Amp troubleshooting

[dfhaii]

I have built the following amp:

I have quite a bit of offset though;

Out left to 0v: 3.5mV
Out right to 0v: 15mV
Out ground to 0v: 150mV
Out left to out ground: ~150mV
Out right to out ground: ~150mV

The 0v is just the centre of a resistor divider, and its running on +/- 6v.
Each opamp and buffer has a 0.1uF ceramic cap between its v+/v- and 0v.

The topside of the board can be seen here http://www.dfhaii.com/images/amp2.jpg

Any suggestions on how to cure this would be most appreciated.
Thanks,
Col

 

[dfhaii]

Sorted, I have put a 47pF cap in the local feedback loop of the ground channel which has bought the dc of that down to 0. Out left to out ground is now 4.0mV and out right to out ground is 20mV. I'll try and get right down but it should be ok with those values?

I have a pair of 637BP's to put in left and right and a 627AP for the ground channel, will it be ok to put these in now or should i try and get the dc down first?

Col

 

[AsuAmo76]

You can see specs here: http://focus.ti.com/lit/ds/symlink/buf634.pdf

1) Buffer input impedance is 100 Om - I would rather short 1k resistors between opamps and buf inputs (see the link above).
2) Why do you need 2 loops on the out-left channel feedback? I wold remove 470k and short 4.7k. If you need gain of 1000, change the resistors to 470 and 470k for example.

 

[dfhaii]

I'll try shorting the 1k resistors between opamps and buffers.

The 2 loops is just a topology, with those values the amp has a gain of just under 10.

I need to get some phono plugs and get it cased up now.
I have got a rail splitter consisting of 2 resistors, an opamp and buffer which works ok, and have 2 * 4700 uF caps between positive and negative. I dont have any capacitance between the +/- rails and 0v after the splitter, would I be wise to put some in there, and if so how much?

Has anyone tried using a capacitance multiplier with a headphone amp, if so how did it work?

Pic of it in its current status

Sorry for all the questions and thanks!

Col

 

[AsuAmo76]

I don't know the theory, but in my design I would put 100 - 330 uF for sure

 

[Garbz]

I think you've gotten a bit to overzealious with your grounding scheme. The use of a ground channel amplifier keeps all the high currents from the actual cicruit ground. But then you also have a buffer on the output of the powersupply to give it some current. This is definatly killing a fly with a tank.

To simplify the design (and arguably simple = good) try and bypass the powersupply buffer. Even the pimeta makes do with the 30mA the TLE can provide in total.

The loop is a Jung Multiloop I believe.

It's also a matter of taste to put caps after the powersupply. We are talking large ones here, not small 0.1uf bypass caps. They do have an audible signature, but ultimatly all required current should be able to get sucked through the opamp and caps aren't needed, however it's usually a good idea to put 0.1uf caps from each power rail to ground right next to each chip.

Finally i'd put the opamps in now. Otherwise chances are that you fix the offset, change opamp and new variables will make new offset. Unless three's a fudemental flaw in the design the DC offset should be the last thing to fix. Also if you have high impednace headphones like sennheisers then the 20mV is hardly an issue.

 

[dfhaii]

The ground channel was going to be used for returns from the headphones ppa style with the ground in being the ground coming from my cdp, but i'll give it a bash without that channel and see which sounds better.

I have the 0.1uF decoupling caps next to the pins on each ic already but will add a 470uF cerafine between each rail and ground. I also plan to replace the 2 4700uF jamicons on the power supply board with 2 3300uF elna rjh's when they arrive.

This will be used with either Grado sr80's or Sennheiser hd477's, both of which are 32 ohm so i would like to get the dc offset as small as possible.

Thanks again
Col

 

[Garbz]

Sorry wasn't paying attention to the picture. I see the caps now

You missunderstood what I meant. The inclusion of the ground channel is good. By including the ground channel you keep the high currents away from the powersupply rails. When you have this it reduces the current demand on the powersupply.

What i'm saying is when you use the ground channel then you don't need a BUF634 in the powersupply, since all it does is provide extra current which isn't actually needed. The BUF634 in the ground channel provides the high current return.

 

[dfhaii]

Ah ok, ill just put a jumper across the buffers socket in the powersupply then

Col