Home dbx User Forum dbx Archive Threads DriverRack PA & DriveRack PA+ PA General Discussion

Digital latency questions

Mike McNanyMike McNany Posts: 22
edited March 2007 in PA General Discussion
It was mentioned in another post that someone had tested using SMAART a drpa equipped system and had found a 7.5ms digital latency delay on the low output when using a LR24dB low pass filter.

Questions on the original measurement stated.
a. Was the measurement performed with a 2x4 or 2x3 drpa setup?
b. Was there no delay on the tops using a LR24dB high pass filter?

Last, could someone with a drpa, Oscope, & some time please run a latency delay check on the unit? Possibly using the other available filters, too?

IF I can use another filter without the latency, fine. IF NOT, for my speaker system, the drpa has just lost 50% of it's usefullness for me and why I chose it with it's delay settings limitation known.

Mike McNany


  • AxisplayerAxisplayer Posts: 6
    edited March 2006
    Unless the group delay is constant between HP and LP filters, you will also have phase problems. Even though the filters have their own phase responses the phase difference between the two output signals is the same at all frequencies. Accordingly, each filter section has the same group delay.

    If the group delay is the same, the only difference is the alignment delay from drivers. 10MS per band should still be enough for your LS808's since the group delay is constant. A measured delay is to be expected on output from the DRPA, but it should measure the same for HP and LP and be a wash.

    Someone correct me if I am wrong here, but I think I am right from what I have read.
  • GadgetGadget Posts: 4,915
    Actually, it was a 260... The document is here:
    You will see that the LF output gets pushed back as the order of crossover increases, but the HF delay remains the same @1.63ms.

    It was Mikey Kovach :idea: , and he intends to do all the different xovers when time permits...and I agree the DRPA has too many limitations... :(
  • Someone at DBX needs to comment on this and clear the air soon.
  • GadgetGadget Posts: 4,915
    I have made the powers tha be know of this issue..... they are \"contemplating the question..... am told.......


    *Note* if you wish to contact or even BITCH to DBX you wil get a whole lot more satisfaction by contacting Tech support dirrectly...
    (801) 566-8800
    [email protected]

    You will most likley get Mike... he is ALWAYS very helpful, and if he doesn't know the answer he gets in contact with those who do... The answers aren't always what we want to hear... but hey...
  • one of the comments at the bottom of the linked word document page did say that this was a natural part of the crossover's operation? Its possible that this is not a digital processing artifact but indeed an unavoidable delay within the filter algorithm.
    To determine if the crossover is working properly would very simply involve summing the outputs electrically (say using a two way crossover, hook the pin 3's together and use 2 off 5k resistors, 1 from each of the pin 2's. Hook a signal generator up to the input to the DRPA. with no eq or other effects and with the crossover set as a 2 way with 0db gain on each and equal crossover points, compare the input signal to the voltage at the join between the 2 resistors. If all's well then the gain at the summation point won't change as the frequency is swept through the crossover point, just the phase might (assuming Linkwitz-Riley X-over, Butterworth will show a 3db gain at the crossover point I believe).
    That procedure can be used to verify that analogue crossovers are working correctly, and should work similarly for digital crossovers. in general, crossovers are not phase neutral (they do change the absolute phase at any frequency) but they should be phase coherent (the phase at the crossover point and for an octave or two either side should be similar). Our ears cannot detect absolute phase but they can clearly detect if two things are out of phase (there's no or less sound volume).
    I suspect that the 7 or so ms delay will put the two speakers in phase, the only effect on the sound will be on transients which is an unavoidable effect of an inline crossover. The 7 ms would be the equivalent of 0ms of delay for the situation being discussed, so the rest of the post-limiter delay is still available.
    I suspect there are ways of making a perfect crossover with no delay between speakers if you had access to the entire program material and could render a track to different speakers beforehand, or at least with a 100 or so ms of pre-processing, but it would be useless for live music.

  • MikeyMikey Posts: 82
    Wow! I didn't realize that this post was here.

    First, let me explain that the timing issues are a result of latency AND group delay. Latency is just the time it takes for the signal to be processed through the DSP.

    Group delay is quite different. Group delay is the \"timing (frequency dependant or phase) offset\" that creates the cancellations required to generate a filter. In this case, that filter is the crossover HPF or LPF.

    Remember high school physics... and the that cute redhead that used to ...anyway, in high school physics, we studied waves that were in phase, out of phase and somewhere in between. Regardless of the phase angle (timing offset), they just added amplitude at any given point in time-add wave one to wave two. Perfectly in phase gave a nice boost to the wave and perfectly out of phase (reverse polarity) gave an output that was essentially zero if the frequencies and initial phase were the same. That is the nature of a filter/crossover-add things that are at screwy phase/timing offsets until you get the filter that you need.

    Confused? Now that I know there is some interest in this, I will try to put together a white paper on crossovers, delays and stuff. It might take a few weeks, but I promise to get to it.

    FWIW, I have only made one contact to dbx, but if they sponsored a workshop for me to go over DriveRack configurations, would anyone attend? I've been asking this question for some time, but would really like to get about 20 or so folks into Indianapolis or some other city to do this. Indy is cheap to get into, so I figure it's a good place to start. I already have a place in mind and they are happy to sponsor the event if dbx will help.

Sign In or Register to comment.