Digital active crossover listening tests

Recently I've participated in some listening tests of various active crossovers over two separate days in two systems.

We compared
Behringer Ultradrive DCX
Behringer Ultradrive DCX with the Pilgrim audio mods
MiniDSP
DEQX

We did two tests:
1. digital conversions
2. digital active crossovers

We used an instant switching level match box that switches both inputs and outputs. For the first test, we let the units simply act as digital converters of an analogue input - ADC then DAC. We compared to a loop through wire.

For the second, we set up with the same settings and measured to ensure we were comparing level matched and with the filters doing the same thing. It's quite time consuming to set up an event like this, so it was not the last word on perfect accuracy. We did not match to within 0.1 db but we took a decent stab at matching.

We didn't do the test blind. While this may disappoint some, but I felt that using instant switching would be revealing enough. We did this test for our own curiosity, but no one had a point to prove. In a previous informal comparison, I found that was enough even with levels obviously not matched. No one in that test could hear a difference with the digital conversions vs a loop through wire, not even one person there who was certain a difference would be heard.

Our extended tests did show differences, but I'd call them subtle. At certain times, I noticed a difference in imaging where there was a small shift. The instant switch during a sustained note is quite revealing. MiniDSP appeared to have increased treble detail compared to DCX, despite mreasurements showing that it rolls off more above 20k. The measurements from what I recall didn't seem to reveal why that might be the case.

The full write up is here in my blog:
Red Spade Audio: Active crossover listening tests
and there are more comments in the linked SNA thread.

My findings if I can sum it up in a nutshell?
MiniDSP is the first one you want to try out of these.

DEQX sadly didn't quite get a fair trial, we had time limits and something wasn't quite right.
 
As a potential MiniDSP customer, this was VERY interesting. Thanks to everyone involved!

I think I still need a little convincing that the 48khz sample rate used internally isn't the limiting factor for me, demonstrated by the sharp roll off above 20khz. I know that we can't hear that high, but I have trouble with the idea that parts of the high definition source material we buy is missing!

In subjective terms at least, did you notice any degradation of the sound using the digital volume control of the MiniDSP?
 
Last edited:
This is just perfect timing-thank you! I'm ready to ditch my noisy DCX and buy MiniDsp. I have some questions which may be a bit off topic but would appreciate advise.
I can buy two miniDSP's to get me 8 channel's = 4 way stereo. I can go either analogue or digital directly from my behringer SRC 2496 to the miniDsp. Digital would mean one less ad/da conversion, but I then need 2 Minidigi's-to get 8 channels out of the 2 Minidsp's. I'd also have the issue of AES/EBU 110ohm out to 75ohm rca coax/toslink in. This mismatch may be worse than analog as I've read it can add jitter/reflections-17' cable.

Rather than continue w/a good tube preamp to the xover, I can go direct and use my Harmon Kardon receiver as a vol control (it has 8 in's and 8 preamp out's), unbalanced and surprisingly good specs for this use. I want to defeat the hiss issue w/the DCX and understand this is best accomplished by feeding a proper level signal in to the MiniDSP w/the vol control after it. I would appreciate opinions on how best to set this up/if I'm going in the right direction.
 
Hi there,

It was my modified DCX that we took along to these listening tests,, and to be honest I had expectations that the modified DCX, with all its upgrades, would out perform stock DCX and miniDSP.. In actual fact they pretty much all sounded the same!! (to me anyway..)

The only obvious thing we proved was to stay away from digital attenuation!!

Cheers,
M
 
G'day Murphy, I recognise you by the Murphy and the avatar! It was quite novel to see a DCX that has a remote!

We measured more top end on the DCX, MiniDSP rolled off above 20k quite sharply, something many audiophiles would be bothered about. But have they had a hearing test? I tried a quick test online and I made it to 19k, not bad for a 35 year old. I'm tempted to blame those crappy PC speakers! (Imagine the worst you can get, then make it even cheaper).

