Another Capacitor Thread (and an apology).

[Guest thesloth]

Quick question.

 

Do the folks that use these "boutique" capacitors in their crossovers worry about everything before the crossover network?

 

I mean why isn't everyone losing sleep knowing that the signal has probably passed through a plethora of aluminum electrolytic caps in the various DAC's, CD players, and amplifiers before the crossover network? I mean when I last serviced my McIntosh C2200 I lost track at how many general purpose lytic's were in the signal path. My point is, if these capacitors added any distortion (at levels that matter) or any sort of terrible sonic attribute from whatever black magic evil voodoo power then certainly this many "cheap" capacitors would render this preamplifier unlistenable. But yet the amp sounds lovely and not one normal person that isn't an audiophile would even notice or care.

[DrWho]

Ya, I was definitely at the Jubilee invite Dean - and I remember our conversation about the Dx38 noise floor, and the cultural conditioning of various artifacts.

 

One of the things I've been exploring recently is the idea of "noise floor" - which is really a very odd way of thinking about audio. The thing I've noticed, but it's incredibly hard to quantify, is how the noise floor in some circuit topologies will ride on top of the signal, whereas in other topologies, the noise floor stays put - for lack of a better way to describe it.

 

There are two very notable flaws to the Dx38:

1) It has gain-ranged inputs: This creates a fluttering type sound (much like the wooshing of air through an undersized port), but it does it at all frequencies

2) It has noisy power-supplies, which is death for AD/DA converter stages (even though it looks good on a spec sheet and a noise floor FFT).

 

I've long been a proponent of active xovers, but the reality is that nobody today is making a "good" consumer grade audiophile xover. I am 99% confident that I could design one to pass the muster of you hardcore analog guys, and I've intended for years to design one, but it just takes a crap ton of time - and the limitations of the parts available today force me into more esoteric type solutions, which take a lot longer to vet properly. Honestly, I get burned out on audio at my day job, so I'm spending more time with cars these days.

 

But just from a pure specs perspective, there is simply no product today good enough for the audiophile realm. The Yamaha SP2060 is a much better unit, but still falls short of what I think would be required of an audiophile solution. Trying to modify these units after the fact could make them a little better, but the real limitations are in their fundamental architectures.

 

And don't get me wrong, many of these units are designed with great skill and engineer finesse, and are totally sufficient for the pro audio realm. So I'm not trying to knock these designs or anything like that - they're just built for different applications.

 

 

 

The true Holy Grail in my mind is to start with a quality digital source (96 kHz, 24-Bit), and then do the active xover functions before the D/A stage. Then you need post D/A volume control with 0.01dB matching, and identical amps for each of the channels. You will also need a studio grade word clock - ideally the same make and model of the studio that created the recording, and then make sure everything is experiencing the same ambient temperatures. The D/A stage is gonna need 130dB of dynamic range, which is gonna be beyond the capability of most analog circuits when you consider the maximum output level will only be a couple volts (to be compatible with consumer gear). Oh, and the outputs should be unbalanced - screw this balanced crap (for short runs).

 

 

Btw, there were artifacts with Marion's passive xovers too. It's not like the analog passive xover is the goal that the active is striving for. The active does a lot of things better than the passive, and I'm not just talking time-alignment either (although that's definitely on the list). The quality of the bass with passive xovers is always more bloated and smeared sounding. Tonal balance is definitely less accurate (which we also see in the frequency response plots), and there's this fuzziness laid over everything - that many would probably refer to as "warmth", but it's really just distortion from the non-linear behaviors of the inductors.

 

 

Anyways, just some random thoughts as I reminisce on all those fun listening sessions.

[Max2]

Why thank you Jay. (-:

 

 

Yeah, I don't laugh much...but the eye contact and info transfer was something that was Seinfeld worthy.

[Guest thesloth]

but it's really just distortion from the non-linear behaviors of the inductors.

 

Speaking of which. I was just PMing Dean about this. Would replacing the ferromagnetic inductors that are in series with the woofer with air core inductors improve this?

Edited October 6, 2015 by thesloth

[Mallette]

Anyways, just some random thoughts as I reminisce on all those fun listening sessions.

 

God, I LOVE it when you talk technical...

 

Dave

[DrWho]

 

 

but it's really just distortion from the non-linear behaviors of the inductors.

