Beyond the Ariel

Hi Zene,

Surely, adding a series R decreases electrical Q, thereby limiting the back-EMF induced current peaks which could trouble lesser amplifiers ?

Adding series R also impairs *dynamic* transduction capabilities and electro-mechanical damping, thereby allowing a cabinet air-spring to have much greater effect upon response - if - this is prefered instead of accurate transduction.

Thus any change in *steady sine* measured F3 can only be due to an enclosure modifying (= distortiing) *music* waveform transduction.

Cheers ......... Graham.
 
Re: Inline resistors

Zene Gillette said:
Lynn ... You may have missed my question when your computer went nuts. May I ask again?

"Would you be kind enough to give the good and bad points of adding inline resistors to gain Q when it is seems necessary to bring the F3 lower?
If already mentioned will you link to post?"
Zene

The motherboard and power supply swap will happen tomorrow, thank goodness, so I can get away from the laptop and the sometimes shaky WiFi connection.

Your question is a good one, and I've been thinking of an answer. I don't have a good one right now.

From an engineering standpoint, the power amplifier source impedance, voice coil resistance, wire resistance, and any intentionally added resistance are all in series. Of all these resistors, the intentionally added one is probably the best!

The source Z of the power amp is intermingled with the amplifier's distortion, since it is a virtual resistor comprised of the nonlinear gain elements of the amplifier and its feedback loop. The VC resistance is slowly modulated by the heating of the voice coil, with a time constant of several seconds. The speaker-wire resistance is slightly affected by corrosion on the surface of the copper wire.

So in theory, intentionally adding resistance should be quite benign sonically, and a simple way to raise Qes and Qts. It does seem a shame to throw away expensive amplifier watts in a resistor, though. This is where twiddling with the feedback of the amplifier to synthesize a higher output resistance has a certain appeal - at least all of the watts get used.

In practice - well - I remember the time Gary Pimm and I were twiddling with his speaker system, measuring and listening. He's using one of those Chinese ribbon tweeters with about 95 dB/metre efficiency, and like other ribbons, the thing is a flat-impedance resistive load, greatly simplifying the crossover design.

As it was, the speaker was quite close to flat, with the ribbon tweeter needing no more than 1 dB of attenuation to make the system almost totally flat. It sounded the way it measured, just a little bright, but a good design overall, and very lively and sparkly sounding. So we added a 0.5 ohm series resistor - not much at all, really - and measured. Ah, perfectly smooth, wonderful.

But the dynamics in the treble - and only in the treble - were obviously flattened, with a noticeable shut-in quality. Gary Pimm and I just looked at each other. This was an unexpected result - the resistor was a top-quality Ohmite wirewound power resistor - and in terms of spectral balance, the sound was more pleasant and better-balanced.

But the loss of dynamics was no illusion. When the resistor was bypassed with cliplead, the dynamics came right back again, along with just a slight amount of brightness. Now with a compression driver, this wouldn't be surprising - the multiple back-EMF resonances of compression drivers are quite touchy about any series resistance between the CD and the amplifier, with a low-Z source being greatly preferable.

But the ribbon tweeter basically didn't have any back-EMF resonances, at least any we could measure. Maybe they were lurking down in the noise somewhere, and were touchy about any series R between them and the amplifier.

Gary and I don't like tipped-up speakers, preferring a warm tonal balance. But we both agreed that choosing between a trace of brightness and vivid dynamics, versus even response and flattened dynamics, was easy. No series resistance in the Pimm loudspeakers. This was something I'd noticed about the Ariel some years back - that no series R sounded best, regardless of balance. Even something as marginal as bi-wiring opened up the sound - and bi-wiring is pretty small potatoes in my book.

So intentionally adding series R is something that should be benign, but in sonic terms, has a subtle, and maybe not-so-subtle, flattening effect. I am a bit suspicious of CD horn shelf equalization for a similar reason - series R does something to the sound that isn't good, but I'm not too sure why, except for interactions with low-level back-EMF resonances.

I guess this comes down to getting burned with the complex crossovers I used to design in the late Seventies. The systems were flat, coloration was very low, considering the dreadful drivers of the day, but the sound was undynamic and tone colors were muted. Not much sparkle or liveliness, in other words.

12 years later, when I was designing the Ariel and its 5 dB greater sensitivity, I discovered that the crossover was exquisitely sensitive to parts quality, especially the capacitors in the HF section. The Sprague 730P and Hovland were the only ones of the day that seemed transparent enough, and the Hovland irritatingly enough required a crossover adjustment to maintain the same subjective balance. I was not happy to make this discovery - parts tweaking is one of my most annoying and least rewarding hifi activities, something I'd rather not do at all.

