Metering
(written in 1997 and first published in “Sound Pro” magazine)Introduction
Meters are to be found on mixing consoles, recorders, effects devices and many other places and so often they show different readings when supposedly displaying the same audio signal! This month some of the reasons for this and then benefits of choosing one type or another will be examined.
Those who have spent many years with one broadcast organisation were protected from alien displays by the local "in house" meter conventions that were used universally. Virtually every audio measuring device was required to be of the standard type if it was intended that anyone took notice of what is was showing! This is no longer true and even national broadcasters find meters of various types in the programme chain and of course others have been faced with obtaining information from a range of meters styles for years!
Most meters are designed to show signal levels but some have other purposes such as phase or spectrum analysis.
Level meters
The prime function of most meters is to show the amplitude of the signal and immediately there is scope for debate! Most meters have "normal" calibrations, often set down in standards documents though local variations are sometimes applied. The main issues that need to be indicated are where the signal lies in relation to five defined points. The levels in the diagram below might be valid for some organisations but must be recognised as being only useful conventions and not absolutes.
The clipping level and the noise floor naturally depend on the equipment being used and are fixed for that system. The clipping level is the point at which the signal will begin to be "squared" and serious distortion occurs and can only be changed by serious re-design work to the equipment involved. With a signal path involving several pieces of equipment, it is of course a function of all the items in the chain as only one device needs to clip before a problem occurs. Metering at different points in the chain is therefore necessary.
The peak and line-up points can be determined for the purpose in hand though many organisations have pre-defined line-up and peak signal levels. The levels of 0 dBu and +8 dBu are conventional in UK broadcasting, corresponding to UK type PPM scale markings of "4" and "8" respectively. Those in Germany and several other places may be familiar with the recording of line-up tone at the peak level of +6 dBu, which reads "0" on the DIN scaled PPM, but not to be confused with "alignment level" which is 9 dB lower! Various US broadcasters have their own specs for meters as have the Nordic countries.
The other threshold is that below which the signal is so low that system noise becomes objectionable. Most meters are unlikely to read such low level signals in their normal modes so although this is an important matter from an engineer's viewpoint, it does not generally become an issue affecting the type of metering and calibration that is used.
Meter scales
The most obvious parameter is that of sensitivity - what deflection is produced for a given signal level. This is also the easiest to "customise" and therefore to cause confusion, particularly when the scaling of the meter is otherwise unchanged from one of the standards. The sensitivity needs to be defined when the meter is reading a sine wave signal and this is where meters can get interesting! The following scalings and sensitivities are commonly found.
They show a bargraph style of meter but similar scales are also used on moving coil meters, though to keep the meters compact, moving coil versions of the DIN and Nordic PPMs normally do not show the -50 or -42 ends of the range and stop at -40 and -36 dB respectively. This is a consequence of the extremely wide range of readings these meters can show, in excess of 50 dB. When that is contrasted with the VU meter's range of just over 23 dB or even the UK PPM at around 28 dB their useful qualities for some purposes are immediately apparent.
Using bargraphs can allow easy reading of a wide dynamic range. If there is an extended quiet section, a UK PPM may hardly display anything at all but this generally doesn't matter as the meter is intended to display signals which are also being monitored by listening and most broadcast material is kept within a limited dynamic range. Admittedly, the original BBC PPM specification helped this slightly by defining a 6 dB gap between the "1" and "2" markings, meaning that it would "twitch" for a signal 2 dB less than a modern version! The UK PPM is therefore not a good choice for a multitrack recording console where for most of the time the metered signals are not listened to individually.
For a given size of display, more range can only be shown at the loss of some resolution and the DIN type of PPM continues to show that a signal is present, even when it is extremely quiet. Despite this, some models of DIN PPM include a 20 dB gain switch which on a "troubled" signal can means the meter displays the system noise floor! The loss of resolution in the upper end of the range can make it more difficult than the UK PPM when setting a 0 dBu or +6 dBu tone level to within a small fraction of a decibel.
For many years, professional analogue audio equipment has had the capability to provide undistorted output levels well in excess of +20 dBu. Whilst many have simply considered this as an emergency headroom which does not need to be included within the range of the meters, the Nordic meter has allowed more of it to be routinely used. This gives the potential to keep the audio signal further away from the noise floor. Each of these scales has its advocates and their arguments are valid. Ultimately there is no alternative but to be familiar with your meter and what it is conveying.
