Paper vs metal cone

[tcpip]

Sorry if this topic has been discussed here -- please point me to it?

I'm asking about an age-old topic, but I am still trying to understand it. How does the cone material impact the sound, in general, for a midrange or midbass driver?

I used to believe that "given all else being similar" a good metal cone driver will have a more problematic cone break-up resonance area, but will reproduce details better and have less distortion. A paper cone (or poly cone) driver will have smoother SPL curves, will have less of the cone break-up challenges, but may have slightly more distortion. "All else being similar". I am not so sure now.

I am currently playing with Dayton RS 180 (metal cone) and RS 180P (paper cone? three-part sandwich cone?) drivers, both mounted on identical enclosures at identical locations. I'm attempting to build two different MTM speakers (homage to Jon Marsh' Cauer-elliptic MTM done a decade ago). I measured their SPL, with the same measurement rig and mic at roughly the same distance in both cases.


This does not tell me anything about distortion, but at least dispels the myth that paper cone drivers have benign and smooth cone break-up jaggies. If anything, in this case the paper-cone break-up region starts earlier, forcing more demands on a crossover to suppress that region. If anyone is interested, I can upload the ARTA PIR files for these measurements too.

My basic question is: what differences does a speaker designer expect to see when he chooses one cone material over another, "all else being roughly equal"?

One common misconception is that metal cone drivers give a "harder", more "hyper-detailed" sound, and paper cones sound warmer. This may be true of raw drivers without crossovers, but with competent crossovers designed to suppress non-linear and resonance regions, I think good metal cone drivers can sound clean, detailed, without any excessive edginess or hardness. I gathered confidence from Jon Marsh' work with that MTM and used metal cone drivers for my Darbari 3-way system, with steep xo slopes, and the sound has no hardness or edginess at all.

[5th element]

The first thing that you've really hit upon is the term 'paper'. It doesn't really mean anything except contains some sort of wood pulp. You can get paper blends in all sorts of varieties, some of them a very stiff whereas some are not.

The main issue with the RS aluminium cone drivers is that not only do they have severe breakup but they also have distortion amplification from the cone resonance.

Zaphs measurements of the RS180 show this.



You can see in the third and 5th harmonics peaks in the HD that are a direct result of the resonant nature of the cone. Typically the peak in the third order is what sets the maximum recommeded xover for a metal cone driver.

Here you would need a steep xover at around 1800Hz maximum to keep the third order peak at ~2.2kHz out of the way.

The idea (one of them at least) of paper, or soft cones, is that they do not exhibit this same amplification of the distortion products. So although the RS180P starts to breakup earlier and has a relatively vicious breakup peak it should not impact the distortion in anywhere close to the same way.

I get the impression that RS have wanted to try and hold on to the qualities of a stiff cone with the paper versions, but eliminate the distortion amplification.

Usually soft cones, regardless of how stiff they are, have resonance issues between 500-1500hz that create a dip that you'd rather wasn't there. This is the price you pay for no distortion amplification.

Of course some drivers are exceptions to this rule and SBs NAC series, with their ribbed cones, do a very good job of damping the cone resonances so that distortion amplification doesn't occur.

[tcpip]

Thanks a lot. This is the kind of insight I was hoping to get.

I'll post the two PIR files here, and will also post the impulse response graphs. The impulse response graphs differ more dramatically than the SPL curves. Maybe some of you can tell me how to read these differences?

Can you please explain the distortion amplification due to the resonance? What does it mean really? I understand resonance, cone break-up, and harmonic and intermodulation distortion at an intuitive level, but can't quite "get" how the resonant peaks at 5 KHz give me 3rd harmonic distortion amplified at 2.2 KHz.

[tcpip]

Impulse measurements

Here are the impulse responses. First, the paper cone:


Now, the metal cone:


The metal cone impulse has a larger number of sharper, narrower waves, but takes roughly the same time to settle down.

Here is an image showing both the impulses in a single graph:


I've also uploaded the impulse files (ARTA PIR files) here. The one named "asawari4" is with the paper cone, and the one named "asawari5" is with the metal cone.

