Audiometer Frequency Tolerance and 

Transducer Response Curve:

How each may contribute to measurement variability

The ANSI S3.6 – 1996 Specification for Audiometers states that the frequency accuracy for each test signal generated by a fixed frequency audiometer be +/- 1% for a type 1 audiometer, +/- 2% for a type 2 and +/- 3% for a type 3, 4 and 5.  

What this means is that when you conduct a pure tone test at a particular frequency (e.g. 1000 Hz), the current ANSI standard allows up to a  1% variation from that frequency (e.g.  1000 Hz +/- 10 Hz) for results to be acceptable for type 1 devices.  However, when conducting tests at 6000 Hz using TDH39 earphones, the permissible tolerance level can and may introduce errors that can seriously affect pure tone results. 

A study by Guelke et al (1977) was conducted to see the effect of frequency tolerance on audiometer accuracy.  They measured the frequency response characteristic of the TDH39 receiver on a type 9A coupler and found a sharp resonance peak at 6000 Hz.  Below is a copy of the frequency response curve obtained by Guelke et al (1977).  Note the exceptional peak in and around the 6000 Hz area.  

One point to keep in mind is that in 1977 the current tolerance according to ANSI S3.6 was 3%.  It was not until 1996 that a reduction of tolerances to +/- 1% (type 1 audiometers) took place in the ANSI S3.6 revision.   

This resonant peak in the TDH39 response curve can introduce a significant amount of measurement variability into pure tone testing.  Because of this resonant peak, any variation around 6000 Hz can cause a significant variation in the dB value from the audiometer.  A standard deviation of 6.2 dB at 6000 Hz has been reported in Guelke et al (1977).  Because the resonance point of the TDH39 earphone is likely to be close to 6000 Hz, it is probable that the precise frequency of the audiometer will affect pure tone results (as well as calibration).  Guelke et al (1977) found that with the current frequency tolerance standard (+/- 3% at that time), a variation of 7 dB at 6000 Hz could occur.  This is to say, if the frequency emitted by the audiometer at 6000 Hz was actually +/- 3% different (i.e. 5820 - 6180 Hz), then a 7 dB difference in SPL can occur due to the resonant peak of the TDH39 receiver.

Audiologists who have their audiometer calibrated using TDH39 earphones need to be aware that a harmonic distortion tolerance of +/- 3% at 6000 Hz may not be sufficient.  An alternative would be to permit only type 1 audiometers to be used for diagnostic purpose or select a more stable (flatter response) transducer during the calibration process.

Guelke at al (1977) suggested using a new receiver, the TDH50 be used in place of the TDH39.  The advantage of the TDH50 receiver is that the resonance at 6000 Hz is damped (not as sharp as the TDH39).  With the damped resonant peak it is no longer necessary to increase the accuracy of the frequency output in the audiometer.  That is, it is less important with the TDH50 that the audiometer frequency at 6000 Hz, be actually 6000 Hz.

Audiologists who have their audiometers calibrated to ANSI S3.6 1989 specification or who are using audiometers classified as type 3 or greater may be allowing measurement variability to enter their test results at 6000 Hz, despite having what they feel to be properly calibrated equipment.  

The take home message here is that Audiologists need to be aware of the frequency response characteristic of the transducer they decide to use. Even though their audiometer may meet the current standard (ANSI S3.6 - 1996), it does not necessarily mean that variation in the signal output will not occur, as indicated in Guelke et al (1977).

For topics related to the calibration of TDH 39 earphones, click on the link below.

Calibration of the TDH 39 earphone

Note:  This site is best viewed using IE 4.X or higher under 800x600 resolution