 

Speaking of which. I was just PMing Dean about this. Would replacing the ferromagnetic inductors that are in series with the woofer with air core inductors improve this?

 

 

 

Ya, definitely....but at some point you run into size, DCR and parallel capacitance limitations. Large air core inductors will also start to act like antennas, and with the efficiency of our horns you might start pulling signals out of the the air too.

 

Resistors themselves have temperature coefficients and low level hysteresis as well. Even for line level signals these effects contribute to the overall distortion. When you think about it, it's pretty bad that even a resistor introduces distortion.

 

So inductors, resistors, and caps all don't behave as expected....and they also all make their own noise even when no signal is present.

 

 

One fun thought extending off of that.....when you have multiple stages in a circuit, the distortion of the previous stage looks like part of the signal to each subsequent stage - so you end up compounding distortion through the signal path. This is why I'm almost a minimalist in all of my designs. I say almost because sometimes you can add parts to make a particular component behave more linear - and with good circuits, even gross non-linearity of the extra components doesn't introduce any new low level non-linearity. That's a complicated way to say it, but things like Nelson Pass's cascoded JFET output stage would be an example of "almost minimalist" - adding another component to offset the linearity of the dominant stage.

 

 

I think this is why we can still hear very small distortion differences in the electronics, even though our mics and speakers are adding multiple percentages of distortion to the signal. It's that multiples of percent getting compounding all the low level electrical gobbly goop.

 

 

Anyways, as you increase the voltages and currents going through any inductor / cap / resistor, the non-linear effects also become a larger percentage of the signal....especially when trying to control size and cost. Obviously there is a point of diminishing returns, but we're talking audiophile here where every last stray electron is something to be avoided. This is why I'm fundamentally opposed to the passive xover in a speaker, but we can't ignore that the passive xover is solving some other very important problems at the same time - like attenuating HF noise from the amplifier, or improving the impedance "matching" (which is really an intentional mismatch, but I digress)....

 

I actually see the idea of voltage encoding as a flaw in the abstraction of the amplifier-speaker interface, but we can't just blindly ignore the industry standard of using voltage to encode the signal. Why don't we use current to encode the signal? Or why not a phase adjusted carrier? Pulse code modulation? PWM? There are all sorts of ways to represent a signal - voltage is the easiest to think about and design with, but I'm beginning to think that it's a bad approach for the amp-speaker interface: especially if we are going to assume that constant directivity is the target.

 

In electrical motor world, we care more about the current because it directly correlates to the torque - the motive force acting on the system. Did you guys know that a current amplifier will inherently lift the HF response of a constant directivity speaker? The downside is it also magnifies any other impedance non-linearity, so you end up with an undamped suspension, but I think the driver can be designed differently to offset that effect - or you just throw in a zobel circuit to damp it electrically. And then if you don't try to beam the HF to make it look flat for voltage, then you don't have to squeeze the sound as much - which is going to reduce the horn distortion too.

 

 

 

At the end of the day we get to choose our tradeoffs, and I totally understand why most people go the passive route. I'd even suggest that I hear what they're hearing, but then it becomes a question of priority. I personally really really like drums, and drums just sound off to me if there is a mis-aligned xover happening in the middle of the drum sounds. I accept the "digital hash" because my source material already has it....none of the music I listen to comes on vinyl, and I wouldn't like it if it did because I haven't learned to tune out the vinyl stuff. That and it's way less work to hit play on my computer than trying to maintain and find physical copies of the source material. It's more than just convenience because improper care affects the sound quality a ton. I don't enjoy the whole taking care of my music process.....but I know a ton of guys really enjoy the whole prepping / cleaning / storage thing of their source material - and in some ways I think the physical habits are important for instilling a mindset of delicacy to the music, and trying to stay true to the source.

 

 

 

Anyways, go with the air core if you can afford the size / cost tradeoffs - but it'll still have whatever distortions are common to resistors....

[pzannucci]

It would be interesting with the large inductors if you can get a large enough size in the resistance range of a ferrite cored inductor.  If you end up increasing the resistance, the values on the other components should change, just as the esr issues and other nuances occur with caps.  Air core should sound better but likely not until close to saturation or in a poor design.  It would be something for an A/B test 

 

I did have some large chokes on the floor for by bass sections and when I would here a hum.  Didn't know what had happened to my amp so I shut it off.  Still had the hum.  Definitely picking up external waves.  Had to reroute a bunch of wires to get rid of the hum.