DSP Xover project (part 2)

[jbator]

Originally Posted by 30m20

@jbator: Yes, I know how to workaround practically. It is only a question of interest, why Nick wrote this - or in other words: why not correct via positive post-gain?

Because of processing with high slopes you will have some small spikes in signal. You can try to post gain but on some material it will fire clipping notification.

[30m20]

@fluid: then the question is, why there are two gains at all: if we are only allowed to diminuish, this could be done with pre-gain alone.....


@shining31: maybe minidsp has an advantage (only this one!): there might be enough headroom for processing.

But: what is enough? - Noone knows, how much gain the user wants for his deep bass frequencies. So, giving responsibility to the user might not even be a disadvantage.

[Shaun]

Originally Posted by 30m20

@shining31: maybe minidsp has an advantage (only this one!): there might be enough headroom for processing.

If by advantage you mean "dumber", then yes. Such is the risk of putting out a product with many tweakabilities and flexibilities: not everyone (i.e., very few) will appreciate it all. However, Nick's caveat still applies to miniDSP. Gain staging is simply about using the available dynamic range to maximum benefit. Nick was mindful to include physical LED indicators as well as graphical indicators in NUC to alert the user to clipping.

[fluid]

Originally Posted by 30m20

@fluid: then the question is, why there are two gains at all: if we are only allowed to diminuish, this could be done with pre-gain alone.....


@shining31: maybe minidsp has an advantage (only this one!): there might be enough headroom for processing.

But: what is enough? - Noone knows, how much gain the user wants for his deep bass frequencies. So, giving responsibility to the user might not even be a disadvantage.

Maybe to cater for the situation for when the input signal is too low coming in to the Najda. Lots of effects channels have "make up gain" settings. Najda is different in architecture than some other DSP processors and it clips hard so care is needed, the last part you know already.

If the signal is represented as a floating point number in the DSP the headroom is effectively unlimited for audio purposes. If / when it is converted back to integer as it must the gain will have to be reduced to avoid clipping.

[30m20]

@shaun: you did not quote to the end: I think we share a similar opinion about miniDSP, but I expressed it a bit "milder".

Originally Posted by 30m20

But: what is enough? - Noone knows, how much gain the user wants for his deep bass frequencies. So, giving responsibility to the user might not even be a disadvantage.

@fluid: O. K., I was just wondering, but I live with this for years and will continue. It´s not a real problem.