All except the "Digital"scaling were first defined for use on analogue systems where there are no absolute limits to the signal amplitude. Several of the calibrations are defined by international standards but there is far from agreement over the "digital" scaling. Moving into the digital domain with a system which uses all "0"s to represent silence, means that the maximum amplitude will be represented by all "1"s. This maximum level is referred to as 0 dBFS and issues of headroom need to be re-thought. It is far from a standard, but several organisations have determined that A-D and D-A converters should be calibrated so that 0 dBFS corresponds to +18 dBu. They have decided that this will provide an adequate headroom and signals that are compatible with the analogue parts of the chain, though different standards may be appropriate for the analogue outputs of CD players.
Rectifiers
To illustrate the difference in behaviour sending a 1 kHz tone at +4 dBu to a selection of meters would give readings of 0 VU, UK PPM 5, DIN PPM -2, Digital -14 etc. By adding or subtracting the appropriate numbers, each meter can be used to give a consistent reading in dBu of the signal level. However as every user knows, sending a programme signal to the same set of meters will no longer give indications that can be equated by a simple addition or subtraction from the scale numbers.
The meters shown in the chart have three fundamentally different characteristics and these need to be understood if the readings are to be correctly interpreted. PPM stands for PEAK programme meter as this is the type of rectifier used by the meter.
The VU meter uses an AVERAGE rectifier to deliberately avoid the short transients which may not contribute much to the general "volume" of the sound and thereby attempts to give an indication of sound volume. It has been suggested that by increasing the sensitivity of VU meters by 4 dB from that shown in the table above, they will enter the red section at about the same signal level as a UK PPM reaching 6! Whilst this is not a generally recognised common calibration it may sometimes have some merits.
The Digital meter is working on a signal which has been digitised and this process may modify the audio content to some degree. However, the characteristics can then be designed to emulate a PPM or VU meter or several other meter types.
A sine wave has an average value of 0.707 of the peak value and VU meters are calibrated so that when displaying sine waves they give readings which are consistent with those on true RMS meters. Peak meters are also calibrated so that on sine waves they will give similar level readings to an RMS meter. It is worth mentioning in passing that modern test equipment usually provides a "true RMS" type of reading, though none of the meter standards described here use this system.
When the meter is reading the more complex wave found in normal programme material, differences will arise.
The average value of this waveform is something like 0.3 of the peak value so a VU meter will show a much lower reading than a peak meter. This might be an advantage if the meter is intended only to give an indication of the programme volume but a serious problem if the transient "t" is likely to cause something to clip. This neatly illustrates the relative merits of the two types of meter. The choice of which to use therefore depends on what is important and how the signal path will react to transient overloads.
Moving coil VU meters use both electrical and mechanical methods to give an average indication. The inertia of the meter mechanism contributes to the behaviour and most movements are designed to be driven from a specific source impedance, normally 3600 ohms to ensure the correct mechanical damping is attained. Moving coil meters also have specific recovery behaviour after large overloads have been applied and this must be emulated if bargraphs are to give a good match to VU characteristics.
PPM meters give an indication of the peak amplitude by using drive circuits which will capture fast transients then hold them long enough to allow a mechanical meter to display the peak and also allow the operator sufficient time to appreciate what level the peak reached. This is done using a circuit after the rectifier with a very short charge time but a longer discharge time. However, any realisable circuit has a limit as to how short the charging time can be made. In addition, there are some very fast transients that are so short they are unlikely to have any significant effect so we can afford to ignore these. This means that the meter is no longer reading the true peak value of the signal but something else that it is convenient to consider as being the peak level. PPMs therefore fall into the category of "quasi peak reading meters".
Some very simple meters found on low cost equipment or devices where the meter tends only to be used to check line up tone levels and not for programme monitoring (e.g. some video recorders) use only half wave rectifiers. This is fine for measuring tone but audio signals can be asymmetrical. For example a bass drum signal is often much larger in either the positive or negative direction than in the other. For this reason full wave rectification is essential in any meter used for audio programme monitoring. For the same reason, listening to a bass drum can be an interesting test for anyone doubting the importance of absolute phase!
Broadcast / Recording meters
Broadcast transmitters cannot be allowed to be over modulated as this will produce badly distorted signals, probably with excessive side bands that can cause breakthrough into other channels and/or affect the video part of television signals. Peak level control is therefore essential and it is interesting to contrast how this has been done in Europe and other parts of the world.
