Studio Design Sound

1
Studio Design - Sound
2
My Background

• Simon Williamson, Freelance Broadcast Engineer
• www.crashrecordtv.co.uk
• Email simonw_at_crashrecordtv.co.uk
• Operations Supervisor, ITV Central News
  (Abingdon)
• Senior Engineer at BBC TV (London)
• BSc Electronic Electrical Engineering
  (Birmingham University)
• Operational Engineering experience in News,
  Studios Facilities

3
Topics for today

• Basic Acoustics
• Microphones
• Wired
• Radio
• Audio wiring / powering conventions
• Sound Consoles / monitoring / peripherals
• Studio terminology and systems
• Digital audio theory and systems (revision?)

4
Basic Acoustics

• The behaviour of sound waves in an enclosed
  space is quite different to when sound travels
  through open air.
• This can be explained by three external
  factors
• REFLECTION
• DIFFRACTION
• ABSORPTION

5
REFLECTION

• If the dimensions of the reflective surface are
  bigger than the wavelength of the incident sound,
  then a degree of reflection will take place. To
  reflect all audible frequencies, it would need to
  be at least
• 12 metres square!
• The nature of the surface will have a bearing on
  the amount of reflection.
• Hard surfaces lots of reflection
• Soft surfaces sound is absorbed
• But remember that certain broadcasts might need
  reverberation, so you dont want total
  absorption, and electronic processing can help,
  too.

6
Standing Waves

• In a room with two or more parallel surfaces,
  sound could be reflected back and forth
  repeatedly between the walls. At some frequency,
  the distance between the walls would be an exact
  number of half wavelengths. The reflected waves
  sum with the arriving ones to cause Resonance,
  also known as Standing Waves.
• These are undesirable in a recording studio, as
  they create pockets of uneven sound distribution
• Making the walls non-parallel would help, but not
  very practical!

7
DIFFRACTION

• Diffraction is a reflective effect caused by a
  standalone object, or a gap in a larger
  surface. Again the wavelength of the sound will
  determine the outcome. For example, a 1 metre
  square board would reflect the short wavelengths
  (high frequencies), but allow the longer
  wavelengths (low frequencies) to diffract around
  it.
• This would create a sound shadow effect, with
  the higher frequencies not being heard in the
  shadow.
• A similar outcome would be experienced close to a
  gap in a reflective surface

8
ABSORPTION

• I mentioned that the composition of the
  reflective surface has a bearing on how sound
  waves travel back towards the source
• Covering a hard surface with cloth or fabric will
  ensure that some frequencies are absorbed
• Creating an uneven surface will cause a
  scattering effect, also known as Diffusion. Such
  a surface could be foam-based with a mottled
  surface.

9
Absorption Coefficients

• Frequency (Hz)
• Material 128 256 512
  1024 2048 4096
• Brick, unpainted 0.024 0.025 0.031
  0.041 0.049 0.07
• Brick, painted 0.012 0.013 0.017
  0.02 0.023 0.025
• Plaster on brick 0.02 - 0.02 - 0.04 -
• Concrete (unpainted) 0.01 0.012 0.016 0.019 0.023
  
• Plaster with air-space 0.02 - 0.1 - 0.04
• Wood boarding - 0.1 - 0.1 -
• Tiles on solid backing 0.01 - 0.01 - 0.02
• Lino on solid floor 0.05 - 0.05 - 0.1 -
• Carpet on boards 0.2 - 0.3 - 0.5 -
• Carpet with underlay 0.11 0.14 0.35 0.42 0.23
• Curtains 0.1 - 0.4 - 0.5 -
• Armchair - 0.35 0.45 0.45 0.5 -
• Adult 0.18 - 0.42 - 0.5 -

10
Studio Acoustics Design

• Three aspects of design need to be considered
  when planning for sound in the Studio environment
• Isolation/Insulation
• Total silence would be impossible to achieve in
  a Studio, but you want to typically achieve
  figures of 20dBA for Radio, 30dBA for TV
  (context living room 40dBA, open office
  45dBAand microphones generate noise at 15dBA!).
  Very expensive solutions involve floating the
  walls and floors on vibration-absorbing blocks,
  within an external shell. If windows are
  necessary, they can be triple-glazed, with the
  internal glass fitted at an angle to minimise
  standing waves. Studio doors can be very heavy,
  with magnetic surrounds to prevent sound leakage
  sometimes an airlock arrangement with two doors
  is used. Remember...external sound can be
  Airborne or Structure-borne.

