The low frequency part of the project box
Though the woofer section of the box isn't the project's main "attraction", there are still important and relevant issues to address here.
Sound reproduction outside, in a less controlled environment, has some different challenges, such as noise, weather, other dwellings around and unlike acoustics: e.g. less room issues such as sound reflexions. A somewhat limited sound coverage pattern may be preferable here, instead of "spilling all over the place". Plus, in free field sound level drops by 6 db per doubling of distance from speaker, per inverse square law.
Original intent was to use a different pair of woofers I have for this project, however upon inspection the foam surround on these needs replacement. Getting proper new foams from another country and do the refoam job would have delayed the project. So another pair on-hand, not first choice here, will have to do the job. And their surround is in good shape: rubberized foam!
That 6.5" woofer (sen. 90 db) is a little larger, therefore becoming directional a little sooner as frequency goes up, isn't necessarily a bad idea in this exterior use context with close-by neighbors in city and the use of a relatively directional piezo tweeter. Could be an indirect bonus helping keep peace with others.
However, the woofer has a rather higher Qts than I would have liked here. The plywood leftovers I have for the project allow making one pair of 15 L. boxes (net, after driver space, bracing, etc.) and Fb stands at 60 hz. The LF response with such Qts, as modeled with 2 Pi in Winspeakerz, is bumpy since the box should have some more volume. Tuning the box lower to get rid of the bump would get vent length too close to the back panel, plus the intent isn't to ask this size driver to reproduce very low frequencies... Then, the idea of creating additional virtual box volume with lots of damping material was put aside in view of the following.
Being mostly an exterior use speaker in a closer to 4 Pi (full space) environment than 2 Pi (half space), the effective LF response also modeled in approx. 4 Pi indicates it should drop by a number of db if there's absence of significant close boundaries where the speakers might be located (note 1). In this case, somewhat "saved" from bumpy sound by the outside acoustics. But if the 4 Pi loss is too high, then maybe shortent the vent a little (tune higher) to increase LF bumping?? Or preferably mitigate that LF drop with some alternative speaker placement compromise, like a table/rear wall distance + or - placement may be sufficient.
Actual on-site testing will give a clue about LF loss level, but for the time being the calculated transition frequency (note 2) at which the 4 Pi effect should START influencing LF response is 260 hz with loss increasing while going further down up to about Fb. Not all woofer response is affected by this 2 Pi/4 Pi placement issue, only the LF part of the spectrum.
Other calculated numbers of interest here: woofer upper frequency bound, where off-axis response BEGINS to fall off (note 3) is 1,176 hz; Directivity frequency (-6 db) (note 4) where DI raised no more than 6 db is 1,872 hz; Note that flat on-axis response can extend beyond these frequencies and it does. Maximum crossover frequency re driver size, as per "traditional formula" (note 5), is 2,948 hz. The initial INTENTION here is to X-over around 2 - 2.5 khz at 6 db/oct. low-pass, as I have inductors on-hand for this, plus to benefit from SOME of the driver's higher response, even if more directional, for reasons explained earlier.
Finally, considering woofer sensitivity, I may have to drop the piezo tweeter's sensitivity (about 94 db?) to minimize unbalanced sound levels between tweeter/woofer. More will follow.
Richard
1- Analogy to JBL's John Eargle 2 Pi vs 4 Pi designed small console top near field monitor speaker systems is interesting here regarding impact on LF response (Handbook of Sound System Design, P. 294-5).
2- Same, P. 22, 295
3- Same, P. 99
4- Same, P. 100
5- Speed of sound (inches per sec.) divided by effective cone diameter (inches) = Frequency
3 Attachment(s)
Motorola 1025A 2x6" mid/high horn pictures
First pic, back cover can be separated into two parts, one side showing cone and resistor, the other side showing the grey piezo ceramic element on top of which there is an orange rubber ring. The pink stuff under the piezo element seems a lot like pink insulation. If I remember well this pink stuff is to damp some resonance. On this model the piezo element is larger at 32 mm dia. than on the 1005A type at 21-2 mm. But this driver does mid/high from about 1.8 khz and and up, the other driver from about 4 khz and up.
Second pic, rear cover removed showing cone or diaphragm, plus resistor with orange and silver lines at 3 o'clock, inserted into driver casing, not a whole lot of space to play there.
Third pic, unit's back cover, on cheap copies sometimes even the screws aren't as nice as these, often no name nor country of origin...
Attachment 82408Attachment 82409Attachment 82410
The piezo horn tweeter 1005a type
* Sorry for the delay to come up with this next part, quite busy these days...
B) SECOND PART
To go a little faster here with typing of the following info, Motorola = Mot, CTS is same, Piezo Source = PS and Jon Risch = JR.
1) Series resistance details
We know from the previous post here there's a 2-3 db or so HF response bump in the 15-20 khz region for 1005A type piezo tweeters (PS and CTS). Plus, that a 2 ohm series resistor won't affect top end response (Mot). So 2 ohms being the lower resistance number, what about the higher resistance value one may use?
