I don't know enough except to add a practice note.
When I adapted a direct-drive Sanders-like amp, I used a large group of resistors as the load. The ESL panels were wired parallel but they had barely any impact on the load the amp saw. As folks have stated, ESLs are actually very efficient by themselves. That made the amp very happy.
Of course, a Sanders amp is a machine to make big voltages while using today's familiar audio amp is quite different. But the notion of resistors in parallel with ESLs panels......
B.
Double Blind Listening Test.
If I was a guru in my previous life, it was on using DBLTs to design speakers & other stuff.
But I've been a real beach bum for the last 2 decades so have problems buying AES papers too.
I've been looking at "4 dimensions" for a very long time. I was instrumental in persuading John Wright to develop the "off-axis waterfall" which is a bell / whistle in Stereophile & and other hi-falutin reviews.
In another thread/forum (?) someone was dissing my suggestion that he should xover at a higher frequency with his selected units and posted an "off-axis waterfall" to show how good his speaker was. Actually it showed terrible off-axis behaviour which would have improved with a higher xover.
We now have much better tools .. some of which I helped develop.
But the knowledge and experience to interpret them is still thin on the ground.
John Atkinson has probably looked at more 'waterfalls' (KEFplots cos KEF invented them as their CDS) than any other person alive.
But in the early 80's I sat down with him over a couple of pints of Tetleys to explain what da wriggles meant in terms of speaker sound.
I've been a real beach bum for the last 2 decades so have no commercial axe to grind. Today, I don't have to be politically correct either so I can call BS if I see it.
But to get back to 'matching directivity' .. there are 3 factors.
- Response AND directivity of lower unit near xover and for at least 2 8ves on either side
- Response AND directivity of high unit bla bla
- Xover itself and the resulting acoustic response
I'm sorry if people thought I meant the directivity of the 2 units have to match EXACTLY at xover.
And there are those who believe brickwall xovers will give good sound. I was involved with early experiments to do just that with DSP at Essex U. Guess what they sound like.
If I keep harping on about the SOUND that's cos I still think making your product sound better helps you sell more. We were very successful with this strategy in da previous Millenium. Perhaps this is naive today as it is almost impossible to get to hear stuff properly in a real shop ... so people rely on 'reviews'.
One thing you find out VERY quickly if you do DBLTs is how many reviewers are deaf ... so there's no point designing stuff to suit their ears. Instead you tell them, loudly & clearly, that your stuff is hand-carved from solid Unobtainium by Virgins.
Thanks for this bolserst. Soon after that, Kellog left the horny handed fields of audio to create his cornflake empire.
Chester Rice went on to make further advances in ESLs though nothing as earth-shattering as his invention with Kellog.
My apologies to golfnut & bolserst who are of course true ESL gurus. I should've recognised this instantly and fallen on my knees in worship. Mea maxima culpa.kgrlee said:Da true ESL gurus will realise I'm being nearly as simplistic as bolserst. eg no mention of near field. I'm waiting for one to turn up on this thread.
PS Do you guys have the facilities to wind the inductors for a Walker/Baxandall type transmission line ESL?
What…no Rice krispies?
I don’t, although you can pick up sets for cheap on ebay from time to time. Speaking of LC transmission lines, Tim Mellow has been working on a similar segmented delay approach that seeks to emulate an oscillating sphere rather than a virtual point source. An oscillating sphere has uniform dipole radiation pattern at all frequencies. This approach also avoids the termination issues Walker wrestled with at the perimeter of the finite sized diaphragm when emulating the radiation from a virtual point source position a distance behind the diaphragm. The tricky part is that you need different delay values for each segment. Patent concept used digital delay and individual amplifiers, although now he has a working passive implementation. His AES and JASA papers on the topic are currently available for free download on his website. The sunflower inspired stator hole pattern is quite mesmerizing.
http://www.mellowacoustics.com/articles/AES_144_Papers_paper_45 (2).pdf
http://www.mellowacoustics.com/articles/Dipole_speaker_with_balanced_directivity.pdf
Book & papers
Products page
US20120033834A1 - Apparatus With Directivity Pattern
- Google Patents
If you can remember, what was approximate piston size, crossover frequency, and frequency of first break-up mode of the beryllium driver?
Can you elaborate on “sounds terrible”?
Do you feel it was related to directivity increase that rolls in for ka>1 with perfect pistons?
Was the terribleness noticeable only when listening to music in rooms? Or was it also noticeable as a “quality” to the sound of pink noise or test tones when measuring the driver in an anechoic chamber?
A lot of questions I know.
Attachments
What you describe gives rise to directivity increase (ie “beaming”) and off axis lobing, but not relevant for this LF topic where wavelength >> piston size.