11
Studio Acoustics Design cont.

• Controlled Absorption/Diffusion
• You could stick fluffy absorbers to your
  Studio walls (something like wire wool), but this
  would have to be impossibly thick (20 foot) to
  be effective at all frequencies. A better
  approach is a combination design, using a
  perforated surface enclosing an absorbent
  material like rock wool. The perforations act as
  a membrane absorber at low frequencies, with the
  rock wool dealing with the high frequencies.

12
Wide-band porous type

• A11

13
Porous Absorbers

• These are made of fibrous material or foam, and
  absorb the energy in the sound waves falling upon
  them by converting the sound energy into heat by
  friction. Porous absorbers are made of fibres or
  of reticulated foam.

14
Commercial Membrane Absorber
15
Studio Acoustics Design cont.

• 3. Reverberation
• This is defined as the rate of decay of a sound
  within a room. The Reverberation Time RT is the
  time taken for a sound to drop 60dB in level i.e.
  a millionth of its original level Cathedrals can
  have RTs of the order of 10 secs, whereas Concert
  Halls built in the 19th Century used materials
  and volumes which led to a time of 2 secs
  indeed orchestral music was then specifically
  composed with this in mind and modern recording
  studios need this reverberation to sound
  authentic.
• The 19th Century physicist Sabine discovered the
  following relationship
• RT 0.16V
• Se
• where V (cubic metres) is the volume of a room
  and Se is the Effective Absorbing Area.
  SeS1a1S2a2.where a refers to Absorption
  Coefficient and S the area of the incident
  surface (wall).

16
Reverberation cont.

• In practical terms, Sabines equation is
  difficult to apply, and it is more usual to
  measure RT using a warbling tone or Pink Noise
  played through a speaker/microphone setup.
• For Broadcast applications, Radio and Music
  Studios will need to be more conscious of
  reverberation. Radio Drama studios may even have
  differently treated rooms, as electronic reverb
  cannot be used for actors getting into the
  part. Television is more forgiving, as the eye
  often overrides the ear.also post-production
  sound dubbing can be utilised to add reverb and
  effects, and a lot of TV Drama is recorded on
  location these days, where good natural sound can
  be recorded.

17
Reverberation Time
60 dB
T60
18
MICROPHONE THEORY

• Microphones belong to a family of devices known
  as Transducers. Their function is to convert one
  form of energy to anotherin this case, sound
  energy into electrical energy.
• Most microphones convert changing air pressure
  (i.e. sound) into a mechanical movement of a
  diaphragmthis is a bit like a loudspeaker in
  reverse! The diaphragm will usually be the face
  of a sealed box arrangement, and can either be
  connected to electrical coils (Moving Coil), or
  be one plate of a condenser-style design
  (Electrostatic).
• The other conversion process involves a pressure
  gradient, and depends upon the diaphragm being
  exposed to sound from two directions. In this
  case the diaphragm might be a narrow strip of
  aluminium foil suspended between the poles of a
  magnet (Ribbon).

19
Polar Response

• Following on from this microphone design
  criteria, it will be seen that the sensitivity of
  a microphone depends upon the direction a sound
  approaches it.
• Most sealed box pressure operation mics will be
  omni-directional.
• Pressure gradient devices (such as a ribbon mic)
  will have a figure-of-eight response.
• Other responses such as cardioid and shotgun can
  be produced by tweaking the design criteria
  during manufacture

20
Microphone Examples

• M58 Reporters Mic
• This is a dynamic mic where the diaphragm is
  attached to a coil of fine wire, located in the
  field of a strong magnet. Air pressure moves the
  diaphragm, moving the coil in the magnetic field,
  which in turn generates a (very small) current.
• Very robust and easy to use (hence the name!).
  Omni-directional. Generally resistant to
  handling noise and distortion caused by very loud
  noises.

21
Microphone Examples cont.

• Sennheiser 416
• This is an example of a condenser type
  microphone. Here the diaphragm is one of two
  plates, the other of which is fixed. A voltage
  bias across these plates sets up a capacitative
  charge effect, which varies as the diaphragm
  moves.
• The 416 uses a tuned RF circuit to amplify the
  signal. A wired voltage must be supplied down
  the cable, but these mics have a wider frequency
  response than M58s
• Shot-gun polar response. Need to be accurately
  pointed.

