Here's the history of the sample rate issue. In
this thread, the problem with high sampling rates was identified. This was with the 2x8 unit, which has a fixed-point processor and was using a plugin having a 96 kHz sample rate. With IIR filters, the higher the sample rate, the greater the potential inaccuracy of the filter response at very low frequencies. This is due to precision considerations in the filter coefficients internal to the DSP chip. In general, with all else equal, floating-point arithmetic has the potential to work better than fixed-point. On one hand, the 96 kHz sample rate of the HD version makes the potential for low-frequency problems greater than the non-HD 2x4 units. On the other hand, the use of 32-bit floating-point processing in the HD makes the potential for low-frequency problems less. The actual behavior at low frequencies of the HD units hasn't been published.
This potential problem can come into play when using e.g. monster low-tuned subs, where you might protect them from excessive driver excursion below the tuning frequency by applying a high-pass filter with a cutoff frequency of around 15 Hz. I spoke to a miniDSP tech via email about this and told him I'd like to see data of the HD unit taken at the analog outputs using a fourth-order Butterworth high-pass filter with a 15 Hz cutoff frequency. We'll see if they do it.
Regarding the original problem reported in the post linked above, miniDSP used a 48 kHz plugin and declared the problem fixed. The response of the original poster to this claim can be found
here.
In the final analysis, I can neither recommend nor recommend against any of the 2x4 units. If I were to see data of the 2x4 HD at low frequencies using a suitable filter and the data looked good, that's what I'd recommend. Sorry I can't be more specific.
