Equipment
  • Software: Latest edition of Arta
  • Mic: Behringer ECM 8000 (calibrated)
  • Sound card: Audioprobe Spartan Cue 110
  • Laptop (Win7 64-bit)
  • Hypex DLCP (sounds cleaner than miniDSP)
  • MiniDSP 2x8 used only during development and for feasibility studies.
  • Yamaha CD-S700

miniDSP 2x8 Mesurements:

miniDSP 2x8 serves me well during speaker developement. I have no major issues with its sound other than a slight hiss. So I was curious how it would perform during some simple lab tests.
For those, miniDSP was fed by a balanced signal on ch2 from my sound card and the output was measured on the unbalanced ch8. This reflects my use model of the system (balanced in, unbalanced out). All filters and PEQs were bypassed.

With approx. 350W, the output impedances of the balanced and unbalanced outputs are rather on the high side.

Where applicable, the measurements were taken in dual channel mode. The spectrum analyzer mode was set to a sampling freq. of 96KHz and harmonic distortion was recorded at -3dBFS.

In the two graphs above, no higher output swing could be reached due to the usage of the unbalanced output. But as expected, the intermodulation distortion components rise with the excitation voltage.

Multitone Testing:

The graph to the left shows the response to a multitone excitation. The tones are 1/3 octave spaced sine signals.
Multitone testing is a rather novel, although established method to look for nonlinearities in electronic and electroacoustic systems. It is also used to test data reduction / compression algorithms such as mp3 and the like.

Traditional measurement techniques like swept sine, THD, DIN / SMPTE two-tone intermodulation distortion measurements are limited in revealing non-linear behavior due to the simplicity of the excitation signal. Besides many other advantages, however, multitone signals are able to approximate the frequency domain energy distribution of typical program material such as music.
As far as I understand the state of affairs, optimization and standardization of excitation signals / patterns are still ongoing.

In Arta you can only look at the complete spectrum, which includes both, the excitation signal and possible system non-linearities. It would be beneficial if the former could be subtracted (since it is a known parameter) and thus the distortion be extracted and plotted against frequency as demonstrated e.g. by Wolfgang Klippel.
Anyway, I think this measurement makes sense and I plan to use it more often.

Last updated 03-Feb-2019

Conclusion:

miniDSP 2x8 is a complex board. For that and the price tag I would say that the results are not bad at all. However, I am nearly convinced that a well designed and dedicated active analogue cross-over would do better in terms of harmonic and intermodulation distortion. There is always the compromise of best possible signal quality vs. flexibility.