I’d recommend not getting hung up on the word “cancellation” which doesn’t describe a dipole very well, and think in terms of the whole radiation pattern. As you said, mostly smooth broad coverage…with minor directivity index of about 4-5dB. Yes, the room can substantially affect the time-averaged sound reaching a listener, but that is not unique to dipoles; monopoles are similarly affected.
I see that you feel it important to add an additional qualifier. Let me re-state and see if you still feel that way. Floyd Toole’s data showed that when people listen to loudspeakers in rooms, they prefer that the reverberant field have roughly the same spectrum as the on-axis response. With EMLs, I only know of one way to make this happen and that is to avoid large shifts in directivity when moving from driver to driver, and avoid operating a single driver over too large a bandwidth.
Was your additional qualifier to cover the possibility that what people are liking is not necessarily that the reverberant field and on-axis response are similar. But instead, the driver compliment needed to make this happen coincidentally (in every case) has some other quality they find desirable? If so, I guess it is possible but not probable considering the many different loudspeaker types and configurations that were tested all supporting the trend.
wrong Kellog - you're thinking of John Harvey and his brother = corn flakes, but no rice crispies, and no wanking, John Harvey Kellogg - Wikipedia
Joke golfnut! Joke! Dun yus bloody Kiwis unnerstan 'joke'
... but Chester did get a couple of later patents on ESLs
Incidentally ... golfnut, I think I've got some PAFplots (not KEFplots. Their CDS are da present day 'waterfalls') of ESL63 done circa 1980 when only KEF & ourselves could do such hi-falutin stuff. It clearly shows the effects of the dustcovers on which you pontificate in another thread.
Last edited:
Hi,
so far I kept shut up and learned from the sheer mega brain powers discussing here
That is almost exactly what I did with my ESELs ... apart from the 2 8ves response (could just achieve 1 8ve) and guess what? .... they sound absolutely awesome with no audible transition between the branches
There´s just one point I tend to disagree .... and that is about the ESL63 and alikes.
Though brilliantly engineered their design concept seems vastly flawed to me.
The flaws beeing:
- that due to the quite small membrane area of each segment dynamics are seriously restricted
- that due to the small sized segment and the current steering the design becomes ridicolously high-ohmic, asking for insanenly high drive voltages (and hence stupidly high transformation factors of the audio trannies) that kill the very last bit of dynamics. I´m not at all surprised that the ESL63 sounds dull, muffled and slow.
- the concept of a pulsating sphere or a globe segment is only a theoretically good concept. In praxis -that is in a room- other directional characteristics like a strip line can be advantageous.
jauu
Calvin
so far I kept shut up and learned from the sheer mega brain powers discussing here
That is almost exactly what I did with my ESELs ... apart from the 2 8ves response (could just achieve 1 8ve) and guess what? .... they sound absolutely awesome with no audible transition between the branches
There´s just one point I tend to disagree .... and that is about the ESL63 and alikes.
Though brilliantly engineered their design concept seems vastly flawed to me.
The flaws beeing:
- that due to the quite small membrane area of each segment dynamics are seriously restricted
- that due to the small sized segment and the current steering the design becomes ridicolously high-ohmic, asking for insanenly high drive voltages (and hence stupidly high transformation factors of the audio trannies) that kill the very last bit of dynamics. I´m not at all surprised that the ESL63 sounds dull, muffled and slow.
- the concept of a pulsating sphere or a globe segment is only a theoretically good concept. In praxis -that is in a room- other directional characteristics like a strip line can be advantageous.
jauu
Calvin
Thanks for this bolserst. I'm not after ESL63 inductors.
Like Mellow, I've got my own crazy idea for a segmented delay approach but to emulate something else. Being a horny handed speaker designer, my target is something much more mundane.
I've done some theoretical work on this but the inductors look like they need to be wound from Unobtainium by Virgins
.. at least that's what Peter Walker said when I discussed it with him.I haven't forgotten this. I'm trying to remember details and I'm hoping to reply in more than a "mine is bigger than yours" fashion.
I've spent most of my working life looking at
- what speaker distortions are audible
- how to measure the important speaker distortion.
But there's a missing link between the oneD measurements like KEFplots ... & Room Interface Profile which is definitely a 3D phenomena.
Your query actually hits at this very question. More to come
Last edited:
You can't hear directivity. There is no perceptual experience called "directivity" as the term is applied to speakers.
You can experience things like tone colour, Doppler, HD, echo, beats, missing fundamental, and transient goodness. In so far as you get better SQ from a speaker with better directivity, some other parameter of sound is mediating it.
If the mediation is just FR in situ, easy enough to address that with EQ.
B.
You can experience things like tone colour, Doppler, HD, echo, beats, missing fundamental, and transient goodness. In so far as you get better SQ from a speaker with better directivity, some other parameter of sound is mediating it.
If the mediation is just FR in situ, easy enough to address that with EQ.
