I want to build a 'speaker to use in the lab. as a source of sound. My thoughts are to have a frequency response that deviates by not more than ±1.8 dB from 100 Hz to 20kHz, or there abouts. A tall order.
I want to base it on one I have been using for about 20 years which uses an Audax bass/mid range with a fibreglass cone, and cross it over to a soft dome Audax tweeter. It will have to be active, although with a passive crossover, only one amplifier per channel would be required. This design I have called a 'pseudo-active design, and uses a common transformer-based power supply, like I'm using in the 'computer 'speaker. The bass/mid units work really well in this small box, and the sound is very near to my Epos M5's, is that good?
Here are the details of the tweeter:
It can be seen the frequency response is very flat above resonance, albeit at 30° off axis, and rising slightly from about 4kHz on axis, but nothing that can't be corrected. It uses a neodymium magnet, which although very powerful, suffers from loss of power when heated, so I think I will attach a heat sink to the back of the unit.
And here is the bass/mid driver:
Again, very smooth between 100Hz and 500Hz, then rising at about 6dB/octave which then dips after 1.5kHz at about 3dB/octave, both of which should be accommodated with simple filters, hopefully. Crossover at 3kHz looks good, which will be carried out by a suitable passive crossover, together with other response shaping, like baffle compensation, and perhaps time delay compensation for the tweeter.
Power. I've decided on using chip amplifiers, for many reasons. Although a very good amplifier design could be used, using discrete components, the amplifiers are for a lab 'speaker, not to listen to music, although they should be fine for this, too. Simple amplifiers using cheap discrete components still need expensive 'iron-ware', that is, the transformer and enclosure and heat sinks (although the last two would probably be made in aluminium!). These are usually expensive items, although bargains can be had from eBay, if one knows what one is doing. I don't expect to be running these 'speakers at high output (a quick calculation, assuming 90dB/watt/m) gives a sound pressure of 105dB at 1 metre, for one loudspeaker, far too much for comfort, and my needs. So, I can probably reduce the heat sink size, and maybe the transformer VA rating. So, I have decided on the chip amplifier, it's the TDA7294. I've bought the kits, including the all important pcb's, and already I have the heat sinks, diode bridge, smoothing caps and a transformer in the project bag.
It will be a pseudo-active design, using one amplifier per channel, built into the loudspeaker enclosure, together with a fully passive crossover, but with a separate power supply, providing ±32 volts DC and other voltages that may be required for frequency shaping duty. The power supply will power both loudspeakers.
I will make use of an existing one, that used to be a mid range enclosure for a KEF Concerto loudspeaker, supplied as a kit from Wilmslow Audio, way back in the '80's. They are made of mdf, and solidly built. Obviously, plain mdf is useless for audio work, as it has very little transmission loss, so that what is generated inside the box will transmit through the side walls, and add confusion to the overall sound. BUT, I have just been given 4 panels of 'Alucobond', about 400mm x 500mm each, in a variety of colours, which are Aluminium/polyethylene/Aluminium laminates, of reasonable damping ability, but maybe when layered onto mdf (with another layer, of, say hardboard), may provide the transmission loss required. Also bought for the inside surfaces is a piece of 25mm thick 'felt', which should be useful when glued to internal surfaces, to minimize reflections.
One problem with mounting the tweeter is that vibrations from the bass/mid-range unit can be transferred to the tweeter unit, and result in intermodulation products. To reduce this happening, a way of mounting the tweeter resiliently will have to be found.
to be continued...........