LM3886 loud pop when turned off

[nige838]

Originally Posted by Th3 uN1Qu3

Then your earlier diagram was misleading. There is no problem in that case, so let's get on and tackle the pop. Insert a 22uF 50v cap in the feedback loop, lift the end of the 680 ohms resistor that goes to ground and insert the cap between it and ground.

Thanks, I'm going to try it tonight. Would it hurt to raise the value above 22uf?

Thanks

[Th3 uN1Qu3]

22u with 680 ohms gives a cutoff frequency of 10Hz, there's little reason to lower this further.

[Michael Bean]

OK, further recommendations. First, Andrew's suggestion to use RF bypassing at the input is a good one, a 220pF NPO/COG ceramic disc across the + and (-) chip inputs will do the trick. The 1k resistor at the + input of the chip will help minimize popping by limiting current into that pin when turning power on and off. And while Th3 uN1qU3 is technically correct about corner frequency (Fc) and using a 22uF cap for DC blocking the feedback network, there are very good reasons for using a much larger value there. A cap that large in value is almost certainly going to be a polarized electrolytic for physical size considerations, and that being the case, using a much larger value will limit the voltage drop across the cap because of the very low corner frequency (the cap behaves like a dead short for frequencies above Fc so "sees" minimal AC voltage across itself), thereby keeping distortion low and extending the life of the cap. Also it's important to make sure that Fc for the feedback network is at least two to three times lower than Fc at the input because you don't want to feed frequencies into the amp that the feedback network can't control. The 100 nF bypass caps on the power supply pins need to have low inductance to be effective in preventing oscillation, use NPO/COG 100 volt caps soldered directly to the chip pins, the under side of the PC board is a good place to put them. The 4.7 ohm resistor and 100nF cap on the output will help prevent high frequency osciillation, and the 10 ohm resistor and .7 uH coil will counter possible issues with excesive capacitance on the output. The resistor/coil can be made by winding about 10 turns of 18 gauge wire around a 1 watt, non-magnetic resistor (carbon composition works well if you can find them). Finally, as always, good circuit layout practice is important, keep high current paths away from low current input signals as much as possible, use separate conductors to common ground as shown on the schematic, and commom current send/return conductors like power supply leads, input leads, output leads, and so on, should be twisted together to keep stray circuit inductances low.

Mike

[stratus46]

Originally Posted by AndrewT

55Vdc is just perfect for a 3875.
It can be single polarity supply 0Vdc to +55Vdc, or 0Vdc to -55Vdc, or dual polarity supply, +27.5Vdc, 0Vdc -27.5Vdc, or dual polarity with the centre voltage not at 0Vdc, eg. +35Vdc, 7.5Vdc, 20Vdc. It can even have the centre voltage not at the arithmetic mean between the two supply polarities.

How does an asymmetrical power supply help an audio amp? I'm not saying it can't be done but why would you? I've used asymmetrical supplies in control systems where the control range wasn't hampered by the supplies but I'm at a loss to see how that would improve an audio amp.

It strikes me the same as trying to draw rectangular coordinates on polar coordinate system. It can be done but what a pain.

G²

[AndrewT]

Originally Posted by Th3 uN1Qu3

22u with 680 ohms gives a cutoff frequency of 10Hz, there's little reason to lower this further.

there are a few reasons to go lower than this.

[AndrewT]

Originally Posted by stratus46

How does an asymmetrical power supply help an audio amp?

I am showing what can be done.
I made no suggestion that it was better or less good to adopt any of my examples, simply that they exist.
Your control example shows where it is an advantage (having different output Vpk in the +ve & -ve polarities) A similar argument can be applied to a single ended amplifier, where equal value dual polarity supplies result in maximum Vpk in either direction being very different.

[Th3 uN1Qu3]

Originally Posted by AndrewT

there are a few reasons to go lower than this.

First he needs to get it to work THEN he can think of more exotic stuff. You have to remember that a big coupling cap will behave almost the same as no cap: there will still be thumps at power on and power off.

[Michael Bean]

Th3 uN1qU3, a large value cap in the feedback network isn't "more exotic stuff", it's good design practice, reasons why were listed in my post #23. A 10 Hz feedback Fc would mean using a Fc of at least 20 to 30 Hz for the input filter which will affect the audible low end response of the amp, along with the other issues listed. And you're correct that it probably won't do much to fix the thumping issue, but why not get the design and implementation as good as possible from the beginning rather than having to go back and re-do stuff later?

Mike

[Th3 uN1Qu3]

Audible range... he never mentioned what he's using this amp on. Contrary to what people like to believe, a higher -3dB frequency makes small speakers sound better. No point in trying to push deeper lows than the speakers can manage - at best you're going to end up with intermodulation distortion, at worst you'll mechanically damage the speakers.

[madtecchy]

Hi .
Are you still using the preamp section of the realistic amplifier ??
Ian