[shake]

I looked at the datasheet for the linear regulators and made a few tweaks to the design.
 

 
The links work for me.  I noticed that the digikey links you post end in .ca.  Maybe that has something to do with it?
 
Fuse - F2418-ND
Input Caps - 495-4209-ND
Decoupling capacitors - 495-1119-ND
Power capacitors - P12409-ND & P12373-ND
Bass boost capacitors - 495-4476-ND
 
Here's a rough sketch of my projected layout:

The dimensions of the breadboard I'm using is 3"x4".
 
EDIT: Fixed schematic.  Potentiometer and capacitor in the bassboost circuit shouldn't be jumpered.
EDIT 2: Changed regulators from +/- 12 to +/-15V.

 

[Avro_Arrow]

Looks good so far.
 
Two of the caps you picked, this one "Input Caps - 495-4209-ND" and
this one "Bass boost capacitors - 495-4476-ND" are "X" caps used
for AC line filtering and have very high voltage ratings.
You may get better performance by choosing caps with a lower
voltage value. I usually stay down around 50 to 63 volts (like this one "Decoupling capacitors - 495-1119-ND").
 
The higher voltage cap just takes up more room to no advantage.
 
You will still need, or maybe have already, a fuse holder for that fuse.

 

[shake]

I know that the voltage rating for the input caps are high.  You're saying that it will still have the same or worse performance as the smaller one?  I can't find any other input caps on digikey right now that are low voltage polypropylene.
I already have a fuse holder
Are there any other suggestions you can make to improve noise reduction/circuit performance?

 

[vixr]

WOW! hats off to Avro_Arrow for the awesome level of support here. This is what its all about for me...simply awesome.

 

[shake]

Yeah, it's like having a private tutor.  I'm very grateful for him and the other DIY pioneers.

 

[Avro_Arrow]

Stop...your gunna make me blush...
 
Try this for an input cap:
http://www.digikey.ca/product-detail/en/ECW-F2105JB/PF2105-ND/248834

 

[wakibaki]

Quote:

Yeah I don't see what R5 and R6 do or where tangent got the values from.

 

I don't think that R6 and R5 are necessary?

 
They limit the bandwidth of the first opamp stage. If the BW of this stage is too high (relative to the driver stage), the amp may oscillate. The stage gain is 470/3.32=141. The gain-bandwidth product of the OPA627 is 16,000,000. Therefore the stage bandwidth (and therefore the BW of the amplifier as a whole) is ~113kHz, a not atypical value. An alternative approach is to use a 'compensating' capacitor to roll off the stage gain @ HF.
 
Bode plots showing the dominant poles of the opamps are useful when determining suitable rolloffs, but a less demanding practise is simply to go for a typically acceptable audio amplifier 20kHz < BW < 200kHz, or to adjust the values depending on whether overshoot is obvious when the amplifier is driven with a (1k?) square wave.
 
w

 

[shake]

So R5 and R6 set a high gain for the opamp to increase stability by limiting its bandwidth?
 
I think I have all the parts ready to go.  I just want to make sure enough noise is being filtered out in the power supply, and the fuse is a good value to prevent explosions or burning out early.

Parts list:

 
The prices of knobs are ridiculous btw.
 
Oh and another question: Where does the signal ground go, just to ground as well?

 

[wakibaki]

So R5 and R6 set a high gain for the opamp to increase stability by limiting its bandwidth?

 
Yes. If you're running it as a unity-gain buffer you get the maximum bandwidth. 'Gain-bandwidth product' and 'unity-gain bandwidth' are synonymous when considering small signal performance. The less feedback you apply, the greater the gain and the narrower the bandwidth.
 
 

Oh and another question: Where does the signal ground go, just to ground as well?

 
I see your first post mentions a 'star' ground. This mean that all the grounded components have an individual connection to one point, that includes the input jack. It's not an absolute requirement that there are separate individual connections, but figuring out where exceptions can be made without impacting performance requires theoretical knowledge and experience. Large fluctuating return currents such as those associated with the output power sections must be kept away from more sensitive sections handling smaller signals, such as the input.
 
w

 

[shake]

My question was because since I'm using a dual power supply now, I won't have a dedicated ground channel.  So would I just stick the signal ground anywhere in the circuit ground or is there a preferred place to stick it?
My current schematic is posted here

 

[Avro_Arrow]

In order to help with answering this question, it help to know what an amplifier is...
An amplifier is just a device that conducts electricity from one side of the power
supply to the other. Power flows from one side of the power supply, through the
output device, through your headphones and back to the other side of the power
supply. All the other circuitry in the amp is just a way to control the output device.
In your circuit, the output device is the BUF634. Power flows from the positive
and negative rails, through the BUF634, through your headphones and back
to the power supply ground. I know I am sounding a bit repetitive, but it is
important to understand how power flows through the amp.
Even the shortest, thickest wires/traces still have some resistance.
When current flows through them, they drop a little voltage. If this
drop in voltage is picked up by the part of the circuitry that controls
the output device, then it becomes part of the signal. This is the concept
behind a star ground and star power. By returning the wires/traces that carry
the most current separately from the wires that control the output device, we
can minimize the interference.
In your case, I would use a star ground consisting of the Headphone Ground, Amplifier Ground
and Signal Ground all returning separately to the point on the power supply where the
regulator grounds are.
 
As a footnote, the same star ground topology applies to the power supply as well...

 

[wakibaki]

A generally accepted scheme for conventional audio grounding uses the point where the main PSU reservoir caps (positive and negative rail) connect together as the star point. The transformer centre tap connects there, as do the input ground (if a screened cable carries the signal to the amplifier proper, connect the amplifier end to the star), output ground, local decoupling and feedback network grounds and any other powered component grounds such as relays. The chassis in this case would connect to the screened cable at the input jack. The mains ground then connects to the chassis.
 
w
 
Sorry A_A, cross post.

 

[shake]

So I wire those ground like this?

 
And I should wire up the power in a similar fashion where each IC get's it's own line to the regulated output?

 

[Avro_Arrow]

Yes, that is the right idea. As wakibaki also suggested, if you are using a shielded cable for your input,
it can ground at the amplifier instead of at the power supply.
 
As you are building your amp on a breadboard and only have two devices that need power, yes, you can
easily use a star power as well.
 
In some high end designs, there is a separate power supply for different parts of the circuit.
D/A converter chips often have separate input for digital power, left analog power and right analog power.
Some amplifies have a separate power supply just for the output devices.

 

[jcx]

a single "star gnd" is usually "too simple" to be "the best" - I use a hierarchy - gnds may be dirty/noisy or clean - to varying degrees
 
the power supply xfmr CT/rectifier/reservoir caps have the largest current spikes, they should be a separate "star", then a branch from there goes to the "clean gnd" "star" point, the Vreg reference terminals can be "between" the two, "Kelvin Sensing" actually runs these to the clean gnd
 
 
 
my output connector star drawing, shown in the 1st post can be about as good as it gets if you only have one channel - or go "dual mono" = two completely separate amps, power supplies, separate R,L connector headphone gnd pins, "balanced" headphone connector
 
 
if you have only a single dual polarity supply, and/or use the shared common R,L gnd TRS connector you have to compromise from that ideal
 
I think a good compromise is to use the TRS common gnd as the "clean gnd star point"
 
when everything can't be at the same point you can sometimes get near the same effect by making the resistance between them low with gnd planes on a PCB or heavy Cu buss - Cu Roof Flashing is the most easily available Cu sheet stock