Loudspeaker Myths: Separating the Scientific Facts from Science Fiction

[Swerd]

… However, there are NPE's (non-polar electrolytic capacitors) around that are rated as low ESR (equivalent series resistance). I just don't know how much lower, or how they compare with poly's. But assuming the NPE's specified as low are low enough not to be problematic, are there are issues with NPE's that would create non-linearity in crossover performance? Perhaps Gene can chime in here.

I would be most concerned with performance in woofer circuits at cross points in the 300 - 600 Hz, range since that's where the real cost savings would lie.

There are so many decent low-uf poly's out there at reasonable prices to make it not worthwhile taking a chance with NPE's in tweeter circuits.

Now that we've gotten off the deep end, let's return to a more practical level on the subject of capacitors in speaker crossovers.

Dennis asked a good question (above in bold). In a 3-way speaker, the woofer-mid crossover frequency is low enough to require a large capacitor in the low-pass filter for the woofer. An example is Dennis's DIY design, the MBOW1 3-way, where that crossover frequency is 450 Hz. Here's the woofer low-pass filter:

The shunt capacitor (C1021) is large, 150 µF. A speaker designer has some choices to make based on performance and price.

A cheap NPE cap (about $4), see the two examples, both are rated at 100 volts (AC or DC is not mentioned, so assume DC):
Erse 150 µF NPE

Bennic 150 µF NPE

Or a moderately expensive ($41) MPP cap made by Solen rated at 400 V:
Solen 150 µF MPP

Or a much higher priced ($113) cap rated at 250 Vdc:
ClarityCap 155 µF MPP

Whether you are building 1 pair or 100 pairs of these speakers, the price range is large. Which one makes the best sense with regard to performance AND price?

And that is only for the woofer filter. The mid range high-pass filter has two more large caps, 110 µF and 120 µF.

 

[gene]

Now that we've gotten off the deep end, let's return to a more practical level on the subject of capacitors in speaker crossovers.

Dennis asked a good question (above in bold). In a 3-way speaker, the woofer-mid crossover frequency is low enough to require a large capacitor in the low-pass filter for the woofer. An example is Dennis's DIY design, the MBOW1 3-way, where that crossover frequency is 450 Hz. Here's the woofer low-pass filter:

The shunt capacitor (C1021) is large, 150 µF. A speaker designer has some choices to make based on performance and price.

A cheap NPE cap (about $4), see the two examples, both are rated at 100 volts (AC or DC is not mentioned, so assume DC):
Erse 150 µF NPE

Bennic 150 µF NPE

Or a moderately expensive ($41) MPP cap made by Solen rated at 400 V:
Solen 150 µF MPP

Or a much higher priced ($113) cap rated at 250 Vdc:
ClarityCap 155 µF MPP

Whether you are building 1 pair or 100 pairs of these speakers, the price range is large. Which one makes the best sense with regard to performance AND price?

And that is only for the woofer filter. The mid range high-pass filter has two more large caps, 110 µF and 120 µF.

ESR is less of an issue when the cap is in parallel with the load so in this case you could probably get away with it. I'd still spend the time modeling the entire circuit in PSPICE while also running Monte Carlo analysis with all tolerances factored in. You can also use smaller value bypass caps to lower ESR if it's even an issue in this case.

Years ago I did a very basic series vs parallel circuit analysis of first order filters to demonstrate how the circuits interact here:
Series vs Parallel Networks - First Order Comparison | Audioholics

 

[Swerd]

ESR is less of an issue when the cap is in parallel with the load so in this case you could probably get away with it. I'd still spend the time modeling the entire circuit in PSPICE while also running Monte Carlo analysis with all tolerances factored in. You can also use smaller value bypass caps to lower ESR if it's even an issue in this case.

Years ago I did a very basic series vs parallel circuit analysis of first order filters to demonstrate how the circuits interact here:
Series vs Parallel Networks - First Order Comparison | Audioholics

Interesting, thanks.