Mot has a "circuit for high roll-off" chart on this and the effect such higher value resistors have on top end response (at 20 khz): 50 ohms/-2.5 db, 100 ohms/-5 db, 150 ohms/-7.5 db. I don't see why someone would want to use as high as 100-150 ohm resistor to correct a 2-3 db or so bump, but if taste requires it the info is here. For response correction of a db amount in between these numbers, a series resistor value in between the above resistances can be used.
As for others' resistor indications, CTS 50 ohms "without noticeably affecting the response", PS 30 ohms, JR "about 30-50 ohms to tame the very top end" (1999) and 50 ohms (2009). So JR mentions higher value series resistor tames the top end, like Mot says. CTS' note re "no" effect on response is curious vs Mot and JR.
JR (1999): "Adding resistance in series with a piezo will actually roll-off the highs a bit, adding more will roll-off the highs noticeably."
My choice, is in the range of 2-15 ohms initially (both of these numbers appear in Mot's spec sheet) for listening test purposes, to minimize impact on HF response, then I'll adjust value higher if need be.
With regards to resistor type and capacity (watts) to use, JR (2009) mentions "can be 5-10 W non-inductive type, metal oxyde, etc." CTS suggests 2 W, PS & Mot don't say. In my book 2 W seems pretty low and the price difference with higher capacity ones being a matter of cents, I'll use 10 Watts here.
I'll go with ceramic type resistors as I have many of those on-hand (10 and 25 watts), though I also have a bunch of 10 Watt metal oxyde type, but prefer to keep them for other projects I have. Btw a 25 Watt resistor will be required later for the crossover.
I tested the resistance of a good sample of both types of resistors (various values) with the 1% ohm accuracy multimeters I have, and even the less "noble", lower cost, ceramic ones showed good tolerance, except for the odd one. So no big surprises here and this is fine. Knowing their accuracy isn't really a critical issue for me since I have a habit of measuring each one I use on a project to assess its tolerance, and discard any donkey.
2) Attenuation (Pad) details
JR (1999) explains that two capacitors in series (piezo tweeter and capacitor) create a voltage divider (piezos are voltage driven) therefore reduce (Pad) tweeter output level. A series capacitor of the same capacitance value as that of the tweeter's (hence my interest in measuring tweeters earlier) will drop the overall tweeter output by 6 db.
Mot also has an "Attenuation without high roll-off" chart with effect on tweeter output level given based on series capacitor value used: 1.0 uF/-1 db; 0.5 uF/-2 db; 0.2 uF/-4 db; 0.1 uF/-7 db (remember their tweeter was rated at approx. 0.15 uF).
CTS also gives some straight level attenuation values using a non-polar series capacitor: 0.1 uF/-6 db; 0.04uF/-12 db (their tweeter was rated at 0.12 uF).
That series capacitor is located right after the "safety" resistor mentioned above. So up to now we have, looking towards the amp, the piezo tweeter, then the series resistor, followed by the series capacitor (if tweeter padding required).
I also tested capacitance on a good number of Solen capacitors on-hand (2%, 3% and 5% tolerances) and these were usually within 2-3%, even for the 5% ones. A few other lower quality/price ones (not Solen) were more like 10%.
3) Variable L-Pad
I have a pair of those L-Pads in a box somewhere but I don't plan on using variable 8 ohm L-Pads on the project's piezos. However, if I did I'd tend to follow JR and Weems' advice on this with the use of a 10 ohm parallel resistor to make a load for the L-Pad, instead of using an 8 ohm resistor shown by Mot. (more later).
For a 10 ohm resistor value with such an L-pad, Weems suggests crossover capacitor values: 2.2 uF/7200 hz; 4.7 uF/3400 hz. JR (2009) also suggests a 2.2 uF series capacitor.
BTW these tweeters have a bump of + 4 db centered at 5 khz that you may want to minimize. Therefore, the choice of the capacitor value for a 6 db/oct. high-pass roll-off, to reduce that bump, should then be kept in mind.
In Mot's and Weems' cases the L-Pad is seen wired before last, looking in the amp's direction: piezo, parallel resistor, variable L-Pad and finally the crossover capacitor. However, in JR's notes the L-Pad is the last item of the bunch: piezo, series resistor, Pad capacitor (if needed), parallel resistor, high-pass filter capacitor and L-Pad...
(David B. Weems, Designing, Building, and Testing Your Own Speaker System, 4th Ed., McGraw-Hill, 1997, P. 101).
4) Listening tests
For listening tests, instead of soldering the components (resistors/capacitors) in place right from the start, I use temporary connections, easy to do and undo, like using a small piece of 5/8-3/4"plywood and 3/8-1/2" screws but with a little larger flat head, to do a turn or two of the components' wires under the screw's head, then screw it down into the wood. The screws act as the junction between and holding the components in place for the time of the listening tests. Holds well, gives a good connection and is easy to undo to try another resistor and/or capacitor. It could even be done on the bottom panel of a cabinet.
More to come.
Richard