lm4766: Discrete changes in gain?

[pulexirritans]

Can anyone tell me how to make discrete changes in gain on the lm4766? The data sheet says gain is determined by resistors Rf and Ri, but does not explain how to design for a particular gain.

I'd like to have a set of resistors controlled by dip switches to replace Rf or Ri so I can bring the gain of whatever tweeter I happen to be using to the level of the woofer. I intend to size the resistors so I can get 1db or 2db increments over a small range.

Thanks!

[AndrewT]

The closed loop gain determines the stability compensation required to allow the chipamp to accurately reproduce the signal.

If you have switchable gain you are introducing a high risk of instability and even oscillation.

The gain of a non-inverting amplifier is 1+ Rf/Ri at frequencies where the impedance of the capacitor is very small relative to Ri. Check datasheet for minimum closed loop gain for stability into a resistive load.
Now, increase the gain to make it adequately stable into a reactive (speaker) load.

Don't use that schematic. It has omitted many of of what National term "optional" components. They are not optional, unless you have a lot of experience in how to troubleshoot potentially unstable amplifiers.

[pulexirritans]

Thanks Andrew.

Somehow I missed the equations for determining gain on pg 17 of the data sheet.

As for optional components, I assume you mean figure 4:



1.Can anyone tell me if I need Rsn and Csn? The amps will be driving 8 ohm speakers with no passive components in line. The inductance of the woofer is ~1mH while the tweeter is very low, on the order of 0.001mH. I'm guessing the tweeter does not need these while the woofer might.

2.No Idea what C1 does, but I'll just assume I need it.

As for the other parts, I think I can get away without them. Please correct me if I'm wrong:

3. I don't think I need Rin or Cin since I'm going to have an input buffer with a decoupling capacitor.

4. I think I can get away without the inductor and the 10 ohm resistor since a.) there will be no passive components/capacitors and b.) the capacitance of the speaker wire should be 170pf (15ft of 18 ga zip), less than the 0.2 uf that the data sheet says is the threshold for ringing. That is assuming the voice coil has no capacitance. Anyone have any experience with this?

[AndrewT]

Do not omit any of those components until you know what effect removing will have.
Add RF attenuation to the input.
Check again what National say about Cs !

There is a lot of leeway for altering values of components, but that is very different from omitting components.

[Redshift187]

Originally Posted by pulexirritans

Thanks Andrew.
2.No Idea what C1 does, but I'll just assume I need it.

If you mean Ci, in series with Ri, it reduces the gain at DC to 1, so that you don't amplify any DC component on the output.

[diptangshu]

Originally Posted by pulexirritans

Thanks Andrew.

Somehow I missed the equations for determining gain on pg 17 of the data sheet.

As for optional components, I assume you mean figure 4:



1.Can anyone tell me if I need Rsn and Csn? The amps will be driving 8 ohm speakers with no passive components in line. The inductance of the woofer is ~1mH while the tweeter is very low, on the order of 0.001mH. I'm guessing the tweeter does not need these while the woofer might.

2.No Idea what C1 does, but I'll just assume I need it.

As for the other parts, I think I can get away without them. Please correct me if I'm wrong:

3. I don't think I need Rin or Cin since I'm going to have an input buffer with a decoupling capacitor.

4. I think I can get away without the inductor and the 10 ohm resistor since a.) there will be no passive components/capacitors and b.) the capacitance of the speaker wire should be 170pf (15ft of 18 ga zip), less than the 0.2 uf that the data sheet says is the threshold for ringing. That is assuming the voice coil has no capacitance. Anyone have any experience with this?

I have tested the chip with the configuration in attached file few days back and I got good enough power and a decent output. You can try this.

Supply voltage: +-28V DC
Filter Cap: 9,400 mfd(4700 X 2) per rail
Xformer current: 2 amp per rail
input: P-P 3V RMS max, from computer sound card
Output: not measured accurately, around 35W in 4 ohm
Heatsink: Xeon server hetsink
Ambient temp: Around 24 - 27 Deg C


You can try and please share your opinion.

[pacificblue]

@diptangshu

Why is the Thiele network reversed? R+C should be next to the IC. You connected them next to the speaker.

[diptangshu]

Originally Posted by pacificblue

@diptangshu

Why is the Thiele network reversed? R+C should be next to the IC. You connected them next to the speaker.

mistake!!! I forgot to mention about the change. I changed that from the diagram. "Thiele network" used before the 10ohm resistor ie next to IC pin. Accordingly I changed that in the PCB layout also.

[AndrewT]

the order of RC or CR in the Zobel part of the Thiele Network does not matter.
Do not put it at the speaker terminals nor at the speaker.
Put the RC very close to the chip with a short connection to the PGND.

[rajudp]

Originally Posted by diptangshu

I have tested the chip with the configuration in attached file few days back and I got good enough power and a decent output. You can try this.



You can try and please share your opinion.

it ia a very good design , i also made this design