Interesting to see such differences in the impulses of two very similar drivers.

[Bear]

I'd bet that the motor is the same for the two, and the differences are probably driven mostly by differences in Mms (moving mass, largely cone and coil weights). But if there's not a significant difference in mass, then the motor structure is the next place to look (assuming no measurement error).

[5th element]

The mechanism is this.

You input a signal to the motor of the loudspeaker. The motor has non linearities and produces distortion products as a by product of the conversation of electrical energy into mechanical.

Let's say you input a 1kHz tone. The motor then produces a distortion product at 3kHz, the 3rd hamonic. If the loudspeaker cone is made of metal and rings like a bell at 3kHz then the 3kHz 3rd harmonic will excite the cone resonance causing it to ring and this amplifies the distortion by how much the cone rings.

For some reason distortion amplification only usually affects the odd order harmonics.

[tcpip]

... If the loudspeaker cone is made of metal and rings like a bell at 3kHz then the 3kHz 3rd harmonic will excite the cone resonance causing it to ring and this amplifies the distortion by how much the cone rings...

Oh, I see. Understood perfectly, thanks. This is why the distortion peak "moves down" the frequency axis... The cone break-up peak shows at 3K on the SPL curve but at 1K on the 3rd harmonic curve. Thanks a lot.

Follow-up question: shouldn't this combination of HD and resonance ringing equally affect all drivers which have similar cone break-up behaviour (and roughly similar levels of HD)? If two drivers have 5KHz shark-tooth jaggedness in SPL curves, shouldn't their THD curves show similar 1.6K peaks?



Sent from my HTC U Ultra using Tapatalk

[JonMarsh]

Oh, I see. Understood perfectly, thanks. This is why the distortion peak "moves down" the frequency axis... The cone break-up peak shows at 3K on the SPL curve but at 1K on the 3rd harmonic curve. Thanks a lot.

Follow-up question: shouldn't this combination of HD and resonance ringing equally affect all drivers which have similar cone break-up behaviour (and roughly similar levels of HD)? If two drivers have 5KHz shark-tooth jaggedness in SPL curves, shouldn't their THD curves show similar 1.6K peaks?



Sent from my HTC U Ultra using Tapatalk

You would expect this, but for reasons I don't fully understand that is not always the case.

Also, a significant impact can be had with the right filtering approach- both for tweeters and cone type drivers. Accuton has let slip that point occasionally when showing distortion plots for filtered drive and unfiltered. It does take some specific crossover techniques to achieve this, and they have been coy about that. The "Tafal" two way by Rainer Feile is a good example of remedying a peaking tweeter's behavior through the crossover filter-

Tafal Unfiltered:



Tafal Filtered:





You only need to look at the distortion spectrum on a number of different Accuton cone drivers to see how much variation there can be, but I do not believe it is just due to having X amplitude breakup mode. There seems to be more at work- certainly the overall driver inductance and impedance curve seems to be a correlating factor, (the old "copper in the gap" issue) with what I'm seeing in a variety of new measurements.

[Zvu]

I pondered a bit on that.

My take on distortion reduction at wide frequency range is that resonance (when excited) pushes the whole dome/cone into nonlinear behaviour and amplifies distortion a little bit everywhere. By damping the resonance, the dome/cone doesn't enter that kind of nonlinear behaviour and the distortion naturally goes down.

[tcpip]

Thanks Zvu and Jon. Am still mulling over it.

In general, taking distortion measurements of unfamiliar drivers and then keeping away from their distortion zones, not just breakup zones, in the crossover design, seems like something to follow. That's as far as I've been able to understand from you so far.

One of the things I'll start doing now is taking distortion measurements of drivers with and without xo, and just see what I get.

On that note: any guidance on how to measure driver distortion when mounted in their enclosure? Shall I just put the mic a couple of inches before the dome/cone and fire away using ARTA STEP? Should I put the mic exactly at the centre of the cone or a bit to one side? Is a couple of inches a good distance?