Nowadays we have many more choices but cap coloration is still an awkward business in a high-resolution speaker - and I found out the hard way that the higher the efficiency, the less forgiving the speaker is of mediocre crossover parts. You can throw junk at a 86 dB/metre speaker and it isn't that audible. 92 dB/metre and up is a different ballgame. I'm expecting 97~100 dB/metre to be completely unforgiving of crossover parts quality - a severe incentive to keep complexity down. These drivers are expensive - the idea of wasting that much money on crossover parts isn't attractive at all.
 
Hi

I made basically the same observations on passive crossover parts sensitivity and the impact of series resistance between amp and speaker. This is also the reason why I go with active designs when ever possible.

Interesting enough that the relatively high impedance resistance of output transformers does not seem to affect the quality of a chain in the same way.

Adding an Ohm or so to the output of a solide state is not really a good possibility to mimic a tube amp sound wise.


Greetings
Michael
 
Hey Lynn, I know you were looking at AMT's originally as a possiblity for a tweeter for this. I guess I'm the cheerleader for these things, so I thought you might want to see this. Evidently Beyma is in the process of releasing a pro audio AMT tweeter. It doesn't look like a dipole (Should I say it looks like it has a cup on the back), but it looks like a pretty potent unit. I hope their pricing is a little more favorable than Mundorf's.

Beyma TPL-150

An externally hosted image should be here but it was not working when we last tested it.
 
JohnL said:
Hey Lynn, I know you were looking at AMT's originally as a possiblity for a tweeter for this. I guess I'm the cheerleader for these things, so I thought you might want to see this. Evidently Beyma is in the process of releasing a pro audio AMT tweeter. It doesn't look like a dipole (Should I say it looks like it has a cup on the back), but it looks like a pretty potent unit. I hope their pricing is a little more favorable than Mundorf's.

Beyma TPL-150

Hmmm ... thanks for the link. Looks very interesting. So now we have three candidates - large-format compression drivers & low-diffraction horns, the twin-RAAL ribbons, and the new Beyma AMT driver. Somehow I doubt it has the insane pricing of the Mundorf. The comprehensive specs, including distortion and vertical directivity, are a welcome change.

Close reading of the specs makes it fairly evident that crossing below 1.5 kHz would be unwise - distortion is rapidly increasing, and the response is starting to go up and down, never a good sign in what should be a piston-response region. The dispersion specs shows that the vertical directivity narrowing starts above 3.5~3.8 kHz.

This prompts the thought that if you stacked two of these things, the upper one could be 1st-order lowpassed at 4 kHz, so there would two working in parallel in the 1.5~4 kHz region, and gradually transitioning to a single driver above that. With a declining magnitude, a 45~90 degree phase relationship between the upper driver and the lower driver, and tipping the upper driver backward by 10~20 degrees, that would prevent directivity nulls from getting too deep.

A series-shunt crossover might be interesting here. To implement the transition between one and two AMT (or ribbon) drivers, both are connected in series, with the upper driver bypassed by a capacitor selected to give a 4 kHz rolloff. This shunts the upper driver above 4 kHz, diverting its energy to the lower driver. The whole array has a 12~24 dB/octave crossover filter between 1.5~1.8 kHz.

(For a 3-driver array, the upper and lower drivers are connected in parallel, and both are shunted with a 3~4 kHz value capacitor. This group is connected in series with the center driver. As before, all drivers work together at the lowest frequencies, and narrow to one at the highest frequencies. You can see the similarity to the LF driver array.)

Most interesting. We have have some fun choices here, well outside the conventional wisdom on how to build loudspeakers.
 
Hi


Wow even 2 new interesting speakers
Tanks JohnL and thanks el`Ol !

BTW, the Beyma link does not work.

My heart is with the AMT's having some of the vivid attributes described by Lynn for the CD's / horn's.
Though I also have some nice memories for the Shackman DIY ELS as I managed to get the schematic of the electronic magazine straight that time. There was a severe fault in the PSU bootstrapping circuit for the supply and polarisation voltage that affected the whole transformerless (transisor-) amp.

I also tried this ESL with a horn but I think I didn't know enough about horn theory to make it work, roughly two decades ago.

The max-SPL plot of the Sonus Novason ESL

http://www.sonus.de/download/pdf/da...df?PHPSESSID=1531e618551baca3599a75128b83408e

is something that was published by pro-manufacturers like EV regularly many years ago but unfortunately became unpopular in the days where simulation software is assessable for anyone .
110 dB / 4m down at 200 Hz is really impressive. Is this ESL really available? The plots are measurements of a prototype as far as I can see.


Greetings
Michael
 
Hi Lynn

The Duelund resistors are designed for crossovers in that as current increases the extra heat generated does not raise the resistance of the resistor thus preserving dynamics .

They come at quite a price , $20.00usd last time I looked . They are used in Peak Consult and a few other highend speakers . Available at Parts Connexion . Maybe this would be an answer to a more neutral but still dynamic speaker ???

