Headphone amp rail splitters

[hejsan]

I've been looking at a lot of headphone amp projects around the web, and the majority of them seem to use a TLE2426CLP "rail splitter" followed by a buffered virtual ground thingy.

Is this rail splitter necessary if I have a dual supply to begin with?

Why are the grounds always virtual and buffered?

Thanks,
hejsan

BTW: Has anyone built Samuel Groner's headphone amp?

 

[owel]

Opamps can be used with single ended power source. The usual way of doing it is deriving the "virtual ground" using 2 resistors of equal values... to split the voltage in half (ie. voltage divider network). This Vin/2 is then used as the reference for the virtual ground, which is then connected to the opamp's input pin.

Problem with 2 resistors is they're not 100% equal so your voltage divider network will be lopsided by a small amount.

The TLE chip will always output half of the input voltage, which you then use as your virtual ground.


> Is this rail splitter necessary if I have a dual supply to begin with?

No. In that case, you just connect the opamp input pin directly to the GND of your split rail PSU.

I think National or TI has a PDF on powering opamps from a single-ended supply. Check it out. Example schematics given for different configurations.

 

[hejsan]

Thanks for the explanation :thumb:
So this is designed as a simple way to get "dual" rails from batteries and wallwarts.

Does it really matter if the two polarities are not 100% equal?

Also, why do I so often see the ground being buffered in headphone amps?
Would I have a problem if I just used a normal dual supply and connected the ground bus straight to ground?

Thanks again,
hejsan

 

[PRR]

> Is this rail splitter necessary if I have a dual supply to begin with?

No.

> Why are the grounds always virtual...

Because they have a single-supply.

> and buffered?

Because two resistors makes big cross-talk, two resistors and big caps makes crosstalk small at high frequencies but rising at low frequency.

Also some popular plans boot-strap their DC errors by the ratio of Rgnd/Rload, which may be 2500/32, making 2mV chip offset into 160mV DC error at the headphone.

The buffer holds Rgnd down to ~~1R and eliminates these effects (however battery consumption rises....).

 

[ChrioN]

2 batteries makes a dual power supply

 

[hejsan]

So if I use a dual supply, you know, the kind we usually use for our projects, like baumans supply: http://www.thediypill.phx.com.br/forumfiles/psu312sch_v1.pdf

would I also have to add a buffer to the ground rail here, or do real dual rail supplies not suffer from crosstalk?

Is this something specific to headphone amps? Do they generate more crosstalk than your average circuit?

Thanks,
hejsan

 

[bcarso]

[quote author="hejsan"]So if I use a dual supply, you know, the kind we usually use for our projects, like baumans supply: http://www.thediypill.phx.com.br/forumfiles/psu312sch_v1.pdf

would I also have to add a buffer to the ground rail here, or do real dual rail supplies not suffer from crosstalk?

Is this something specific to headphone amps? Do they generate more crosstalk than your average circuit?

Thanks,
hejsan[/quote]

No buffer. It's not about crosstalk, it's about the impedance of a bypassed resistive divider that one calls "ground" when using a single-rail supply.

Headphone amps are nothing special.

 

[hejsan]

ok cool, that simplifies things for me.

Thanks a bunch! :guinness: drinks for all