European broadcasters have not been in the habit of expecting limiters and other automatic level control devices to clamp transients. The preference has been for a skilled operator to observe a PPM and make manual level adjustments, leaving "protection limiters" to catch only those things the operator misses. Elsewhere, automatic control of transmitter level has been a fact of life for a long time and the operator's job is to leave peak level control to pre-set limiting devices and to concentrate only on the programme volume. VU meters are therefore preferred by these broadcasters.
When signals are not broadcast, they are recorded on some type of media and this will influence the most appropriate type of metering. We now take harmonic distortion figures of 0.00x% on analogue consoles and other devices very much for granted. It can be worth remembering that even the best analogue tape systems can have harmonic distortion which is one or more orders greater than this! Analogue tape has highly non-linear magnetisation characteristics and impressively low distortion is achieved by several techniques but non linearity increases as the magnetisation levels increase. This has the effect of providing a form of limiting whereby the recorded amplitude ceases to be proportional to the recorded signal. Unless extreme overloads are applied, this limiting effect is a "soft" one and can pass unnoticed to all but the most finely tuned ears. It is therefore reasonable not to worry too much about metering many short transients as the "tape limiting effect" will cause few seriously audible problems.
Digital systems
A different mechanism comes into play when the recording medium is digital. Some semi-professional grade DAT and MiniDisc systems have a degree of "soft limiting" built in to make them more forgiving of overloads but this necessarily introduces a non-linearity which is unwelcome in the most demanding circles! Overload of truly linear A-D converters produces a very harsh sound. The precise behaviour depends on the type of conversion being used but many converter circuits use a "2s complement" output. The diagram shows a four bit system to make for a clearer illustration but the same principle applies no matter how many bits are used.
The blue line is the input waveform and the horizontal lines show sixteen quantising levels numbered 0 to 15. Four of the samples have levels greater than the maximum level of 15. Some types of converter may simply consider these four samples all to be at maximum level but many may attempt to add one more most significant digit which does not exist and the sample values are therefore output which represent quite different levels. The purple line shows the resulting output wave which is obviously very different from the input signal.
A difficult issue arises when using an analogue console with a digital recorder. Many console manufacturers provide a "peak" mode on the console meters without making any claim that the console meters are any of the standard PPM formats. Indeed, unless it is clearly stated in the console specifications, it is fair to assume that the bargraph meters of most mid price analogue consoles have only an approximation to a standard PPM format! The digital recorder meters will almost certainly display the signal after A-D conversion and it is almost inevitable that this will give a different impression of level from that shown on the console meters.
Most digital recorder meter systems are concerned to avoid digital overloads and normally use an LED, LCD, gas plasma or other electronic display rather than a moving coil meter movement. This means that they can be made to respond in highly controlled ways.
To the right is the same signal envelope as earlier but now magnified to show a shorter length of time in more detail. Some more very fast transients ("spikes") can now be seen, though these would be too narrow to be shown on even a fast attacking analogue PPM. It is interesting to notice that these spikes are also not symmetrical and might even be audio "defects" rather than genuine programme but they must still be metered.
With a normal 48 Kbit/sec digital system, the sampling circuit "captures" a segment of audio roughly every 21 microseconds. Each of these high level samples is converted and some of these "spikes" will get digitised and produce large digital values. Because the meters on digital equipment "hold" the highest reading for a short while, a spike which lasted for only one sample can cause a high reading which may be displayed for a second or more. This tends to make the digital meter read even higher than the console's analogue peak meter.
Some early digital multitrack recorders (when very few mixing consoles were digital!) had the option to remove their meters from the machine and be placed on the console and reconnected via an extender cable. This gave a useful insight into the levels being handled by the digital machine but is less ne cessary now more consoles are digital and include digital meter systems. Adopting the "safe" approach of being guided by the highest reading meter will certainly help to avoid clipping but may also result in signals being recorded at a lower level than is absolutely necessary. This therefore places the programme nearer to the noise floor which, especially on sixteen bit systems, is not as silent as one might at first assume! As always, the ear is too important a tool to be ignored!
Phase meters
Phase meters now come in so many varieties that they almost deserve a separate article to themselves so this can only be a brief summary! The essence of a stereo signal is that the left and right channels contain different information. The more different the left and right signals - the wider the stereo image.
