Beginner question - speaker cone movement during attenuation

lhaymehr

Member
2009-10-05 4:43 pm
Hello,

I'm a let's say a developing intermediate at electronics and understand a few advanced concepts :). I recently built a simple two channel amp based on LM3886 with a p2p construction. The amp works fine and I'm satisified with the sound. Very low noise, great performance and no artifacts, except one..

When I turn the volume control the speaker cones move in and/or out quite a bit depending on if I'm turning it up or down, and in under a second they move back to initial position. There isn't any audible effect on the sound but this cannot be right.

I'm suspecting it's a DC decoupling issue somewhere. Can someone please explain why and where does this phenomenon usually occurs.

P.S. I tried picking the best section to post this, failed and decided to post it in the chipamp forum since this is a LM3886 build. Feel free to move the thread if I missed.

Thanks
 

lhaymehr

Member
2009-10-05 4:43 pm
Here's the schematic. The aux was solved with a shunt later on instead of a simple divider and the output lowpass is not implemented. I believe everything else is accurate. There's another identical channel path coming into R14 as well and not shown in circuit.

0h02Ffy.png
 

infinia

Member
2005-05-15 9:51 am
SoCal
install a DC blocking cap in series with Rb. Problem is DC offset from U1, when atten pot U2 is adjusted to minimum, the inputs become unbalanced, ideally both inputs seen by U1 will be 22K. also check DC at input when source/s is connected almost all sources are zero DC except certain 'lab' sources. DC block ~ 4.7uF to 1uF film cap, low voltage = small fitment size
 
Last edited:

lhaymehr

Member
2009-10-05 4:43 pm
You need an input coupling cap. Let the 22K Raf bias the input stage.

At least you have a feedback roll off cap. A lot of people think you don't need it. I bet you'd have some real woofer cone flapping if you had eliminated that capacitor.

Yeah and I experimented with that rolloff cap. Wanted to really have the lowest possible HP point while still controlling oscillations. I just love how the cone reacts to a 1Hz test signal :D. No amp I had ever did that! :D

Put a cap between U8 and U2. This will also extend the life of the volume control. Something like 1u at 50V or more.

You're right, didn't cross my mind at all the buffer would put a heavy load on the pot. Was eager to get it running.

install a DC blocking cap in series with Rb. Problem is DC offset from U1, when atten pot U2 is adjusted to minimum, the inputs become unbalanced, ideally both inputs seen by U1 will be 22K. also check DC at input when source/s is connected almost all sources are zero DC except certain 'lab' sources. DC block ~ 4.7uF to 1uF film cap, low voltage = small fitment size

Alright. That's probably gonna eliminate the issue. Now that I think of it the phenomenon started after I added the buffer. In general, I just wanted to get away with as few caps as possible.

Thanks everyone!
 
Yeah and I experimented with that rolloff cap. Wanted to really have the lowest possible HP point while still controlling oscillations.

This is a poor design practice that is pushed by woo peddlers. The fact is that a properly implemented coupling capacitor will not affect the audible frequency response and will make your amplifier safer and more practical. Shoot for a -3 dB cutoff of 2-5 Hz and you'll never know it's there.
 
Here's the schematic. The aux was solved with a shunt later on instead of a simple divider and the output lowpass is not implemented. I believe everything else is accurate. There's another identical channel path coming into R14 as well and not shown in circuit.

[IMGDEAD]https://i.imgur.com/0h02Ffy.png[/IMGDEAD]
You have mixed AC & DC coupling for your amplifier.

Look at the source resistances presented to the -IN & +IN pins of the 3886.
 
Your SubOut (low bass) does not work.
Your existing circuit mixes Left and Right and creates unwanted crosstalk.
The Q of the passive filter is extremely low. Much lower than adopted for any crossover that I have ever seen.
http://sim.okawa-denshi.jp/en/CRCRtool.php

Use an inverting opamp as your channel summing stage.
Then apply active filtering to create the roll-off you require.
If you adopt an MFB filter stage it can be used as a single opamp that sums the two channels and achieves a 2pole low pass filter. Yes, one opamp channel that sums AND filters !
http://sim.okawa-denshi.jp/en/OPtazyuLowkeisan.htm
Duplicate R1 to create a two input summing circuit.
 
Last edited:

lhaymehr

Member
2009-10-05 4:43 pm
Your SubOut (low bass) does not work.
Your existing circuit mixes Left and Right and creates unwanted crosstalk.
The Q of the passive filter is extremely low. Much lower than adopted for any crossover that I have ever seen.
(Sample) 2nd order CR Low-pass Filter Design Tool - Result -

Use an inverting opamp as your channel summing stage.
Then apply active filtering to create the roll-off you require.
If you adopt an MFB filter stage it can be used as a single opamp that sums the two channels and achieves a 2pole low pass filter. Yes, one opamp channel that sums AND filters !
Multiple Feedback Low-pass Filter Design Tool
Duplicate R1 to create a two input summing circuit.

I know, it's a quick and dirty solution like the rest of the amp. I did look into it and with time and money I'll put in a true effort in making a new one; PCBs, chassis, < 1% tolerance components and all. For now, the amp is a simple p2p construct and going solid ;).

UWdrbtJ.png