Headphone Power Buffer

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Member
Joined 2011
Paid Member
Hey Guys (& Gals),

I'm designing a cool little thing kinda like a desktop "amp" that is just a current buffer for headphones.
I suppose it could be run from batteries if you really wanted it to be.....
Basically what I am doing is massively paralleling NJM4556's to provide some external grunt to wimpy "headphone amps" that you find all over the place.
What I am designing here will accept basically any gain stage to be attached to the front end and allowing plenty of power to come out the tail end to drive just about any headphone you might want to connect.

I'm looking for a bit of input on my design since I'm learning Eagle at the same time.
I know my layout may be a bit wonky but it makes sense to me.
I am really wondering if there are any glaring errors that I have not noticed that would be a detriment to the final product?

Thank You,
Rich
 

Attachments

  • Headphone Buffer.png
    Headphone Buffer.png
    26.3 KB · Views: 1,033
Member
Joined 2011
Paid Member
Power Supply Board Complete

I was able to rearrange the power supply board to make it single sided.
I figured I would take a shot at making my own PCB.
I used the toner transfer method and it worked out GREAT, on my first try even.
I have some PCB drill bits on the way so I can drill the small holes I need.

I don't know if I will be able to do the double sided buffer circuit board homestyle or not.....that might take more skill and luck than I have.
 

Attachments

  • PSU PCB.jpg
    PSU PCB.jpg
    670 KB · Views: 791
I would suggest to use a master/slave configuration instead of "dumb" paralleling: it will greatly improve performances at no cost:

I experienced three configurations of four paralleled buffering sections of NE5532 :

An externally hosted image should be here but it was not working when we last tested it.


I found no differences of distortion between them.

However, testing with an unrealistic pure capacitive load of 220 nF,
buffer #1 needed a 47 Ohm resistor output to maintain stability (no overshoot at all), buffer #2 needed a 10 Ohm resistor, and buffer #3 no resistor at all.

Response of buffer #3 to a 10 kHz square ware, 100 kHz band limited :

An externally hosted image should be here but it was not working when we last tested it.


For my 32 Ohm Senneheiser P100 heaphone, 1 Vrms gives a much higher level
than that I need.

Distorstion at 1.2 Vrms :
H2 : -115 dB no load, -110 dB 32 Ohm load
H3 : -110 dB no load, -108 dB 32 Ohm load.

The circuit can be easily built (and modified) on a prototyping board :

An externally hosted image should be here but it was not working when we last tested it.
 
The #1 has the advantage of FB extending to the output, meaning ~=0 impedance, better control of non-linear loads, etc, but as with most FB amplifiers, the additional pole created by a cap direct on the output generates instabilities.
The other configurations benefit from an isolation resistor outside the FB loop, and can thus tolerate any amount of reactive loading.

There are other, less obvious benefit for the #1: for example, the slaves can be pushed to their limits, without fear of distortion if the master is kept in its most linear current range. This can be achieved by unbalancing the summing resistors, making that of the master 1.5 or 2 times larger than the others.
This will effectively reduce distortion.

I see no advantage in config #2, except perhaps smaller circulation currents in case of differing offsets between the opamps.

If the amp has to drive a capacitive load, the best solution is probably to use an inductive Zobel, just like any regular power amp
 
There are other, less obvious benefit for the #1: for example, the slaves can be pushed to their limits, without fear of distortion if the master is kept in its most linear current range. This can be achieved by unbalancing the summing resistors, making that of the master 1.5 or 2 times larger than the others.This will effectively reduce distortion.

I saw config #1 a long time ago in "Circuits Ideas" published in Wireless World. It was called "Poor man's dumper" which undoubtedly makes think to the Quad 405 amp.

It would be an interesting exercise to try to implement an error correction to this kind of buffers. However the distortion of these circuits is already extremely low. A voltage follower NE5532 section can deliver 16 mA(rms) with less than 0.001% harmonic distortion at 10 kHz .

In his Elektor power amp, Self has used an array of paralleled voltage followers made of NE5532s, each one with a 1 Ohm in series at its output before the common point power output. There is no global feedback including the voltage gain stages, but an output inductance, as usual in power amps, has been retained. This is probably the audio amp which contains the greatest number of transistors in the world.
 
Member
Joined 2011
Paid Member
The reason I have chosen to do the buffer in this way is that my O2 headphone amp has just a single 4556 per channel paralleled to itself.
I figured that doubling the current output that an O2 can deliver would be capable of driving any headphone you desire as long as the gain stage in front of it has enough volts to swing it.
This circuit topology works beautifully in my O2 and now I am just looking to beef it up and drive a similar circuit from different gain stages to experiment.

Thank You for all of the suggestions.
 
Member
Joined 2011
Paid Member
Here is my current layout for the dual paralleled NJM4556 headphone buffer with it's power supply.
I was able to have a friend adjust the layout for use of a SIP8 package of the same opamp to gain an additional 200mW dissipation per channel.
AND Mouser has like 5000 of them in stock.
I may do a group buy for the PCB's if the prototype ends up working as planned.
 

Attachments

  • Headphone Buffer SIP8.jpg
    Headphone Buffer SIP8.jpg
    544.2 KB · Views: 378
TPA6120A2DP is cheaper, slightly higher V

and its $5.44 in singles at Digikey

for DSL specs the TPA6120 is the same chip as the THS6012 but its popularity as the TPA6120 audio speced headphone/audio line driver lets us get the cheaper price

I've used the soic 20-wide powerPad package upside down with a cpu cooler clamped to its power pad to run at over 5W per chip


I don't understand the price objection - you would need a dozen 5532 to come close to the output current which is going to take up space, assy time for a objectively worse result given the ancient semi process used


these DSL pretty much need pcb, nice projects spend more on PS, case, pot than all semiconductors together
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.