AnsweredAssumed Answered

matching delays between outputs

Question asked by twest820 on Dec 7, 2011
Latest reply on Dec 14, 2011 by twest820

Hi, I've a fairly typical configuration where a stereo input runs through a crossover block. Each of the crossover's outputs goes through a parametric EQ block and then to I2S. This means each of the outputs passes though some number of filters in the crossover and then some some number in its specific parametric EQ, depending on how the blocks are configured. I'm curious if one needs to do anything to avoid introducing skew between the outputs. I'm not finding the latency in samples of first order filters or biquads on a search of the SigmaStudio docs or EngineerZone so I suspect it's negligible.  Am I correct all the filters in each signal path would be calculated all the way through on every sample and hence no skew is introduced?  I'm most interested in the 144x parts but the closest thread I've been able to find only mentions 1701 latency when the core's just passing through data.  That thread hints filters add additional delay.

 

I recognize latency isn't a hard thing to measure but I'm currently in an evaluation and capacity planning phase and hence don't have silicon handy to put under the scope. It would be helpful to have this info at design time.  But if I need to wait 'till verification time to measure the details it's not the end of the world----worst case would seem to be a few extra biquads with b0 = 1 and a1 = b1 = a2 = b2 = 0 would be needed.  Or maybe b0 = a1 = b1 = a2 = b2 = 1 if there's optimization.

 

EDIT: Looks like there'll be splitters involved too as I've need for the output of one of the parameteric EQs to appear on two I2S links.  SigmaStudio allows me to pull a second wire from an output or parametric EQ terminal but then won't let the other end attach to anything.  So it seems a splitter's required to get things hooked up even though the gain will always be unity.

 

SigmaDspRouting.png

Outcomes