Lamb's Wool Tweak - Dayton Tweeters

The Dayton 275-070 tweeters responded a bit differently to the chamber than did the Vifa D27TG-45 with which I experimented. I have in essence three versions of the Dayton, those being the original, an updated one with a different dome doping compound and a vent cap (not a true chamber) and a third of my own making (The Hybrid), covered on another page.

I attempted to add a chamber to the original version, with less than spectacular results. Others who have done this had better success than I, so I suppose that my attempt was not optimal. Still, this same chamber had worked very well on the Vifa, the results of which are shown in the section on adding a chamber.

However, all versions of the Dayton showed improvement with the Lamb's Wool felt added to the pole piece vent. The impedance graph below is for the orignial version of the driver, before and after the modification. It has a piece of closed-cell foam over the pole piece vent on the backplate and nothing in the vent. Adding the wool (yellow and cyan) slightly increased the peak and lowered the Fs. The Q remained similar.

The difference between the yellow and cyan curves is the results of differing amounts of wool in the vent. The cyan curve shows the case of the highest density. It looks better, eliminating the glitch at 4.5khz. The SPL was smoothed there as well, at the detriment of a lowered sensitivity in the rolloff region, not necessary a problem.

As you can see in the next graph, the wool linearized the SPL a lot in the old version from 3-20khz. The low end rolloff is slightly extended. This is not too surprising, since the low end rolloff was already very low. I can only guess that the Dayton tweeter, the old and new version, have T/S parameters which result in little benefit from a chamber (or the effective apparent increase in volume which wool should provide). Either that or I chose the wrong type or amount of stuffing material when I tested the chamber.

I have two of the original versions now modified with lamb's wool, shown below. I am maintaining this pair as is for some listening tests.

I only have one sample of the new version with which to work at this time. The next graph may not be representative of current samples as I've read that few others have seen the non-linearities below 10K shown in the graph below. That said, the low frequency rolloff response changes should be similar. It does show significant improvement in the midrange, from 3-5khz, when the wool is added.

I even used some isopropyl alcohol and Q-Tips to remove some of the doping compound, which made little difference. But the real improvement didn't occur until I closely inspected the new version due to the anomolies after some discussion on the Parts Express Tech Talk board. I finally located part of the problem, at least for the 8khz dip. The surround was not properly glued onto the faceplate. There is a significant portion which lifts off of it when the coil former is pressed from the back side. This is usually glued and has the foam on top for additional security.

I cut a piece of the foam from the outer perimeter and stuck it onto the area described (see arrows above). The faceplate then helps to hold it down, since it presses onto the foam. This improved the performance fairly significantly. This is, I believe, the whole reason why this one, and likely others, performed poorly. Above 10khz I believe it be a function strictly of the dome/surround itself.

Next are the CSD (waterfall) plots for the original driver without and with the lamb's wool. The midrange falls off much cleaner. Resonance ridges are nearly eliminated at 5khz and 10khz. The low frequency decay is much faster and smoother and the high frequency hash is greatly reduced. I consider it a significant improvement.

The CSD for the new version looks much like the old version. I haven't been able to figure out just what makes the difference in the SPL. I do not see any significant resonances specifically associated with the dips, so I initially made an assumption that they are due to way too much damping in the dome, likely from the new doping compound used in the new version. This damping is apparently frequency dependent. Either it's the material or, more likely, the application process.

The next CSD is one in which I stopped the plotting after the second time curve. This shows that there is some sort of significant out-of-phase problem occurring. I've never seen a dip such as this one has at about 9.5khz. I don't know what causes it, but it must be what is responsible for the main problems.

Last, but not least, are the step responses. The first two below are for the original driver, without, then with, the lamb's wool in the vent. Differences appear to be very minor. The improvement is easy to see in the smoothness and the good transitions with the lamb's wool when compared to the original without it.

This last graph is for the new unit with wool in the vent. It's easy to see the relationship with the high frequency non-linearities. The kinks in the first return segment show how strong the effect is of whatever is causing it. The combined modifications have made a significant improvement in this one as well.

Dave's Speaker Pages