LM3886 Servo Design

[Redshift187]

So for my next project, I'm trying to design my own board (and learning to use Eagle at the same time). As the title suggests, it's a LM3886 with servos for removing DC offset. I used the datasheet for the LM3886 and the non-inverting servo design from AN-1192. I have finished a first revision of the schematic design.

Notes:
- there is no input shunt resistor as I plan on using a passive attenuator
- all capacitor values are in uF

Let me know what you think!

[peranders]

It's a no-no to have a floating JFet input. Tie it down with anything, 100k, 1M or whatever. You can get "unwanted" effects if the input of some reason leaves open.

D1 and D2 don't have to be schottkys, use regular diodes such as 1N4148,

[Redshift187]

Ok, so I can add the input resistor.

The app-note lists "low leakage" diodes for the servo. It might help if I actually understood what they do in this situation.

[infinia]

Originally posted by peranders


D1 and D2 don't have to be schottkys, use regular diodes such as 1N4148,

Originally posted by Redshift187


The app-note lists "low leakage" diodes for the servo. It might help if I actually understood what they do in this situation.

Hi D1,2 are protection for the integrator opamp. The chip amp outputs more than the opamp supplies (Vcc,Vee > +/-15 V), so it could turn on the LF411 substrate diode. Normally schottkys are prefered for this, but since R2 is so large to limit damage.
EDIT>signal diodes provide the lowest leakage, so are best here.

[AndrewT]

Hi,
the input impedance of the non-inverting 3886 is very high.
If you add that Zin resistor that Peranders talks about then that will define the input impedance of the amplifier. It will also supply the input offset current.

You do not need an input buffer to feed the Zin of this amp. Delete it.

Now, choose whether the attenuator needs a buffer to drive the interconnect.
Similarly decide if your source needs a buffer to drive their interconnects.

The 411 is excellent for the servo duty.
I suspect many opamps can provide a far better buffer than the 411. Change it.

Some low power transistors (I'm not calling them signal transistors deliberately) are even better when connected as diodes than diodes are with respect to leakage.
I think it was Jung that discussed particularly good transistors for this duty.

Finally,
a DC coupled amplifier with servo control of the output offset is great as long as all components are working properly.
Add a DC detect and output isolation to protect your speakers in case anything becomes faulty. Consider also adding an input mute activated by that same DC detect circuit.

[aparatusonitus]

A right way to protect an J-Fet servo opamp input which I've learned recently from ilimzn

[Redshift187]

Ok, starting on the input side. I've been reading more about servos, and it was recommended that servo amps have a low pass filter at the input so they don't oscillate at frequencies above the audio range. Does this sound like a good idea?

In the meantime I'll do more research about transistors for servo protection and DC detection circuitry.

[Redshift187]

Ok, finally got around to updating this. Here is the amp circuit with the shunt resistor, input pot, input buffer removed, transistors for op-amp protection.

In order to keep the schematics easier to read, I'm making the DC detector and protection circuit a separate schematic.

[Redshift187]

Finally got back to working on this. I decided to put the voltage regulators on board, and keep the protection circuitry off board for this version. In version 2 I'll put the protection circuitry on board too.

Schematic:

[Redshift187]

Silkscreen and component placement: