I bought the new Behringer ECMPRO a few months ago but hadn't hooked it up.
I took a bunch of measurements 2 weeks ago before the EW started acting funny. Wanting to get this project done I will try to use the Behringer to wrap things up. My measurement rig is setup to be able to make pretty repeatable measurements. The ECMPRO does not have a calibration file and is supposed to measure out to 40kHz. I figured I would measure the speaker and make a calibration file to finish things up.

What the F@#!
Earthworks M23R $600 in blue
Behringer ECMPRO $40 in red - NO CALIBRATION FILE
I'll do more thorough testing and test distortion when I get the EW back, but just had to post this comparison.

Impedance overlay:

Model overlay:

Sample 1 parameters:

Sample 2 parameters:

Sample 1 has a production date of 8/19/15 and sample 2 production date is 1/11/16 so from two different batches. Kudos to Meniscus and Misco.

Are horns/waveguides the answer here?

TIA.

For an apples to apples comparison, I set the output of both units to 1.2V. This is somewhat arbitrary, but also provides about -3dBFS on the Steinberg with the input gain knobs at 12 o'clock which is how I generally start with it when I use it.

Checking the output on the Steinberg it gets a bit weird. It has a TRS output but was only providing an unbalanced signal. Tip at 1.2V, ring and sheild were 0V. Hmm, wierd, though not so much of an issue since I generally adapt the signal for RCA connection to an amplifier anyway. I didn't troubleshoot it at all, but did compare the Motu both balanced and unbalanced at 1.2V using both the TRS and RCA outputs available. The Motu did provide a balanced signal on the TRS output.

I'll mention as well that the input gain on the Motu appears to be a fair bit higher, as I only needed about 9 o'clock position on the input knobs for the same -3dBFS signal. As well, the unbalanced test used a different cable (RCA with TS adapter), while the Motu balanced and Steinberg test both used the same pair of TRS patch cables for the loopback.

This simple comparison was completed using Rightmark Audio Analyzer. Both units were tested at 24/96. Nevermind the typo in the titles.

Frequency Response. Point Steinberg.

Noise. Point Motu.

THD 1kHz. Point Motu. Steinberg is dominated by 2nd and 3rd harmonic, 6th+ harmonics are non-existent on the Steinberg where they do appear on the Motu.

IMD. Point Motu.

Crosstalk. Point Motu.

The frequency response of the Motu is unfortunate, the Steinberg is much better in this regard, but otherwise the Motu Excels in raw performance in this loopback D/A and A/D comparison. Both are very capable units, one is just a bit better

Just measured the T/S for the pair of SIG180-4 drivers I received as a door prize in August.
I am pretty surprised at how far off some of these numbers are.....

And the AMT, just as they really should be...

I'll be testing and posting data for the pair of original AMT's I got from @LifeOfBrian at the IowaDIY event. (EDIT: Please note that these are VERY old ESS AMT's, dating back to the early 1970's, that I am testing!) I know that others also have the newer versions of these tweeters, so please feel free to post your own measurements for comparison purposes.

I'm currently gluing up a tower enclosure, using Menard's value shelving, and I expect to start testing later next week. I still need to run out to the hardware store and pick up some 3/8" x 7" long bolts and T-nuts to complete the mounting. Hopefully they will have some in stainless steel.

I plan to do full 180 degree spinorama type SPL's, both vertically and horizontally, using Soundeasy and my dual channel mic setup. Will also do distortion measurements and run impedance curves with OmniMic and DATS V2.

Gonna need some aloe for that....

The plugin is pretty straight forward, just select the audio track, adjust the desired quantity of each harmonic, and apply. For a quick test I generated a 1kHz tone, and processed through REW for spectrum analysis .

1kHz pure sine wave, just to show that REW analysis is free of harmonics.

0.1% H2 added, other harmonics are still completely absent.

0.1% H3 added, other harmonics compeltely absent.

So it does what it says on the tin. Now you can add any combination of harmonics to any audio you like and hear the result first hand. Best used with the lowest distortion equipment you have available. Have fun!

PART 1: IMPEDANCE TESTS

Below is the published impedance curve by Dayton Audio.