22
Microphone Examples cont.

• Personal mics (e.g. ECM55)
• Another example of a condenser mic, known as an
  electret. Here the biasing charge for the
  capacitor is stored in the diaphragm/backplate
  during manufacture. Still need a voltage from a
  battery or the cable from the mixer to drive the
  pre-amp.
• ECM55s are very small, and can be attached
  discreetly to presenters or interviewees.
• Omni-directional response. Not good in noisy
  environment. Delicate design, can be damaged
  quite easily.

23
Microphone Examples cont.

• AKG C451
• Another condenser type microphone.
• The cardioid polar response of this mic means it
  picks up sound from a general arc in front, but
  rejects noise behind it.
• Because condenser mics generally dont like being
  handled, the AKG works best in a press conference
  situation, set up on a table stand and directed
  at the participants.

24
RADIO MICROPHONES

• Typical frequencies used in the UK are VHF
  (150MHz) and UHF (600Khz). These are
  licensed by OFCOM, to prevent interference
  between users.
• Typical working range is 10-50 metres. Could
  probably get 200m outdoors, but not reliably.
• Diversity Receivers have dual aerial receive
  paths, which can switch automatically to the
  stronger incoming signal.
• Squelch circuitry is used to filter out unwanted
  interference from other radio frequencies.

25
AUDIO WIRING CONVENTIONS

• Balanced / Unbalanced wiring
• Professional systems use cable with 3 wires two
  for the signal path, one for an outer (earth)
  screen. External interference, like an inductive
  spike caused by machinery, lightning, etc is
  significantly reduced because the amplifier is
  only interested in differences in the signal
  paththe interference is the same in both legs.
  This is known as Common mode rejection.
  Unbalanced wiring uses just a single signal wire
  and an earth connection. Works fine in domestic
  or controlled setupsnot a good idea for Outside
  Broadcast!
• Phantom Powering
• As discussed already, condenser microphones need
  a voltage supplied down the cable from the audio
  mixer to bias the plates. The standard voltage
  used is 48 volts seems quite high, but enables
  long cable runs to be used effectively.
• And this voltage is ignored at the amplifier end
  due to common mode rejection.
• Some condenser mics can use other forms of
  voltage. The electret (e.g. ECM55) can use a
  small battery built into the capsule, and gun
  mics (e.g. 416) can work with 12 volts (known as
  T-Power)

26
TV Sound Desk Parameters

• TV Sound Desk design has remained fairly
  unchanged in the last 30 years, apart from
  processing the signal in digital rather than
  analogue.
• Most desks have the following features
• Programme Chain, which will enable all available
  sources to be equalised and routed to various
  destinations
• Monitoring, using high quality loudspeakers and
  PPM metering,
• and again some sort of routing system
• Auxiliary Facilities, which enable the desk to be
  interfaced to various external systems

27
Programme Chain

• The elements which make up a typical programme
  chain are as follows
• Front-end channels, usually split into Hi-level
  sources (video tape machines, CD players, outside
  broadcasts, etc.) and Low-level sources
  (microphones)
• Group channels, perhaps to accommodate the above
• Master (main) output channel
• All of these would have associated faders,
  equalisation and router switching.

28
Typical Channel facilities
29
Group and Main configuration
30
Sound Desk Monitoring

• Typically the Sound Engineer will want to be able
  hear any source, and check its level. A
  selection of speakers and meters will be used,
  working with various switching matrices. Terms
  to look out for
• PFLstands for Pre Fade Listen. This is a
  means of checking there is a signal before a
  fader...very useful for live working! Sometimes
  referred to as Pre Hear.
• AFLAfter Fade Listen. This is a quality
  check, before the Main fader, used for monitoring
  the effect of EQ (equalisation), or for
  fault-finding analysis.

