Im going to be feeding my chipamp lm3886 with a buffered preamp. This means no pot. Right now Rin is 22k. Ive seen Rin anywhere from 47k to 10k. Do i even need to worry about Rin being 22k instead of 10k if im cap coupling?
And 2 last things. Do i need the electrolytic feedback cap if i dont have dc offset? And how low can i go in gain before i start worrying about stability? right now my stock gain is 33, but my dac at full tilt will do 4vpp, so that would make the output double my rail voltage
And 2 last things. Do i need the electrolytic feedback cap if i dont have dc offset? And how low can i go in gain before i start worrying about stability? right now my stock gain is 33, but my dac at full tilt will do 4vpp, so that would make the output double my rail voltage
What's the rest of the circuit?. The resistance from the input to ground should be equal to the value of the feedback cap from the output to the inverting input - this gives you (well should give you) no DC offset - it is based on the assumptions that the input currents of the + and - inputs is the same - but it also assumes no other path to ground for the feedback, which requires the cap in the feedback circuit. If you by pass the cap, you will increase the DC offset. (this is why servo circuits are popular - no cap in the feedback loop). AFAIK, you can't really go below 10 for the gain (20db) before causing problems with stability - unless you want to add some capacitance between the input terminals (lag compensation which IMO will cause the quality to decrease. Someone may have other solutions, but usually the degrade the quality of the circuit. All compromises with gain is an integral part of the designing process. Good luck with it.
You brought up new info and lost me.
right now my ri is 680 and my rf is 22k. my Ci is 47uf and my Rin is 22k. I think you just told me to unbypass Ci and convert Rin to 10k.
Im going to leave gain alone because if i use a 2volt source i might not be able to do full output.
http://sound.westhost.com/project72.htm
hmm Rod uses the same values as what i got for the input/feedback.
right now my ri is 680 and my rf is 22k. my Ci is 47uf and my Rin is 22k. I think you just told me to unbypass Ci and convert Rin to 10k.
Im going to leave gain alone because if i use a 2volt source i might not be able to do full output.
http://sound.westhost.com/project72.htm
hmm Rod uses the same values as what i got for the input/feedback.
Last edited:
Oops - I said feedback cap when I meant feedback resistor. Rin = 22k - you want Rf to be the same - 22k. Un bypassing Ci should result in lower offset. The only 10 I mentioned was the minimum gain - which is determined by the ration of Ri to Ri+Rf.
Hope this clarifies things.
BTW ESP is one of the best information sites for DIY on the net.
Last edited:
Hi,
Fig1 of the ESP link shows R5 = R2 = 22k.
But R3 = 2k2 ruins this match.
R3+R2 must equal R5. If this is maintained, then Rin can be any sensible value from 2k0 to 1M0.
C3 = 47uF is too small. It will develop some AC voltage across it when very low frequencies are fed to the amplifier. This creates avoidable distortion.
C3 >= C1 * [R1+R2] * sqrt(2) / R4 gives 2u2 * [2k2+22k] * 1.414 / 1k0 >=75uF
Use 100uF or better 150uF.
If you use 150uF or 220uF you can increase C1 slightly to 3u9 (<=4u3F)
The gain is [22+1] / 22 = 23 times = +27.2dB, comfortably above the minimum of +20dB
Fig1 of the ESP link shows R5 = R2 = 22k.
But R3 = 2k2 ruins this match.
R3+R2 must equal R5. If this is maintained, then Rin can be any sensible value from 2k0 to 1M0.
C3 = 47uF is too small. It will develop some AC voltage across it when very low frequencies are fed to the amplifier. This creates avoidable distortion.
C3 >= C1 * [R1+R2] * sqrt(2) / R4 gives 2u2 * [2k2+22k] * 1.414 / 1k0 >=75uF
Use 100uF or better 150uF.
If you use 150uF or 220uF you can increase C1 slightly to 3u9 (<=4u3F)
The gain is [22+1] / 22 = 23 times = +27.2dB, comfortably above the minimum of +20dB
Last edited:
Hey andrew while i got you on the horn. If my dac has a 30khz sonic filter inbetween the dac output and the balanced to unbalanced opamp converter, is cap coupled to a 10k pot, uses a 1.5-2meter interconnect, do i still need to install the extra components on the chip amp? the 10k pot and cable combined should make a 60-80khz sonic filter, and the pot is ac coupled to take care of the sub sonic filter and i hope dc offset.
Ok using the stock lm3886 schematic:
Arg how i hate that everyone has diff formulas.
stock is Rin 10k, Rf 20k, Ri 1k, and Ci 22uf.
chipamp uses Rin 25kpot in parallel with a 22k resistor, Rf22k, Ri 680, Ci 47uf.
Now i got 2 people saying Ri should equal Rin. Right now that is sort of the case; but i am using an external dc coupled 10k pot (the pot is cap coupled to the dac).
So am i cool as is, other then changing the Ci to 150uf? In my mind the 10k external pot is paralleled with the 22k Rin which would make Rin like 7k. Combine that with the fact that with the volume pot all the way down, Rin is basically a short to ground, my brain turns to mush.
Thinking alittle abstractly with the datasheet and Rin being a pot, Rin needs to be 0 - 10k. Or with you 2, Rin needs to be 0-22k.
The problem is the input currents. According to the datasheet, the LM3886 is .2 (typical) to 1 uA (Maximum). The inputs are NPN, so the current has to flow into the IC, so there is normally a small negative voltage at the + and - inputs. If you keep the resistances equal and keep the coupling cap in, the circuits "balance out" and the output offset is zero volts. If you crunch the numbers, you will get 2 to about 10 mV with a 10k pot on the input and 22K resistor for feedback, and use the coupling cap. If you by pass the cap, the output offsets go up by a factor of about 20, which I would consider unacceptable. Also, means there may be a small amount of noise in the level control as its moved. Another option is to use a fixed input resistance, and cap couple the volume control (22k and 1 uF film type cap works good - 4.7uF film type would be better YMMV) If you don't like to crunch the numbers - try assembling it and seeing how it sounds and measures. Good Luck.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.