[Dave Bullet]

REW seems to do a pretty good job of HD measurements with sine wave sweeps. In terms of distance I think near and farfield are required for larger drivers. This is so you can pick up the full spectrum of cone and suspension induced distortion that may be swamped by nearfield? Happy to be corrected

[Juhazi]

The major problem with nearfield (1" or so) is that the mic gets overloaded too easily. 1' (30cm) might be a compromise, but you will get baffle diffractions with it. Or the box resonances if not open baffle.

I have chosen my drivers based on published by factory and independent data. When I have drivers in my (proto) baffle/box, I measure them at many distance and spl also off-axis to find out how response and distortion behave. This way I can see low end roll-off and distortion and resonances/peaks in upper end. Remember to run a high spl sweep from both/all units!

90dB at 1m distance is one standard, but I test woofers up to 105dB/1m. My mic is UMIK-1 with calibration file and I use REW and ARTA.

[tcpip]

Thanks both. Let me do some actual attempts at measurements and get back. I've never used LIMP, or done distortion measurements using ARTA. Straight off the bat, I have three drivers of my Darbari in their enclosures, which I can measure with and without their crossovers. Should I start another thread on just the distortion measurements?

Sent from my HTC U Ultra using Tapatalk

[Dave Bullet]

I found Room EQ wizard very good and easy to setup to take harmonic distortion measurements via sine wave sweeps that only take seconds. Compare that with SpeakerWorkshop which used to do several discrete MLS tone bursts and take forever if it was to be fine grained enough to see the actual breakup. My current limitation (pun intended) is that my soundcard in loopback is overdriving the card input as I turn the soundcard volume up to test distortion at volume. Might need a voltage divider?

In anycase, I could quite accurately measure the primary breakup of a Seas L15 (~ 8100Hz) and resulting even and odd order HD spikes further down (peak = primary / order), e.g. 3rd order measured at 2700Hz.

You then apply your filter and / or notches and remeasure and see how the HD is attentuated.

[tcpip]

I'll have to set up STEPS and try some measurements. Let me see how they turn out. I feel, for no good reason, that it's much better doing distortion measurements using stepped sine than with periodic noise or MLS. Speaker Workshop was in that respect a one-trick pony: it would do MLS even for impedance measurements. I'm glad I have the ARTA suite now.

Sent from my HTC U Ultra using Tapatalk

[5th element]

As far as I know you have to use a swept sine stimuli for distortion measurements in all those other programs, not noise/MLS.

[Dave Bullet]

As far as I know you have to use a swept sine stimuli for distortion measurements in all those other programs, not noise/MLS.

Speakerworkshop I think forced noise/MLS for distortion measurements - and it did it via a "pseudo sweep" via multi-tone IMD measurements, then stitching together the results. It also takes eons to execute being discrete tones. And this increases the chance of environmental noise (slamming doors, family members etc... from interrupting said measurements.

PS: Does anyone have an equal loudness curve for non-linear, specifically harmonic distortion?

[Dave Bullet]

@tcpip - this is good:

http://zaphaudio.com/nonlinear.html

In your case - re the aluminium and paper RS180 versions - take a read of the last paragraph. If you create 2 identical FR response speakers, with identical acoustic slope rolloffs, then the difference is non-linear distortion (note: this will mean different crossovers to achieve the same acoustic slopes). If you built 2 speakers - one with Paper, one with aluminium - the difference will be non-linear distortion you can hear.

[Zvu]

Difference will also be in different load that two crossovers present to the amplifier. For same acoustic slopes for paper and aluminium cone the crossovers can be quite different. DSP makes this kind of stuff much easier to compare because there isn't crossover between woofer and amplifier.

Cheers

[Dave Bullet]

Difference will also be in different load that two crossovers present to the amplifier. For same acoustic slopes for paper and aluminium cone the crossovers can be quite different. DSP makes this kind of stuff much easier to compare because there isn't crossover between woofer and amplifier.

