Hello André@homeswinghome
As promised, Christian,
Notes:
This is TSP's on the Dayton DAEX30HESF-4 driver. I'm measuring these HESF's because they are the only ones lying around at the moment.
I've used 2.5g added mass which seems to be the same as what Billionsound uses in their measurement of this driver.
Impedance axis is logarithmic.
View attachment 1137426 View attachment 1137427
Note Re=2.5ohms?? Yes, I did check my reference resistor a few times.
Note the F3 and Le-at-F3 numbers...
Here's what I think is the OEM driver from BillionSound. https://www.billionsound.com/product/177.html?productCateId=18
Their curves are virtually identical to mine.
The numbers below their graphs are almost too low-res to read...
View attachment 1137429
Comments?
The results are so clean that I wonder why I don't have a test gear ready beside me to do that! Near to me currently an old class AB amp no more used (most of the time I use a small bridged class D for testing which is not suitable for impedance measurement. Hmm.
Your results seem consistent and in line with the curves of the documentations.
The strange point is the difference between the Le you get (0.1mH) and the one in the Dayton spec (0.3mH) which itself is not consistent with the impedance curve. About F3 I have no idea as it is defined as sqrt(2)*Fs.
Christian
No Paul, From what I know from the equivalent models, we can fully use them to evaluate the electro-mechanical elements. The answer should be some where else. Looking more carefully what is strange is the low value of the impedance curve at low frequency. This is not in the physics of the exciter. The curve should be almost flat here just below 4 Ohms.
@Andre Bellwood : would it be possible you have a kind of band pass limitation in the LF? Serial capacitor? Have you checked the Re with an ohmmeter?
PS : sorry I just have seen your answer André : Re = 4.3Ohms which make sense.
TS parameters relate more at the end to the volume calculation and alignment choice (FR shape). Here it is the electro-mechanical model of an electrodynamic motor which is used. Of course as you said André, the closed box method can't work but the mass yes.
When a panel is added everything changes as the mechanical impedance (or mobility?) of the panel comes in parallel to the LRC parallel network of the model pondered by a Bl² factor. the RLC figures the voice coil mass, spider compliance, spider loss.
André, if you have the opportunity to make an impedance curve of an exciter with a panel it would be interesting to share its shape with the peaks at the main resonances of the panel.
Last edited:
So people will actually remove the cone to make a similar test on a conventional speaker driver?
couple questions.
1) for the square or rectangle foam board insulation piece many attach their Dayton exciters to, is there any benefit to cutting the edges so that no sides are parallel to each other?
2) On an old thread here: https://techtalk.parts-express.com/forum/tech-talk-forum/64791-diy-flat-panel-speaker-love/page12
fred: "The only difference is I have glued a wood frame to the perimeter and have the exciter mounted to a spline. This makes the panels truely full range with deep base and no panel ratting"
There's a picture of the exciter "mounted to a spline" there, but I can't figure out what he means by "mounted". Glued with epoxy or something? has anyone tried firmly mounting the exciter chassis so that only the attached material is moving?
1) for the square or rectangle foam board insulation piece many attach their Dayton exciters to, is there any benefit to cutting the edges so that no sides are parallel to each other?
2) On an old thread here: https://techtalk.parts-express.com/forum/tech-talk-forum/64791-diy-flat-panel-speaker-love/page12
fred: "The only difference is I have glued a wood frame to the perimeter and have the exciter mounted to a spline. This makes the panels truely full range with deep base and no panel ratting"
There's a picture of the exciter "mounted to a spline" there, but I can't figure out what he means by "mounted". Glued with epoxy or something? has anyone tried firmly mounting the exciter chassis so that only the attached material is moving?
Sorry, no. It was a bad choice of words on my part. As far as an exciter goes, it is indeed nothing more than a a naked, unloaded voice coil (and a spider, and a magnet etc.)
Doing added-mass TS params on a conventional loudspeaker would obviously include the cone, no box, ie unloaded, with the speaker being suspended by a wire from a support.
On the other hand, the sealed box method for TS measurement requires the speaker to be mounted into a carefully measured box rather than adding weights to the cone.
You can mount the exciter to the spine by using any type of adhesive or even screws. Look at AllenB's pictures.
https://techtalk.parts-express.com/forum/tech-talk-forum/64791-diy-flat-panel-speaker-love/page99
That's my point! I may still be wrong of course, but at the risk of belabouring the obvious, let me be perfectly clear:
The obvious difference between conventional drivers and exciters is the cone. So the obvious prima-facie culprit for your low value of Re, is the cone. The cone moves air, and so the motor does mechanical work (W = F.s) So it expends energy. All energy expenditures in electrical circuits look like resistances. So if the software is inferring DC resistance from a dynamic test, predicated upon conventional drivers (ie with cones), it stands to reason that it might infer a lower value of dc resistance because its sees a lower energy expenditure. Because there is no cone.
I note the area you input is quite small (~8cm^2), but that would be the area of the exciter bobbin if it were covered over with a 'cone'. The effective area of a naked exciter is smaller still.
