Here's the first:-
 

 
It's no more than the headphone amp reference design from the LME49600 datasheet with the addition of dual regulated power supplies. There is a slight departure in that the servos don't use amps from a second package, but rather each channel comprises one LME49600 + one LME49720. This simplifies and compacts the layout somewhat:-
 

 
 
This is a step on the way to a more comprehensive integrated desktop amp (including volume control, PSU and offset protection on a single PCB), but I think it makes an interesting stand-alone module. It'll probably fit in 50 * 50 mm.
 
Here's ground:-
 

 
And +/- rails shown for L/R channels for clarity:-
 

 
I'd obviously prefer to route the power rails superimposed, but the tracks are hardly extensive and no PCB is without compromises.
 
Unlike a lot of stuff I do, it's not really suited to homebrew PCBs due to the PTHs and via-in-pad construction employed, despite the fact that all but one of the second layer links are straight. I still have to get the LME49600s in my hand to verify the footprint.
 
I'm toying with the idea of doing one with only LM317s aboard but since I'm currently aiming at using a single AC wallwart I don't have the luxury of a dual-secondary-type supply.
 
w 
 

Personally, I wouldn't run the LME49600 (or BUF634) in wide band mode in a headphone amp.
You are more interested in the slew rate and that does not change with the bandwidth.

I took this from the datasheet A_A, I'm firmly of the opinion that the manufacturer's recommendations are there to be followed, particularly in the light of no previous experience of the part, which is what I've got.
 
w

If you make a space for a PTH resistor, you can leave it open, jumper it or put whatever
resistance you like to try...just a thought...
 
Edit: It doesn't need to be a PTH resistor, an SMD will serve the same purpose.

If you want to do an all 317 power supply, consider 338 instead...it has lower output impedance.

If you do a small run of boards I might be interested in one depending on the final price and shipping.

Quote:

avro_arrow said:

If you make a space for a PTH resistor, you can leave it open, jumper it or put whatever
resistance you like to try...just a thought...
 
Edit: It doesn't need to be a PTH resistor, an SMD will serve the same purpose.

Click to expand...

Thanks for the input. That's easy to accommodate, R4 & R7:-
 

 
I might get some made. I can get 100x100mm boards at a very reasonable price. I have to cut them up myself, I haven't decided how to fill the other 3 25x25mm sections or whether to go with a single integrated PCB. I think I could get everything into the 100x80 that fits into an O2 enclosure.
 
Here's the protection circuit I'm planning to use. I haven't verified it in simulation yet, but I'm pretty sure it will work, it's simple scheme and straightforward to understand.
 

 
Presettable +ve and -ve voltage references are presented to an LM2902 configured as a comparator. They are compared with LP filtered voltages derived from the amplifier outputs. The comparison is configured so that the outputs are low in normal operation. Resistors and diodes constrain the voltages input to the bases of the 4 transistors to a maximum of -0.6V to avoid exceeding the permitted transistor base-emitter voltage. The transistors are normally cut-off. This results in the voltage at the base of the 5th. transistor rising at switch on, albeit slowly due to the capacitor there. Eventually the transistor turns on and the relay closes.
 
If the DC element of the LP filtered output voltage exceeds the reference in either direction, the comparator output changes polarity, turning on one of the 4 transistors, turning off the 5th and causing the relay to open. The LP filter values are not set in stone.
 
This is where I'm at with the layout:-
 

 
The existing amp has a gain of three. A 3rd. module might comprise a volume control and preceding additional gain stage, and the 4th. would be the PSU.
 
w

I drew up my own version to muck around with:
 

 
Here's the "I'm crazy" part.
I want to use either a BUF634 TO-220 or the LME49600 TO-263
Here's what I'm going to do...glue the TO-263 package to
a heat sink and mount it like the TO-220. With a little reforming
of the leads it just might fit.
 
The other "crazy" thing I was thinking about was making a non-symmetrical
op amp from an OPA132 and LME49990 on a Brown Dog adapter.
OPA132 for the servo and LME49990 for the signal.

 
Quote:

avro_arrow said:

I want to use either a BUF634 TO-220 or the LME49600 TO-263
Here's what I'm going to do...glue the TO-263 package to
a heat sink and mount it like the TO-220. With a little reforming
of the leads it just might fit.
 
The other "crazy" thing I was thinking about was making a non-symmetrical
op amp from an OPA132 and LME49990 on a Brown Dog adapter.
OPA132 for the servo and LME49990 for the signal.

Click to expand...

I think you can do that with the heat sink, and the asymmetrical opamp is an interesting idea, but I'd rather steer clear of the adapter, it's bound to introduce some strays. You can see I've removed the ground plane around the amplifier pins as per the layout advice for the TPA6120.
 
I kind of fancied the OPA2277 for the servo role, but in the end I went with the layout resulting from using the second amp from the LME49720 package. When all is said and done, and without figuring the offsets in the circuit given, the '49720 is still pretty hot in a DC precision application. 
 
w

I would solder the Brown Dog in just like a DIP to minimize strays.
 
Of course the "proper" thing to do would be to just use single SOICs
right in the layout...

Here's the general topology for the integrated amp:-
 

 
The component values are a bit small for comfortable reading, but you get to see the general idea.
 
A half-wave rectified supply provides +/-15V regulated. CRCRC filters mean that ripple to the regulators is already fairly well controlled. Ferrites should help in keeping any diode switching noise off the wallwart cable.
 
The amp itself is provided with switchable gain, a modest selection between 3x and 5x. A conventional volume pot of 10k is employed. I have an existing design using a TI PGA23xx digital volume control, but this provides a massive 35dB of gain which is unnecessary and would result in clearly audible hiss (undriven) at the top of the range in sensitive IEMs and a damaging output level with even a comparatively quiet source.
 
The protection circuit will certainly obviate any thump at switch-on and hopefully at switch-off too.
 
The layout so far:-
 

 
w

I know this digital pot would take up more space but it is easy
to implement with no uController.
 
Looking good so far...

Well, I've got room, I haven't even started on the other side of the board.
 
Here's the one with the regular pot:-
 

 
Gotta get some sleep...
 
w

This link shows a simple way to power the DS1802.
You can take from it what you will. You will probably
have a much better idea than me for it.
DS1802 is also supposed to work with a rotary encoder though
I have not done it myself.

The board looks really good.
It's nice to see a professional layout.