As you may have read, Phil Graham and I are putting together a curriculum for an advanced measurement and array seminar to be taught in early 2012. I sat down this afternoon to try and pound out some measurements of electrical filter behavior: different high and low pass classes and slopes, combining behavior, phase differences, etc. Since I don't have any DSP lying around I borrowed a dbx DriveRack PA+ from a company I was working for last week. My first step was to double check its electrical response with no filters engaged, and what I saw was so bad I double checked my own interface to make sure I wasn't crazy! This is a fresh program loaded with no settings of any kind: XO, EQ, Delay, etc are all disabled.
You'll probably want to click on that image to blow it up and see clearly... The reasonable looking trace is a hard wire loopback test of my Smaart I/O, which is about 30° laggy and 1dB down at 13Hz; totally acceptable for anything I use it for.
The purple trace is the LOW output of the DriveRack PA+ which, as you can see, has totally wonky phase behavior and is polarity inverted! The green trace is the HIGH output of the DSP which has different phase behavior in the high end (with the same delay time!) and isn't polarity inverted, so the two are about 90° out of phase at 16kHz and 180° out of phase over essentially the rest of the unit's bandwidth. FYI, the word "polarity" doesn't even appear in the manual, as far as I can tell there is no way to invert the polarity of an output, purposefully or not... not like I ever do that when I'm building a loudspeaker preset!
So dbx has managed to value engineer a DSP so hard that its behavior is atrocious compared to even the first DSPs from the early 1990s. I can't use this processor to generate plots for my slides, as the phase behavior is just as important as the magnitude to our discussions and this POS warps it by a totally unreal amount. How can a company like dbx let these things out the door? Who is doing their product design?
P.S. This explains the DriveRack PA trace I got back when I was collecting DSP measurements for my article, which was also polarity inverted and also had funny HF behavior. I wrote it off as a cabling error and since I didn't end up using it in the article didn't worry about it. The goddamn Behringer DSP measures WAY better than these things.