Sound level meter Calibration
what is it ?

Calibration – what is it.

‘Calibration’ means many different things to different people and it is very important to be sure that when it is being discussed that everyone means the same thing.

The main uses of the ‘C’ word, in the order that they occur in a production instrument’s life, are:-

Common usage	Correct description	Who does it?
Model Calibration by a National Laboratory	Properly called ‘Pattern Evaluation’	An approved National Laboratory, using IEC 61672 part 2
Initial Calibration during manufacture	Part of the product production testing	The manufacturer
Calibration before and every each usage	Checking overall sensitivity @ 1kHz	The actual user of the instrument, every time he uses it
Annual Calibration	Properly called ‘Routine Verification’	National Laboratory OR an approved independent lab, using IEC 61672 part 3.

The first and the last items on the list, ‘Pattern Evaluation’ and ‘Routine Verification’ have – for the first time – been enshrined in an International standard, namely IEC 61672. Before this standard was produced, anyone could claim compliance with the IEC standards but in some cases the manufacturer made no real attempt to comply. For example, it is possible to buy a sound level meter, typically made in the Far East, that claims compliance and yet it is offered at a LOWER price than the cost of testing it. Such a unit will never comply.

The four items are explained below.

Pattern Evaluation

The object of the exercise is to have an approved National Laboratory fully test the instrument to its claimed performance standards and if it passes, to formally endorse it, when it becomes a “Pattern Approved’ instrument. Typical laboratories in the EU are the German Physikalisch-Technische Bundesanstalt (PTB), the British National Physical Laboratory (NPL) or the French Laboratoire National d’Essais (LNE). Of these, the PTB and NPL are felt to be the most rigorous, but of these two only the PTB currently has an approval program. While each National laboratory has its own way of performing these tests, the following is a typical method.

The manufacturer provides five models to the laboratory, which selects three of these, and returns the other two. Of the three, one is kept as a reference and usually not used again. The second of the three is actually tested, while the third is kept in case the one being tested develops a fault. The Device Under Test (DUT) is then tested against the requirements of the relevant standard both as an electrical device but also as an acoustical device. Electrical testing covers such things as time and frequency weightings, the dynamic span, the response to impulsive signals etc. and is performed by removing the microphone and fitting in its place a ‘dummy microphone’. However, such testing is incomplete and so the device is then placed in an anechoic chamber and tested with a variety of acoustic signals to ensure that the directional response is in compliance and that there are no artefacts of the instrument case that cause the acoustical frequency response to differ significantly from the electrical one. These acoustic tests are true ‘type’ measurements as they are mainly affected by the physical attributes of the case, including the microphone and tend not to change from model to model. Finally, the units is tested over the whole temperature and humidity range required by the standard as well as its immunity to external radiation.

If the whole instrument is found to comply with all the requirements of the standard, the authority will issue a ‘Pattern Approval Certificate’ for that instrument type. Any subsequent changes to the instrument usually require that it be re-tested.

This Pattern Approval process is VERY expensive and for a company making only 250 of one model, $90 or about €80 can be added to the price of each instrument. Because of the high cost and the very long time taken, manufacturers will rarely have every model in their range formally approved, especially those with a very small annual sales volume; but if they do not have a single model with formal approval, it may well be because they are unable to pass the tests. Caveat Emptor!!

‘Routine Verification’

Because a sound level meter or PSEM is used for measurements that may have a legal significance, it is critical that the instrument remains in a near perfect state. Because it is hard to tell if a sound level meter is accurate or not by visual means, the Routine Verification is a simplified test – usually done annually – to check that the performance has not changed, or at least has not drifted outside the specified parameters. Many things, such as the acoustic shape will not change, nor usually will the microphone polar response, so these are not checked, but almost all the electrical parameters are re-tested, even if not as exhaustively as in the original Pattern Approval test.

With the coming of IEC 61672 part 3, the process of Routine Verification is to be – for the first time - being defined, although IEC working group 4 have not yet totally agree on the details of the test. One huge stumbling block is that the working group, being composed mainly of Government owned Test Houses, wishes to have the same tests for both class 1 and class 2 instruments. The result is likely to be that the cost of doing a routine verification on a class 2 instrument is a huge proportion of its original cost, if the test is done by a Public Laboratory. This of course will totally defeat the object of testing as very few users will be prepared to accept such a cost.

Pulsar Instruments – in common with several of the major manufacturers expect to be approved to do routine verification tests, if and when the EU determines what such approvals entail.

Manufacturing Tests

To the buyer, these are very important indeed. The object of the manufacturing testing is to ensure as far as is possible that each and every instrument delivered to a user meets its claims 100%. However, despite what some manufacturers appear to claim, to test for 100% compliance on every unit delivered is in reality not possible. To get 100% compliance, about 4/5th of the Pattern Approval tests must be done on very unit and this will cost in the region of $2,400 or about €2,100 per instrument, assuming it passes the first time. If it fails once and has to be re-tested, the cost naturally escalates. As a Leq meter from Pulsar Instruments or one of the major manufacturers costs less than this, it is clear that 100% testing would more than double the price, resulting in no units being sold.

Most reputable manufacturers use a statistical method of testing to aim for an insignificant number of parameters not being in compliance. For example, all Pulsar integrating meters actually MEASURE Sound Exposure and from this CALCULATE, first SEL, then Leq and then the other integral derivatives. It follows that if the last derivative, say LEP,d is correct, it is almost inconceivable that the metrics from which it is derived are incorrect. A Pattern Approval laboratory cannot do this as they cannot make assumptions about how an instrument works and have to test everything. Another example is the performance over the required temperature range of -10º Celsius to +40º Celsius. If tests on a sample batch of units’ shows that they are all very similar and every one passes; most engineers would accept such batch testing as indicative of the whole production.

Further savings can be made with such matters as the electrical frequency response. If testing shows that 20Hz, 125Hz, 1 kHz, 8 kHz and 20 kHz are in specification, because of the weighting method used, the chance of an error at intermediate frequencies is extremely low indeed. Thus tests are not done at each third octave frequency as they need to be for Pattern Approval and this cuts the number of frequency tests down from about 30 to 6 – a huge saving in both time and money.

At Pulsar Instruments, each instrument type designed after 1995 is tested completely automatically by a computer-controlled system that applies the required electrical signals to each individual unit and reads the results. This not only tests more parameters than would be possible by manual means but also records each and every parameter for future reference. It also has the advantage that Pulsar can set limits even tighter than those required by the IEC standard and thus be reasonably sure that after long usage, the instrument will still meet the claimed specification.

Calibration before and every each usage

In many ways this is the most important calibration. Every time the sound level meter is used, an acoustic calibrator should be used to check the sensitivity at 1 kHz, both before and after the measurement. If it is correct both times, there is a good chance that the measurement will be acceptable, but if these checks are not done, it is not possible to have any idea of the measurement accuracy. The acoustic calibrator should match the performance of the sound level meter; that is a Class 1 calibrator should be used with a class 1 (or Type 1) meter and the calibrator and the meter should be from the same manufacturer. This is very important as with a calibrator from manufacturer ‘A’ and a meter from manufacturer ‘B’, the user has no idea of the microphone correction and very significant errors can result. If manufacturers A and B have mutually agreed their respective microphone correction data, then their products can be used together. In case this ‘correction’ figure is thought to be unimportant, in some cases it can be as high as 1,4dB and that would give rise to huge measurement errors.

Many users with obsolete sound level meters still keep their old calibrators “because they fit the microphone on the new meter”. This is a very foolish policy as not only may the old calibrator not meet current standards – no who made it – but it will probably not match the new sound level meter and huge errors may occur.