Beyond the Ariel

[john k...]

These may be of interest.

Discussion of stored energy in drivers.

Stored energy in crossovers.

Impulse response of dipoles with emphasis on the mid frequency range, baffle and source type.

[Pano]

Thanks John!

[Pano]

Originally posted by Lynn Olson
'''a scaled-up Fonken seems like a good candidate for the 414, ...

AKA "Le Petit Onken".

[Lynn Olson]

What I like about the Fonken are 45-degree slant sides, similar to the RCA LC-1A Harry F. Olson design. It wouldn't be rocket science to re-calculate the box volume to comply with a modern minimum-group-delay alignment (tending towards a Bessel/Gaussian highpass characteristic), and use slant sides a la Fonken/LC-1A practice.

The Bessel/Gaussian highpass function isn't all that exotic - my first subwoofer in 1979 used that alignment - and yes, they do sound better, mostly for the simple reason it is less sensitive to driver parameter shifts than the more common alignments. Conventional 4th-order highpass filters (vented boxes) are quite touchy about parameter shifts, and low-Q alignments can help quite a bit. If your simulation software has a group-delay visualization feature, I'd certainly use it. Minimizing group-delay variation is a lot more important than trying to chase the last few Hz out of a given driver.

JohnK, thanks for the links. In my usual contrarian way, I mostly agree but not 100%. I've seen some very unusual behavior in the waterfall displays that makes me doubt how linear drivers are in the breakup region. (By "linear" I mean capable of being modeled by minimum-phase RLC networks).

Here's an example from Page Five of the MLSSA Gallery (shown below). It's an Eton 4-203 showing what can only described as perverse behavior - a resonance that starts at 4.5 kHz at T=0 and slides upwards to 5 kHz by T=1.5 mSec. No conceivable RLC network can do this; it takes a mechanical device - like a piano or a driver in breakup - to have resonances behave in this way.

This was a highly regarded driver back in its day. But how would you design a notch filter around a sliding resonance? If it were tuned to the starting point, 4.5 kHz, the resonance would "fade in" after time, an extremely unnatural result. If the filter were tuned to 5 kHz, the resonance would fade in and out, like a very fast shortwave broadcast with lots of multipath skip.

Other breakup behavior, for lack of a better word, looks chaotic, and does not respond to in-band equalization. The tip-off to chaotic behavior is when the waterfall looks radically different when the microphone is moved a few inches to one side or the other. A lot of the "exotic" rigid cone materials have trouble with this - I've seen it with Kevlar, carbon-fiber, and the various composites that get thrown into paper cones.

[planet10]

Originally posted by Lynn Olson

The Bessel/Gaussian highpass function isn't all that exotic - my first subwoofer in 1979 used that alignment - and yes, they do sound better, mostly for the simple reason it is less sensitive to driver parameter shifts than the more common alignments.

I don't know if i hit bessel with the Fonken, but one of the goals was to get a bass alignment that was less sensitive to parameter shifts. We've been able to extend the success we had with the Fonken to a whole range of boxes.

dave

[Lynn Olson]

Originally posted by planet10

I don't know if i hit bessel with the Fonken, but one of the goals was to get a bass alignment that was less sensitive to parameter shifts. We've been able to extend the success we had with the Fonken to a whole range of boxes.

dave

You've hit it, or are very close. Congratulations. The insensitivity to parameter shifts means the inherent 2-pole filters within the 4th-order highpass function are low-Q. If they had high-Q elements you'd be back to conventional alignments, with their notorious sensitivity to driver parameter shifts.

Continuing the previous post, here's a pix of some high-end speaker (costing more than $7,500 a pair) I took some years ago. I don't remember its name, but it was one of $tereophile's Class A or Class B favorites, for what it's worth. These curves, by any definition, are pretty bad. The frequency response (as seen at the rear of the graph), is not good, and the waterfall is pretty gross as well. I would not expect this speaker to have low coloration, no matter what the reviewers said.

[JoshK]

How do you go about modeling a priori for this alignment? Can you model this in something like Unibox or WinISD?

[Lynn Olson]

No idea. I just fiddled around with various alignments until the group-delay looked smooth around the F3 region. I'd like to know if there is software that has group-delay optimization - I haven't seen it yet, but there's lots out there I don't know about.

One alignment I avoid is the Extended Bass Shelf - the ones I've seen have not one, but TWO group-delay bumps, one at the F3 corner and a higher one at the sometimes fairly abrupt shelf-transition frequency.

The nasty thing about LF group-delay variation is the physical number is very large - an apparent front-to-back shift of many feet in the span of a few Hz. Is it audible? People have been wrangling about this for decades - you'll see AES papers written in the 1970's that argue the merits one way or the other. Recordings have huge phase shifts at LF, but these are typically pretty smooth, unlike the sharp highpass characteristics of vented loudspeakers.

By contrast, HF group-delay variation isn't pretty to look on a graph, but the physical numbers might be very small - an inch or less. Maybe it's audible, maybe it's not. Musicians usually wriggle around more than that when they're performing.

[JoMoCo]

Originally posted by planet10


I don't know if i hit bessel with the Fonken, but one of the goals was to get a bass alignment that was less sensitive to parameter shifts. We've been able to extend the success we had with the Fonken to a whole range of boxes.

dave

Dave, why not post or link the dims here for an Onken / RCA LC-1A with the Altec 414 driver for our digestion & DIY and we can proto some cabs for testing...or do you have a mature product availible for this driver & or plan to market one in the near future?

[planet10]

Originally posted by JoMoCo



Dave, why not post or link the dims here for an Onken / RCA LC-1A with the Altec 414 driver for our digestion & DIY and we can proto some cabs for testing...or do you have a mature product availible for this driver & or plan to market one in the near future?

Unfortunately i have little interest in this driver and at least a solid month of designs & drawings in the queue already. I have discussed the philosopy behind the Fonken design in at least a couple threads... probably the most recent in a thread started by Bigun.

dave