressure will be measured at the Ear Reference Point [P.64, clause 11].
When performing an end to end test using artificial speech. The speech shall be according to [P.50], band limited to 200Hz and 4KHz at a wideband Sound Pressure Level of -4.7dBPa. The speech will be in 3 10 second segments, the first is ignored and the second and third (male and female) used for the measurement. The 1/3 octave filtered long term average signal spectrum is computed. The difference between the 1/3 octave input power and output power is then computed.
The sending frequency response is measured in according to [GSM 03.50, Annex C.3.2] and the receive sensitivity according to [GSM 03.50, Annex C.3.4]. Note that artificial speech is used throughout these tests.
The method of test has to be defined. In general it should be in accordance with [P.38].
| Frequency(Hz) | Upper Limit(dB) | Lower limit(dB) |
|---|---|---|
| 100 | -9 | |
| 200 | 0 | |
| 300 | 0 | -14 |
| 800 | 0 | -10 |
| 2000 | 4 | -8 |
| 3400 | 4 | -11 |
| Note 1: The limits at intermediate frequencies lie on straight lines drawn between the given values on a logarithmic(Hz) - linear(dB) scale | ||
| Note 2: All dB levels are on an arbitrary scale | ||
| Note 3: The sampling rate of the A/D converters and filtering will ensure that the level will not approach the upper limit above about 3KHz | ||
| Frequency(Hz) | Upper Limit(dB) | Lower limit(dB) |
|---|---|---|
| 100 | -10 | |
| 200 | 2 | |
| 300 | -9 | |
| 1000 | -7 | |
| 3400 | -12 | |
| Note 1: The limits at intermediate frequencies lie on straight lines drawn between the given values on a logarithmic(Hz) - linear(dB) scale | ||
| Note 2: All dB levels are on an arbitrary scale | ||
| Note 3: The sampling rate of the D/A converters and filtering will ensure that the level will not approach the upper limit above about 3KHz | ||
Copyright ©2002 Andy Fletcher andy@x31.com