A pearl from the Bobfather

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all,
Last week I spent some time with Bob Carver at the annual Carverfest retreat in NC. We discussed the OB hybrid ESLs that I had brought with me, and Bob had an interesting suggestion which I will share below.

I had mentioned to Bob that, with hybrid ESLs, the woofer exciting the diaphragm’s resonance, and phasing errors between the two are major concerns. I had also mentioned that I use a DSP to time-align the woofer and ESL.

Bob replied that time-alignment after the fact can’t make the two drivers work together as one, if they aren’t in proper physical alignment to start with. And he followed up with a suggestion on how to find the correct physical alignment between the two drivers, in the design phase:

With ESL mounted on a baffle of design dimensions, and the woofer hand-held at its mounting location:

1. Paint a white dot in the center of the diaphragm for reference.
2. Using a tone generator; play an 18-20Hz sine wave through the woofer only.
3. Observe the white dot and resulting diaphragm motion induced by the woofer.
4. Find the point of physical alignment by moving the woofer forward or backward until the diaphragm motion stops or is maximally reduced.
 
I think the assertions in the OP post need some examination, starting with why pay attention to the thoughts of Carver who seems addicted to "imaginative" theories and designs.

Next I have to say, I've always just taken for granted that DSP time alignment works roughly as advertised, at least if you do the set-up empirically.

B.
 
I figured this thread might spur an interesting discussion. Hopefully, others will jump in with their thoughts as well.

Bob certainly has a flair for unusual designs and catchy labels to market them. That said; over the years I've grown to admire his imagination and intellect-- especially in his Sunfire amp designs. He's definitely old school though--more often than not opting for analog over DSP.

Regarding Bob's assertion that the woofer and ESL diaphragm must be physically aligned before they can act together as a single element, I too was initially skeptical, but after further thought I'm prepared to concede that he may be right-- insofar as time alignment with and without physical alignment are not equivalent, and physical alignment may offer an advantage.

Consider the case where the woofer is rear-mounted on the baffle, as in the speaker shown below which started this whole discussion with Bob.

In this case, the two drivers are not on the same plane, and a digitally applied time delay is required to shift their outputs into phase at the listening position. In this case the woofer must start ahead of the ESL because its output must travel further to reach the listening position. I believe Bob was asserting that the drivers can't work together as one when one must start before the other, for their outputs to reach the listening position at the same time.

Now consider the case where the woofer is moved forward into physical alignment with the diaphragm. In this case the woofer and diaphragm are simultaneously in physical and time-alignment, both moving in sync, and their outputs reaching the listening position at the same time. Or in Bob's words; working together as one.

The questions now are whether the two cases noted above are equivalent, and whether case offers an advantage over the other.

Video

IMG_0906.jpg
 
Last edited:
...In this case, the two drivers are not on the same plane, and a digitally applied time delay is required to shift their outputs into phase at the listening position. In this case the woofer must start ahead of the ESL ...[/IMG]
Yes, obviously requires quantum physics analysis. Or what has Carver been drinking?

Granted, alignment can only be perfect at one location; but that location is broad and the alignment only slightly erroneous elsewhere.

B.
 
I'm pretty sure there won't be a null no matter how you translate the woofer as long as the motion of the woofer is perpendicular to the plane of the ESL.

Unless you are crossing over really high, the woofer is going to be pretty close to phase aligned with the panel, due to the shortest wavelengths of the woofer being several meters long or more

sheldon
 
I'm pretty sure there won't be a null no matter how you translate the woofer as long as the motion of the woofer is perpendicular to the plane of the ESL.

Unless you are crossing over really high, the woofer is going to be pretty close to phase aligned with the panel, due to the shortest wavelengths of the woofer being several meters long or more

sheldon

The concern really was the woofer driving diaphragm, rather than any audible phasing difference between the two-- and whether having the two drivers in exact physical alignment would be advantageous to that. The speaker sounds really good, so the concern is more theoretical than actual.
 
…Bob replied that time-alignment after the fact can’t make the two drivers work together as one...
Bob is right that time-alignment by delaying signal to one of the sources will not make them work together as one the way physical alignment will. It can provided identical results for one point in space (like the listening position) which can improve things, but not all points in space. Taking your thought experiment in post#3 a bit further...think about what the signal phases would be like in the reward direction once you had them delayed for in-phase arrival in front of the speakers. Adding delay to correct things in the front will actually make things worse toward the rear! Fortunately, as has already mentioned, the physical misalignment is a small fraction of the wavelength at your low crossover frequency so not a practical concern.

