I took a brief look at the posted schematic. One thing I noticed is that the signal going into
- input of the LM3886 is AC coupled. The device requires some sort of DC connection to support input bias current for both the + input and - input and as the schematic is now, the + input bias comes from the pot and there is no path for - input bias.
- input of the LM3886 is AC coupled. The device requires some sort of DC connection to support input bias current for both the + input and - input and as the schematic is now, the + input bias comes from the pot and there is no path for - input bias.
Looking at the data sheet for the lme49710na, it looks like the max input voltage is 0.7V and the max output is 3V RMS. The implementation of the feedback loop on my board will give a gain of 6. There is a series resistor of 532 ohms after the a 56.2k ohm resistor running to ground. This doesn't make a voltage divider, but when I added the additional 50k in front of the input, and therefore in front of the 56.2k resistor I created a voltage divider that dropped my voltage to around 1V. It seems that if I drop in around 150k ohms then I will reduce the input voltage much closer to the .5V it should be for an output at or below the max 3V the lme49710na is capable of. Hopefully my analysis is correct and I can avoid all the extra distortion that the stepped potentiometer seems to be adding. When I get the 50k resistors in I'll have to try it with 1, 2, and 3 of them to see how it affects the sound.
You asked "what do you mean by no dc inputs?". Post #22 points out that the schematic does not show any DC path on the - inputs of the LM3886. Both + and - inputs require some sort of bias current. The example app circuit shows 20K to ground on the + input and a 20K feedback resistor to the - input. Not saying that it must be done that way. Just that there needs to be means to provide input current. The datasheet shows the inputs to be going to the base of NPN transistors. They will require some small current.
The implementation works absolutely perfect with no pot when the lme49710 is set up as a follower, but the 2V from the minidsp is not enough to push the lm3886's in this configuration. And, if I run the lme49710 with the feedback loop then the minidsp is too much voltage going into it and the noise gets amplified. I was hoping to add a pot so I could lower the voltage at the input and use the lme49710 to drive the lm3886. When I had wired the pot up incorrectly without a ground connection i was blown away with the sound quality. This was with the pot level set to 50k resistance across the input wires. There was a little bit of noise, a "shhhh" sound, coming through the tweeters though. When I reconnected the pot the correct way I got rid of the noise at levels below about half way, but a big increase in measured distortion below 100Hz. I feel like I also lost a lot of dynamic range as well, although that doesn't show up in my measurements.
Hmm... I'm not really sure how this DC Servo circuit even works. I've tried to find more information about it but closest I've come is that I think the Modulus 86 uses one too. It looks like it's primarily for the feedback loop of the lm3886. Thanks for clarifying, btw. I definitely still consider myself a noob at this.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.