31
Auxiliary Facilities

• These are extra features which need to be
  available to the Sound Mixer
• Foldback
• In a News Studio, this will be a loudspeaker of
  all sources apart from the Studio microphones. On
  entertainment/music shows, foldback tends to
  carry a backing track or special mix for the
  artistes.
• PA (Public Address)
• Another loudspeaker feed, this time mixed for a
  Studio audience.
• Clean Feed
• This is the Main Output of the Desk minus one of
  the inputs its called Mix Minus in the
  States. Useful for feeding back to a contributor
  in a remote studio or outside broadcast. Also
  used to provide individual feeds for other
  Broadcasters i.e. a sports feed without
  commentary. A Multi-Way Working Matrix is a
  routing system for multiple clean feeds.

32
Auxiliary Facilities cont.

• It is important that the Production team working
  in a Studio and at external locations can
  communicate with one another.
• Talkback
• This usually refers to an open microphone in the
  Control Room, used by the Director or P.A to
  instructions, timed counts, etc. Can be relayed
  to Presenters, Crews, OBs by wired or radio
  means. Doesnt usually end up on a loudspeaker
  due to the swearing! Reverse Talkback is a
  return mic feed at another location (not used
  very often)
• IFB
• Interrupted Feedbackvery confusing term!
  This is usually a Clean Feed which can be
  overridden by a switched version of talkback.
• Obviously these days it is very common for
  mobile and satellite telephones to be used for
  adhoc comms.

33
Telephone Balance Units

• These are very useful devices for utilising
  telephone circuits (wired or mobile) for live
  broadcast contributions. The basic design
  centres around converting a 4 wire system,
  comprising of a send and return path, into the 2
  wire telephony system. In practice, balanced
  audio from the Sound desk will be converted into
  signals suitable for passing down a telephone
  line.
• Need to be able to minimise outgoing side tones
  interfering with the incoming signal, and to be
  able to optimise for different standards used in
  foreign telephone exchanges.
• More recently ISDN circuits have been used to
  improve the bandwidth of the signals travelling
  both ways. Often used to provide two way
  talkback from an OB Truck. A Satellite News
  Gathering (SNG) vehicle might use spare carriers
  within the uplink and downlink frequencies.

34
EQUALISATION

• Most audio equalisation is some kind of band-pass
  or band-limiting filtering. The control on the
  desk will usually have a level setting and
  frequency combined on a dual-gang pot.
• Typically there are 3 flavours
• Shelf controls for low and high frequencies,
  typical slope 6dB/octave and range 18dB
• Mid-frequency (Bell) control
• Low or High pass, with a typical slope
  18dB/octave

35
Equalisation cont.

36
Limiters and Compressors

• Sound limiters are used to automatically prevent
  signals exceeding a pre-determined level.
  Typically they have a protective role, ensuring
  devices such as recorders and transmitters are
  not overloaded
• Compressors are creative limiters. Use the
  same circuitry,
• but can be set up for example to modify the
  dynamics of a music performance in different
  circumstances, such as TV commercials or
  listening on a car radio. Other uses are for DJ
  voiceovers, de-essing and noise reduction

37
Digital Audio

• An analogue audio signal is simply an electronic
  representation of a sound wave the changes in
  air pressure are converted into changes in
  electrical voltage. Such a signal can easily be
  attenuated or distorted, and needs a certain
  amount of bandwidth for accurate transmission.
• A digital audio signal consists of a series of
  measurements of an analogue signal (samples),
  taken at regular intervals. Only the numbers are
  transmitted or recorded. Such a pulse chain can
  be more immune to distortion and generally
  resilient, especially when combined with
  effective error correction. It can also be
  packaged efficiently for economic transmission.
• The next few slides recap sampling as per your
  video lectures.

38
Sampling and Quantising

• The audio signal has to be turned into a digital
  representation.
• The analogue waveform to be sampled.

39

• Being sampled and resolved into 9 discreet levels

The Nyquist Sampling theorem states that the
sampling frequency must be just over two times
bigger than the highest signal frequency
40
Quantising Errors

• The blue line shows the difference between the
  original waveform in RED and the re-constituted
  waveform in BLUE, the quantising errors

41

• By increasing the sampling rate and increasing
  the resolution, the quantising errors are reduced.

42
Sampling systems and ratios

• So whats the best sampling frequency to use for
  audio? Oh, theres about nine of them! As usual
  with technical standards, different countries and
  industries have over the years decided on
  different values
• 48.000 KHz. Widely used in Broadcasting and
  Film, as its well clear of top end audio (20k)
  and it provides a whole number of frames for most
  video formats. You want the audio to be
  synchronised with the video.
• 44.100 KHz. Established as the sampling rate for
  Compact Disc and most commercially released
  digital media.
• 32.000 KHz. Used for NICAM (Stereo) transmission
  in the UK, as 15 KHz is the top analogue signal
  in the terrestrial television system.
• And the computer industry has come up with much
  lower numbers (down to 6 KHz!) to save space on a
  drive or CD ROM.

