Stereophile's JA discusses Caps!

[---k---]

I got a good chuckle while reading the August issue of Stereophile (I only read it for the measurements! ). On page 76 and 77 of John Atkinson's (whom I respect) review of the Esoteric MG-20 speakers(which I recently spent 30 minutes listen to at a local dealer), he states:

The MG-20's hardwired crossover comprises two inductors (one air-cored and the other with a core of iron laminate), two resistors, and four high-quality polyproplene-dielectric capacitors. (An ICW Clarity capacitor is used in the tweeter feed; the other caps are marked "Bennic," a respected Taiwanese brand.)

There you have it folks. Stereophile has called the lowly Bennic a "High Quality Cap"! The debate is now over. :P

Thomas should probably just go ahead an lock this now. I don't post this to get into a flame war over better caps versus worse. I also don't want to get into a Sterophile sucks thread. I'm not here to argue or even passionately debate. I really don't care that much.

But to all those "noobs" out there wondering whether it is worth it to use anything other than Bennics and Dayton caps, I think this really does answer the question. If it is good enough for a $10,000 pair of speakers and Stereophile, it is likely good enough for them.

BTW, I found the speakers to be very good. They weren't perfect, but other than the Khans, what are? :P JA also liked the speakers very much and says they are likely going to be Class A (Limited Base) recommended.

[Amphiprion]

Bennic is respected. Tolerance is the only reason I might buy something else. Polypropylene, polyethylene, it's all good. If anyone REALLY wants to delve into capacitors, search out Cyril Bateman's website and read his series of articles on capacitors that was published in Wireless World. VERY enlightening.

The major flaw in the speaker crossover he popped open was the iron laminate inductor. It is FAR, FAR more 'non-ideal' than any metallized film capacitor in the speaker. Even NPO ceramics IMO are better than any inductor, but you'll only find those in active circuitry. I frequently wonder why capacitors are such a subject of debate while the real non-ideal components get little if any discussion.

[Saurav]

The major flaw in the speaker crossover he popped open was the iron laminate inductor.

Is this because of non-linearities in the BH curve? Near the zero-crossing or close to saturation? All/none of the above? Are the problems worse at low signal levels or high?

[Amphiprion]

DCR is an immediate problem with all inductors. Hysteresis will be present at any signal level (B-H curve). Saturation comes into effect with larger inductor currents. Problems will be worse at high signal levels. Inductance changes with frequency over about 1KHz depending on the quality of the core (iron core < laminated steel < ferrite, but ferrite saturates faster IIRC). Oh, and they radiate fields that interact with other inductors. I had an old speaker buddy who tested various inductor layouts and found mutual coupling could cause crosstalk at levels orders of magnitude higher than any distortion a cap would generate. I'll see if I can find it.

Inductors are the real POS in any passive crossover circuit, save for non-film caps (electrolytics). They invariably deviate from ideal in so many ways I can't believe audiophiles don't fret about them more.

Frankly, in a 10K speaker, I would be appalled to see a laminated core. 10 or 12 gauge air core far away from any other inductor would be much more appropriate depending on DCR. But DCR can be worked into the crossover, nonlinear distortion can't.

[Dennis H]

Well, Mark, get ready to be appalled. Revel Ultima Salon. I don't know what they cost now but they were $15K back in 1999 when Stereophile reviewed them (quite favorably BTW).

[Amphiprion]

Well, at least the inductors are kinda far apart... :/

ETA: Are those axial electrolytics???

[Hank]

I only see one iron-core, but those purple jobs are electrolytic. Even if they're non-polarized, they shouldn't be in a $15,000 speaker. They'd probably be $20,000 today 8O

[---k---]

And I think that is my point. You can build a very very very good speaker using cheap caps and inductors. "Noob" shouldn't be scared of using cheap caps. Using fancier parts, is just that last bit of optimization for perfectionists.

[chasw98]

These are really for perfectionists.......

http://estore.websitepros.com/1736754/Categories.bok?category=DUELUND%3AVSF+Capacitors

[Amphiprion]

Cheap yes, but of the wrong kind no The tolerance on bipolar electrolytics is terrible, at least for the ones I've seen.

[ThomasW]

These are really for perfectionists.......
http://estore.websitepros.com/173675...VSF+Capacitors

Hey don't be so cheap.... :B

http://www.soniccraft.com/sonicap_platinum.htm

[Dennis H]

I see 2 iron cores, 9 air cores, 6 film caps and 5 electrolytics. It's a 4-way so most likely they are following the Zaph philosophy of using the cheap parts on the woofers and the lower mid and using better stuff on the upper mid and the tweeter.

I suppose there could be some justification other than price when you're looking at that lowest XO. A film cap could be the size of a beer can and a 12 ga. air core could weigh 20 pounds.

[---k---]

I suppose there could be some justification other than price when you're looking at that lowest XO. A film cap could be the size of a beer can and a 12 ga. air core could weigh 20 pounds.