BTW using electrical delay to compensate for physical separation may not be ideal for your situation, but it can be useful for creating directional speakers…in particular subwoofers. Basically, you separate two monopole woofers by a distance and then invert signal to the rear woofer and delay it by the separation distance. The result is a cardioid radiation pattern with a null toward the rear. See attached excerpt from 1972 Olson AES paper. Internet searches for “cardioid bass array” will uncover more details if interested. This is becoming a popular pro-audio solution for steering bass.
 

Attachments

  • cardiod_sub_array_Olson.png
    cardiod_sub_array_Olson.png
    13.8 KB · Views: 288
…Find the point of physical alignment by moving the woofer forward or backward until the diaphragm motion stops or is maximally reduced.
Bob is not correct about the null.
See attached plot of pressure and particle velocity trends around a dipole speaker. It is interesting to see the direction of the particle velocity flip around as you move from in front to along side the woofer. If you happen to have a ribbon microphone, you can actually measure its magnitude and direction.

Notice that although there is a very clear pressure null to the side of the dipole piston, the particle velocity(what is exciting your ESL diaphragm) does not. In fact, if you look just at Vx, the X-axis component of velocity (ie the velocity normal to the plane of the ESL diaphragm) you can see that it would actually be slight larger when your ESL is in perfect physical alignment with the dipole woofer pressure null than if it was a bit forward or aft. It’s a really shallow trend though, not at all like the distinct pressure null. One big thing to note from the equations is that at low frequencies the particle velocity will fall as the cube of the distance from the dipole source. So small additions of separation distance can make a big difference in how much the woofer is exciting the diaphragm.

One further thing that interested me, was that at a 54.7 degree off-axis angle from the source Vx = 0, so that is where you would need to place the ESL to find a null, or minimum in excitation. The reason this interested me actually has nothing to do with ESLs or Acoustics, but antique radios! Back in the 1920s the “magic” 54.7 degree angle was patented by Hazeltine (US 1577421) for use in cascaded RF amplifiers in early unshielded radios.
 

Attachments

  • Dipole_pressure-velocity_field.png
    Dipole_pressure-velocity_field.png
    49.1 KB · Views: 297
  • neutrodyne_01.png
    neutrodyne_01.png
    240.7 KB · Views: 307
  • neutrodyne_02.png
    neutrodyne_02.png
    182.4 KB · Views: 306
Bob is not correct about the null.......


Thanks for the excellent info. I'm not sure I entirely follow your extrapolations from the theory?



It seems to me that for an ESL membrane to be excited by air there needs to be a pressure differential. A pressure differential can obviously created by air movement creating a negative pressure via a differential laminar flow, or positive pressure with eddies. However the membrane movement force created by velocity of air flow across a panel is surely directly related to pressure effects. It seems to me your positing that air flow can move or excite the diaphragm in a pressure null. That appears to be logically inconsistent over a diaphragm? It seems the theoretical basis is modelling free air effects rather then what will occur over a diaphragm?


Practically the design issue with the OB driver adjacent to the ESL membrane seems to be that the membrane becomes part of the dipole baffle with clear pressure and velocity effects impinged on the mylar. However most of the pressure and velocity effects can intuitively be expected to occur close to the driver down low and rapidly dissipate with distance leaving most of the ESL membrane relatively unaffected. This would explain the sound quality.


Still it appears to me the ideal design would have a slot between the OB bass driver and the ESL to allow for unimpeded air flow under the ESL, allowing the dipole pressure null to fully envelope the ESL panel?
 
It seems to me that for an ESL membrane to be excited by air there needs to be a pressure differential.

Being excited by or sensitive to pressure differential, pressure gradient, or velocity are all ways of saying the same thing. Like ribbon microphones, ESL membranes are velocity sensitive. The velocity field and pressure field are related to each other. If you are mathematically inclined, most acoustic texts(ie Beranek, Olson, etc) provide examples of how to calculate one from the other for a variety of source types. Keep in mind that the pressure at a location in space is a scalar quantity whereas the particle velocity has both magnitude and direction.

It seems to me your positing that air flow can move or excite the diaphragm in a pressure null. That appears to be logically inconsistent over a diaphragm?

That is exactly it, but think air particle vibration rather than flow. Take a pressure sensitive microphone and move it forward and backward past a dipole woofer and you will easily find a null or minimum pressure point(your ear is pressure sensitive). Do the same thing with a velocity sensitive microphone (oriented parallel to woofer piston) and you will find that there is no null as you sweep past the woofer. In fact the velocity is slightly higher at the point of the pressure null than at positions just in front or behind it. Since ESL membranes are velocity sensitive, as odd as it may sound there will NOT be a null in the excitation by the dipole woofer even with the ESL positioned at the location of the pressure null. Since the velocity vector is parallel to the woofer motion, rotating the microphone(or ESL) 90 degrees with respect to the woofer while at the null location will effectively remove the acoustic excitation from a velocity sensitive device.