So I take it this would not apply for the two large caps in the mid range circuit (C2011 120 µF and C2031 110 µF). They are in series with the driver, would more likely benefit from spending money on MPP caps, such as Daytons or Solens.

 

[cutedaddy]

Dsp xo

Now that we have DSP, how come soldering a crossover isn't a thing of the past? When we're talking about tweaking, upgrading or bypassing standard OEM crossovers, why bother with Caps and Resistors etc. when you can adjust your xo frequency, order/slope and electromagnetic properties of the drivers real time, while listening to the result (and, when desired, while also watching your measurement graphs) realtime? With the added advantage of adjusting for driver sensitivities..

You can even add to that the freedom of using different xo's for different music material (that requires different directivity patterns of the loudspeaker), or for different listening rooms (although adjusting in the electronic domain, for anomalies in the acoustic domain, is to be kept to a minimum IMO). Combined with the numerous advantages of active amplification per driver-way, made affordable by incredibly accurate class-D amps, i don't see the point in passive crossovers anymore.

Please contradict me (using valid arguments)? Tx!

 

[chilipalm]

Another myth is that $1000 speakers are better than $200 speakers, $5,000 speakers are better than $1000 speakers, etc. They're only better if your ears tell you they're better. Don't be deceived by higher prices by "better" brands.

 

[cutedaddy]

Well isn't freedom of speech a beautiful thing!

Dear Gene,

Although i liked your article a lot, i did think it seems to vent a sentiment against DB or ABX testing, whereas your actual point, when reading on, rather seems to be the fact ABX tests are generally executed in a flawed fashion. A good ABX testing practice is still valid, and as 'Enfant Terrible' Krabapple mentions, not used just in medics but rather in ALL FIELDS OF EXACT SCIENCE.

So Gene, as you would find from my posts (would you have time to read 'em, which i expect you don't), i find disagreeing much more educational than agreeing, but that doesn't mean i don't value your expert opinions! Tx. for your giving direction to us, bunch of wandering audio addicts...

And to mr. Krabapple: the more antagonistic the tone, the weaker the plea. Why not stick to pure content?

Tx everyone.

 

[big2bird]

Now that we have DSP, how come soldering a crossover isn't a thing of the past? When we're talking about tweaking, upgrading or bypassing standard OEM crossovers, why bother with Caps and Resistors etc. when you can adjust your xo frequency, order/slope and electromagnetic properties of the drivers real time, while listening to the result (and, when desired, while also watching your measurement graphs) realtime? With the added advantage of adjusting for driver sensitivities..

You can even add to that the freedom of using different xo's for different music material (that requires different directivity patterns of the loudspeaker), or for different listening rooms (although adjusting in the electronic domain, for anomalies in the acoustic domain, is to be kept to a minimum IMO). Combined with the numerous advantages of active amplification per driver-way, made affordable by incredibly accurate class-D amps, i don't see the point in passive crossovers anymore.

Please contradict me (using valid arguments)? Tx!

Humans as a species are reluctant to change. Industry is slow to change. Change is NOT always good.
A DSP in the hands of a novice is just an elaborate tone control.
(More later when I ponder the subject longer).

 

[Dennis Murphy]

Now that we have DSP, how come soldering a crossover isn't a thing of the past? When we're talking about tweaking, upgrading or bypassing standard OEM crossovers, why bother with Caps and Resistors etc. when you can adjust your xo frequency, order/slope and electromagnetic properties of the drivers real time, while listening to the result (and, when desired, while also watching your measurement graphs) realtime? With the added advantage of adjusting for driver sensitivities..

You can even add to that the freedom of using different xo's for different music material (that requires different directivity patterns of the loudspeaker), or for different listening rooms (although adjusting in the electronic domain, for anomalies in the acoustic domain, is to be kept to a minimum IMO). Combined with the numerous advantages of active amplification per driver-way, made affordable by incredibly accurate class-D amps, i don't see the point in passive crossovers anymore.

