Effects of surface mounting versus flush mounting drivers
When it comes to DIY speaker design it is often asked if it is really necessary to go through the extra work required to flush mount a driver when it comes to overall performance. I'll try to address this issue with some theoretical concepts as well as actual measured data. The problem with surface mounting a driver is that the thickness of the frame of the driver causes a transition between the plane of the driver frame and the front baffle of the speaker. Diffraction will occur at this boundary and the severity of diffraction will be based upon the thickness of the driver frame meaning a thicker frame will produce worse diffraction effects. The effect of diffraction on the frequency response is a constructive and destructive combination of waveforms that produce dips and peaks in the response. The frequency at which these dips and peaks occur is based upon the distance between the edge of the driver diaphragm and the edge of the driver frame. This distance corresponds to a wavelength at which diffraction effects occur and shorter distances cause higher frequency diffraction effects. Drivers with circular faceplates will produce the most extreme diffraction effects because the distance to the edge of the frame is the same in all directions which reinforces the diffraction effects at the same frequency (wavelength) causing a more severe dip/peak at that frequency.
Now when it comes to flush mounting drivers it doesn't really matter much if you flush mount lower frequency drivers like woofers because the distance to the frame of the woofer corresponds to a diffraction frequency that occurs higher than the passband of the driver and doesn't really affect the response. So it is not essential to flush mount woofers although there is one benefit to doing so. Flush mounting a woofer will eliminate diffraction effects between the tweeter and the edge of the woofer frame but these diffraction effects are very minor for several reasons. The main reason why tweeter diffraction with the woofer frame is not an issue is because the distance between the tweeter diaphragm and the woofer's frame edge is not constant which spreads the frequencies that diffraction occurs at apart and minimizes diffraction effects.
Flush mounting a tweeter is another story. Since tweeters operate at higher frequencies and the distance between the tweeter diaphragm to the edge of the tweeter faceplate corresponds to a wavelength/frequency that is in the passband of the tweeter diffraction effects will come into play if a tweeter is not flush mounted. This effect will be most pronounced for tweeters that have round faceplates for reasons previously mentioned. Most tweeters have a diaphragm to edge distance between 1 and 2 inches which corresponds to frequencies around 13000Hz and 6500Hz respectively. So the wider the tweeter frame, the lower the frequency that tweeter diffraction effects will occur.
Below is a comparison of the frequency response of the Hiquphon OW1 tweeter mounted in my Hyperion speaker cabinet with and without flush mounting. As you can see the flush mounted tweeter has a much flatter response especially above 5kHz. The peak/dip at 2kHz/3kHz is due to cabinet diffraction effects cause by the edge of the front of the cabinet. Above that point there is a peak around 5kHz, a dip around 9kHz and another peak around 13kHz in the surface mounted tweeter's response. The dip around 9kHz corresponds to a distance of about 1.5 inches which is exactly the distance between the edge of the Hiquphon tweeter dome diaphragm and the edge of the tweeter faceplate. As you can see there is a +1/-4dB swing in the response caused by not flush mounting the tweeter, however, off-axis these dips and peaks shouldn't be as prominent.
So the main point of this demonstration is to convince you that it is necessary to flush mount tweeters to obtain a flatter frequency response but woofers do not necessarily have to be flush mounted. The problem with flush mounting drivers is it usually requires skill with a router and the proper bits and circle cutting jigs. There is a way to flush mount drivers without using a router that I've used several times including in the example above. I cut out a frame from 1/8" masonite or some other type of thin plywood that matches the thickness of the driver frame and then glue it to the front baffle of the cabinet. This method was used for every speaker design shown in my projects section and is very effective. Furthermore it makes it easier to flush mount odd shaped drivers.
The only other thing that I'd like to add is that you can apply some of the lessons learned in this tutorial to driver positioning on the baffle. Just like diffraction effects between the tweeter and its faceplate, there are diffraction effects between the driver and the edge of the cabinet. Rounding over the edges of the baffle is one way of reducing these effects but it is very hard to completely eliminate them. The best way of reducing diffraction effects is to locate the tweeter at different distances to each side of the baffle so that the frequencies that diffraction effects occur are spread out across the frequency response. This is why you often see tweeters offset to one side of the cabinet in many designs. Ideally the best way to minimize diffraction effects between the tweeter and the cabinet edges is to locate the tweeter according to distances based on the golden ratio of 0.6/1/1.6. If you can't do this, at least try not to use distances which are multiples of each other. Now if you don't feel like moving your tweeter around to reduce the effects of diffraction then the other alternative is to strategically place acoustically absorbant felt around the tweeter near the edges of the faceplate. This should help reduce diffraction effects but is a whole other topic of discussion.
For more information on this topic there is a great writeup written by John Krutke at www.zaphaudio.com and I highly recommend checking this out this link. Also browse around the rest of his site if you have time because there are a lot of other interesting discussions and projects there as well.