43
Analogue to Digital Conversion

• Bit Rate. Once youve chosen the sampling rate,
  you need to decide on the number of bits to use.
  A good rule of thumb is one bit for every 6dB of
  dynamic range. Top of the range systems are 24
  bit a good domestic set up would be 16 bit, as
  this offers a pretty wide dynamic range (96dB).
• Anti-Aliasing Filters. High input frequencies,
  even inaudible ones, can generate harmonics. A
  filter at the front end is required

44
Error Correction

• In a typical PCM signal, say 16 bit, an error in
  the LSB (Least Significant Bit) would be
  inaudibleif it happens in the MSB (Most
  Significant Bit) you stand to lose half your
  signal! Thats a big click. Error Protection can
  be factored in by a process of detection,
  followed by either correction or concealment.
• Parity Check
• This involves adding an extra bit to make the
  number of ones either even or odd. Not
  originally very powerful, only really worked for
  NICAM. However powerful mathematical models
  (Reed-Solomon) have led to good results on CD.
• Interpolation
• This is an example of concealment you look at
  the signal before and after the error, and
  replace it with an average value.

45
Error Correction cont.

• Interleaving
• Because errors can be induced at a physical
  location (e.g. tape or disk damage), shuffling
  around the bit words before recording or
  transmission helps to mitigate the effects of
  local dropouts. In the example below, the
  errors on words 4, 7, 11 and 15 are spread out in
  the restored sequence, and can be corrected by
  traditional means (like parity checking).

46
Bit Rate Reduction

• Audio has mirrored video in going down the path
  of compression and redundancy to achieve cheaper
  means of transmission. You take a snapshot of
  the bit-stream and throw away stuff that isnt
  changing much, in the hope that the ear wont
  notice! You can save a bit more by reducing the
  sampling rate too.
• NICAM
• Near Instantaneous Companding Audio Multiplex.
  14 bits and 32k sampling, which achieves a 25
  data rate reduction.
• G.722
• System used for ISDN. 7.5k channel can be sent
  down a 64kbit/sec telephone channel.
• MPEG
• Various flavours, 1-7. Audio variants use
  perceptual coding, again throwing away stuff
  below the threshold of hearing. Layer II has
  compression of 81 and was used in the Musicam
  standard. Layer III uses compression 121.

47
Any Questions?
48
Sound Recap

• You will all have studied Digital Audio as a
  compulsory pre-requisite for this module
• A brief reminder.
• Sound waves are pressure waves travelling through
  the air. You cannot hear sound in a vacuum.
• The range of frequencies heard by the human ear
  is about 20Hz to 20KHz
• The analogue TV system permits a frequency range
  of 50Hz to 15KHz
• The Human ear is not a linear deviceit is more
  sensitive to changes at low levels than those at
  higher levels. The ear recognizes a doubling of
  sound level as a similar increase whether the
  sound is quiet or loud.

49
Threshold of Human Hearing
Threshold of Human Hearing
50
Measurement

• The way the ear perceives changes is reflected in
  the way sound is controlled. Therefore there is a
  need for a way to measure the changes that
  related to the human perception.
• The Decibel.
• The Decibel follows a logarithmic law and
  provides numerical values that can be used to
  quantify changes in sound level.
• It is a unit of comparison, rather than an
  absolute value.
• From the previous chart, the threshold for human
  hearing was defined as 0dB (at 3000Hz), then the
  relative values of some common sounds are as
  follows

51
Relative Sound Levels
52
Line-up Level

• In recording and broadcasting, sound is commonly
  referred to a reference known as Zero Level
• In actual voltage, it is 775mV and is used
  throughout Europe and much of the rest of the
  world to align equipment
• If correctly written, it is shown as dBu.
• If this voltage is fed into a load of 600Ohms it
  will produce a power output of 1mWatt
• This reference is known as a dBm.

