Active vs Passive: an experiment

[augerpro]

I know not many here run active systems, but I thought I'd share this since Jon is probably one of the few who can shed light on the causes of the differences I heard.

So I've been working on some small TM's for the last year and finally came to a place where I was happy with the voicing (actively via minidsp) and translated the active response into a passive network for the final speaker. All the modeling was done on Soundeasy which does both active and passive so this is fairly trivial to do. It occurred to me this would be a good time to hear for myself the difference between two active and passive responses which have identical responses.

About the gear: the speakers are using Scan Disco 15W woofers and Scan Disco HDS tweeters. LR4 acoustic response at 2300hz. Both speakers are run through the minidsp so I can apply level matching as the passive speaker is slightly more sensitive (passive networks in a TM often induce a hump between 100-250hz, so this area is my reference for flat. The overall effect when level matched is less bass for the passive network) and do quick A/B switches. I used Jantzen Z-Superior cap, shunt air coil, and Mills resistor on the tweeter. On the woofer is an Erse Super Q coil, Clarity PX shunt cap, and air coil. Connected by 20g speaker wire and Cardas Quad Eutectic solder. Decent components.

The differences were small, but pretty much immediately noticeable on good material. The active network was clearer and cleaner. The image was more 3D and separated better from the speaker. On the other hand the passive network had a pleasing (to me) body or fullness to the sound, almost resonant or chesty. Switching back to the active network almost seemed light weight in comparison, but again clear and somehow more relaxed where the passive sounded a bit busier on dense music.

So now I'm thinking ok, something went wrong in the translation and the two responses are not identical as the software predicted. So I took some measurements. The results were exactly as Soundeasy predicted. The difference in the treble is due solely to it being two different tweeters, and the discovery that the HDS is fairly sensitive to how flush mounted it is, it actually measures smoothest slightly deeper than flush otherwise that wiggle from 2-6khz gets bigger. One set of measurements is 6ms gate and no smoothing, and the other is 200ms gate and 1/6 octave smoothing. Offset for clarity.



Now...I'm guessing I can more easily dial in some fullness to the active network, while I just can't dial in clarity to a passive network so this was certainly interesting. And actually it was very music dependent which I liked more - the fullness was pleasing, especially on acoustic stuff, but the clarity, purity and relaxed nature of the active network was a clear step up in fidelity.

Thoughts?

[5th element]

Amplifier linearity is affected considerably by going active too. How considerably depends upon the amplifier but it's worth noting. Most amplifiers performance degrades the harder they are being driven. Going active reduces this demand quite considerably, especially for the tweeter. Another aspect is IM distortion. When your amp is playing the full audio spectrum it's going to generate IM products right across the range. With bandwidth limiting in place before the amplifier this is no longer going to be the case.

The series pass inductor to the woofers will add in its own distortion as will it also have its own insertion losses. The insertion losses will modify the box alignment by a small amount and alter any port tunings. Electrical damping will also be reduced by the presence of the inductor to the woofers.

People usually say that active gives you cleaner, clearer and tighter sound.

[JonMarsh]

Matt covered the bases pretty well... the multi-amp narrow frequency band thing really becomes evident when you go three way, as ThomasW's living room system was setup. You didn't mention what amplifiers you are using.

In my case, the reason I don't go active more often boils down to DAC channel outputs- I have that last version of the Arvo physically built, but no time to do the development, and it would use my Metric Halo ULN-8 for the crossovers, EQ, and DAC. Yeah, that's about $5K for 8 channels, but it's a reference quality DAC, only bettered by much more expensive items for the most part (think TotalDAC-D1), and has very good A/D also. It's a bargain for an 8 channel unit, and with the DAW functions and DSP (which only can be programmed from a Mac, unfortunately for most of the rest of you- once programmed, unit can operate stand alone) can do some very cool things.

So, here I am still building passive crossovers, with expensive parts, because if you want a sufficiently high end front end, that's the only way to get there (you can can a TotalDAC-D1 Dual like mine configured with the extra channel not as a complementary output for balanced out, but as a single ended output with electronic two way crossover- no EQ and DSP, you speakers need to be flat in their passbands).

I would emphasize that a lot of amplifiers will do much better direct connected to the driver and with a reduced pass band. At many price points, this is a very reasonable choice to make, going DSP with active speakers. Hyped sells stuff for doing this, based around the UcD amps.