If you are interested in measuring the velocity field but don’t have a ribbon microphone, a small magnetic planar like the BG Neo3 or Neo8 can be used as a microphone in a pinch. You can also use the ESL itself as a microphone to measure exactly what is happening. Peter Walker first suggested this in his AES paper and after a little prodding from golfnut I can say it works rather well for quantifying what is going on at low frequencies. On a related note, standing waves in pipes produce maximum particle velocity at pressure nulls/nodes.

It seems the theoretical basis is modelling free air effects rather then what will occur over a diaphragm? Practically the design issue with the OB driver adjacent to the ESL membrane seems to be that the membrane becomes part of the dipole baffle with clear pressure and velocity effects impinged on the mylar.

If you can model a woofer’s velocity field covering the location of the ESL diaphragm without the ESL in place, you will find the velocity imparted to the ESL diaphragm is nearly identical to what was calculated. Borrowing words from Peter Walker, the ESL diaphragm is for all practical purpose acoustically transparent. Peter Walker demonstrated this during AES talk by holding up a stretched diaphragm in front of his face while continuing to talk. For low frequencies, you can demonstrate it to yourself by testing a small dipole woofer in 3 ways:

1) with minimal baffling
2) with a large flat wooden baffle
3) with a thin frame the size of the baffle in 2) but with a diaphragm stretched on it.

You will find that 3) will measure nearly identically to 1) rather than having the increased low end response from 2).
The only differences will show up at the resonance frequency of the diaphragm.

…most of the pressure and velocity effects can intuitively be expected to occur close to the driver down low and rapidly dissipate with distance leaving most of the ESL membrane relatively unaffected…
Correct. As mentioned in the previous post, for low frequencies the velocity field falls as the cube of the distance. So, double distance and velocity falls by factor of 8. Distance is your friend when trying to minimize acoustic excitation of membrane.

...it appears to me the ideal design would have a slot between the OB bass driver and the ESL to allow for unimpeded air flow under the ESL, allowing the dipole pressure null to fully envelope the ESL panel?
The gap will be more helpful in a line array situation with a column of dipole woofers adjacent to the ESL panel than with a lone woofer below the panel. Even then, its effect is much smaller than anticipated…not much more than just the added distance would give on its own. This surprised me the first time i tried it. But, the gap is also beneficial for reduction of mechanical vibration which may be equally important. Adding rearward baffle extensions on either(or both) sides of the gap to increase the path length will reduce the particle velocity local to the ESL diaphragm.
 
...velocity imparted to the ESL diaphragm is nearly identical to what was calculated...

... On a related note, standing waves in pipes produce maximum particle velocity at pressure nulls/nodes.

Appreciate the analysis, as always, not that I am able to understand it fully.

But back to the question of diaphragm movement, it seems the diaphragm isn't getting too much impetus to move after all. But for whatever imbalance of pressure it is experiencing from the woofer, an ESL panel seems transparent because it is moving the same as the air would be in the absence of the film. Not standing still. So wouldn't the motion be undesirable, imparting Doppler-like and other distortions?

About the second quote, is that right?

B.
 
… an ESL panel seems transparent because it is moving the same as the air would be in the absence of the film... So wouldn't the motion be undesirable, imparting Doppler-like and other distortions?
Yes. Any motion at LF imparted to the diaphragm by an adjacent dynamic dipole will produce IM sidebands around HF tones. This is no different than when a full range ESL is generating the LF diaphragm motion itself. But, with most hybrid ESLs, we are talking about 1/16” of motion or less so the Doppler related distortion tends to be a non-problem since sideband tones generated from that small motion are far down in level.

About the second quote, is that right?
Yes. A couple websites on the topic.
Standing Sound Waves (Longitudinal Standing Waves)
http://www.physics.mcgill.ca/~guymoore/ph224/notes/lecture20.pdf

The Animation provided at the first link is rather nice. Easy to see pressure nodes at maximum motion points and vice versa.
Note also that the stationary end of the tube on the right is a reminder of why placing foam on a wall(where there is little air particle motion) does little to damp room modes. You need thick foam or foam spaced away from the wall to add some acoustic resistance where it will do some good.
standing.gif
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.