Dayton-Wright XG-10

[bentoronto]

Here's a trace for a 1984 Dayton-Wright ESL more-or-less XG-10 with a full-range signal input. The speaker was sold with a booster tweeter although this trace doesn't include one (which should be apparent). Earlier models with similar panels didn't need extra tweeting.

The box resonance is just south of 80 Hz. With no sub, these have pretty handsome bass. I think as a result of the cleanliness of the sound, it gives your ear no clue when your ear is adding the fundamentals back in... that's my theory.

REW software, RS SPL meter with correction, 1/12 8ave smoothing.

Traces for this speaker as I currently use it tri-amped appear in another thread, post 1394.

Multiple Small Subs - Geddes Approach

[planet10]

I certainly enjoyed the 2 systems of the guys i sold XG8s (late 70s, one set was a stacked pair).

dave

PS: i fixed your link.

[bentoronto]

Thanks for link help.

Yeah, I'm a big fan of these speakers, and on a good day, the many imaginative products of this mercurial man. The size of these speakers and their 17kv bias supply make other drivers with puny 1 or 2 mm spacing seem, well, puny. Pity their heyday was before amps putting 400 watts into 2 ohms were easy to find*. I have a few dozen spare elements sitting in a trunk that I might turn into a "wall of sound" some day.

I neglected to say this curve is at 2 meters (79 inches) where my head ordinarily is, both speakers playing.

(Maybe I can figure out how to do one of those anechoic-like up-close time-gated curves to reduce room influence on the mic.)

*Tri-amping, even leaving the ESLs a quite wide and seamless 5 8aves across the middle, greatly helps the amp challenge - not to mention the sound as a whole. Even so, I am sometimes pumping big gobs of power using my Kenwood Basic class-G M2A, with 525 momentary watts per channel into 4 ohms.

[bentoronto]

Here is a trace taken an inch from the grill cloth.... OK, I poked around till I had a really nice curve but didn't "cook" it otherwise. Full-range input to a single Dayton-Wright, 1/12 8ave smoothing, which provides about the right degree of meaningful data and not too much meaningless micro levels. At least to my eyes.

The results are within a few dB or my house-curve (or the theatre X-curve) from 42 Hz to maybe 15 kHz, more or less. Pretty nice for an old speaker or a new one.

The second attachment shows yesterday's ear-location curve (both speakers) compared to the up-close curve. Is one curve more valid than the other? One sure is prettier than the other.

[bolserst]

Originally Posted by bentoronto

Here is a trace taken an inch from the grill cloth.... OK, I poked around till I had a really nice curve but didn't "cook" it otherwise. Full-range input to a single Dayton-Wright, 1/12 8ave smoothing, which provides about the right degree of meaningful data and not too much meaningless micro levels.

Hello bentoronto,

Very nice near field measurement
The first thing I noticed was the absence of any high-Q resonant peak near the bottom of the LF bandwidth. You mentioned a resonance of 80Hz in post #1 but it is barely a blip here with useful LF extension nearly on octave lower. I read somewhere that the transformer primary inductance was resonated with a capacitor to produce a low Q circuit at about 45 Hz to lift the response...perhaps that is what we are seeing.

Do you know how the diaphragm resonance is damped? Is it just from the use of the heavy SF6 gas? or is there damping material applied near the stators. Or electrical notch filter perhaps?

Speaking of the SF6 gas, is the speaker filled such that the internal pressure puffs the front and rear gas containment diaphragms outward? How do you know if some of the gas has leaked out and you are in danger of arcing.

Is it easy to get the grills off of your XG-10s? would love to see a pic of the inner workings.
I'm assuming it is similar to that attached pic I grabbed from the D-W patent.

My apologies for all the questions...kept meaning to ask you for some details on your rather uniquely designed ESLs and kinda just all came out in one post

[bolserst]

Originally Posted by bentoronto

TThe size of these speakers and their 17kv bias supply make other drivers with puny 1 or 2 mm spacing seem, well, puny.

Any idea what the diaphragm to stator spacing is on your panels? Wiki says 0.2" = 5mm. Does this sound correct?

