So I've been experimenting with the B&W FST, on and off, for a while and I've finally come up with a combination that I am pretty happy with. It certainly isn't where I thought I'd end up!
The basic idea started when I saw the new SEAS coax drivers, I figured I'd like to give those a go, but didn't really have any real reason to do so, except to satisfy my curiosity. Then I got to thinking that the FST has the perfect surround for decent coaxial performance and the phase plug is easily removable. Indeed, unscrewing it leaves you with this.
A nice flat pole piece with an M6 hole tapped through the middle and a decent amount of depth in which to sink a tweeter. The only trouble is that the internal coil former requires a tweeter around 29mm in diameter to fit properly. This isn't easy to find. I first tried out the OT19 from Vifa. I already knew that this would be too wide to fit properly but I thought that it might be possible to mill down the outside to make it fit - no go, unless you want to risk damaging the tweeter. The only other options I could find would be to acquire some KEF replacement domes, some very unknown car tweeters from Ciare, or to try giving the ND16 from Dayton a go. I wasn't even sure if the idea would do what I wanted it to do, let alone work properly, so the inexpensive Dayton's fit perfectly for an experiment. I had already used these in a small loudspeaker and I had some concerns about their low end capabilities as the design would necessitate a cross no higher than 2500Hz. In a non wave-guide loaded design this tweeter really requires an xover around 3.5kHz, so the hope was that the gain, provided by the cone loading, would help it in this regard.
Now the ND16 isn't perfect for the job as it's external diameter is bigger than 29mm, but whereas the OT19 cannot be milled down, the ND16 can. Then a pair of contacts need to be transferred from the back of the tweeter through the front.
Here is a picture of the milled down tweeter and illustrates how I got the contacts through to the front.
And then here's a rear shot with an M6 screw glued to the back.
And two shots with the tweeter mounted. (Yes the FST sphere still needs finishing ops
Now I was very curious to see how the tweeter measured, all things considered wave guides and coaxes tend to have some issues, so this was always a bit of a gamble. The on axis curve was quite ugly, but they always tend to be. So slightly off axis things looked like this.
The only real issue is the dip at 12.5kHz, otherwise it looks great. The 12.5kHz issue varies in how bad it is with how far off axis you go.
Filtering this was surprisingly easy with only one notch being required along with the usual high pass elements.
Now I figured that the top octave issue would be fairly audible but it's far less of an issue than I thought it would be. I have read a number of comments, from wave guide people, saying that it is often best to ignore top octave stuff so long as what's below is absolutely fine, so maybe this is a prime example of that in practise.
That aside it sounds great, a $10 tweeter with a $200 mid isn't what would usually be prescribed but in this case it works very well together. I've measured the distortion at realistic listening levels and aside from a slight increase in 2nd and 3rd order around the lower end of the tweeters range, it seems to cope absolutely fine. There isn't an increase in 4th and 5th, which is what I was mainly concerned about. It'd be nice if Scanspeak would release a tiny Be tweeter, but I can't see that happening any time soon!
There is obviously some room for improvement and over the next few days I am going to be experimenting with DIY phase plugs in an attempt to help with the top octave issue as these are usually created by throat/tweeter diaphragm issues.
The basic idea started when I saw the new SEAS coax drivers, I figured I'd like to give those a go, but didn't really have any real reason to do so, except to satisfy my curiosity. Then I got to thinking that the FST has the perfect surround for decent coaxial performance and the phase plug is easily removable. Indeed, unscrewing it leaves you with this.
A nice flat pole piece with an M6 hole tapped through the middle and a decent amount of depth in which to sink a tweeter. The only trouble is that the internal coil former requires a tweeter around 29mm in diameter to fit properly. This isn't easy to find. I first tried out the OT19 from Vifa. I already knew that this would be too wide to fit properly but I thought that it might be possible to mill down the outside to make it fit - no go, unless you want to risk damaging the tweeter. The only other options I could find would be to acquire some KEF replacement domes, some very unknown car tweeters from Ciare, or to try giving the ND16 from Dayton a go. I wasn't even sure if the idea would do what I wanted it to do, let alone work properly, so the inexpensive Dayton's fit perfectly for an experiment. I had already used these in a small loudspeaker and I had some concerns about their low end capabilities as the design would necessitate a cross no higher than 2500Hz. In a non wave-guide loaded design this tweeter really requires an xover around 3.5kHz, so the hope was that the gain, provided by the cone loading, would help it in this regard.
Now the ND16 isn't perfect for the job as it's external diameter is bigger than 29mm, but whereas the OT19 cannot be milled down, the ND16 can. Then a pair of contacts need to be transferred from the back of the tweeter through the front.
Here is a picture of the milled down tweeter and illustrates how I got the contacts through to the front.
And then here's a rear shot with an M6 screw glued to the back.
And two shots with the tweeter mounted. (Yes the FST sphere still needs finishing ops
Now I was very curious to see how the tweeter measured, all things considered wave guides and coaxes tend to have some issues, so this was always a bit of a gamble. The on axis curve was quite ugly, but they always tend to be. So slightly off axis things looked like this.
The only real issue is the dip at 12.5kHz, otherwise it looks great. The 12.5kHz issue varies in how bad it is with how far off axis you go.
Filtering this was surprisingly easy with only one notch being required along with the usual high pass elements.
Now I figured that the top octave issue would be fairly audible but it's far less of an issue than I thought it would be. I have read a number of comments, from wave guide people, saying that it is often best to ignore top octave stuff so long as what's below is absolutely fine, so maybe this is a prime example of that in practise.
That aside it sounds great, a $10 tweeter with a $200 mid isn't what would usually be prescribed but in this case it works very well together. I've measured the distortion at realistic listening levels and aside from a slight increase in 2nd and 3rd order around the lower end of the tweeters range, it seems to cope absolutely fine. There isn't an increase in 4th and 5th, which is what I was mainly concerned about. It'd be nice if Scanspeak would release a tiny Be tweeter, but I can't see that happening any time soon!
There is obviously some room for improvement and over the next few days I am going to be experimenting with DIY phase plugs in an attempt to help with the top octave issue as these are usually created by throat/tweeter diaphragm issues.
Comment