I can weld but can't decide what I am going to do
3mm steel hemispheres
Fibreglass and resin
Carbon fibre and resin
Cast concrete (using molds)
They all have their own pro's and cons. Everything has a resonance frequency. Light but very strong is very appealing. Mass stores energy. Steel is attractive due to its weld ability but it has potential to ring. Concrete is very rigid but heavy in the thickness required. Acrylic would be nice but it is expensive. If cost was no object then acrylic would have been ordered but it is not easy to machine without melting or cracking.
I am starting to think about trying "allegedly" better mid bass drivers
The Visaton tweeters are good and I see no reason to explore further. I am even starting to think about trying a dopeless approach. There are 5.25 " drivers with a well behaved response and a higher sensitivity that when used firing up and down should be good with no doping or electrical filters.
The other thing going through my head for quite a while now is trying a three way approach. Although I currently have three drivers per speaker if is still only a two way as the up and down firing mid bass have exactly the same output. A 2.5 way in the up and down firing configuration makes no sense as that is only used for baffle step compensation and the opposing bipole approach already has BSC.
I have two nice B&W cast frame 6.5" mid bass from my old transmission lines that could be used as down firing bass drivers. They should have suitable Qts for a sealed cabinet. The Vas and required box volume is about right Although I don't know these values exactly. The sensitivity is 90 dB. Using a larger bass driver is not an option without the sphere becoming too physically big.
The big thing with this is I would need an electrical filter on the 6.5" bass driver
Quite a departure from the doped approach.. I don't want a speaker level low pass filter and I don't want an active one either. That leaves me with using a first order passive line level low pass filter. I don't want the insertion loss of adding a resistor and capacitor between pre and power amp plus extra components is against what I am trying to achieve.
Luckily the power amp boards have a low pass filter already built in to prevent the amplification of radio frequencies etc. All that needs to be done is to change the capacitor value to change the knee frequency down to what I would need. A 35cm sphere will have a baffle step of 6dB just like any other enclosure. The 3dB (middle) point of this for a 35cm sphere will be 330 Hz approx. That will be the xover frequency.
I lose the opposing bipole force cancellation doing a three way but it might be a trade off worth doing (or it might not)
So I then require an up firing mid range driver from 330 Hz up. Relieving the up firing mid of low bass duties is the big win here. The low xover frequency rules out the use of a dome and anyway, I don't want the xover from bass to mid any higher. I have found a suitable 4" cone midrange with a lovely response that will not need a low pass filter. It will require a high pass filter though at 330 Hz. That will be done passively at line level too. No added components required as all it will need is the change of a single capacitor value.
This 4" mid is good up to 10 kHz so I will need to increase the existing high pass filter knee frequency that I already have. Having relieved the tweeter of more lower frequency output and potential distortion I may try a single forward firing tweeter again and compare. Having a single point source for tweeter may have enough advantage over the multiple tweeter approach but that trade off can only be assessed by listening and comparing.
There are lots of changes of approach here and a few trade off's effected. Only comparisons and listening will sort this out.