Two other FR ESLs I know of that used similar sized D/S spacing:
- The SoundLab A-1s I got to inspect have D/S = 6mm and run the HV bias at about 14kV.
- The KOSS 1A woofer panels have triple diaphragms with D/S = 7mm. HV bias is run at a lowly 7.5kV but they get the force generated for 22.5kV since they have three diaphragms pushing in tandem. This is another way besides the use of SF6 to get higher efficiency. The only drawback to this technique besides the complicated construction is that the spaced multiple diaphragms limit the HF bandwidth.

[bentoronto]

Many good questions from Bolserst. Here are today’s best answers.

The panels used in the XG-10 are a bit evolved from those of earlier models. There’s a mysterious perforated metal (reinforcement?) plate. Frankly, I prefer the earlier models because they have better treble and don’t need an auxiliary tweeter. But then beaming needs to be addressed.

It has been a while since I peeked inside an XG-10 and so my comments are based on my collection of earlier panels.

The HV is rated something like 17kV, but I used to keep it lowish in the summer and bit higher in the winter. Which means re-adjusting the amp gains. I used a DW bias supply on my direct-drive amp and the amp added 2.4kV B+ to the stators and then maybe -10kV bias. I can’t remember values when I used the direct-drive amp with free-air DW panels.

As best as I can measure it, the spacing is 5 or 6 mm each side of the membrane. Could be less or more. That’s big but shouldn’t that lead to a cleaner output and bigger maximum output? Whatever words of theory, I might hear from the True Believers in Engineering, there’s something to be said for big gaps.

My speakers have never had their gas seal compromised, that I know of, and there are no sparks in humid weather. And I don’t plan on having cats around ever.
The sealing membranes move in and out with the seasons - I don’t know why and it isn’t just room temperature. Currently, -20C (-5F) out of doors, and they are sucked in an inch or two at the centre, maybe more. In August, often popping out an inch or two. About 5 times over the years, I’ve turned them upside down and bled the air (the gas falls and the air rises - so only air evacuates) to avoid rupturing the envelope film.

Listening to glide tones, I thought I sensed a peak at 80 Hz and decline below. Indeed, that would be true of the curve of my present listening position. You may be right about the basic, gas-filled resonance of the whole box as being 45 Hz. They make pretty nice full-range speakers, unless you are keen for the top treble and bottom bass (and you’d have to add both since it sounds bad to add just one or the other to a speaker). Funny thing, full range they certainly don’t have ESL sizzle or any boom in the bass: they are, in fact, about as characterless as any speaker could be. But my Dennesen tweeters can add all the sizzle I can handle.

Depending on the room, I sometimes have them a foot off the carpet and sometimes with a 4 x 4 foot absorbent panel behind. In the present small damped room, just 2 inches off the carpet and angled up a bit!

I have given some thought to the gas concept. If you have certain ambitions, it is brilliant trick. But for ordinary domestic ESLs, the drawbacks are too much.

[bolserst]

Originally Posted by bentoronto

The HV is rated something like 17kV, but I used to keep it lowish in the summer and bit higher in the winter.

Interesting…I figured with the ESL assembly operating inside a sealed bag you would be able to keep the HV bias set the same no matter what the humidity was outside the bag. Good point to know.

The sealing membranes move in and out with the seasons - I don’t know why and it isn’t just room temperature. Currently, -20C (-5F) out of doors, and they are sucked in an inch or two at the centre, maybe more. In August, often popping out an inch or two.

Besides the obvious thermal expansion/contraction of the SF6 that you noted, could also be seasonal change in barometric pressure perhaps? When you are playing music loudly with lots of LF energy, how much excursion do you see in the sealing membranes?

Listening to glide tones, I thought I sensed a peak at 80 Hz and decline below. Indeed, that would be true of the curve of my present listening position. You may be right about the basic, gas-filled resonance of the whole box as being 45 Hz.

I scanned thru the D-W Patent US3778562 and noted a mention that the resonant frequency with ESLs in the heavy SF6 gas is very low and in fact can be set subsonic. This might explain the non-peaky nature of your near field measurement. Most ESL dipole panels when swept downward in frequency will have an extremely obvious resonant peak unless some form of damping is used.

As best as I can measure it, the spacing is 5 or 6 mm each side of the membrane. Could be less or more. That’s big but shouldn’t that lead to a cleaner output and bigger maximum output? Whatever words of theory, I might hear from the True Believers in Engineering, there’s something to be said for big gaps.