Given that they tend to pick favorable scaling and overly smooth their SPL curves, it is odd that they scaled the impedance from 2 to 10 ohms. I looked at a LOT of tweeters at Parts Express and Madisound and couldn't find anything zoomed in this much. Given that Dayton has the capability to design (in house, or outsource) good drivers, I wondered if this was even accurate. JR also noted that sometimes removing the faceplate and putting it back on can improve matters. So, I ordered a pair and played around with them.

Tweeter...

Without faceplate...

Just the faceplate...

Here are the impedances for tweeter #1 and #2. (Impedance measurements using DATS v3) Yep, that spike is really there. Axis are hard to read...impedance is 0 to 30 ohms and frequency is 20Hz to 20Khz.

Here is the impedance for tweeter #1. Green line with the spike is stock tweeter. Yellow is with the faceplate removed and put back on, maybe a little tighter than orginal, but it didn't do much. Flatter green line (I couldn't figure out how to change the colors in DATS) is with the foam ring.

Here is the impedance of tweeter #2, same scenarios. Pink is stock tweeter. Orange is faceplate removed and put back on. Green is with foam ring.

Here is impedance of tweeter #1 and #2 with foam ring.

I thought the foam ring looked promising. So after all of the other measurements, I came back tried to seal it up even better using some DAP Kwik Seal caulk. It comes out white but dries clear. I also put a little gasket tape over the holes near the terminal connections.

I only did this with driver #1 and the results were unimpressive. Hopefully I can remove the caulk tomorrow. Stock tweeter green and caulked tweeter pink.

Deep drawn stamped basket, plenty of mechanical throw, doesn't bottom out when pressing all the way down on the cone, bumped and vented back plate, 8mm top plate.

Not unattractive to my eyes - although I have always liked how the Infinity CMMD drivers looked, not too mention these are from what I call the "golden age" of Infinity (The early 2000s). Loved how their products looked and performed from that period.

Nice modeling driver (with 0.8 ohms series resistance added - simulating what a 3-way crossover would add)

Also bought the matching 4" midrange/midbass drivers so could make a pretty cool Interlude IL40 tribute build.

Very intriguing....

I know now for certain that they do indeed use shorting rings, and the phase plug is the second shorting path. If you committed my few previous statements to knowledge, please rewrite your data with this positive proof. Phase plug is not shown. My apologies...

I have a baffle design formulating in my mind but need some real data/parameters to work with.

1). Is it the point where the frequency's wavelength becomes greater than the baffle's width or at some lesser value?

2). Does the shape of the driver's cone (Dome?) play an additional roll in determining this?

3). Does the use of a waveguide reduce or eliminate the need for baffle edge curvature?

IOW - at what point/situation do curved/arched baffle edges become a requirement?

TIA.

This, along with my other document detailing the difference between single and dual channel measurements, should make it clear why this is my recommended process, and with VituixCAD in mind, why the instructions call for dual channel measurements and do not detail any minimum phase processing between measuring and processing the response data.

My other supporting documents (VituixCAD oriented) can be found here. Hopefully someone finds them useful.
https://drive.google.com/drive/folders/1KDyECkAIVAuNEtKVobKmsfvtFczGqOQw?usp=share_link

Any feedback is welcome, I will update the document if something is not clear and needs more detail or rewording, or if something is incorrect.

Happy Holidays

I will be doing in-box testing using one of the TM cabinets I just finished machining for heads up testing.

The big XT25 is not the same as the little XT25 with the exception of the extra magnet. Examining the data sheet it appears the motor has a lot more in common with the Beast Mode than it does with its little brother, so I am curious about its performance. It is also slightly more sensitive than the single magnet version. The Beast Mode advertises 0.1mm of Xmax, but even with the back cup removed in Brad's bigass open baffle speaker it held its composure reasonably well at Barbie Girl volumes - the Max Fidelity midrange was crapping out before the tweeter did. They both bowed out before the subwoofers did. It proved my moniker of Beast Mode this weekend.

Winner goes in the 2.5 way speaker, loser goes in the MTM. No reason, just decided that right now.