Doesn't stop Jon and Thomas! :P

[JJones]

Well, Mark, get ready to be appalled. Revel Ultima Salon. I don't know what they cost now but they were $15K back in 1999 when Stereophile reviewed them (quite favorably BTW).

After two days of auditioning speakers in Manhattan audio boutiques, the Revel Salons were actually the only speakers I found enjoyable to listen to (not fatiguing within a few minutes). Other speakers we listened to included B&W's higher end range and a half dozen other makes. I didn't find any of the B&Ws tolerable to listen to. I would want to have them [Revel Salon Ultimas] in my house for a couple days before I'd be willing to write a fair review of them, however.

I don't think capacitor tolerance really matters, in fact, I like to get a variety, that way it's more likely I'll be able to put together exactly the right total by mixing and matching. I wouldn't buy expensive caps unless I got them cheap (certainly not more than a few $ each).

[Saurav]

I don't think capacitor tolerance really matters

...

that way it's more likely I'll be able to put together exactly the right total by mixing and matching

Are you measuring each capacitor? If not, it's actual value is within +/- 10% (or whatever the cap tolerance is) of the number printed on its side. So your 'exactly the right total' will be as 'exact' as the tolerance of the caps you buy.

In reality, it might be a little closer, because as you combine multiple caps, you increase your statistical chances of individual tolerance errors canceling each other out. I.e. it's unlikely that the caps you choose will all be at +10% tolerance.

[JJones]

Are you measuring each capacitor? If not, it's actual value is within +/- 10% (or whatever the cap tolerance is) of the number printed on its side. So your 'exactly the right total' will be as 'exact' as the tolerance of the caps you buy.

Certainly, just got my LC meter at the PE Tent Sale, and, I'll also use the Clio Win 7 demo (free) which comes with an LCR meter for your PC using your sound card (can also be done with other programs, but the Clio Win one looks very straightforward and easy to use).

I had fun measuring a few dozen poly caps last night, some brands were VERY consistant and others were not - but - I assume the value of poly caps won't change much over time? Once you measure them, you know? Is that a safe assumption?

In reality, it might be a little closer, because as you combine multiple caps, you increase your statistical chances of individual tolerance errors canceling each other out. I.e. it's unlikely that the caps you choose will all be at +10% tolerance.

Yep, and, you prob. want to sit at your PC doing sweeps lots of times while you tweak the crossover anyway, until it looks real nice, then measure to see what values you've ended up with?

Some veteran speaker designers can calculate crossovers in their head and build it at the workbench, watching the test results, trying a little more here, a little less there, until they get a nice crossover w/o even using a PC.

[Dennis H]

Doesn't stop Jon and Thomas! :P

True enough but we all know their mental stability has been questioned in the past. If some is good, more is better and too much is just barely enough..... :lol:

Seriously though, the Salon's lowest XO is 125 Hz, 4th order and I think Jon and Thomas would probably choose to go active that low. Commercial guys don't have that option so they bite the bullet and use 'lytics and steel cores. As long as the steel cores aren't saturating, I doubt anyone could hear the difference that low.

Oh and Stereophile claimed Revel individually fine tunes each XO to get it within spec. Maybe because of the variability of those electrolytics?

[---k---]

Dennis,
Those are very good points. I agree with you. I just couldn't help put take a swipe at our fearless leaders.

[Dennis H]

Certainly, just got my LC meter at the PE Tent Sale, and, I'll also use the Clio Win 7 demo (free) which comes with an LCR meter for your PC using your sound card (can also be done with other programs, but the Clio Win one looks very straightforward and easy to use).

ARTA (actually LIMP) looks like it has a nice one too. It not only gives you the L or C but it also gives you the series resistance. Hook it up just like you were measuring the impedance of a driver.

[ThomasW]

True enough but we all know their mental stability has been questioned in the past.

Hey! I resemble that remark.. :??

[Kal Rubinson]

Oh and Stereophile claimed Revel individually fine tunes each XO to get it within spec. Maybe because of the variability of those electrolytics?

Yup and the variability of almost everything else. They keep records of the measurements/tweaks of each and every unit.

Kal

[chasw98]

Yup and the variability of almost everything else. They keep records of the measurements/tweaks of each and every unit.

Kal

Would this indicate that each individual unit is tweaked to meet a certain measurement specification? And do they meet the specification by swapping different values of parts in and out of circuit to match the units to a desired value. That would then take into account the variances in tolerance from one component to another. But it also seems very labor intensive. Maybe that is where some of the money goes for the cost of these speakers?

[Kal Rubinson]

Would this indicate that each individual unit is tweaked to meet a certain measurement specification? And do they meet the specification by swapping different values of parts in and out of circuit to match the units to a desired value. That would then take into account the variances in tolerance from one component to another. But it also seems very labor intensive. Maybe that is where some of the money goes for the cost of these speakers?

You would have to ask them about what they do now but, at the time of the original Ultimas, the answer would be yes. They said that this assured them that all speakers met the same standard, that matching speakers in the field was not an issue and that repair/replacements in the field could be accomplished (without shipping the entire speaker system to the factory) by sending a new driver tweaked to match.