Please contradict me (using valid arguments)? Tx!

I actually agree with you, and I'm trying to get my designs converted over to active configuration, at least as an option. But so far, stand-alone active designs for dedicated music listening (as opposed to computer monitors) haven't passed a market test. I think we just need to find ways to get the cost down and simplify use. The hardware is all there--someone just has to get the right pieces put together. I'm trying to get a local EE to help me out, but people are busy. Particularly the retired ones.

 

[3db]

I actually agree with you, and I'm trying to get my designs converted over to active configuration, at least as an option. But so far, stand-alone active designs for dedicated music listening (as opposed to computer monitors) haven't passed a market test. I think we just need to find ways to get the cost down and simplify use. The hardware is all there--someone just has to get the right pieces put together. I'm trying to get a local EE to help me out, but people are busy. Particularly the retired ones.

Yeah, good luck yanking them away from golf

 

[krabapple]

Sean Olive, has, btw, commented on this article ....specifically on what he (Sean) takes to be Gene's four main issues with loudspeaker DBTs:

Initially Gene mentioned no companies, basically lumping all companies that practiced double-blind tests as nefarious. Harman is only given credit because Floyd Toole has the time and patience to correct his mis-statements on how we do subjective and objective loudspeaker measurements. Gene's main issues with double-blind testing seem to be the following:

1. Companies use employees as listening subjects, therefore they are biased.
My Answer: Yes, we train employees as listeners but we also repeat and validate our results using external naive listeners: we get much the same results, which tells me the trained employees are not biased. They are just more discriminating and reliable in fact about 5 times better than the average audio reviewer we've tested.

2. Companies use single speaker comparisons -- not stereo comparisons -- hence they are invalid.
My Answer: We have done both mono and stereo comparisons and the results track well. Single speaker comparisons tend to generate more discriminating responses and less noise, so we prefer to conduct the tests in mono. We also control loudspeaker positional biases via our automated speaker mover, which most companies or reviewers don't bother to do.

3. Speaker companies seem to always win the tests - so they must be flawed
My Answer: When speakers lose the tests, the test results aren't reported. Companies usually only report test results that are positive. No surprise there. I have actually reported results in AES papers where our products didn't place first. No one knows because the results are coded. The point of our tests is to use the information to improve the speaker until places first. Otherwise, why bother to do the test in the first place?

4. Companies report the tests are DB when they are SB because either the subjects have learned the sound of the speakers, the experimenters knows the identities of the speakers during the test and when they do the statistical analysis.

My Answer: Maybe the experimenter knows the speakers in the test; but during the test, the computer randomizes the order in which the speakers are presented and the listener controls the switching. The experimenter has no control over the running of the test. The computer also does the statistical analysis of results, so the experimenter really has no influence on the analysis of the test results. If the data is coded then the experimenter really can't tweak the results. Again, the tests can always be repeated with external listeners to validate the results.
With our Virtual Headphone tests that run on an iPad we have been doing DBT on headphones in different countries using external listeners, and found that we get essentially the same results as using our trained internal listeners. So, I feel comfortable that our internal listeners are not giving us biased results.

http://www.hydrogenaud.io/forums/index.php?showtopic=105148&pid=862026&st=0&#entry862026

(NB the thread title was not Sean's , 'twas mine)

 

[GranteedEV]

Now that we have DSP, how come soldering a crossover isn't a thing of the past? When we're talking about tweaking, upgrading or bypassing standard OEM crossovers, why bother with Caps and Resistors etc. when you can adjust your xo frequency, order/slope and electromagnetic properties of the drivers real time, while listening to the result (and, when desired, while also watching your measurement graphs) realtime? With the added advantage of adjusting for driver sensitivities..

