A Pneumatic Loudspeaker Experiment Traveling Through Time: How Would You Measure It?

[Mnyb]

My unprofessional view ( not my kind of engineering ) it’s still a loudspeaker the working principle is not that relevant ?

It still produces sound ? The measurement system won’t know how the sound wave was formed.

The loudspeaker experts here migth help you with the typical problems of measuring low frequency properly ?

 

[DeltaSignum]

The measurement system won’t know how the sound wave was formed

Yes, exactly. The mic doesn't care how the sound was made.

My only problem is how to measure the full low-frequency response properly when the prototype is just 2".
Of course another option is for someone with proper measurement gear to print and build the same speaker - single module costs less than $20. All STL files are on

Delta Signum · Dynamic Zero

deltasignum.org

and GitHub.

But first I need to show that this DZ pneumatic thing is actually worth looking into, and not just another snake oil project.
If anyone wants to print it, happy to share my experience with 3D printing.

 

[Vladimir Filevski]

Same 2" driver, sealed box, 1 cm, 1.5V as suggested. Looks different.
Infra looks similar - apparently a UMIK-1 noise artifact.

The red curve? Frequency response shape is exactly the same as curves for 1cm, 1.5V, open and closed. But SPL level is 6-7dB lower, which is impossible - you made some errors in measurements.
UMIK-1 is not calibrated for frequencies below 10Hz. Disregard everything below 10Hz, that is pure junk.

 

[somebodyelse]

I don't see why this can't be modelled with conventional tools - I think I saw one with this arrangement for a small desktop speaker on diyaudio some years ago.

 

[olieb]

The main problem is your inexperience in measuring loudspeakers.

I do not think so, the problem is more in the interpretation of the (competent) measuring results.
Obviously the result shows the behaviour of a 2nd order high pass (with some low frequency noise), exactly as a sealed box will do, but OP sees things like “pneumatics“, “engines“, “variable mass“, hidden frequency response and so on, when the by far simplest and convincing explanation is that this super long, lowly tuned, modified bass reflex port just does not do anything meaningful.

Looks different.

Well, yes. The sealed box you built has obviously a smaller effective volume than the original one with “exhaust“.
This increases the f_s and the same happens to the Q value.
Compared to the original enclosure the increase seems to be about 20%.
And for some reason the red curve (SEALED) is ≈1.5dB lower in level.

The pic shows the resulting high pass SPLs.

 

[DeltaSignum]

The sealed box you built has obviously a smaller effective volume

No, it is the same enclosure. Just plain PLA - a leftover from my filament experiments that fortunately I hadn't thrown away. I only sealed the internal channel.

You can try it yourself - everything is open source. I've already done enough experiments with the 10" version back in 2000

 

[DeltaSignum]

you made some errors in measurements

If there are measurement errors, the same errors are present in the DZ measurements, so I'm not sure I would call them errors in this comparison.

And deliberately manipulating the measurements would be rather pointless after publishing the STL files and BOM for anyone to reproduce the model.

 

[Vladimir Filevski]

I do not think so, the problem is more in the interpretation of the (competent) measuring results.
Obviously the result shows the behaviour of a 2nd order high pass (with some low frequency noise), exactly as a sealed box will do, but OP sees things like “pneumatics“, “engines“, “variable mass“, hidden frequency response and so on, when the by far simplest and convincing explanation is that this super long, lowly tuned, modified bass reflex port just does not do anything meaningful.

Well, yes. The sealed box you built has obviously a smaller effective volume than the original one with “exhaust“.
This increases the f_s and the same happens to the Q value.
Compared to the original enclosure the increase seems to be about 20%.
And for some reason the red curve (SEALED) is ≈1.5dB lower in level.

Inexperience of the OP is obvious.
I concur - that super low tuning doesn't do anything useful. Volume of the new sealed box is much smaller than the original (why?), so Qtc is much higher. But that doesn't explain 4dB lower SPL at and above 300Hz - it must be the same. Hence, OP made some errors during measurement.

 

[DeltaSignum]

Vladimir, did you read my previous reply? It is not a new smaller box. It is the same physical enclosure; I only sealed the internal channel. What exactly do you mean by “volume of the new sealed box is much smaller”?