B.
Last edited:
Thank you Ben for your pedantry. I'm sure it advances this discussion though I'm not sure how.
And your point is ? As we are playing pedant, could you explain 'transient goodness' and how it is 'experienced'? Is it somehow related to my 'audible'?
Instead of my highfalutin' pedant's word "hear" which as you abrasively imply is within the grasp only of persons with graduate degrees in philosophy, please substitute the common parlance everyday familiar word, "audible".
There.... now aren't you happy?
B.
Footnote: let's say you are showing off your HiFi to a friend and they said, "Gosh, can you turn down the directivity, it really is too much." What would you do? Perhaps you would start by asking them what the heck they were talking about.
Last edited:
Thanks for this Ben.
Could you please elaborate on your "transient goodness"?
This is not a trick question. As an ex DBLT guru, understanding terms that Golden Pinnae (and unwashed masses) may use to describe what they hear is vital as we are using DBLTs as tools to design better speakers.
I suppose I should repeat my question about your point but I think it has been answered.
Last edited:
Yeeeeah…I shoulda listened to my wife when she tells me not to attempt humor in public because no matter the joke, I’ll likely screw it up. I started a post to correct kgrlee on Edward Kellog vs the cornflakes Kellog but then thought he must be joking and with the name “Rice” sitting right there…
Can you point to any of them? Or recall what aspects they addressed?
The only other significant transducer related patent of his I was aware of was the 1929 one on adding shorting rings in the gap to reduce voice coil inductance effects.
What does PAF stand for?
Most measurement software today uses burst decay or wavelet analysis to provide improved time-vs-frequency resolution tradeoffs.
No rush, looking forward to it.
We’ve covered this topic several times in other threads so I’ll try to keep response brief.
There are at least 2 issues covered under the umbrella term of directivity.
The first is pretty much ESL specific, and that is the extreme narrowing of directivity in the upper octaves with large uniformly driven flat panels. The result is the “head in a vise” sensation of tonal and image shifts with minor adjustments of posture or head tilting. When asked to turn directivity down, I pause and re-engage the segmentation on my ESL line source. Have yet to have somebody want to keep listening to 15” wide ESLs uniformly driven when the broader-smoother polar response from segmentation can be switched back to.
The second is what we have been talking about here, the relationship between power response and on-axis response and how it affects ambient or time-averaged response compared to first-arrival response. This is only relevant when listening in rooms. I know you feel all that is required is to plop most any speaker down and EQ at the listening seat for desired tonal balance or time-averaged house curve. But, you can’t EQ power response separate from on-axis. So at best, you are coming up with a best compromise of first arrival and ambient sound which our brain processes differently. Manipulating the directivity provides a lever to improve this compromise, as does room treatment.
You might explore a “reductio ad absurdum” experiment (thought or real) of your comment that there is no perceptual experience of this aspect of directivity(relationship between power response and on-axis response). Sit in your listening chair with time-averaged FR adjusted via EQ to your desired house curve. Enjoy a few of your favorite tracks. Now get up and point your left speaker at the back left corner and the right speaker to midway along the right wall(or left wall, or directly at the listener). Sit back down and EQ for the same house curve. Feel free to EQ each speaker separately or as a stereo pair. Then re-listen to those same tracks. Pause again to randomly shift speaker angles and repeat EQ and listening. All equally enjoyable experience?
Are you talking about your hybrid configuration of an ESL line source crossing to a line array of dipole woofers?…That is almost exactly what I did with my ESELs ... apart from the 2 8ves response (could just achieve 1 8ve) and guess what? .... they sound absolutely awesome with no audible transition between the branches
This certainly was the most seemless sounding hybrid configuration I have experienced for line source ESLs crossed in the 500Hz – 150Hz range.
purist esl MK2 - calvins-audio-pages
Attachments
Last edited:
Sheesh! You & golfnut are really digging into my senile memories. In case it isn't obvious, I'm a real beach bum.
Might have been in Bell Systems Tech. Journ. rather than a patent.Peter Arthur Fryer AES E-Library >> Analogue Loudspeaker Measurement with -3-D- Display
It was based on BBC work by Shorter & Harwood in the early 70s on Delayed Resonances though that had 'time' resolution rather than 'cycles' which was Peter's slant. Tone burst based. You need an anechoic cos the improved LF resolution. To get the anechoic 'good' enough, we had to remove the steel mesh floor which didn't endear us to da speaker designers.
This Millenium, I think Don Keele has fancy wavelet maths analysing this 'method'. I shud lern 2 reed en rite cos his papers are on his website .. specially wid da pedants on this forum
For us, PAFplots were 'cycle' resolution. KEFplots da 'waterfall's in 21st century hi-falutin reviews. There were even JAPplots from a Japanese Co. which presented KEF-like data but without more resolution.
Last edited: