Keith's Electronics Blog

I’m interested in faithful audio reproduction on my home stereo, not just sheer loudness; and even if I don’t end up choosing to stay there, I’ve always wanted to start with a flat frequency response — that is, every pitch that’s played back coming out of the speakers (more to the point, reaching the listener) at the same volume.

My Sony TA-E9000ES preamp / processor has a white noise generator for calibrating the volume of the surround speakers — it sends “static” to each speaker one at a time going around the room. Because I can hear a distinct difference in the “color” of the static when it goes from my main front speakers to my center channel, I already assumed that my different speakers are not providing the same reproduction of the signal going into them — which means that at least some of them (and probably all of them) are not delivering a flat playback.

While poking around on Behringer’s web site, I came across their discontinued DSP8024 digital equalizer. It offers thirty-one 1/3-octave bands of graphic equalization from 20Hz to 20kHz — but more importantly, it offers real-time analysis of your audio system. Connect a reference microphone and turn on auto-equalization and it plays pink noise through your speakers and adjusts the EQ to give you flat response.

And because it’s discontinued, I figured a used one would be a cheap way to get into real-time analysis and flat frequency response. Within a week or so I had picked one up on eBay with the ECM8000 reference microphone.

After flattening my room response, the sound coming through my DSP8024 is simultaneously absolutely awful and absolutely glorious.

Setting It Up

Immediately upon putting the EQ inline between my preamp and amp, I noticed a significant hiss coming out of my speakers that’s never been there before. When I put the EQ into digital bypass, the hiss is still there; only when I power off the EQ to make it engage its physical bypass relays does the hiss disappear.

Wow.

Oh, and my amplifiers’ pop-suppression relays do suppress power-on / power-off pops perfectly, but the Behringer EQ manual’s statement that

Fail-safe relays have been incorporated into the design of the BEHRINGER ULTRA-CURVE PRO, which automatically and silently [emphasis mine] bypass the unit in the event of power supply disconnection or failure.

is most wildly inaccurate. DO NOT power on or off the EQ with the downstream amplifier on. Holy crap; I only did that once.

So already I wondered whether I had a bad unit, and I hadn’t even started putting music through it yet.

On to the auto-EQ.

I find the DSP8024 operating manual to be one of Behringer’s worst; it took me quite a long time to figure out how to turn on the pink noise generator at the proper signal level and get the auto-EQ function started — which you would think would be pretty simple as it’s one of the core functions of the unit.

The correct order of operations for running auto-equalization — which I’m listing from memory as it would take me literally ten minutes to find all of this scattered randomly throughout the manual even though I’ve read it clear through multiple times — is:

1. Reset the entire EQ to (flat) factory defaults, including disabling the advanced functions like feedback suppression.
2. Double-check that the EQ is set for flat response.

   This is important because the EQ is in the output path of the pink noise generator and the auto-EQ adaptively disables frequency bands in which it doesn’t “hear” anything (if it can’t hear it, it must be outside your system’s ability to reproduce); so a band that you (or a previous owner) set low might get inadvertently suppressed.

3. Connect the reference mike and select it as the real-time analysis (RTA) input source.
4. Set the AUTO-Q target curve to FLAT.

   The auto-EQ can generate EQ settings to make your room sound like any desired equalization curve; but why target anything other than flat? (That’s a rhetorical question, and I can think of answers that pertain to situations other than my own.)

5. Put on ear protection and prepare for a physical assault (at least with my system).
6. Turn on RTA pink noise generation at -20dB.

   Even though the input gain is set for auto, it seemed as though running with pink noise at -48dB resulted in an EQ curve and master volume set much higher than bypass levels. Running the RTA with pink noise at -20dB gave me settings that deliver about the same overall sound volume whether the EQ was bypassed or in the loop. I know this doesn’t make sense and I’m willing to run it again to retest.

7. Start the auto-EQ, watch the on-screen spectrum analysis slowly flatten out as it figures out your room, and listen to the “color” of the noise changing.
8. Shut off the %^#! pink noise.
9. The new equalization curve is automatically transferred to the live EQ settings.

Here’s what it got me:

Yow. Looks like I have some pretty serious room and/or woofer resonance problems in the 80Hz range, and a surprisingly jagged set of needs in the mid and upper frequency ranges.

And the Sound?

I started with Bon Jovi’s Lost Highway because it was close at hand. I had the preamp at a relatively low level, and the bass notes sounded like they were playing out a speaker with a shredded woofer — very buzzy. Wow, wow, wow.

And then I turned up the preamp to louder than normal listening volume. And I just couldn’t stop. Oh. My. Word.

Once I got up to a high enough signal level to swamp the hiss in the background and get past the low-frequency artifacting (more on this in a bit), the sound was unbelievable. I stood there with my jaw dropped switching the EQ in and out of the mix, and I couldn’t believe I’ve been listening to the way the stereo was for all these years with all that music missing. It was unbelievable.

I’ll come back to how it was good and how good it was; but first, I need to address the low-frequency artifacting. The buzzing noise does go away when the equalizer is in digital bypass and it also goes away when the EQ is active but with a flat curve, so it’s something to do with the digital equalization.

The manual warns against too great a difference between the settings of adjacent equalizer bands — that it can cause “overflow” — and obviously I have a pretty extreme difference between neighboring bands in the bass range. I tried smoothing out the lower end; but even after a fair bit of tinkering, I can’t find the point between the curve set by the auto-EQ and a flat response at which the low-end buzzing goes away.

Even if I could locate bass settings that eliminated the buzzing, they would likely be such a compromise that they’d defeat much of the purpose of having the equalizer in the first place. So chalk one up to the inability of the digital signal processing algorithm to handle extreme differences between neighboring bands.

Additionally, this EQ is made to be used with professional equipment running at a +4dBu (1.23V) signal level, not with consumer equipment at -10dBV (.316V). I’m giving it a much lower amplitude input than it’s really made for; and especially as I hear the buzzing only in low-volume portions of the music with the preamp at a low level, I can’t help but think that the DSP algorithm is impacted even more adversely when dealing with low-bit-length samples.

The buzzing does not happen when the same portion of music is played at a higher amplitude, so it’s not just a matter of higher volume in another range aurally masking the artifacting; it’s really about the input amplitude. In fact, I can turn down the digital master EQ output level, turn up the input amplitude, and largely eliminate both the hiss and the buzz at the same output amplitude.

It Sounds Awful, So I Bought Another One

That is to say, I really wanted to know whether the hiss and buzz were a design flaw or a faulty unit. So I bought another one — and when it arrived, it simply would not do auto-EQ. Whuh??? I spent about an hour reading and re-reading the manual to find out why every time I pushed the AUTO-Q button, it just came right back without doing anything.

Well it turns out, the second one was running 1.2B firmware, the first was running 1.1, and the Behringer web site has the 1.3 firmware EPROM image available for free download. Go Behringer! Rather than take the second one back to match the first, I figured I’d go forward and hope against hope that the new firmware might even address some of the DSP issues.

The firmware goes on a 27C256 EPROM, and I didn’t have any on hand (not sure why), but Joel did. He burned me a new EPROM in exchange for a burger and malt, and I was ready to swap.

Obligatory “guts” photo of the DSP8024 main board.

The new EPROM with foil tape protecting the UV erasure window.

I hot-glued the factory PROM into the case, both because I didn’t know what else to do with it and in case someone else ever owns this EQ and is weird about wanting it to be exactly the way it came from the factory (like I am with arcade games). I know the hot glue is at risk of popping off the metal case; in the long term, I may put the PROM in static foam in a baggie and tie it to something.

And the Sound?

Exactly the same as the other. The flaws are design, not damage.

How It Was Good and How Good It Was

I had described the difference in sound to Jeremy, and he and I have similar enough impressions of sound reproduction that he believed me and wanted to hear for himself.

He too was blown away.

It has to be partly the boosts in the mid and treble ranges clarifying the vocalists’ enunciation, bringing detail to the cymbals, and filling out the timbre of mid-range instruments; but it really is as though Jeremy and I had never heard the music before. Of course the correction in the bass range helps with the enjoyment as well.

I let Jeremy run the controls and I handed him CDs to try. Sometimes he’d ask which track I wanted, but most of the time he just picked the same one I had in mind. Sometimes he’d hit the bypass button to take the EQ in and out of the mix; sometimes he’d just wince and couldn’t bring himself to bypass it.

A few remarks from the session:

On “Maxine”: If it makes this sound good, it’s gotta be great! [That's the one song on The Nightfly that Jeremy doesn't love. I love them all. Favorite album ever.]

“Only Over You”: [bypass] Car driving by on the street. [EQ] Now you’re in the car. With the band.

“Foreplay” / “Long Time”: You can hear the organ growl underneath the guitars!

“Jessica”: I just want to listen to the whole album.

“Rosanna”: Wow, it even makes 80s pop music sound good.

And many, many others.

Finally from Jeremy: I just want to sit and listen to all of my music all over again. Also: They should just label this button, “What we meant.”

Me again: On most albums, the difference is so prominent as to make the bypassed signal sound like “swimming pool speakers.” Only very, very rarely does it not make much difference — interestingly enough, specifically including much of Chicago 16.

What Next?

Because of the extra equipment to power on and the extra effort required to get a good sound out of the overall system at normal listening levels, I’m not using the EQ most of the time, so I might as well take it out of the system.

When I move the amps down to the basement, I intend to use +4dbU balanced transmission to get the signal from the preamp to the basement. It seems plausible to me that the EQ will be a lot happier with a higher input amplitude, and I’ll be happy to take the time to calibrate the output level appropriately. I’ll also build a full power sequencer for all the equipment, so I could power up the EQ and automatically wait 5-10 seconds before powering up the amps (to wait for the horrible EQ power-on pop to get over with).

I should try the auto-EQ with my speakers in a different room and/or different speakers in my family room, to see how much of the bass correction is needed because of the speakers and how much because of the room. I want to build new (triamped) speakers anyway; and if that alleviates the need for such dramatic bass correction which in turn eliminates the buzzy artifacting, I’ll be delighted.

I intended the DSP8024 as an entry-level experiment with RTA and auto-EQ, and for that it has served me well. If I’m still not satisfied with the sound in my finished system, I’m not at all opposed to buying a different EQ.

The RTA has interesting potential even where the DSP8024 is not going to be the EQ in the system. The RTA can generate pink noise, fed through a different EQ into the amps, and display the spectrum analysis of what’s received at the reference mike. It could thus be used strictly as a setup instrument to set an external EQ (graphic or parametric) for as flat a response as can be achieved with as many bands as the external EQ has available, then removed from the system.

Jeremy’s already asking about using this setup to fine-tune his car stereo — although we both have doubts about having enough and narrow enough frequency bands on an car EQ (even Jeremy’s) to get the signal as flat as the DSP8024 itself is capable of.

This entry was posted on Monday, June 8th, 2009 at 12:51 pm by Keith Neufeld and is filed under Audio, Repairs. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.