2017 POLL: classic Passive or Active/DSP/EQ ?

[JonBocani]

Hi Jon,

I agree about DAC quality, but not about the said mid-highest frequencies !

It's also night and day with a good one with mid-bass and bass area : night and day : but these ones are even rarer ! And frankly, at least from my experience : it changes the game ! And which is hard is : you can not investigate according to me as far you have not a sota source be it vynil or DAC ! And hard to winn le "je ne sais quoi" with a DAC ! there are always Something wrong : too much dynamic in the dynamic, too shiny or too mudy sound, too much transparency instead of some quitness against brighness ! ANd the most difficult today is always in the bottom range ! But it's stays a whole thing and good equilibred DAC are not so many imho !
I didn't know myself the Forsell so can not argue anymore...

By experience, the most audible effects from the DAC comes from the high frequencies, and it's particulary audible with drivers at ease in 12-15khz+ (as the RAAL ribbons)

That being said, i'm not sure it would be day & night in an blind test. But i think it would be easier to spot than different EQd drivers...

 

[mondogenerator]

I use both

But I prefer analogue active.

Minimal slopes while aligning phase closely as possible (though not absolute as I don't use DSP)

Minimal EQ. BSC. Convenient EQ you may say...using a small passband ripple present in the highest Q stage to conveniently fill in the Peak or troughs around XO point, or to fill ripple caused by whatever baffle diffraction I can't reduce further.

So I guess....kind of minimalist approach?

>96dB filters are just a bit too much (unless your anti aliasing...)

Besides....the room is king and will defeat your best efforts in all but one listening spot.

 

[Pano]

Just how much passive EQ is available on the market, or being built by DIYers?
Obviously there is still a lot of used pro gear around for graphic or parametric EQ, how many people use it. I know only one person on the forum who does.

Or are we counting EQ that is built into a passive crossover, such as traps, shelf filters and such?

 

[JonBocani]

>96dB filters are just a bit too much (unless your anti aliasing...)

Besides....the room is king and will defeat your best efforts in all but one listening spot.

I'm using 300db/oct linear phase for years (DEQX dsp) and i prefer that to any other slopes i tried, 99% of the time.

I don't see (hear) any downside by having such brickwall crossovers. On the contrary. It feels more dynamic, more precise and better controlled. Also, i can push the drivers at their limits of their comfort zones (such as a subwoofer crossed as high as 150hz or so).

 

[Pano]

For me it's the opposite. Brickwall filters sound odd to me. Like the character of each driver is more obvious then with more gentle slopes. 4th order acoustic usually works for me. We all notice and are bothered by different things. It's nice to have choice to achieve the results we want.

 

[plasnu]

I'm using 300db/oct linear phase for years (DEQX dsp) and i prefer that to any other slopes i tried, 99% of the time.

I don't see (hear) any downside by having such brickwall crossovers. On the contrary. It feels more dynamic, more precise and better controlled. Also, i can push the drivers at their limits of their comfort zones (such as a subwoofer crossed as high as 150hz or so).

I agree that the very steep filter can sound clearer, dynamic and more controlled, but I find it can sound sterile, cold and somewhat artificial. I'm not surprised if someone prefer it over conventional slope speakers, though. I'm currently using 24 or 48dB linear phase, and I have to say both have pros and cons depends on the music. One is not absolutely better than the other, I mean.

 

[Drewan77]

I'm using 300db/oct linear phase for years (DEQX dsp) and i prefer that to any other slopes i tried, 99% of the time.

I don't see (hear) any downside by having such brickwall crossovers. On the contrary. It feels more dynamic, more precise and better controlled. Also, i can push the drivers at their limits of their comfort zones (such as a subwoofer crossed as high as 150hz or so).

I generally agree & use the steepest slopes I can until I sense music becoming slightly sterile or cold - Jon may be correct in his earlier post that this is a limitation of the drivers I currently use. However with DEQX I would clarify for non-users that a sub crossed at 150hz cannot use a 300dB slope but nevertheless, still fairly 'brick wall':

@150hz Xover:
Linear Phase <108dB slope (300dB slope can only be >401hz which is impractical for a sub).
LR <96dB slope
Butterworth <60dB slope

I currently cross subs at 100hz/72dB, the steepest slope available at that frequency. The new speakers I am working on may need sub reinforcement below 50hz only - for that I will be limited to LR @ 24dB.

 

[bcodemz]

Spent probably over 2000 hours playing with the DEQX and close to 1000 hours playing with various miniDSP (plate amps, 4x10HD, 2x4, etc..) and there is things i like that i have on the DEQX and other things that i prefer on the miniDSP plug-ins.

And i'm not talking about D/A or A/D converters, because they're not very good on both, even though the ones on DEQX are much better... Still not enough for my taste.

But, DEQX forte is the linear phase xover up to 300db/oct which is very nice and works better 99% of time than smoother slopes. But the downside of the DEQX is the limitation of 10 EQ points. Also, it's 3 way max, and you cannot overlap nor make more than 1 bandpass.

miniDSP on the other hand, is a bit more ''poweruser'' and flexible, but a bit more limited. 48db/oct max, no linear phase only butterworth and LR, also a larger total of EQ points BUT limited to 5 or 6 per channel.

Try a miniDSP 2x4 HD. The DAC is much better than the 2x4.

 

[mondogenerator]

For me it's the opposite. Brickwall filters sound odd to me. Like the character of each driver is more obvious then with more gentle slopes. 4th order acoustic usually works for me. We all notice and are bothered by different things. It's nice to have choice to achieve the results we want.

This pretty much sums up my feelings.

I've used high order filters in software when mastering and I didn't like the outcome. I'm neither clever enough at using them nor expert enough to be able to quantify why that is, but they always sounded a little odd to me.

However, going from passive 2nd order to 4th order acoustic and removing phase inversion definitely sounds better to my ears despite a cycle of delay.

No doubt I could go to 48dB and have more improvement at least if I used DSP but 24dB is simpler easier and avoids unnecessary DA AD conversions and some region of my brain likes that idea. No rationale again but it works for me.

Besides that if I used DSP I'd never stop fiddling to enjoy the music long enough

 

[Drewan77]

Besides that if I used DSP I'd never stop fiddling to enjoy the music long enough

That's a fair point

I spent nearly a year 'fiddling' with DEQX settings and configs back in 2012 and eventually settled on 16th order as the highest I could go whilst still sounding transparent & smooth in my system (2-way active OBs +subs).

After 3+ years just listening to music, an itch needs scratching so I'm just starting on a huge 4-way active OB project so no doubt I will be faffing about with measurements, frequencies, crossovers, configs, time alignment etc for many months again.

 

[bcodemz]

For me it's the opposite. Brickwall filters sound odd to me. Like the character of each driver is more obvious then with more gentle slopes. 4th order acoustic usually works for me. We all notice and are bothered by different things. It's nice to have choice to achieve the results we want.

I think the problem is the big change in directivity as well as severe group delay (non linear phase crossover) or pre-ringing (linear phase crossover) arising from use of such steep slopes. Pre-ringing can be largely corrected, but drivers need to have matching directivity. If these 2 conditions are met, I think a brickwall crossover has the potential to sound better than any sloped filter.

 

[JonBocani]

Try a miniDSP 2x4 HD. The DAC is much better than the 2x4.

ok thanks, good to know.

 

[Lojzek]

If brickwall filtering were any good, we'd be all looking like sponge bob, since we aren't, then the most natural solution is gradual roll off.

 

[Pano]

I think the problem is the big change in directivity as well as severe group delay (non linear phase crossover) or pre-ringing (linear phase crossover) arising from use of such steep slopes.

Yes, I think those are important factors. But there is something else, too. Something I rarely see anyone talking about in crossovers - harmonics. All drivers have them, and how they blend and sound together is not intuitive.