The narrowest possible image is when left and right channel carry identical signals i.e. double mono. The widest image is when the right channel carries a polarity inverted inversion of the left channel - i.e. the stereo signal is "out of phase" mono. This wide signal is completely non compatible with mono listening as in mono the left and right add to produce silence! Phase meters are there to give some indication of how wide the stereo image might be and therefore how much of a problem it can be for mono listeners.
UK broadcast has traditionally relied on the having a pair of PPMs, often sharing a common scale, to read the level of mono sum and the difference signal. Simple double mono reads as a large difference ("M") and no sum ("S") signal. Out of phase double mono reads as a large difference ("S") and no sum ("M") signal. For normal stereo signals, the mono sum should remain a few dB greater than the difference signal for the vast majority of the time to ensure a reasonable image.
Single needle phase meters can use a similar technique but have a centre zero point. When the sum signal is greater the needle moves right of centre indicating that the in phase material is dominant. Deflection to left of centre indicates that out of phase material is predominant and some action may need to be taken. This type of meter can give clear phase indications, even when the signal levels vary over a considerable range. The M/S PPM can give little information once the signal level drops.
Phase oscilloscopes and LCD systems are popular in some quarters and use the sum of left and right signals (L+R) to provide vertical deflection and the difference between them (L-R) to give horizontal deflection (or vice versa, depending on the brand). On tones a constant line can be seen at varying angles which indicates the relative levels of the left and right signals. If only the signal is only on the left or right, the indication is a line pointing at 45 degrees toward the left or right. When the line broadens to an ellipse, a phase shift between left and right is indicated. An extreme case is when there is tone on left and right but with a 90 degree phase shift which displays as a circle. Normal "in-phase" stereo shows a "ball" pattern. The image width is indicated by the width of the ball. For most of the time, broadcast signals show a predominantly vertical band. If one channel is polarity reversed to create a problem signal, the image becomes wide with little height.
A further item of information, particularly valuable with digital signals is if there is a DC offset. This displaces the indication up or down the vertical axis. Highly reverberant programme material may produce an almost circular "ball" pattern. Whatever type of phase meter is available, do not rely on it but use ears! An infamous case when this was not done is of a baseball match encoded at the OB for Dolby ProLogic Surround. Signals routed to the surround tend to produce high levels of "out of phase" information in the encoded two channel signal. The sound supervisor / director decided it would be exciting to place a lot of the crowd noise in the surround and these crowd effects formed the opening of the programme. Someone at the studio centre noticed their phase meter showing a large "out of phase" component and assumed that the OB links had a phase reversal which they "corrected" by throwing a phase reverse switch. The meter then looked better until the commentator spoke - panned to dead centre. By this time, the OB link had been well established and the meter was no longer being observed. Sadly the signal was not being critically listened to either as the broadcast continued with the commentator as out of phase mono for those listening in stereo, perhaps sitting somewhere towards the Dolby rear but was not heard at all by those listening in mono!
Keep an open mind!
The familiar styles of meter are handy reference points, but many "non standard" meters also have a useful role to play. For example, the simple "overload" LED or more advanced channel LED bargraph on the console input channel can be informative, but only if you know what it is reading. This means manufacturers must clearly show from where in the signal path it is driven. For example if it is pre-equaliser it might not warn that clipping caused by large amounts of lift in the equaliser was imminent! If it is post equaliser, it might not warn of input stage overload when the equaliser has significant amounts of cut!
One form of meter that is seeing increased use is the "loudness" meter. Loudness is a complex issue which reflects the humans ear's perception of sound level which varies with frequency. Loudness meters therefore use frequency weighting methods to process audio levels so that the meter is no longer a simple display of level. Companies such as Dorroughs have been offering this type of meter for many years and they use a multi colour display to give information not only on the signal loudness, but also peak and average levels. Many of us brought up on UK or DIN systems have been somewhat reluctant to embrace new systems whilst our traditional PPMs etc. have been so universal. Digital systems have highlighted the limitations of some existing analogue meter systems and this should encourage us to look at alternative meters with a more open mind and not only for digital systems.
All material is copyright PHM © 2004.
P H M (P H Music) :
Ramsbottom : UK
tel: +44 (0)7799 621954
email: info@phmusic.co.uk