I wanted to do a deeper dive on driver Re to maintain my sanity. From another recent thread, I had questioned how useful it is to get out a multimeter and measure the DC resistance of a driver over just letting the software estimate it. This test compares various T/S extraction software and their ability to extract an Re value, if that ability exists. As well, I compare the cabinet model of T/S determination using different software calculations of the exact same impedance data.

I took my Fluke 179 meter and measured a driver, the meter only has 1 decimal place of precision so +/- 0.1 ohm is as good as it gets, and measured 5.7 ohms. I have another low end Wavetek meter that also measured 5.7 ohms. Shorted leads measured 0.1 ohms. If we really want to dig down to the precision here, the measurement is +/- 0.5 ohms due to the resolution, so the measurement is really 5.65 to 5.75, and the shorted leads are 0.05 to 0.15 ohms, making the driver Re by the meter anywhere from 5.5 to 5.7 ohms. For the sake of this example, I am just using 5.7 ohms as the baseline.

I don't have a Dayton woofer tester anymore, but I have ARTA, REW, VituixCAD and SoundEasy available to me for this comparison.

ARTA
The biggest problem I see is the Re estimation in ARTA. My measurement is clean down to 5Hz, and I zoomed right in on it, however can't see why the software would determine 6.28 ohms given that the impedance at 5Hz is in the 5.75 ohm range. Fail for ARTA, recommend manual entry of Re for this software. If the estimated Re of 6.28 ohms is used, all other T/S parameters are skewed as a result. Using a manual entry of 5.7 ohms, T/S is similar to the rest.

VituixCAD
VituixCAD doesn't do any estimation of Re, it requires you to manually enter this value. T/S result is very similar to ARTA when using the same values and measurements.

REW
REW is quite nice that it shows the curve fitting down to 2Hz, but also requires a manual DCR entry for Re estimation, its a bit odd but the estimated Re is not the same value as the manual meter reading. For consistency I imported the measurements from ARTA, the speaker was not re-measured. In this case, it came up with a value of 5.428 ohms, which is a bit on the low side compared to the multimeter.

SoundEasy
SoundEasy also requires manual entry for Re, so it intends for this to be provided from a manual multimeter reading. Again, for consistency I imported the measurements from ARTA so it utilized the exact same impedance measurement for it's calculations.

How do they compare?
For comparison I entered all the T/S parameters into VituixCAD. VituixCAD supports pasting information directly copied from ARTA and REW, so the process takes only seconds here, SoundEeasy is the odd one out requiring manual entries.

Apart from ARTA's failure to estimate Re properly, all parameters modelled within 0.5dB of each other. Top to bottom is SoundEasy, REW, VituixCAD, ARTA (manually entered Re), then the solid line at the bottom is ARTA with Re estimation. If I learned anything from this exercise, it's to measure with ARTA, then use something else to determine T/S parameters as it's results are vary the most from the other methods. REW and ARTA are the only options used here that allow for estimation of Re, and while REW estimated a value on the low side, the cabinet simulation is indifferent from the others.

But what about that meter resolution?
Right, I used 5.7 ohms here, that could be as low as 5.5 ohms. If I recalculate using 5.5 ohms instead of 5.7, this 3.6% difference in Re makes for a 0.3dB difference in sensitivity of the driver.

I'm obviously not up on the gear as I don't own one, but this would really upset me had I been using one.

this is at 94db, 12.1v

Would value this forums input as well.

Thanks guys.

Related discussion somewhat recently on this forum:
https://diy.midwestaudio.club/discussion/1830/box-sizing-for-sb15crc30-4-woofers

T/S parameters are defined as parameters for the driver "at rest". In other words, they are low voltage by definition. However, there is not a well defined specification for the conditions under which T/S parameters are measured, and there is not agreement from different manufacturers on the process either, so the inclination from a manufacturer may be to use whatever method provides the most marketable result.

Factors that influence the parameter results include voltage level for test, constant voltage test vs voltage divider test, ambient temperature and humidity, driver break in, probably more.

