Read JBL TiK Master Sub Service Manual online
MLS benefits from good tolerance of poor SNR, so a high level stimulus is not required. MLS
analysis uses a Pseudo Random Binary Sequence (PRBS) which is pink-filtered so as to
adequately stimulate the LF end of the spectrum without over-driving HF drivers. The stimulus
sounds just like pink noise.
The stimulus is ‘played’ over the system, and is synchronously recorded via the measurement
microphone into memory on the DSP card.
At the first (primary) microphone position, the main system and the sub system are independently
stimulated, and their levels and magnitude responses are measured (see Level Matching). A gain
factor is then calculated for the sub system to match the level of the main system. The rest of the
auto-EQ process is carried out with both the main system and the sub system active.
Several sequences are repeated, and the result averaged. The recorded sequence is then
correlated with the original stimulus in such a way as to derive an impulse response for the
system. With knowledge of the parameters of the pink filter etc, an impulse response for the room
can be calculated from which the magnitude response parameters can be obtained using an FFT
with a window function. The linear-frequency domain FFT data is converted to log-spaced 'bands'
for EQ response fitting.
At the first (primary) microphone position, the magnitude response is evaluated (see below) firstly
with normal polarity, and then with reverse polarity. The responses are then compared over the
range to be EQ’d, and whichever polarity setting has fewer points beyond the Max Boost / Max
Cut limits is used for the rest of the auto-EQ process, and is retained for normal operation. If both
responses lie within this range, the response with the smallest absolute error is chosen.
The remaining microphone positions are then measured in a similar way, and their magnitude
The resulting response is then inverted to obtain a required EQ response, and smoothed to take
out noise and narrow peaks or dips, using moving average filtering. A target curve is then
subtracted to compensate (where possible) for the LF response of the Main speakers. The
parameters of a bank of Parametric equalisers are then manipulated to give a best-fit to the
required response. The mapping of the required response to EQ parameters is done by looking
for the frequencies with the worst error, assigning an EQ with the appropriate amount to boost to
cancel the error at that frequency, then widening the bandwidth until the error within the range of
the filter changes sign (when the skirts of the EQ bell tend to go over the target line). The errors
are then reassessed and further filters assigned etc until all filters are used up.
audio sample rate. This is not a limiting factor for room equalisation, which rarely extends beyond
level of the sub is accurately matched to the level of the main system so that the combined LF
response curve lies within the limits set by the maximum allowed eq cut and boost . The process
is carried out in 4 steps: -
output gain so as to achieve a reference level at the Mic (-12dBFS). This gain (
MainGainRef )is saved for later calculations, as is the difference between the final
meter reading and the desired target level(MainRefDiff). An FFT of the main system is
room response, is measured (MainZeroLine ).
before, to achieve a reference level at the Mic. This gain (SubGainRef) is stored, as is
the difference between the final meter reading and the desired target
level(SubRefDiff). An FFT is then carried out on the Sub and the average level is
difference between the level measured at the input (Fed from main amp “Sub Out”)
and the output level is stored as MainSysGain and is a measure of the position of the
Volume control on the users amplifier. The signal is band-limited to ensure that an
amplifier with a band-limited sub output will read the same as an amplifier that does
not band-limit it’s sub output.
amplifier, (fig. 3) the mean sub level is at the same amplitude as the reference level in
the main response.