Last edited:
Ermm, I'm sorry Paul, Re in a speaker is DCR, that is, the DC Resistance of a stationary coil. It's not an impedance. So it doesn't matter whether the coil is loaded or not, and it doesn't even matter if it's in a magnetic field or not as long as it's all completely stationary. One could strip a cone speaker, take out the coil, and measure it with an ohm meter, and it will still get the same Re as when it's fully assembled.
The Re in my REW measurements is flat-out wrong because my meter measures it to be 4.1ohms DC (But maybe my meter is not quite accurate, and this will influence ALL of the TS readings. I might have to get it calibrated.)
The Ze on the other hand is usually the lowest impedance reading at that frequency after the resonant peak before the inductance rise, and this is influenced by coil loading, and the coil will obviously be vibrating at that frequency. See the pic below.
@homeswinghome
Christian,
I checked out the lower frequencies. I think there's a problem with the circuitry in my USB converter. I might have to find schematics and take out the HPF parts of the buffers.
Side-note. It appears that REW measures Re at 20Hz.
Yes, that is the surface area of a 32mm disc IF the voice-coil was covered.
But I think you've put your finger on an important point there.... The VC IS covered and sealed by panel material when one mounts it to a panel. Therefore the driver is probably designed to accommodate an acoustic boost in the VC cavity at some higher frequency (I think I picked up a 1.2kHz notch in some previous acoustic measurement with the VC uncovered. Ah, here it is: Post #9056.) But it's an acoustic resonance, not a mechanical one. And besides this, if one adds a cover (which is also an added mass) to the VC then the whole structure of the VC, its suspension and the moving mass also changes, and this will shift the mechanical Fs downwards.
Therefore, it appears that the right thing to do would be exactly that, i.e. cover the VC with the appropriate diameter of whatever material the panel is made of for base-line TS reference measurements.
The manufacturer's (dynamic) measurements in this case would be null and void.
Last edited:
Yes, Im aware of that. Im not saying the REW got it right, only that maybe its inferring the DC reisitance based upon dynamic measurements plus some assumptions it may be making about the kind of device is being measured.
If its actually measuring the resistance at DC, then so be it. But maybe it's deriving it as a difference. I remember reading the REW document saying that accurate low-ohm measurments are difficult to make (you would need a wheatstone bridge I guess). So I presume its calculating DC resistance from other quantities and inferring a more accurate DC resistance. But it will only be accurate if its assumptions about the driver are correct.
Ill stop now, because I probably AM wrong about it, I just wanted to be clear about what I meant.
Last edited:
Yes, I think you're quite correct there. As I said to Homeswing, it appears that REW infers the Re from the Ze at 20hz. But it's also obvious that there's something wrong with low-frequency readings in my setup probably due to high-pass filters in the USB converters.
sorry if I open a parenthesis in your speeches, I wanted to go back to the characteristic of the larsen for a moment, given that the panel in nomex carbon construction with its structure, even if in a rather precarious way, works well indeed very well, I wanted to ask if you have any suggestions or video tests anything to better understand since I don't see any benefit in reducing feedback, so if this thing has to happen because it's a feature of the dml why don't I see it? what am I doing wrong? exciters position or what? in the end I think this thing should work regardless of many factors, yet if something has to work it has been overlooked. Again, any suggestion is welcome
Pixel I'm not sure who you're addressing, but I'll answer anyway.
What's Larsen??
I work on stage with my speakers right behind me, less than an arms-length away from the microphone. And I play quite loud.
Look, DMLs are not immune to feedback. They are more resistant. That means you can turn them up a bit higher than a cone speaker before you get feedback.
You have to take marketing claims with a grain of salt. As far as I know, there is no general agreement that DML tech is immune to feedback. I assume you are talking about the video posted earlier with the singer in front of the panel. The claim that DML is relatively less susceptible to feedback may have some merit, in that different notes tend to be created at different places on a panel, and are so are effectively 'uncorrelated'. But if you are unlucky enough to have a resonance peak on the panel at the frequency your amp likes to oscillate, I guess they could be MORE prone to feedback. If you are using DSP, perhaps you could eliminate feedback with EQ or if all else fails a notch filter at the oscillation frequency?
If you want ideas, here are a few. Are you singing? That may probably be less inclined to trigger feedback than purer sustained notes from a guitar or other instrument. How are you oriented with respect to the panel? In the video the singer was facing away, so maybe was protecting the mic from direct, correlated wavefront. How big is the room? In a small room, I guess the many reflections each have a chance to make the amp oscillate. What type of mic? Ideally I suppose it would be directional so you as far as possible only pick up the source you want and exclude the amplified sound.
Last edited:
yes everything is very clear, it's exactly what I thought but I wanted confirmation, I take this opportunity to ask for advice, the project as inspiration is based on the 500 tecntoni dm which provides for 4 fixed points and seals on the sides for most of the perimeter, currently for testing I have used a draft excluder that is used for doors, it is quite wide and I am testing how tight it should be and how much leeway to leave, the question is what is the best system and material in your experience?