As already uttered here on kvr, we used a collection of test files to check our Satin plugin against the real world. These files might also be of some value for others.
Here it is, in a zip (a single wav file, 44k1 24bits stereo, 5:40min, feel free to convert to other formats):
http://www.u-he.com/Sascha/_temp/Tape_Test_Tones.zip
Details:
Code: Select all
-----------------------------------
- sin 1k @ -12dB, 10sec (-> adjust to 0dBVU on the target machine)
- dirac (3x: -12, -6, 0)
- square 500Hz, -6dB
- sine ramp up -inf..0dB, 20sec @1kHz
- sine steps (log.) 20..20kHz, 30sec:
- left ch -12dB
- right ch -12dB
- sine sweep (log.) 20..20kHz, 20sec:
- left ch -12dB
- right ch -12dB
- left ch -6dB
- right ch -6dB
- left ch 0dB
- right ch 0dB
- white noise ramp up -inf..0dB, 20sec
- drum track @ -12, -6, 0dB
-----------------------------------This way, we have a dynamic range from -inf to +12dBVU (quite a lot).
After this reference level is set, just apply the entire file in one pass. The test scenario should be complete to inspect frequency response, distortion, dynamics, wow&flutter and crosstalk, hopefully.
If you want, you can make several passes, different speeds, different bias settings (if possible; min/nominal/over).
Have a look at the results on an analyzer, like Voxengo span. Using span, you can set the mode to 'max' and have a frequency response plot. This would also reveal itself in the waveform display envelope, but the FFT is surely better.
Using the static sines and sweeps, you can easily reveal interesting facets of your device, like for instance permanent head magnetism (2nd harmonic, like 2k for 1k input), head bump etc.
For digital emulations, the sine sweeps are great to find out about aliasing.
Bias check:
Some plugins and actual machines offer adjustable bias. Proper biasing reveals itself in two ways: frequency response and THD figures.
Biasing is usually performed with 1k, 10k and 16k sines. 'Optimal' bias has an even spectrum and the least harmonics for a given input. At least in theory. In real life, a compromise must be found, since the least THD comes at a price of diminished treble response.
Should your machine show an increase of treble reponse, it is usually underbiased. This would also result in added distortion (watch the harmonic series).
[In case increasing bias current does not lead to less THD and less treble, the device under test is apparently out of specs, misaligned or otherwise 'broken'.]
Call for presets:
If you have a good recording machine, it'll be nice if we could get some presets. You may try creating these through 'null tests': put the machine's output file on one track, flip its phase, and put the original file on a 2nd track containing Satin on the insert. Adjust the plugin until the output diminishes as much as possible. Then save the setting as preset, along with some info.
Cheers,
Sascha
