Posted by John Roberts on December 04, 2003 at 03:44:39:
In Reply to: Re: Graphic EQ Shootout -- specific tests posted by Bink on December 03, 2003 at 23:18:39:
:
Thanks for taking the time to give this some thought, JR. I will try to implement every single test that time permits, given the types of test equipment that folks can commit.
======
cool.
------------
: As far as THD+N goes I hadn't thought of using a parametric EQ and measuring/listening to the resulting noise. That's pretty cool. My neighbors in Oakland would call that method 'ghetto' -- an expression of endearment given to MacGyver-style resourcefulness. Perhaps I'll have an offer of a lab-quality THD+N measurement device and I won't need to use the PEQ cheat. At least one of the local audio engineering houses has such a device. I've still got some people getting back to me about the availability of their gear.
==========
The product output from a proper distortion analyzer would be best. Null tests have a few problems when trying to reveal noise floor issues. First the combination of two channels out of phase will cancel coherent signal, but the noise floor of both units will add together. So be sure to benchmark the noise floor of the reference unit. Also when trying to null out sine waves even small amounts of phase shift will compromise the depth of null. This may actually be useful to parse out phase response differences between different designs.
-------------
: AFAIK, every digital EQ is made with companding. A few digital EQs will be there including Sabine's GRQ3102 and Rane's new DEQ 60L. I will synthesize a pulse CD to test tracking and linearity. What do you mean by 'widely spaced tone bursts or clicks'? If I know exactly what to put on CD then it's much easier. Wouldn't I need an o-scope to watch the results? There's an HP oscilloscope that has been confirmed for the day.
=========
If the DigEQs use companding they may also use pre/de-emphasis. Your two-tone 19-20kHz might also highlight potential IM from HF O/L.
I am more in favor of tests that generate in band audible artifacts. Interpretation of numbers from unfamiliar (and sometimes familiar) measurements requires relying upon someone's judgment. If something sounds nasty it is (another "Roberts" rule?).
Regarding timing I would suggest at least a few hundred milli-Seconds between pulses/bursts to allow the compander time to gain range. Short duration clip like artifacts are often caused by momentary overload while a gain cell is trying to ramp between full gain and unity, low frequency artifacts are also generated due to control voltage feed through, more of problem with cheap VCA/gain cells.
This requires careful listening of processed and unprocessed bursts or clicks as even a clean short burst of a pure tone will sound clicky or thumpy from the on/off gating. Back when I designed companding noise reductions (late '70s) and compressors, I actually designed my own tone burst set. This is one case where listening tests are superior to specifications from conventional measurements.
--------
: What specific 31-band units did you have a hand in designing? Peavey's GEQ model numbers confuse me. CEQ, EQ, GME, Q, QF coupled with 31, 131, 231, 431 and F or FX -- I have a hard time keeping track of each revision. Oh, and I haven't had a single participant offer a Peavey EQ yet. Somebody please speak up!
=============
I was not the engineer of record on any of the EQs. When I managed Mixer engineering the EQs were designed by the Analog group. I was already working in product management when I came up with the FLS invention. For that matter, it wasn't even my product category. As any one who knows me will attest, little things like that don't keep me from nosing around the engineering lab and getting a patent here and there (2 of my 7 at PV were conceived after I left engineering). I argued with the actual prod manager of the EQs when FLS was introduced regarding sensitivity and finally got it right later when I got that product category too.
It's been a few years and I can't help you with the all the pre/suf-fixes. They look like maybe more than one series is scrambled. My recollection is that 31 or 131 would be a single 31b EQ, 231 is a dual, 431 is a single (don't ask why "4" cause I don't know). F stands for FLS, and FX was the top of the line, no compromise, LEDs in the sliders, XLR bal, etc. unit. If I was still at PV marketing these I'd bust a gut to get a unit into a fair unbiased test. However I realize that even if it stomps the higher priced competition it will never overcome the stigma of the brand.
----------------
: As far as wishing someone had done this years ago I'm fairly certain each manufacturer kept tabs on the competition in a similar manner but it required the internet (and LAB) to achieve the kind of critical mass for users to take the ball and run with it as a group. I was very impressed by Dan Mortensen's strictly-defined PNW AES subwoofer shootout and by other fun or sales-based shootouts held around the world but they all seem to involve speakers. There's no denying speakers are sexier and will draw a bigger crowd. Testing line-level devices is more of a 'plugger' gig. We'll still have a fun day and I hope we will establish a template for future myth-busting (or myth-confirming) get-togethers organized by end users.
: Thanks again for the notes!
: -Bink
============
Such unbiased and well executed shootouts are an opportunity to separate the wheat from the chaff... However at this point I can't get very excited about even a positive outcome. Graphic EQs are mature (read obsolete) technology. Feedback killers will get better and in-ear monitors cheaper to the point where 1/3 octaves will become moot. Of course somebody will sell a few between now and then :-)
JR