LM3886 as Tweeter high pass amplifier.

[Hayk]

I knew this IC for its particular high frequency character. It disappears when bridged or in composite, to say that its nonlinearty makes it sound nice.
This chip is used as tweeter amp in many powered speakers of high quality bearing famous brands as JBL, FOCAL.
To use this amp without crossover and speaker protection, I have designed a decade ago for my personal use, a single supply 48v 2khz 3rd Butterworth. I have lost the original circuit but I designed another one today, maybe it might interest someone else.
The LM3886 has a secret not explicitly advertised in the DS.
This chip is not ordinary silicon semiconductor, it is aerospace grade silicon carbide with 250°C junction max temperature and the outputs are protected not to exceed and the low level, to shutdown at 165°C for long term reliability reason. Normal silicon semiconductors which are 150°C junction, are designed for 120°C average operating temperature to derate with power increase. The chip has 150°C operating temperature and a table on page 11 derates to be 114+30=144°C for 30w dissipation and 114°C the tab temperature. The purpose of such technique is to use smaller heatsink at high temperature.
The SQ gets still better at higher operating temperature as the Hfe of the transistors increases up to 120°C junction temperature.
This why the heatsink should be kept hot about 60-70°C with a heater resistor and thermostat switch to get the best sonic character of this amp.

 

[Hayk]

I will transform the circuit into Sallen-Key high pass filter which is much easier to calculate for a required frequency, see the formula bellow.
For Butterworth 2nd order, the Q =1/alpha is 0.7 and 3rd is 1 with output capacitor and load making the 3rd zero. The H in the formula is the gain fixed 20.
The Q is adjusted by C2/R1 keeping C2xR1 constant.

 

[Hayk]

This is much simpler circuit to adjust. The cut frequency here is about 2khz and adjusted by C3=C4, inversely to frequency. C1 is dependent to the load to form high pass at the same cut frequency. If 2nd order is required, the C1 is 10 times higher and the Q adjusted to get -3db at cut frequency.
The Q is adjusted by R6 and very sensitive so it doesn't act much upon the cut frequency.
I'll make a kicad detailed schematic.

 

[Hayk]

This is the starting circuit to be developed according the SQ.
I omitted the virtual ground transistor as I have done numerous circuits with floating ground without it, successfully.
The lm3886 is mounted at the bottom side flat, to have a ground shield protection from EMI, as it is very sensitive to.
It will require a small heatsink with heater resistor and thermostat.
I used once 100x25x10mm heatsink with this chip for 30w continues with 90°C tab temperature. With this circuit, I will use the TF, 2°C/w and probably silicon isolator to run the chip at sufficient temperature. A tweeter can need high peak power, but the average heating power for most hifi tweeters is less than 5w if not 2w for small voice coils.

 

[Hayk]

The board is very small, to give more value, I will add an optional slew rate limiter which can also act as buffer to get the 3rd RC high pass instead of the output capacitor.
Why a slew rate limiter. Many audiophiles spend lot of money on tube amplifiers to get that particular sweet, smoother sound. It has nothing to to with the tube character, it is power limited because of output transformer frequency response. Take the Williamson 15w amplifier, 20db NFB with 100khz -3db. It must have an output transformer of 10khz-3db. This means at 10khz, the maximum output power before slew rate limiting is 7.5w and quarter of it at 20khz. Normally with music spectrum, you never reach that power but the complex wave does reach the slew limit. The sound of saxophone particularly gets artificially, very agreeable as some female voices.
The circuit bellow can be adjusted by the resistor R2 50k or by passed by cutting the diodes.

 

[Hayk]

I replaced the diodes by transistors to keep the input clean which can be common with others. The potentiometer is rearranged to to keep the response constant.
This is the circuit at this point for 3rd order Butterworth 2khz.

 

[Hayk]

I made the PCB so that the slew rate limiter can be cut off if not needed or to be mounted vertically if space is needed. The slew rate limiter is autonomous and can be used for other projects.