Protection Schemes
Protection of the $225 15" woofer is important. It is also necessary to protect the power transistors from short circuits, both from shorting across the bridged outputs and either side to ground.
Several of the protective measures are latching, & these generally turn off the +-24V supplies. If there is a fault, the question is, what was the fault? An Atmel CPLD is used for amplitude control of the 65.4Hz sine wave, & it has enough spare to latch faults. There may be enough logic to report out an LED hex digit, as follows. (Faults through 5 are shown by 7-segment display in the supply box, faults beyond 6 are shown by 7-segment display in the amplifier.)
0 fail to turn on + or - 24V supply, state 1110 was encountered on the way to state 0100 in the supplies' state machine (a separate CPLD in supply box) (report which supply it is by single LEDs.)
1 + or - 24V supply lost power after successfully attaining 0010, state 0110 in one of the state machines
2 OFF pushbutton was pushed, push START button at any time
3 an auxiliary supply is not good
4 overtemp in 6ohm (not latched, pull bias spreader to ground)
5 an inrush thermistor is too hot for supplies to be started (not latched, just push START button when cool)
7 bias spreaders are pulled to ground but outputs are not at ground (engage 6ohm protective resistors in supply box)
8 overtemp in oil (pull bias spreaders to ground, not latched)
9 short to ground, or across the bridge outputs (overcurrent in one of four driver transistors)
A speaker attenuator is active (relay) (not latched)
b meltwater level is too high
c power transistor is shorted, applying too much voltage to a speaker terminal: one of the upper left transistors
d same, lower left
E same, upper right
F same, lower right
May 15: additional protection possibility, alarm #7 in the list above, for the amplifier box was realized May 13. Just after +-24V startup, when state machines have both attained 0010, it is possible to do a two-stage amplifier check (on both sides of bridge). 1) Starting with bias spreaders grounded and analog switches grounding the inputs, check that outputs are near ground. 2) Then release the bias-spreader groundings & see if the outputs continue to be near ground. Finally, release the analog-switch groundings & let program material get through.
It may be time to change the hardware base for amplifier-box protection, from Atmel CPLD to Arduino Uno. The fault states reported by 7-segment display may increase a lot if the two-stage grounding scheme is realized; I have 14-segment alphanumeric displays that would do a fine job, & Arduino would be a great way to handle the added complexity, & also maybe prioritize the reporting of multiple faults at once. With Arduino, it would be possible to have an Ethernet shield that could report faults to a network, & report the thermistor temperature of the left 4uH inductor, the one that simulates speaker voice-coil temperature.
If the amplifier box changes from CPLD to Arduino, the question is whether Arduino can handle the trimming of the loop gain of the 65.4Hz sine oscillator. But in August, I did the 65.4Hz oscillator with 4011 NAND chips, and that is successful as a standalone module, no Arduino or CPLD.