Cheers

Does that matter to reproduction if you level match the 2 speakers? I mean unless amplifier distortion changes with load (and we're talking minor differences I assume) - would that be negligible?

[Zvu]

Depends on amplifier competence to deal with complex impedance loads and listening level. In my experience, very few things will differ in that situation with first few watts.

[tktran]

@tcpip - this is good:

http://zaphaudio.com/nonlinear.html

In your case - re the aluminium and paper RS180 versions - take a read of the last paragraph. If you create 2 identical FR response speakers, with identical acoustic slope rolloffs, then the difference is non-linear distortion (note: this will mean different crossovers to achieve the same acoustic slopes). If you built 2 speakers - one with Paper, one with aluminium - the difference will be non-linear distortion you can hear.

Partly. But I think the influence of the non-linear distortion is overstated. Here's why:

We can correct for ONE-axis only. Using DSP, one could correct for the on-axis, for the paper and metal cone, to within +/-0.5 dB (depending on the number of taps available and DSP power)

But we can't do anything about the other axes. So there's a big difference in the total radiated sound. A cone/dome is going to radiate sound in 2Pi, and taking measurements from every degree from 1 degree to 90 degrees off-axis, they are going to be markedly different.

Here's a look at the SEAS Excel W12CY001:

https://pkaudio.webnode.cz/excellia/

Here's a paper cone:

https://pkaudio.webnode.cz/illu3/

And we know from a lot of research by Toole. et al, that first reflections and total radiated sound are really important factors in terms of perceived sound qualities.

My view is that this is likely a big difference, and why drivers like AudioTechnology, smooth as a baby's bottom up to and past 10Khz on and off axis sound as well
as they do. They may do moderately well in, but not not stellar in harmonic/IMD distortion.

The other things about harmonic distortion is masking effects. Listening to the H3 of an 800Hz tone at -40dB (ie. 2400Khz -40dB) down in the metal cone vs 50dB down in the paper cone, well may be hard, and also because 800Hz is an unpleasant tone to listen to (masking effects)

[5th element]

Off axis response differences aside. No one should be using any driver high enough up that it's off axis behavior (except tweeters) actually plays into the designs overall sound. We should be crossing low enough that our bass drivers, mid/bass drivers or midrange drivers haven't started beaming yet. (Unless you're deliberately crossing for a directivity match).

We all know that the wide-narrow-wide dispersion profile of a poorly implemented two way is a bad idea but lots of people continue to cross over too high.

Besides, a drivers off axis behavior is largely dictated by its geometry and any influence the cone material has is usually very small. Even in designs where the manufacturers have directly extolled their cones benefits as being superior off axis response due to carefully controlled breakup etc. I've never seen one of these where it allows you actually crossover usefully higher to a tweeter. The cone geometry always dominates.

[Zvu]

Partly. But I think the influence of the non-linear distortion is overstated. Here's why:

We can correct for ONE-axis only. Using DSP, one could correct for the on-axis, for the paper and metal cone, to within +/-0.5 dB (depending on the number of taps available and DSP power)

But we can't do anything about the other axes. So there's a big difference in the total radiated sound. A cone/dome is going to radiate sound in 2Pi, and taking measurements from every degree from 1 degree to 90 degrees off-axis, they are going to be markedly different....

If we are talking about break up, this is not true. Break up peak treated as it should on axis will not show up off axis.

If we are talking about trying to make the paper and aluminium cone midwoofers sound the same (all other things being equal and both being pistonic in intended pass band) you should aim for the same power response not frequency response.

[cochinada]

...and why drivers like AudioTechnology, smooth as a baby's bottom up to and past 10Khz on and off axis sound as well
as they do. They may do moderately well in, but not not stellar in harmonic/IMD distortion.

I don't remember I have ever seen any independent measurements of AudioTechnology on and off axis and much less harmonic distortion. Actually their site is pretty awful. Can you show where one can see that please?

[Zvu]

http://www.audioexcite.com/?page_id=2683

http://www.audioexcite.com/?page_id=1587

http://www.audioexcite.com/?page_id=2802