53

• Some voltage decibel ratios are

54
Decibel Ratios

• To compare two signal powers the value in dB
  would be
• 10 log (P1/P2) where log is to base10.
• For example, the difference in power between 20
  watts and 10 watts is
• 10 log 3dB
• To compare voltage ratios, the value in dB would
  be
• The ratio between 1.55volts and 0775volts is

55
Dynamic Range

• Real life produces a very large range of sound
  levels requiring careful control if satisfactory
  recordings are to be made. It is not sufficient
  to simply balance all sounds to the same level.
  The dynamic range of the real world would be
  lost.
• You need to consider the conditions in which the
  material will be viewed and heard
• If the programme is to viewed at home, it can be
  seen that the level cannot drop below the level
  of quiet conversation or the sound may be lost to
  outside noises, kids playing, aircraft noise, nor
  can it go so loud as to cause distortion,
  discomfort and annoyance.

56
Levels

• The routing and recording of the audio signals
  require the signals to be correctly maintained
  within the specification of the equipment being
  used.
• There are two main types of metering
• The VU (Volume Unit) meter, and
• The PPM (Peak Programme Meter)
• and both are available as mechanical movement
  meters, LED Bargraph meters and In-Picture
  displays

57
Levels cont.

• In all cases, the decibel is used to indicate the
  relative levels of the audio signal
• The Peak Level is simply the maximum voltage the
  signal reaches. For any piece of equipment, that
  is the maximum acceptable level that can be
  accepted before distortion will occur.
• The peak level is not important to the viewer,
  unless it causes distortion. The viewer is more
  interested in perceived loudness, so loudspeaker
  monitoring is all important

58
Level Meters

• Meters which monitor audio levels are typically
  one of two varieties
• VU (Volume Unit) or PPM (Peak Program Meters).
• Though both perform the same function, they
  accomplish the function in very different
  manners.
• A VU meter displays the average volume level of
  an audio signal.
• A PPM displays the peak volume level of an audio
  signal.
• Analogy The average height of the Himalayan
  Mountains is 18,000 feet (VU), but Mt. Everest's
  peak is 29,000 feet (PPM).
• For a steady state sine wave tone, the difference
  between the average level (VU) and the peak level
  (PPM) is about 3 dB.
• But for a complex audio signal (speech or music),
  the difference between the average level (VU) and
  the peak level (PPM) can be 10 to 12 dB.

59
Meters cont.

• VU meter and PPM also have different ballistics
  (acceleration/deceleration rates). If a 1kHz
  steady state tone is fed into a VU meter, it
  takes 300 milliseconds (0.300 seconds) for the
  meter to stabilize. However, the PPM stabilizes
  within 10 milliseconds (0.010 seconds). As the VU
  meter displays an average volume of the audio
  signal, it must "sample" the audio signal over a
  longer time period than the PPM.
• Because of the crest factor and the difference in
  ballistics, a VU meter and a PPM will display the
  same speech/music audio signal in very different
  ways. Therefore, using a steady state tone to
  line up a VU meter with a PPM is not effective
  unless these differences are taken into
  consideration. Analogy A mini-van (VU) and a
  sport car (PPM) will cruise side by side at a
  constant 60 MPH (steady state tone). But they
  will not cruise side by side if each vehicle
  accelerates to 100 MPH and brakes to 20 MPH many
  times and as quickly as possible (speech/music
  signal). Though both vehicles are performing the
  same function, the location of each vehicle
  (position of each meter indicator) will be very
  different.

60
Meters cont.

• The VU meter closely corresponds to the level
  sensing mechanism of the human ear. It provides a
  useful indication of the subjective loudness of
  different programs and is very useful when
  matching levels between programs. But the VU
  meter does not give an accurate indication of
  peak signal levels because of its relatively slow
  ballistics. In practice, a VU meter will
  under-indicate the peak signal level by 8 to 20
  dB.
• When the VU meter indicates "0" (typically a 4
  dBm level), the PPM should be set to read 20 dB
  below its maximum full scale reading. For
  example, when the VU meter reads "0", the PPM on
  a Sony Beta Cam with a "12" full scale reading
  should be set to read at "-8". Like any rule of
  thumb, this one may vary depending on the actual
  specifications of the products in use.
• Note Meters marked with the symbols "VU" or
  "PPM" may not actually meet the international
  standards for such meters. The best advice is to
  listen critically while recording and not rely
  solely on meter readings.