Hooray! A link! Unfortunately, there's no info in there on sound output from speaker cabinet walls - the only panel he measured was an OB.
...but it is intriguing. Maybe I should pick up a contact mic for comparative tests. At $20, it is a pretty cheap tool.
I use a cheap (around 10-15 Euro) violin bridge piezo pickup. It is quite insensitive to airborn vibrations compared to the contact induced vibrations, so speaker radiation is not really an issue.
I connect the pickup to one channel of an oscilloscope for observing the panel vibrations at different panel wall positions, while on the other channel I connect a link to the loudspeaker excitation signal.
If I want to make a recording, I use the pickup channel and the Y-Out of the oscilloscope to feed one channel of a soundcard . The other channel of the soundcard is connected to the excitation signal
For fast comparative observations a stethoscope is a great tool
George
I connect the pickup to one channel of an oscilloscope for observing the panel vibrations at different panel wall positions, while on the other channel I connect a link to the loudspeaker excitation signal.
If I want to make a recording, I use the pickup channel and the Y-Out of the oscilloscope to feed one channel of a soundcard . The other channel of the soundcard is connected to the excitation signal
For fast comparative observations a stethoscope is a great tool
George
Maybe I should forget the tests and simply ask: Dr Geddes, how would you build a standard box (setting aside drive units, directivity etc), if you were limited to relatively commonplace tools + materials?
My purpose is to build good boxes that are lighter than my overbuilt backbreaker boxes, so I want a good attenuation:weight ratio from the cabinet walls.
In "The Geddes Loudspeakers Design Philosophy", your have ideas in common with that old BBC report:
1) make the shell rigid
2) add a layer to make the shell non-resonant
...but you also:
3) add bracing that damps it some more
4) use lots of rounding on the baffle edges (to minimise diffraction)
I was thinking, for my shell:
Build roundovers as a stepped curve, as illustrated.
The shell would be ply (4mm F27)* onto which I'd laminate an aesthetic layer (5mm strand woven bamboo). I'd line it internally with a thin layer of fibreglass. For a light ~10mm shell, that's about as strong as I think I could make it.
For damping:
Glue 10mm rubber matting (recycled car tires) over all internal surfaces.
Is there anything reasonable (no PhD or complex equipment required) that would substantially improve on this?
*Australia uses F grades F4 to F34: E.g. F14 indicates that the basic working stress (in bending) for that timber is around 14 MPa.
Attachments
I'm often dismayed at the bad judgement you see with some construction projects at DIYaudio. The principles are very simple and self-evident, and with just a wee bit of intuition are easy to apply.
Just for a little sampling of principles, you want bracing between opposite walls, minimize unbraced surface expanses, avoid too much symmetry and repeated dimensions, choose materials wisely, etc.
One trick is to epoxy lengths of metal electrical conduit tubing (or hardwood dowels or old hockey sticks) to opposite sides. Light and rigid.
I kind of agree with Earl that there's no need to over-do cabinet construction as appealing as that seems intellectually. On the other hand, I've never walked by a speaker box that I didn't wrap my knuckles on.
B.
You are confusing things here a bit. What is good about the BBC reports is that they are based on basic engineering principles and evidence. I am aware of little else that is. Unfortunately, the BBC objectives were not to build the ultimate cabinet but ones that were just good enough to do the job they wanted to get done. This involved modest weight to lug around, modest cost and straightforward to build using materials and processes common 40-50 years ago for speaker companies. They wouldn't design a speaker cabinet like this today even though the principles obviously still hold.
1) The BBC designed their shells to be the opposite of rigid. This is important given their objective to maximise midrange damping using relatively inefficient extensional damping.
2) They added a significant amount of extensional damping and its relevance seems to be missed by many speaker DIY folk.
3) Bracing does not add damping. Bracing done properly adds stiffness. This may or may not be useful depending on how the cabinet is designed to work.
4) Rounding on the baffle is related to external acoustics.
If you wish to design a good cabinet of modest weight then you need to understand how to get the most of the materials used. That is, where to use stiffness, where to use flexibility, where to use damping and where to use mass. Lightness is a common driver in the engineering industry these days leading to a wide range of composite materials that are relatively light and designed to do more than one job. For example, good support and good damping which are rarely provided by a single material.
Given your design is using layers then it would seem natural to use the more efficient constrained layer damping rather than extensional damping. However, for it to work well the damping layer needs to have reasonably appropriate properties and it needs to bond strongly to the structural and constraining layers.
My students tests were done a long time ago (decades), way before I started to make speakers.
My early speakers were fiberglass with rigid wood linings and CLD throughout. They weighed a ton. As time went on I scaled back the design and gave up on fiberglass as too expensive. I tried to measure the effects of my changes and found that I was not able to. I wanted to know if all that I was doing was important.
When I gave up on fiberglass I used MDF, but quickly decided that it wasn't any good because of its dimensional stability. The seams would always crack, so unless MDF is layered with a veneer it is not usable. But honestly this is mostly a cosmetic problem.
I then started using polyurethane boards (Renshape), which I came to love as they machined very well, glued nicely, and had superior dimensional stability, waterproof even. I retained the CLD baffle and back panel as these were the largest two surfaces and used a cross brace (+) between the sides and front-back. As time went on I dampened this brace. The knuckle test on these cabinets is very good. But Renshape is very expensive and hard to get, but it will make the best and lightest cabinets that you could make.