Voltage Divider Method
The common DIY method for testing a driver is to use a "voltage divider" method, because it is very simply done by placing a resistor in series with the driver and then measuring the voltage across the resistor. This "sense resistor" is typically in the range of 10-100 ohms.

The "problem" with this method is that the voltage that is present at the driver is not constant, the voltage present at the driver is proportional to the impedance of the driver. Easily observed in VituixCAD, the voltage across the sense resistor is a mirror image of the voltage present at the driver, neither are the original constant voltage sourse from the amplifier.

Constant voltage Method
A solution to the voltage divider problem is to design an amplifier specifically for this task. What is required is that the sense resistor is placed inside the feedback loop of the amplifier, which will compensate for the voltage dependency of the voltage divider. In this operation, the gain of the amp now varies with the impedance of the driver in order to provide a constant voltage over frequency.

While this does "solve the problem", the question remains, was it even a problem to begin with. If we think about what is occurring in this constant voltage method, essentially the drive voltage is being increased when the impedance of the driver is increased, so especially at Fs the voltage is greater compared to the voltage divider method. So what if we just increased the drive voltage overall and keep with the simple voltage divider method?

Elevated drive voltage
To illustrate the effect of a difference in drive voltage on a driver, it is easily done in LIMP. Use PN noise and turn off averaging for a somewhat real-time measurement, then simply adjust the gain on your amp. The difference will be more easily observed in crappy drivers, but you should see Fs decreasing as you increase the voltage applied. So with this in mind, what is the "correct" Fs, and does any of this even matter?

To answer this, I simply used a cheap 3" TB fullrange driver, no copper in the motor I am sure, and completed full T/S measurement at a voltage of 250mVrms, and again at 1.5Vrms.

At elevated voltage, Fs is indeed decreased by nearly 5Hz, but also Qts is lower, Cms higher, Vas higher, BL slightly lower.

250mV test:

1.5V test:

Comparing the cabinet model shows basically no difference at all:

So what was the point of all this? The knowledge to be gained is that there are many factors that affect T/S parameters, and direct comparison of one's own measurements to the manufacturer data is futile, especially if focusing on single parameters like Fs for example. What does matter, is that the cabinet model is accurate, so any comparison of whether a T/S measurement is good or "matches" should be made using a cabinet model, since something simple as the test voltage can change the T/S parameter results, but the cabinet model can remain unaffected.

I always knew the Ominimic had a poor noise floor, so distortion measurements always had to be completed at a fairly high SPL just to get the harmonic products above the noise floor. It is just a basic integrated USB audio codec after all.

Comparing to the OM1, I completed sine sweeps at the same SPL, and adjusted the scaling and colours in ARTA so that the graphs can be directly compared. The SPL was fairly low, but just high enough to see something discernible above the noise floor in the Omnimic. At low frequency, the result is similar, but at high frequency there is a good 10dB+ advantage to the OM1.

As a double check of the noise floor, I simply ran the measurements with the output disconnected. There is probably even some further room for improvement on the OM1 with some gain adjustments to maximize SNR.

Dayton could really do with an update to the Omnimic, include a better USB codec like miniDSP did with their UMIK, and update the software to fix some of the bugs and reduce some of the annoyances. For example I shouldn't have to rerun a measurement to change the smoothing, and I should be able to select "play from soundcard" without hitting the play button first.

ND25FW is a low cost dome tweeter with a nice small waveguide. As of today, it's cost is USD$22 making it one of the cheapest waveguide tweeter options available.

It was installed in a bookshelf cabinet along with a SB PFCR series woofer:

Measured frequency response is very promising.

You may be concerned that the high Fs of ~1400Hz is a problem, but it is a fairly small peak as this tweeter uses ferrofluid:

For a basic crossover example, a simple 2nd order filter was built.

And here's the transfer function of that filter. As you can see, the impedance creates a bit of a bump in the knee of the response, but not enough to be too concerned about, especially at $22.

WEBINAR AGENDA:

Free Field Measurements
Acoustic Response
Quasi-Anechoic Techniques and the Time Window
Combining Near Field & Far Field Data
Impedance Thiele Small
Loudspeaker Production Test
Rub & Buzz