I purchased a pair of use MDT-30 tweeters a while back. I've not used them extensively, though I did exchange them for the Scan-Speak D2905/9300 tweeters I have in my 3-way for a while and was pleased with them.
However, the two didn't match perfectly, but were close, especially above 2khz. Below that the SPL levels crossed as one sloped a little more than the other. But the most obvious difference was that the impedance measurements were very different in the area of the Fs. This is what lead to my investigation.
When I first received them I noticed some excess wire under the faceplate on one of them. I removed the faceplate and found that there was about a 1/2" piece of voice coil lead lying there. It wasn't a problem where it was, but could have been trouble if it had come in contact with the coil leads. This was the first hint at a small quality control problem, although it didn't affect its performance.
I inspected the second one, which seemed OK, so I put everything back together. The impedance difference remained. I left them as they were and used them for a brief time this way. They didn't seem as good as the Scan-Speak D2905/9300s I was using, but then I never fully optimized them other than to set them at the same point as I used for the 9300s. I plan to test them more thoroughly at some point.
The Dayton tweeters I've been working with showed how the impedance can change due to ferrofluid re-distribution, so I became curious about the Morels again. I decided to open them both up, but made sure to make before and after measurements, especially of the impedance, to track any changes. What I found was a surprise.
The first graph below shows the impedance measurements of both drivers, before and after opening them for inspection. The first thing you may notice is the red curve. It is definitely not normal. Morel drivers which I've seen typically have a 2-humped impedance curve. I'm not sure as of this writing of the reason for this, but I suspect that it is the manner in which the chamber is stuffed. Without the chamber it is the likely resonance without a chamber, but with just the vent stuffed, as is done in tweeters with no chamber. The felt pad is placed such that there is a small amount of space between it and the vent opening. It's easy to slide it around by reaching through the vent with a pencil or other non-magnetic probe when the dome is removed.
This doesn't explain the odd behaviour. What I first found when I removed the dome was a small piece of foam in the vent. This foam is disc-shaped when removed and allowed to return to its original shape. If I remember correctly it's about 1" in diameter and maybe 1/8" thick. Everything looked typical except for the gap menioned above. This gap may explain the double-hump in the curve. My guess is that it allows a bit more air movement into the chamber for the lower frequency absorption at the expense of a bit less attenuation of the higher frequency.
Whatever the reason, it all looked good. So I replaced it. It had not been inserted as well as the second driver's foam disc had been. I made sure that the edges were all the way down against the felt pad in the chamber. The foam is inserted by folding down the outer circumference and placing that in first, to make something of a dome shape with the peak towards the front.
When I measured them again, what I found really surprised me. The tweeter with the strange impedance curve had changed dramatically! It now looked like the other one of the pair.
Compare the red and dark blue curves above. These are the curves before opening them. Then compare the cyan and yellow curves. These are the new ones. The Fs differs by no more than 100hz, but the magnitude of the two peaks are very close. The problem with the odd impedance was apparently the incorrectly placed foam.
The next graph shows the two drivers measured some time ago, before I opened them up. I no longer have the original files, so I've had to use the graph file I generated then. Please be aware that these were taken on a 7" x 12" baffle, but they can still be used to show the relative difference between the drivers. These two curves have 1/6 octave smoothing applied. Even given all of this, they are very close above 2khz. Note the variations below 2khz, where the impedance oddity could account for the differing SPL levels. The difference is 1db or more, primarily at the points where the odd driver impedance anomolies are located.
The next two SPL measurements were made at a distance of 0.5m on my "quasi-IEC" baffle, which is 1m x 1.2m with the driver centered on the baffle. This helps to lower the frequency at which the measurement is still valid and excludes any diffraction effects when windowed properly. There is no smoothing applied in those below.
Note the closeness of the two curves, especially below 2khz. I can only conclude that this is due to the change made in the foam placement in the vent.
On a final and separate note, compare the curves below. I've included this to show the impact of a baffle on a driver. There are three curves with no smoothing applied. The yellow one is from the quasi-IEC baffle. The cyan curve is an on-axis curve on the 7" x 12" baffle with no felt treatment, so all of the diffraction impact is present. Finally, the red curve is an average of three measurements, -15/0/+15 degrees, which I use in an attempt to approximate the power response. Note that there are 2db to 3db peaks and dips, these right in the most common crossover area. This can make it very difficult to get a good optimization. A simple on-axis measurement, if there is no baffle diffraction treatment, in my opinion is not proper for use in an optimization.