 

[JonBocani]

I think different drivers overlapping is worse than anything a brickwall slope can bring as negative.

 

[Pano]

Why would you think that?

I strongly disagree. In theory, the super steep slopes might seem like a good idea. But in reality, they sound odd and call too much attention to the drivers.

 

[JonBocani]

Why would you think that?

I strongly disagree. In theory, the super steep slopes might seem like a good idea. But in reality, they sound odd and call too much attention to the drivers.

That would be interesting to blind test

Very easy to try with a DEQX, up to 4 presets: 6db butterworth or LR... up to 300db LP.

 

[Pano]

I have done blind and sighted tests of it using DSP. I think the super steep filters don't sound right, but not in a way I expected.

But yes, it's good to do the tests yourself and with various types of drivers. With DSP that's pretty easy.

 

[KJ_Knowles]

That would be interesting to blind test

I did exactly this at the 2014 Lone Star Audio Fest in Dallas. Compared a three way active system with presets for Fourth Order LR (IIR) and 1792 tap Kaiser-Bessel (FIR).

Crossover points were identical and the drivers were chosen to assure that they would be within their bandwidth sweet spots in both filter sets. Some EQ was applied to make the overall amplitude response as close as possible between the two settings.

The IIR setting was delayed to match the filter delay of the 1792 tap FIR. This allowed switching at will with almost no perceptible interruption of the source material. I say "almost" because sometimes the DSP would be a bit unsettled and produce a small click or fraction of a second silent period.

The test subjects (victims) were given the switch and were encouraged to switch when they felt like it. There was also a "coin toss" randomization set up using a logic engine function on the DSP that was executed between subjects, so that I didn't know which they were selecting until the test was over. That was probably overkill, but since the hardware allowed, why not? Most subjects used their own source material. Otherwise, they could select from a server I had set up by means of a tablet menu.

During the event I had 118 subjects. The final tally was:

21 - No Preference
47 - Fourth Order LR IIR
50 - 1792 Tap Kaiser-Bessel FIR

Statistically, no result. One interesting side effect was how many subjects mentioned that they sometimes liked one setting better for a particular track, but the other for different tracks. I didn't expect that, but it isn't really surprising, either.

That said, this was using FIR for XOV functions only, which was the variable that this test was all about. Had a full FIR implementation been used for all filter functions, the results would probably have been more defined.

An externally hosted image should be here but it was not working when we last tested it.

Nothing like an opportunity to geek out just for the fun of it! Off topic, but I have an idea for a 2017 LSAF experiment that will incorporate separate DSP control of the front and rear wavefronts (drivers on front and rear of cabinet, separate internal enclosures), allowing adjustment between bipolar, cardioid, quasi-dipole, etc. Mid and HF are to be plain dipole. I want to play with adjusting the LF radiation pattern and EQ to optimize according to room placement and acoustics.

Which reminds me...less than four months. I need to get to work!

 

[Greebster]

I use active DSP on my three way TL's. The mid is a push pull planar magnetic (quasi) fullrange. Crossover points are 330 and 3300. The 3300Hz point was chosen to match the polar response of the tweeter. I want a nice balanced off axis response, not just on axis.

But more to the point, in listening tests have noticed that out of band abnormal response shape has a negative impact. Take your typical 6.5" midwoofer as an example, around 4-5kHz the response normally exhibits a peak. This peak I then pay special attention to and tweak it out. This way the roll off of the filter I'm using is nice and smooth, a text book acoustic curve. Doing this results in a natural sounding splice with a tweet. That tweet also may require it's bottom end response tweaked to produce a match in acoustic profile with the midwoofer. The end result is a seamless transition from one driver to the next. If you do it right across all three drivers, correct for time alignment, etc. it sounds as near as one can get to a point source without the limitations all full ranges exhibit.

In practice most of the eqing done is out of band, as long as you utilize respectable drivers and implement them properly.