Joey B
 
Let's try again. The original Beyma link downloads fine for me, but it did appear on my desktop with a malformed extension, "TPL150.pd", instead of "TPL150.pdf", as it should have. If your browser is set to auto-launch instead of downloading the target file to the desktop, I could see how this would happen.

As it was, I just retyped the extension, and it became visible to Adobe Acrobat. (My OS is set up to always display extensions in file-names.)

Beyma TPL-150
 
For a 3-driver array, the upper and lower drivers are connected in parallel, and both are shunted with a 3~4 kHz value capacitor.

Lynn,
This approach works well for increased headroom but you may have meant in series...otherwise the upper and lower drivers carry a lesser "LF" burden. Still, only one cap is needed with the (physical) middle driver wired (electrically) on either end of the 3 driver chain.
Paul
 
Resistors

Thanks, Lynn and guys for help on the inline resistor problem. It's been bugging me for years.
Michael Percy carries a heat sinkable mounting resistor. Looks pretty good.
One more question (Colombo)?
If I use one that adds only pure (OK, more pure) resistance will I defeat most of the negative aspects, especially dynamics problems?
I will do some testing, but didn't want to spend a fortune on resistors and find they were not worth the effort.
I'm building two systems at the same time, one all horn and other OB.
It will all be modular so I can mix and match. For instance OB coaxes with my horn bins. That is where the resistor idea originated so I can
manipulate the crossover regions. Incidentally, system will be all active as I have an Ashley stereo 4 way crossover to make experimenting a lot easier. Plus I am finishing an 18" sub that I can sneak over from HT to hi fi room if I want to try something radical. Combinations will be many. I'm not immune to trying something stupid.

Zene
 
Hope the people that need to are making notes as this thread gets longer, I forget from a couple of pages ago let alone the beginning!!!

I realise things are not finalised yet (are we close??) but was just idly wondering to myself what the estimated cost of this is at the moment, things are starting to gel and variations are lessening so that may make a preliminary costing feasible.
 
Hi Zene,

The voltage dropped by a series resistor is proportional to current flow.

Current flow through a LS is NOT proportional to applied voltage, not even with a tiny ribbon, and especially not with a horn/compression driver. Current flow through a LS varies in time with applied waveform according to reactance and priorly excited energy storage mechanisms.

If you put a resistor in series with an ultra-low distortion power amplifier (voltage output) then the voltage on the LS side of a series resistor (or longer cable with its own inductance) will be distorted by both the passive and dynamic characteristics of the LS.

This would not apply to an amplifier having current output when only a single driver is in use because inserting series R will merely waste output.


Cheers ......... Graham.
 
Hi

The pro audio ELS just made me thinking abut some side aspects of highly asymmetric directivity loudspeakers like line arrays, small long ribbons / AMT's or the multiple speakers of the current OB project radiating within the same frequency band.

One of the effects here is that the SPL does not seem to fall at –6 dB with double distance but only around –3 dB du to the cylinder shape of the wave front.
Lynn made a side note elsewhere that there might be a possibility to virtually HEAR the shape of the wave front.

Correct FR equalisation presumed, I doubt that this is possible in free field but within a room I see a possible explanation.
Air is also a frequency dependant absorber which is varying with pressure, temperature and humidity. With around –6 dB at 90 m and 20 kHz at normal temperatures and humidity (needing 4 times the distance for any half frequency to achieve the same –6 dB) this effect is NOT very distinct but may be enough to add a slight coloration that can be detected.



humid.gif

http://www.doctorproaudio.com/doctor/temas/ref_humd.htm

http://www.earthworksaudio.com/tech/hf_sound.pdf


The reflected wave fronts from the walls and furniture preserve a fraction of the original wave front depending on the reflection area with respect to the wave length. The oftener the sound was reflected the smaller the fractions will get. Frequencies below that wave lengths most certainly will change their decay behaviour back from –3 dB to –6 dB each distance doubling.

The interesting point now is that we should be able to compare the characteristic of the original wave front to the characteristic of the decay part.

In an normal listening environment this means that the decay attenuation of the sound due to diffraction of the cylinder shape will be lowest for the highest frequencies.

This overlays the "normal" absorption characteristic of any room and should provide a possibility to discriminate different wave front shapes originating form the speaker.
For highly asymmetric directivity speakers I would expect a brighter sound perception in rooms with same FR equalisation.


Maybe I am just splitting hairs here.
Any thoughts on this?

Greetings
Michael
 
Lynn Olson said:
Close reading of the specs makes it fairly evident that crossing below 1.5 kHz would be unwise
...
This prompts the thought that if you stacked two of these things, the upper one could be 1st-order lowpassed at 4 kHz, so there would two working in parallel in the 1.5~4 kHz region, and gradually transitioning to a single driver above that.


Meaning that those two Beyma's can be crossed at 1.5 kHz to e.g. 12NDA520 (LF supported by two 15NMB420), just before cone breakup ((mis)reading the 12NDA520 curves there...).