 

[Vladimir Filevski]

No, it is the same enclosure. Just plain PLA - a leftover from my filament experiments that fortunately I hadn't thrown away. I only sealed the internal channel.

You sealed the internal channel? That restricts available volume to the driver, compared to the original. That is an error - you compared apples to oranges.
Make a new empty sealed box with the same volume as the total internal volume of the original, then measure again. Apples to apples comparison.

 

[Vladimir Filevski]

If there are measurement errors, the same errors are present in the DZ measurements, so I'm not sure I would call them errors in this comparison.

No, in your earlier measurements you have just been lucky.
You made some error during new measurement, unfortunately.

 

[Vladimir Filevski]

Vladimir, did you read my previous reply? It is not a new smaller box. It is the same physical enclosure; I only sealed the internal channel. What exactly do you mean by “volume of the new sealed box is much smaller”?

I peek in your github project - there are two chambers internally connected via channel. Sealing the internal channel cuts off the volume from the second chamber, so only the first chamber volume is available to the driver - that volume is smaller than the original volume of the two chambers combined. You have to remove the internal division between two chambers, to let the driver use whole internal volume. Or build a new sealed box, empty, without internal chambers and channels.

 

[DeltaSignum]

You made some error during new measurement, unfortunately.

The total volume of the entire pneumatic system is 0.47 L. However, the main chamber is pneumatically separated from the rest of the system, so an effective volume of approximately 0.35 L for that chamber is entirely reasonable.

The table also shows that enclosure volume is not particularly critical here: even doubling the volume from 0.25 L to 0.5 L changes Fc only from 203 Hz to 183 Hz, and F3 from 141 Hz to 129 Hz.

So even a 50% difference in assumed chamber volume would produce a relatively small effect. There is no need to re-prove basic T/S sealed-box theory - the numbers already show the scale of the effect.

 

[Vladimir Filevski]

Difference between 0.47 liters and 0.37 liters is 34% - not so big, but measurable: hIgher Qtc and F3. But not lower SPL, especially above 300 Hz - as in your measurement. Hence, you must have done some error in measurement.

 

[DeltaSignum]

The measurements were made under the same conditions, so if the difference is repeatable, it should not simply be dismissed as an error.
If there are measurement errors, they happen to coincide with the T/S predictions.

 

[Vladimir Filevski]

TS modeling deals with low frequencies only. Whatever the type of the enclosure, frequency response and SPL must be the same above 300Hz. If difference is repeatable, then you are repeating the same error, unfortunately.

 

[DeltaSignum]

If I were interested only in what “must be,” I simply wouldn't measure it at all. I could just sit and theorize about what it is supposed to be.

Same input, same distance, same speaker. That's it. Всё остальное звучит как «не всё так однозначно».

 

[Vladimir Filevski]

The same frequency response above midbass (give or take) for any type of enclosure is not theorizing, but proven fact.
I tried to be helpful, but no avail. I am out.

 

[olieb]

I could just sit and theorize about what it is supposed to be.

Well, isn't that exactly what you do? Theorising all sorts of funny stuff that you think it is supposed to be?
And then you measure and it turns out the result is exactly as expected without the funny stuff.
Next, you think the measurements are wrong, keeping up the theorising.
That is not exactly the scientific method.

But well, you do what you do. Just wondering what, after all, you were asking about here?
Have a nice time.

 

[SSS]

Yes, exactly. The mic doesn't care how the sound was made.

My only problem is how to measure the full low-frequency response properly when the prototype is just 2".
Of course another option is for someone with proper measurement gear to print and build the same speaker - single module costs less than $20. All STL files are on

Delta Signum · Dynamic Zero

deltasignum.org

and GitHub.

But first I need to show that this DZ pneumatic thing is actually worth looking into, and not just another snake oil project.
If anyone wants to print it, happy to share my experience with 3D printing.

Bruel&Kjaer 4145 1* microphone goes down to 2.6 Hz. Use such one with the right pre-amp and a voltmeter which should also go down to 2 Hz. Soundcard I think won't do it.