If anything I felt the MiniDSP had just a touch more treble detail, even though the measurements we took didn't indicate that this should be the case.

AES - you might run into a problem there. My situation is the opposite. I can't run SPDIF into DCX as the level is far too high, although last time I tried it, I had it going into DEQ first.

BTW, I use DCX (2 of us use a stock version), and don't find the noise to be an issue, although some are more fussy or have different results. I often have the Plasma running as well and it makes more noise - sadly it is a bit of a pest. You can easily miss that kind of thing in a retail store with too much noise going on all the time.

One thing that did come out of the 2 days. Sometimes to pick subtle differences, you actually have to deliberately listen for something specific. Keith was quite good at this, who hosted the first day. He mentioned towards the end that he would listen for just one thing then block out the rest of the music and focus on that. If you just listen to the music overall, your sensory acuity will drop right down and you'll always miss things. So with the subtle things, you have to zero in and listen for the noise floor, or pay attention to the treble detail, or the image location of one instrument in particular. That's where you listen very carefully at the moment of the instant switch. Then you can decide how significant that change is. I think for most of us, the differences were small enough to pass under the radar when listening for enjoyment rather than listening critically.

Now I just wish MiniDSP could match the power of DEQX for dealing with phase. If you hear a system well set up on DEQX, it's definitely an experience. The most incredible imaging I've ever heard so far resides at Bathurst (thanks Terry). There is a seat over a spot marked X with masking tape and it all happens there.

PS - let me know Murphy if that digital input mod idea of yours works on DCX!
 
Hi Paul,

You do have a 'nack' for recalling accurately in your posts.. Great post above!

Just for the others, I will explain that I too can't seem to use Digital Input on my DCX as each digital source I use seems to 'clip' (Red LED constantly flashing) the DCX input!! Upon study of the circuit diagram, I found a 110 ohm resistor across the digital input that sets up the 110 ohm transmission impedance of AES/EBU digital standard.. I am hoping that if I replace this with a 75 ohm resistor, I can better "tune" to SPDIF standards, and maybe solve issue with digital source clipping the input??

I will let you know how it goes :)

Cheers,
Murphy
 
Murphy, I'm hoping that one works out, because I'd like to be able to try digital input. The pest of course is that I can only try it out until I get multi channel volume control. Wait - does your DCX do that with the mods?

Yes, amongst others, it is part of the "active i/o" mod..

8 channel (2 for analog inputs) preamp style analog volume control, including on/off, mute, main volume, separate Low/Hi volume and left/right balance,, with memory save/recall..

The only 'catch' is that although it's 8 channel volume, it's stereo pairs!!
So you might say 4 channel stereo volume..

Eg, the pairs are:
-input A / input B
-Output 1 / 2
-Output 3/ 4
-Output 5 / 6

With the separate Low/Hi volume being for output 1&2, and 5&6 respectively..

Cheers..
M
 

Pano

Administrator
Paid Member
2004-10-07 6:05 am
Panama
Thanks very much to Paul, Murphy and all who participated in this test. I read the posting over at RedSpade. As a potential miniDSP customer, this sort of thing is great to read.

Some thoughts and ideas of my own, FWIW, IMO, YMMV, etc.

The reason the miniDSP seems to have more treble or more air even when the FR plot says no, may be because of greater distortion - harmonic and inharmonic - and phase problems at the top end. That can sound like more top end even if it does not show on the FR plot. Don't know if that is the case, but it's worth a look.

The DCX. If it's clipping on a digital input and your inputs and EQ are set to 0dB, then it's the digital signal that is clipping. Have a close look at what is upstream. DEQ, playback software, or the sound file itself. There are plenty out there that are badly clipped. I see them flash the red lights on my DCX, too. Many software players also have digital gain and replay gain leveling. Check those.

I can post a couple of test tones if you want to double check this. it is not related to the 110 ohm termination. Going to 75 ohms is fine, but won't change the clipping - unless there is something really odd going on.

