I do have 0.1uF at both positive and negative rail but not at the IC legs. They are quite far, please see C13, C14 on PCB first post. I guess they don't count?
(also, yes pin 5 and pin 1 are connected, I've updated the schematics now)
Sorry for the confusion. My schematic is meant for my own use, hence the pins are not labeled.
The 27K is the muting resistor and is the connected to Pin 8,
Due to lack of space, Pin 7 to Ground is not drawn.
Neither is Pin 4 to V-
Also not shown are the de-coupling caps of 100uF with 0,1uF at the V+ and V-.
Apart from these, the circuit is complete.
Mike
The 27K is the muting resistor and is the connected to Pin 8,
Due to lack of space, Pin 7 to Ground is not drawn.
Neither is Pin 4 to V-
Also not shown are the de-coupling caps of 100uF with 0,1uF at the V+ and V-.
Apart from these, the circuit is complete.
Mike
No, not really. I did spot those capacitors in the picture in Post 1. The reason they "don't count" is that there's significant inductance (trace length) between them and the IC. This means the capacitor + trace will act inductively at frequencies where they need to be capacitive in order for the LM3886 to remain stable. This is pretty fundamental decoupling, actually. The good news is that you can easily solder the decoupling capacitors to the bottom of the board. Just add a capacitor from V+ (say pin 1) to ground and another from V- (pin 4) to ground. You can find a good grounding point at C5/C9 (and similar for the other channel). A 100 nF, 50 V film cap or 1.0 uF, 50 V X5R or X7R ceramic will work well here.
I'm guessing/hoping you have pins 9, 10 reversed in the schematic in Post 20. The circuit as shown will oscillate because it has positive feedback instead of negative.
Tom
This looks to be a good writeup on decoupling: https://www.analog.com/media/en/training-seminars/tutorials/MT-101.pdf
Note that this particular writeup focuses on power supply noise. In your case the issue is more power supply inductance. The techniques and underlying theory is the same in the two scenarios though.
Tom
Note that this particular writeup focuses on power supply noise. In your case the issue is more power supply inductance. The techniques and underlying theory is the same in the two scenarios though.
Tom
Thanks again Tom. I made mistake, here's corrected schematic...
I'll try to add coupling C later this weekend.
You have Cmute connected incorrectly in that schematic. It should connect from the MUTE pin to ground. Then 33 kΩ from the MUTE pin to V-. I'm thinking that's just a schematic error and the PCB is implemented correctly.
If the LM3886 still oscillates after you add the decoupling capacitors, I suggest reducing the two 7.5 kΩ to 1.0 kΩ. Scrap R3, C3 and replace them with 20 kΩ in parallel with (47 pF + 20 kΩ).
Tom
If the LM3886 still oscillates after you add the decoupling capacitors, I suggest reducing the two 7.5 kΩ to 1.0 kΩ. Scrap R3, C3 and replace them with 20 kΩ in parallel with (47 pF + 20 kΩ).
Tom
Thanks again Tom. I've read the stability article on your website. It is great and concise.
I wonder in this case, since I already have the board, is it worth to remedy the issue by changing the physical location of feedback resistor to underside, right across IC pins? (I have to put coupling capacitors underside anyway). Would it make sense to add a "fix" like this in all not-so-good layout on aliexpress/ebay board? Of course it wouldn't be as good as well-engineered board, but should be sufficient to tame the oscillation?
I wonder in this case, since I already have the board, is it worth to remedy the issue by changing the physical location of feedback resistor to underside, right across IC pins? (I have to put coupling capacitors underside anyway). Would it make sense to add a "fix" like this in all not-so-good layout on aliexpress/ebay board? Of course it wouldn't be as good as well-engineered board, but should be sufficient to tame the oscillation?
I doubt the oscillation is due to the layout. It could be, but I doubt it. For now my money is on the decoupling. My second guess is the feedback network. For now I would use the PCB as-is and swap the resistors as I mentioned in Post #26. Just solder in the series combination of a 20 kΩ resistor and a 47 pF C0G/NP0 ceramic capacitor in place of C3 on the board. If that gives you a stable amp leave it be.
Tom
Tom