Still to Do
In brief:
This whole system is a work in progress. Although operational from early 2018 many more-or-less minor things have to be finished or can be improved.
On short terms:
The control amplifier has no separate 5V for the digital circuits. Consider that. Does it improve the noise performance?
There is also some audible beep when the volume control is turned up. Is that from the Arduino?
The volume control goes to early into mute when the knob is rotated CCW. Improve that.
Find out where the audible random noise in FLT-TPA comes from. At the usual listening position it is not audible, but it should be much better.
Install the Master-Slave clocking of the TPA boards.
Measure the speaker signals during normal operation. Which output levels are actually used?
Do measurements to determine the optimal position of the cone. Use tick sounds, a microphone and an oscilloscope.
Likely the studs for the tweeter unit must be made longer.
Check the properties of the step-up trafo.
Mount fuses in the mains input, both in the loudspeakers and in the control amplifier.
Start a new construction for the ESL's. The Solosound modules are not very "open" and there are barriers in the center. A 1-piece stator would probably be better. During the measurements on the transformer I noticed some damage on more than 1 module.
Consider a high voltage unit which is not directly connected to the mains voltage.
Consider a step-up transformer with lower parasitic capacitance.
Find out if there is any difference in bass reproduction when the rumblefilter in the Control Amplifier is bridged or brought to a lower order.
Make it switchable.
Investigate effects of returned energy.
When a loudspeaker returns energy back to the amplifier (damping) an analog amplifier will dissipate it in the power transistors. A class D amplifier however will return the energy to the supply rails. Analog, and even most of the switching power supplies cannot return such energy to the mains supply, so the rail voltage will rise if other loads draw not enough current to use it up.
Do we have such effects? Measure the rail current and voltage under reactive load conditions.
Cosmetics:
Spray-paint in the final color. Must be done outdoors, or in a spray-cabin.
Black -anodize the amplifier panels.
While adjusting the volume matching of Bass and Midrange by the ears it turns out that the settings for Bass and Mid must be quit low, approximately -12 to -16 dB. How does this fit with the gains in the system and the sensitivities of the loudspeakers?
Gains in the Tweeter channel: The high-pass fiter has + 11.2 dB gain, but that is compensated in the correction circuit. The receiver on the TPA-board has +10.3 dB gain and the tweeter is calculated to have 86 dB sensitivity. This adds up to 96.3 dB
The Bass and Midrange channels both have +12.7 dB gain for frequencies that do not need correction and both speakers are specified as 89 dB sensitivity. Makes 101.7 dB. The bass speaker however is 4 Ohm which adds another 1.5 dB.
Now we have T = 96.3, M = 101.7 and B = 103.2 so a correction of -5.4 dB for Mid and -7 dB for Bass should do it.
It is not much a surprise that the tweeters appear to produce much less sound level than calculated. First they are over 30 years old and the assumptions for my calculations may be inadequate. F.t.t.b. I will live with it until the new ESL modules are operational.