Murphy. Why do you say that you found it good to keep away from digital attenuation? Can you explain further?

Thanks guys.
 
Murphy. Why do you say that you found it good to keep away from digital attenuation? Can you explain further

Hi Panomaniac,

Well, where do I start..

The only conclusive reults we got between DSP's was that whenever you attenuate the signal (move below 0dB) using the DSP controls, the sound quality is reduced to the point where it's easily noticable in our "instant switching" listening tests..

Sound becomes grainy and quite noisy, Loss of dynamic range, reduced frequency response, loss of detail in the sound.. Etc..

It's like listening to a song that's been compressed using the lowest bit rate possible!! This is not surprising as these DSP's use 'bit stripping' for signal attenuation, so in affect, you are compressing the signal..

For our second listening test, one of the guys wired in an analog pot for level control, that way he could leave the attenuation on his MiniDSP at 0dB, and it worked a treat!!

The stock DCX and MiniDSP both suffered the same result when using digital attanuation. (Modded DCX uses analog volume control)

Hope this helps..

Murphy
 
Hi Panomaniac,
For our second listening test, one of the guys wired in an analog pot for level control, that way he could leave the attenuation on his MiniDSP at 0dB, and it worked a treat!!

The stock DCX and MiniDSP both suffered the same result when using digital attanuation. (Modded DCX uses analog volume control)

Can I assume that the pot for the miniDSP was post DAC?
I wonder about a hybrid system for active crossover attenuation, where most of the attenuation is done post DAC via analog pots, but the day-to-day volume control is done in the digital domain.

Was the DCX analog volume control using PGA chips?
 
We didn't do the test blind. While this may disappoint some, but I felt that using instant switching would be revealing enough. We did this test for our own curiosity, but no one had a point to prove. In a previous informal comparison, I found that was enough even with levels obviously not matched. No one in that test could hear a difference with the digital conversions vs a loop through wire, not even one person there who was certain a difference would be heard.

In other words this is not a valid test. It is just a whole lot of subjective opinions from people who may or may not have a preconceived bias. I'd hardly call that conclusive :(

Also every CD ever sold comes with built in quantisation errors due to the finite resolution of the binary number system and there is nothing you can do about it no matter how good you think your analog reproduction system is !!

regards
trevor
 
Last edited:
Can I assume that the pot for the miniDSP was post DAC?
I wonder about a hybrid system for active crossover attenuation, where most of the attenuation is done post DAC via analog pots, but the day-to-day volume control is done in the digital domain.

Was the DCX analog volume control using PGA chips?

Hmmm, I don't actually know if the pot was pre or post DAC.. I will check with Gainphile,, but I'd say to keep s/n ratio good, the pot would have been post DAC..

The analog volume control in the modified DCX is a CS3318..

Cheers..
 
In other words this is not a valid test. It is just a whole lot of subjective opinions from people who may or may not have a preconceived bias. I'd hardly call that conclusive :(

Also every CD ever sold comes with built in quantisation errors due to the finite resolution of the binary number system and there is nothing you can do about it no matter how good you think your analog reproduction system is !!

regards
trevor

Hi Tevor,

Yes, we did do blind tests.. Literally,, the "switching person" was behind a curtain with a big pillow over the switch box.. That way you couldn't hear the click of the switch, AND, you couldn't see his arms moving while switching.. I'd definately say it was a blind test :)

But really, blind or not blind, "instant switching" is a fantastic aid for this sort of stuff..

Cheers,
Murphy
 
right, see this is where systems like the sabre's integrated digital volume are a different ballgame, the attenuation is done in the dac registers themselves. I use this for my 2 channel rigs with overwhelming superiority to analogue to my ears and shortly the 8 channel ackodac will allow this sort of system for crossovers. this is where i'm headed also adjusting the gain in the spdif stream itself pre-dac is transparent with something like an RME9652