Some background information on the
whys and wherefores.
Test and Design
Test and Measurement hardware:
HP 8903B Audio Analyzer, 20 Hz to 100 kHz
HP 4274A LCR Meter, 100 Hz to 100 kHz
HP 3314A Function Generator, millie Hz to 20
Tektronix 465 oscilloscope.
EMu 1212M Soundcard.
E-Mu 1212M Soundcard
My audio web pages include a large number of FFT
graphs, many of which show substantial 50 Hz mains supply and harmonics
as well as pickup from the computer and monitors power supplies.
Many of the tests are conducted in less than
ideal lab conditions on an open bench in front of my computer. These
tests are pragmatic and reflect the reality of many DIY constructors who
do not have access to the sorts of equipment or facilities of large Hi-Fi
manufactures or review magazines. Where possible I include a picture of
the test setup (flying wires and all) to demonstrate the basic setup and
to show that I am not doing anything special that any competent DIY
constructor cannot do themselves.
For the FFTs I plot the full range of the 24bit
sound-card resolution to show signals down at the -140 dB range which is
normally considered inaudible. I readily note that below -120 dB much of
the harmonic data is as much influenced by the sound card as it is from
the unit under test which is which I include test system loop-back plots
as well. See my E-Mu
1212M Soundcard FFT Tests.
The basic noise floor, or the resolution of the
24 bit converters, is discernable and I differentiate between noise and
interference such as the mains harmonics and the switcher pickup.
Despite these extraneous signals I believe that
the low level information still has meaning and not only can give a
flavour of how the unit is operating but also gives clues as to other
issues which might not otherwise be apparent from refined data.
One example is the effect of poor grounding and
this shows up in one series of tests on the
Zeus Single Ended MOSFET Preamp / Line Driver as a large symmetrical
lift around the fundament. As this peaks at -120 dB most standard
measurement setups would not show it.
Another example is the inter-modulation effects
where there are apparent signals created between a fundament and it's
harmonics, not just between two strong tones, in the
Zeus 6C33C PP Line Driver tests.
Given that long serious and often heated debate
goes on about such things as capacitor distortion and cable effects
(much of which is not apparently measurable with standard audio tests)
these things which do show up will I believe have an effect on the sonic
signature of the system.
So I present my FFT graphs "warts n' all".
Susan - November