Yes, what there is to be said for big gaps is that they are required for Full Range ESLs if you plan on putting out any useful SPL level at low frequencies.

Theory and practical experiment absolutely agree concerning gap and SPL. As I’ve mentioned before, I built test panels with 1/16”, 1/8”, and 1/4” gaps to test the Baxandall theory that the max SPL and efficiency for ESLs is dependent only on Area and the Breakdown Voltage Gradient of the gas in which they are operating. Gap size only determines how low in frequency you can put out this max SPL. From reading Sander’s book I was originally under the impression that smaller gaps always led to higher output. But, this is only true if you don’t have the capability to design a step-up transformer to match the gap. I think you are probably the only person I have run in to that believes the opposite, that larger gaps lead to higher SPL capability.

For hybrids you simply don’t need the excursion capability.
I see no reason for using gaps larger than 1mm – 2mm in hybrid ESLs.
You gain nothing by using larger gaps besides the headaches of designing a higher step-up ratio transformer and insulating things properly to keep the higher voltages from going where they shouldn’t.

For full range ESLs on the other hand, large gaps are absolutely necessary as they define your low frequency output capability.

[bentoronto]

Many changes (and production goofs) over the years, but to the best of my knowledge (until corrected by wiser people):

1. gas bag is mylar and glues resemble Pliobond contact cement - good at keeping SF6 inside and but some slight permeability to humidity; not sure I've ever heard sparking, not sure summer is worse, and not sure how damaging it is, and if there are holes in the diaphragm I am not sure how fatal that is, but I superstitiously lower the bias each summer

2. never played these speakers much without the grill cloth (which sits about an inch from the "bag"); my guess is you could readily see/feel it vibrate just as you can see light interference on a diaphragm

3. expansion of gas bag seems seasonal, not just momentary temperature expansion or weekly barometric pressure

4. my many opinions on ESL theory are entirely amateur! a big gap (which of course needs a big bias voltage) is needed for moving lots of air on loud sounds, likewise in my amateur view, a big driver loafing is better than a small driver working very hard,or at least that is the case when you are shaking blocks of cardboard to make sound as with Rice-Kellogg devices, but certainly no need to have too much gap (my Dennesen tweeters have maybe 1 mm)

5. having tested lots of matching transformers long ago, and feeling the magnificent weight of two 39 lb transformers in the DW box, I am very skeptical about the little mains toroids the DIYers are using today, can somebody show me evidence these work well compared to a Dynaco Stereo-70 tube transformer hooked up backward

[bolserst]

Originally Posted by bentoronto

4. my many opinions on ESL theory are entirely amateur! a big gap (which of course needs a big bias voltage) is needed for moving lots of air on loud sounds

But what good is a big gap if you can't produce enough force on the diaphragm too move it very far. Perhaps if you think about the fact that the bias voltage is only half of the equation. Force on the diaphragm is proportional to (Vbias)*(Vstator)/(gap^2). If you double the gap and then boost the bias voltage to the point of ionization in the gap you will still lose 6dB of output capability. This means that at a given frequency the diaphragm will only move half as far as it did in the smaller gap. You would need to double the drive signal to the stators as well to gain back the lost output capability. That is the hard/expensive part about increasing gap size...building/buying a transformer with a higher step-up ratio that will still operate over the bandwidth you desire.

likewise in my amateur view, a big driver loafing is better than a small driver working very hard,or at least that is the case when you are shaking blocks of cardboard to make sound as with Rice-Kellogg devices, but certainly no need to have too much gap (my Dennesen tweeters have maybe 1 mm)

I'd agree that in general for a given output, large area with small excursion tends to produce lower distortion output with fewer extraneous noises than a small area with large excursion.

5. having tested lots of matching transformers long ago, and feeling the magnificent weight of two 39 lb transformers in the DW box, I am very skeptical about the little mains toroids the DIYers are using today, can somebody show me evidence these work well compared to a Dynaco Stereo-70 tube transformer hooked up backward

The mains toroids that folks are using work well for hybrid ESLs because their winding geometry has low leakage inductance. This is partly because they have fewer primary turns than the ST70 backwards transformer arrangement and partly because of the longer winding length afforded by the toroidal core shape. However, your ST70 transformers will be able to drive the ESLs lower in frequency and higher in output before the cores saturate.