Before:

After:

The replacement boards are for a late 70s bi-amplified powered speaker made by Advent.  They were funded by the speaker’s owner, Christopher Jensen, who wanted a set of amplifiers with more stability in high temperature conditions; The finished boards represent a couple months of collaborative effort. Our solution was using five National Semiconductor LM3886 Overture chip-amps (a higher output version of the LM3875 “Gainclone” chips): one chip (50W RMS) for the tweeter driver and four in bridged-parallel configuration (200W RMS) to drive the woofer. Later it was decided to build the crossover into the system instead of using the original system’s crossover board.

The design for the high channel is a straight forward LM3886 non-inverting amplifier with an AC coupled input circuit. For the bridged parallel amplifier we used the Fig. 11 BPA-200 design from National’s BPA-200 Application Note, which includes the servo amplifiers for output voltage offset compensation and uses the LM3886 in both inverting and non-inverting modes.

The boards were done in 2oz. copper and black soldermask, with integrated fuses and reverse voltage protection diodes. All high power speaker connections were done with .250″ Fast-On connector tabs. Signal inputs use RCA jacks. Rails for the LM3886 are about 1/2″ wide, running along the edges of the PCB, with a pair of 1000uF Panasonic FM series caps on each rail at each chip-amp. 2200uF FC series caps will fit into the same location. Power for the servos and buffer amp was provided by fixed +15V and -15V linear voltage regulators. The unregulated supply was also replaced, using a 500VA Avel toroid transformer, and very large electrolytic capacitors (they’re like soda cans).

The crossover circuit is integrated into the amp as a module with 0.1″ pitch headers for signal and power and locations for standoff mounting. The power section board can be used with or without the crossover, both amplifiers and the crossover have RCA input connectors. Two different crossover module boards were designed: one based off a 2-pole Linkwitz-Riley active crossover circuit, another using a 4-pole Linkwitz-Riley active crossover. Both include an input buffer before the active filters, and a simple comparator circuit with adjustable reference for driving a clip detector LED.

During design, a potentiometer for volume was added to the inputs when the crossover module was added. This makes the LPF designed into input change roll-off frequency with adjustments of volume, although in later designs this can be fixed by replacing the amp’s input resistor with the attenuator pot, unlike how it is now, before the input circuit filters and coupling. The LPF in the input circuit can also be sensitive to different source impedance. This circuit can easily be eliminated with 1206 jumper and omitting a capacitor. Also Christopher used a cap in series with the feedback divider on the high channel power amp to get its output voltage offset similar to the less than 1mV values of the servo-stabilized bridged-parallel low channel amp. He stood the 1206 resistor and capacitor up so both they fit onto the resistor pad. For this to work on the lower frequency amp or full range amps, you would need a very large value bi-polar capacitor.

I’m told pictures of the re-assembled powered speaker will be available as soon as the silicone sealer stuff dries. I guess for the rear panel, or speakers, or something. Those pics and possibly also circuit schematic and PCB layouts in later posts.

Despite the lack of posts, work on Darker Technologies has progressed between the mundane tasks of daily living. The next few posts will cover some of the work I have completed on the Curiously Strong and GrowLEDs projects, as well as related contracted work and research involving audio PCBs, and of new things to come.

Put on your seatbelts.

cs_regulators_schema.png

shifty_vu_controller_schema.png

Short term testing of hardware observations:
- Normal room temp.
- 25V unregulated psu.
- All circuits function as intended.
- Red and blue series transistors warm to touch.
- White series transistor hot set at 1/2 normal current.

Conclusions:
- Multiple rail PSU, <0.8V overhead at max string Vf. Blue rail is an example. Benefit is less dissipation per output.
- SOT223 instead of SOT23 3906 series transistors. Same dissipation but within spec. Benefit is single-rail lower-cost psu and a sexier transistor package.
- Dropping resistors for white and red LED rails. electrically same as SOT223 dissipation, retain cheaper SOT23 package.
- Higher dissipation SOT23 PNP for red rail viability without resistor. Must test at higher ambient.(MMBT4403?)
- Three channels current feedback SMPS. Much less dissipation, psu same level complication as tri rail. Cost benefit unknown.

Paper Test:

LED Vf(max)
Red =2.4
Blue =3.6
White =3.6

LED Vf(typ)
Red =1.7
Blue =2.8
White =2.8

String Vf(max)
8x Red =19.2
6x Blue =21.6
2x White =7.2

String Vf(typ)
8x Red =13.6
6x Blue =16.8
2x White =5.6

String W(max) @ 20mA
8x Red =0.384
6x Blue =0.432
2x White =0.114

String W(typ) @ 20mA
8x Red =0.272
6x Blue =0.336
2x White =0.112

System Output W(max) @ 20mA
4x Red Strings =1.536
3x Blue Strings =1.296
1x White Strings =0.228
RB Total =2.832
RBW Total =3.060

System Output W(typ) @ 20mA
4x Red Strings =1.088
3x Blue Strings =1.008
1x White Strings =0.288
RB Total =2.096
RBW Total =2.384

MMBT3904 Vce(max) @ 23V V[supply]-Vf[string_max]-Vbe[0.6V]
Red =3.2
Blue =0.8
White =15.2

MMBT3904 Vce(typ) @ 23V V[supply]-Vf[string_typ]-Vbe[0.6V]
Red =8.8
Blue =5.6
White =16.8

MMBT3904 mW(max) @ 23V/20mA
Red =64
Blue =16
White =304

MMBT3904 mW(typ) @ 23V/20mA
Red =176
Blue =112
White =336

MMBT2904 Temp(max) @ 23V/20mA 357C/W
Red =22.848
Blue =5.712
White =108.528

MMBT3904 Temp(typ) @ 23V/20mA 357C/W
Red =62.832
Blue =39.984
White =119.952

MMBT3904 mW_derate(max) 23V/20mA Ambient 25C 2.8mW/C
Red =63.9744
Blue =15.9936
White =303.8784

MMBT3904 mW_derate(typ) 23V/20mA Ambient 25C 2.8mW/C
Red =175.9296
Blue =111.9552
White =335.8656

MMBT3906 mW_overhead(max) 23V/20mA/25C 350mW[SOT23]-mW_derating-mW_used
Red =223
Blue =319
White =-257(!)

MMBT3906 mW_overhead(typ) 23V/20mA/25C 350mW SOT23
Red =-1
Blue =127
White =-321(!)

MMBT3906 Source, Vf(max)/If=20mA/23V/25C/77F
Red = OK
Blue = OK
White = FAIL

MMBT3906 Source, Vf(typ)/If=20mA/23V/25C/77F
Red = ALMOST, BUT FAILED
Blue = OK
White = FAIL

• dts7had01_schema.png
• dts7had01_parts.txt

Going to swap the blue 0603 LED for a red one. Maybe amber. Will possibly swap the 0402 resistor with a larger one. Try to kill the sniper beacon blink effect. Alternative is a thick film or thin paper light filter behind the translucent white button.

The grid is .250″.