Metric Halo SpectraFoo - Manual (Page 64)

The Transfer Function
64
You can easily measure your D/A/D chain by looping back your system’s analog output directly into it’s analog
input. Make a measurement using the internally looped-back signal generator as the source and the system’s
analog input as the response. The measured transfer function is the response of your D/A/D chain.
Figure 4.17: Setup for checking your D/A/D chain
If the response is sufficiently flat, you can use the internal loopback method. If it is not, then you must use
an external loopback of the source as the source for the transfer function (since this will remove the D/A/D
chain from the measurement).
One of the strengths of SpectraFoo’s Transfer Function Measurement system is that it allows you to do Mu-
sic-Based Measurement (MBM). This means that the system can automatically apply a threshold to the source
signal to make sure that it is only measuring the response when the source is actually stimulating the SUT.
MBM allows you to make accurate transfer function measurements even when the source signal is not station-
ary (like music) and when the signal “comes and goes” in various frequency bands (like music). MBM builds
the transfer function up over time.
In order to be able to create an accurate measurement of the system response when the test signal is non-
stationary, the response signal and the source signal have to be synchronized in time. If the signals are not
synchronized the relative measurement of response will be wrong. For stationary signals (like pink and white
noise), even if the source and response are somewhat unsynchronized the power measurement will be pretty
good, but the phase measurement is useless.
Figure 4.18: Acoustic test setup
Under normal circumstances the response signal will always be delayed from the source signal. Sometimes
this delay can be very short (for example, if the signal source is external and the SUT is analog and contains no
acoustic elements, the delay can be less than a sample). If the SUT contains acoustic elements (like a speaker),
there is always the acoustic propagation delay. In any case, the delay must be compensated to achieve an
accurate measurement.
When we make a typical acoustic measurement we are measuring all of the elements that are not common
between the source path and the response path. In the case illustrated above, the speaker, speaker amplifier,
acoustic space, measurement microphone, and preamp are not common. The amplifier, speaker and acoustic