The back panels, since they are not painted with the rest of the cabinet, have always been MDF. Two 1/2" layers with poly glue and glass beads as the CLD.
Short of Renshape I would use baltic birch plywood, cross brace and CLD the front panel and the rear panels. That, to me, is the minimum that I would do.
But I have to admit that I was never able to convince myself with actual data that all this was necessary, it just seemed like good engineering in the absence of any data.
If only we had a small transducer, sensitive in the audio range, which would contribute minimal mass to the DUT... like a phono cartridge! I think Siegfried Linkwitz did that to measure panel resonance in box speakers, there is an article out there somewhere.
Measuring the vibrations of panels is easy enough, but that doesn't tell us what is actually radiated. A small vibration of the right form could radiate more than a large vibration of a non-radiating form. It's just not a 1:1 situation, making the vibration measurements hard to interpret.
Construction of a loudspeaker enclosure is never that critical as an unexperienced DIYer might believe it to be. A straight forward solution is the one from Harwood paper on panel resonances along with Dr.Geddes' recipe for cld method and bracing.
Right nezbleu. Violin bridges are very light structures. The bridge pickups are small and light devices.
One can also use a thin naked piezo disk.
In every case the sensor cable has to be thin , light, very flexible.
That’s very true.
IMO apart from measurement, one has to use his senses to feel the effect.
The knuckle test, a sensitive hand touch on the panel with the fingers lightly touching at different points (this is important as it increases the sensitivity through comparison action) while the speaker is operating, or an ear in close proximity with the panel (this is also highly sensitive comparison method as the other ear is exposed only to the speaker radiation and the brain does the comparison), the stethoscope.
These all provide qualitative but meaningful results
George
Attachments
I recall reading on Linkwitz' site that he had mounted a phono cart with stylus against the cabinet side to evaluate resonances there. It was in the very old '70s article...
I would agree with that - don't obsess over the cabinet. Obsess over the design, not the drivers or the cabinet. Will the design yield constant directivity across a wide bandwidth and it the frequency response along some axis smooth.
Problem: pistonic drivers cannot yield constant directivity.
Do you consider the effect as minimum or you suggest the solution is feasible?
These are two different things.
If you think the effect of enclosure/panel radiated sound is unimportant but you don’t hold it as a belief, I suggest the following test.
Use a single small FR driver. Think of a simple shape for an open baffle. Built this baffle out of three four very different materials/techniques. Use the same way to attach the FR driver on each baffle and take a listen to the same piece of music of your choice with each baffle.
George
It sounds like you've tried this? What would you recommend for a very large baffle? I have OB WAW but would like to try a HOBO 😉
I would expect no difference. Do you have test results, blind, recorded, or measurements that show otherwise.
Personal listening doesn't count as you will just create the difference that you seek.
George, I am a practical man and my experience has taught me that enclosures are not an issue as long as you follow common sense rules in making one. Those rules are so common that one would naturally follow these as if one was born with it. The game really begins when you start dealing with linear distortion ( frequency response ) at the points in space where it is important and these need to be discovered. There you need to make some sound wise important decisions.
Mr. Geddes
With this post you are contradicting your #28
Personal experience gained through listening worth a lot, isn’t something to be discarded.
It is obvious that I have done what I suggested. I don’t intend to influence the potential experimenter with my observations.
Besides the subject of construction is OT here.
If participants are interested, please visit the Build & Design section, there are numerous dedicated threads
http://www.diyaudio.com/forums/construction-tips/
Me too 🙂
George
Everybody opines this. I've braced my boxes cos everyone says it is good to brace boxes. But who rigorously measures it?
As Dr Geddes says:
The BBC did publish some actual data. Their philosophy was to drop resonances out of the easily audible range, so putting stiffening braces everywhere would be a bad thing, for their approach.
Ludwig did some measurements. On his page (linked post one), he says that "What I found is that bracing helps only at the lowest frequencies, below 100 Hz".
If one was to follow this reasoning, it would make sense to build mains + subs, and do standard stiffening braces only on the subs.
This has been suggested before. Not much data, though (e.g. would 100Hz be the best cross point, for this approach? Has anyone ever made a system this way, varied the crossover point, and measured the difference?)
Exactly. That was the list I put down: the basic BBC formula plus a couple of modern updates.
1) make the shell rigid
2) add a layer to make the shell non-resonant
3) add bracing that damps it some more
4) use lots of rounding on the baffle edges (to minimise diffraction)
If I recall correctly, the BBC report is also a bit dated in that it has no mention of fibreglass (or similar) stuffing.
Emphasis mine:
Yes. The BBC philosophy was to drop resonances out of the easily audible range, so stiffness-adding braces would be a bad thing, in that approach.
Dr Geddes has stated the opposite of your point 3, emphasis mine:
Non-stiffening bracing is something I'd like to know more about. I'm delving into the KEF white paper now.
https://www.kefdirect.com/media/wysiwyg/documents/ls50/ls50_white_paper.pdf
I found this referenced when I was looking for info on Geddes' braces:
This KEF white paper is a good example of what I was asking for, in post 1.
It looks like their bracing approach is essentially the 'normal' panel based CLD, rotated 90 degrees.
- Status
- Not open for further replies.