I've built a few LM3886s essentially based on the datasheet, and Tom's useful web pages regarding the amplifier.
For my first builds, I used an input resistor of 20K, a DC blocking capacitor of 4.7uf, a feedback resistor of 20K, and the current balancing resistor of 1K at the input.
This works and sounds good, but I'm wondering if I can do a bit better, while making a few changes.
1) The 4.7uF capacitor is massive. It takes up about 1/4 of the PCB by itself. I wouldn't mind lowering it to 2.2uF as this gives me more breathing space on the PCB to move things around
2) Tom's guide 'Complete Guide to Design and Build a Hi-Fi LM3886 Amplifier' says you want to balance the input currents against the feedback current.. I have 20K + 1K on the input, and 20K on the feedback, so while it works, I wouldn't mind making this exact on the next build.
To that end, I've debated the following changes.
1) Change the input resistor from 20K to 23.7K
2) Change the feedback resistor from 20K to 24.9K
3) Change the gain setting resistor Ri from 1.0K to 1.2K, keeping the gain really close to what it was before
4) Change the input current resistor Rb from 1.0K to 1.2K, making the current exactly balanced (23.7K + 1.2K = 24.9K)
I calculated that the change in the input configuration would drop the frequency response at 30Hz by about 0.03dB compared to before, and I can live with that. My question is regarding the feedback compensation network (the optional components). Since I've kept the gain pretty much the same, do I need to worry about adjusting/simulating those, or will they likely be ok as they are now (20K + 47p I believe)? Anything else I might need to worry about with this change?
For my first builds, I used an input resistor of 20K, a DC blocking capacitor of 4.7uf, a feedback resistor of 20K, and the current balancing resistor of 1K at the input.
This works and sounds good, but I'm wondering if I can do a bit better, while making a few changes.
1) The 4.7uF capacitor is massive. It takes up about 1/4 of the PCB by itself. I wouldn't mind lowering it to 2.2uF as this gives me more breathing space on the PCB to move things around
2) Tom's guide 'Complete Guide to Design and Build a Hi-Fi LM3886 Amplifier' says you want to balance the input currents against the feedback current.. I have 20K + 1K on the input, and 20K on the feedback, so while it works, I wouldn't mind making this exact on the next build.
To that end, I've debated the following changes.
1) Change the input resistor from 20K to 23.7K
2) Change the feedback resistor from 20K to 24.9K
3) Change the gain setting resistor Ri from 1.0K to 1.2K, keeping the gain really close to what it was before
4) Change the input current resistor Rb from 1.0K to 1.2K, making the current exactly balanced (23.7K + 1.2K = 24.9K)
I calculated that the change in the input configuration would drop the frequency response at 30Hz by about 0.03dB compared to before, and I can live with that. My question is regarding the feedback compensation network (the optional components). Since I've kept the gain pretty much the same, do I need to worry about adjusting/simulating those, or will they likely be ok as they are now (20K + 47p I believe)? Anything else I might need to worry about with this change?
You can try this circuit. For the input capacitor, we can use a 100V rate and the size will be smaller. For the feedback compensation, we can just use a 5pF capacitor with the feedback resistor is ok.
100pf is too low as RF blocker capacitor and it should be placed after the 1Kohm resistor. 470pf is a good value to start. As input capacitor 4.7uf is unnecessary, 2.2uf is good enough.
Also the input cap has too high voltage rating unnecessarily large. If as usual, a volume pot will be placed, the DC potential across it is negigible, and 16V cap sufices.
The error due to this mismatch is 5 to 50 times smaller than the offset voltage. You could even leave-out the 100uFd cap and current-induced error would still be less than voltage error.
You'll want a minimum of 100 nF || 10 uF || 470 uF for the decoupling. You're missing the 10 uF in the schematic. Without it you'll get oscillations at higher output currents.
You don't need AC coupling on the input, but that's a nice thing to do given that the amp does not have a DC servo. You could use a Nichicon UES-series bipolar electrolytic on the input. 10 uF, 16 V would be tiny compared to a film cap.
Your changes are pretty small so I'd think your circuit would work just fine.
Tom
You don't need AC coupling on the input, but that's a nice thing to do given that the amp does not have a DC servo. You could use a Nichicon UES-series bipolar electrolytic on the input. 10 uF, 16 V would be tiny compared to a film cap.
Your changes are pretty small so I'd think your circuit would work just fine.
Tom