You can even add to that the freedom of using different xo's for different music material (that requires different directivity patterns of the loudspeaker), or for different listening rooms (although adjusting in the electronic domain, for anomalies in the acoustic domain, is to be kept to a minimum IMO). Combined with the numerous advantages of active amplification per driver-way, made affordable by incredibly accurate class-D amps, i don't see the point in passive crossovers anymore.

Please contradict me (using valid arguments)? Tx!

I can't speak to the high-end of the market, but "the bottom" is where the changes would invariably need to start to "trickle up", because that's where trends start.

What would need to happen is dirt-cheap receivers would need pre-outs, and dedicated processors would need to be priced even less - economy of features not economy of scale. If everyone has access to pre-outs, then everyone has access to powered speakers.

Further, I'm of the opinion that to an extent passive makes more sense for surrounds because of wiring (easier to wire just speaker cable, than interconnects + power cables). So my ideal "dedicated pre-pro" still has surround amplification. Which, then, is a receiver. And so the loop begins again.

 

[wlmmn]

Gene, I love these videos that offer a refreshing break from the intellectual dishonesty of marketing brochures. I love hearing marketing BS get a kick in the teeth with cold hard facts! As someone with extensive schooling and field experience as a live sound and recording engineer, I feel like I don't hear or read enough "reality check" time in home-theater-oriented websites. I would've gone on an angry 10-minute rant on the "digital" speaker part. Like you said in another video, you don't hear a lot of this stuff in the pro audio market because it's not tolerated among technical people who have studied signal flow and the physics of sound.

 

[gene]

Gene, I love these videos that offer a refreshing break from the intellectual dishonesty of marketing brochures. I love hearing marketing BS get a kick in the teeth with cold hard facts! As someone with extensive schooling and field experience as a live sound and recording engineer, I feel like I don't hear or read enough "reality check" time in home-theater-oriented websites. I would've gone on an angry 10-minute rant on the "digital" speaker part. Like you said in another video, you don't hear a lot of this stuff in the pro audio market because it's not tolerated among technical people who have studied signal flow and the physics of sound.

Thank you very much for your compliments and insightful feedback. Unfortunately its a challenge to speak your mind these days without offending people (either manufacturers or fanboys of the products you are referring to). We try to remain brand neutral in these videos but there always seems to be offended parties nonetheless thinking we are specifically talking about them more so than a trend we may be observing. Hugo is pretty new to this industry and he couldn't help notice all the marketing fluff around speaker cables and speakers. So we just let the camera roll and I spoke my thoughts. My apologies to anyone that suffered hurtfeelioma

 

[ImcLoud]

I like the videos too, I can tell Hugo has everyone at AH working out a little more, even Nimoy is starting to look more defined while Shatner can't be helped...

 

[hifitommy]

i am glad to see that this site doesn't embrace the foolish use of DBT for audio evaluation. i will admit that if such tests are "properly" conducted, the results can be meaningful. BUT this takes a lot of skill, personnel, and equipment to execute. for this reason i commend the NRC in Canada for making their facilities available to Canadian manufacturers. the results speak for themselves. still, DBT cannot be the biggest factor in speaker and electronics evaluation.

listener training can help immensely like having the potential listener/buyer learn what real instruments sound like in live music situations of all types. i was able to make my own decisions after i acquired these skills and relating the sound of speakers and electronics to those learned real sounds.

i still strive to hone my perceptive and correlative skills. do i believe in speaker break-in? well, yes. i may not subscribe to the multi THOUSAND hour period but perhaps greater than one hundred.

wire? yes, they are different but miniscule in difference compared to speakers or electronics. the better (cleaner, that is lower distortion) the electronics and speakers, the more you can hear. big bucks on wire isn't an option for me and i have never purchased any for more than $100. i do own a couple of $1k interconnects that were won at raffles and those ARE better than most.

that will be enough from me for now.

 

[Audio Architech]

CANADA ROCKS EH !