[TEK]

as ThomasW's living room system was setup

Where is ThomasW? Hope he is well, I have not seen any post from he at least for a year or so

because if you want a sufficiently high end front end, that's the only way to get there

I think that the explanation for this was somewhere in that post - but I was unable to grab the essence in the reasoning that resulted in this conclusion.
Could you simplify and repeat the main arguments for a non-technical like myself?

[BobEllis]

TEK, I read that as if you want a signal that does justice to your Wavecor Ardents you need really good DACs for each channel you generate in a digital crossover. While a lower end sound card or mini-DSP may be fine for some you should benefit from better DACs. Not a criticism of mini-DSP or those who use them, rather a compliment to the resolving power of the drivers chosen for the Ardent. I have a mini-DSP and think it sounds great with my current speakers. (Seas Excel based)

Anothe route Jon hinted at elsewhere is using an analog active crossover to provide the signal shaping and filtering so you can "get away with" a stereo DAC and have the other benefits of active crossovers. Really good op-ampsor discrete buffers along with good quality capacitors and resistors will improve the end product. As a bonus, they aren't nearly as expensive as the parts Jon uses in his passive crossovers.

[augerpro]

Jeff Bagby commented on the inductance hump in the upper bass common to nearly all passive speakers, as the reason for the difference and I was suspecting this was the major culprit too. Even though it measures near identical I wonder if the resonant nature of the bump makes it sound different to our ears. I would need an anechoic chamber to do a CSD on frequencies that low though, so I'll never know if this resonance is in fact being measured but my setup is limited from "seeing" it. What do you guys think?

Along those same lines a similar phenomenon was a primary design consideration here, where I worked the baffle to lessen the peak/dip of the woofer at and above the BD peak, and the similar peak/dip for the tweeter. Both are typically in the crossover region and it is very easy to use a high Q filter to fill those dips, but I've wondered at what cost to sound quality? And even if the final response is flatter, is the time performance the same? Is it not more electrically resonant (for lack of more technical description)?

[Juhazi]

Minidsp4x10HD or Nanodigi are the recommended dsps for hifi multiway speakers, I have the first and a friend the latter. UIs are nearly the same . 192k input is not possible with 4x10.
It is easy to adjust eq for baffle step and cone resonance, set delay for time alignment, adjust efficiencies, boost the low bass with closed cabs etc. Easy to "voice" the speaker and modify if a driver is changed. Etc. etc.

[JonMarsh]

Jeff Bagby commented on the inductance hump in the upper bass common to nearly all passive speakers, as the reason for the difference and I was suspecting this was the major culprit too. Even though it measures near identical I wonder if the resonant nature of the bump makes it sound different to our ears. I would need an anechoic chamber to do a CSD on frequencies that low though, so I'll never know if this resonance is in fact being measured but my setup is limited from "seeing" it. What do you guys think?

This is quite true, as long as one doesn't impedance compensate the box Fb, and the box Fb is high enough to interact with the crossover. you can see this, for example, with most two way crossovers I've done with LspCAD; it predicts that bump.

OTOH, if the Fb effects are sufficiently low in frequency OR if they have conjugate networks placed in parallel to flatten the impedance curve, then one can wind up with a response free of that effect with a passive crossover. Oddly, I'd cite that new design with the Seas C18EN001 I've been working on as a good example of this; the midrange LF impedance bump is compensated with a conjugate network, and the woofer impedance bump is so low in frequency that I don't measure or hear any audible interaction or impact.


Mid Impedance conjugate network:




SW223BD02 + SS PR impedance:

[augerpro]

Yes SE will predict that effect also. I had played with a conjugate network early to see if there was any benefit and it wasn't anything I was too impressed with for this design, certainly not for the parts expense. But then I haven't worked with those before so I might not have been doing it right. Still that brings me back to the questions of why if they measure the same - even in that region - do they sound different?

[Ray_D]

Levels?

Yes SE will predict that effect also. I had played with a conjugate network early to see if there was any benefit and it wasn't anything I was too impressed with for this design, certainly not for the parts expense. But then I haven't worked with those before so I might not have been doing it right. Still that brings me back to the questions of why if they measure the same - even in that region - do they sound different?

How do your listening levels compare with your measurement levels? I suspect your amps just like to power the drivers alone at the higher sound levels typical of listening.