[Pibben]

Hi all,

I'm designing a compact headphone amp. I'm planning on using the LMH6321 buffer in the virtual ground circuit. However, I'm confused as the LMH6321 itself has a GND connection. According to the datasheet: "this pin must be set to a voltage of one V_BE (∼0.7V) or greater, or more commonly, mid rail, by a stiff, low impedance source. This precludes applying a resistive voltage divider to the GND pin for this purpose." But the only reference to GND in the datasheet is for the Current Limit function and "if the V_CL pin is left open, the output short circuit current will default to about 700mA." So perhaps the GND can also be left open i this case? Or should it be connected to the input or output of this circuit?

 

[tangent]

I'm planning on using the LMH6321 buffer in the virtual ground circuit.

See the first page of the PIMETA v2 schematic and the documentation of its R11 resistor.

 

[Pibben]

I think I may have been unclear. I plan to use the LMH6321 as a rail splitter to create the virtual ground. In the PiMETA it looks like the LMH6321 is used as an output buffer and the virtual ground is created by other means.

 

[tangent]

In the PiMETA it looks like the LMH6321 is used as an output buffer and the virtual ground is created by other means.

That doesn't matter. You configure the buffer the same way.

Replace the TLE2426 with a resistor divider, and you get the same effect.

 

[Pibben]

Thanks! I'm confused because the datasheet explicitly says that a resistor divider cannot be used to create the mid-rail voltage for the GND connection.

 

[tangent]

Thanks! I'm confused because the datasheet explicitly says that a resistor divider cannot be used to create the mid-rail voltage for the GND connection.

They're telling you not to connect the resistor divider to either the GND or Vcl pins because they require a low impedance input.

If you replace the TLE2426 in the PIMETA with a resistor divider, you will need to make other modifications, such as to remove the input resistor on the ground channel's AD8610. It probably also prevents the IG/OG split that's a hallmark of that style of amp.

I suggest that you post your proposed schematic before ordering parts or PCBs.

 

[barsk]

Any particular reason the LMH6321 is used instead of a BUF634 in PIMETA?

I'm also building a headphone amp, and need a solid, simple and battery friendly solution for a virtual ground. Not aiming for audiophile quality since it will be embedded with a raspberry pi and hence a rather noisy environment, so the aim is cheap, solid, and "decent" quality. Any recommendations?
The LMH6321 seems to be EOL... And a BUF634 is quite expensive, but I'll go there if needed.

 

[tangent]

Any particular reason the LMH6321 is used instead of a BUF634 in PIMETA?

https://tangentsoft.com/audio/pimeta2/history.html

The LMH6321 seems to be EOL...

Not according to TI.

Maybe you're looking at the old non-RoHS version, which is obsolete, whereas the NOPB variant is still active?

 

[barsk]

https://tangentsoft.com/audio/pimeta2/history.html
Not according to TI.
Maybe you're looking at the old non-RoHS version, which is obsolete, whereas the NOPB variant is still active?

Sorry, I did just that! The LMH6321 is NOT eol...

And thanks for the link to the Pimeta history. Explains it all.
A suggestion though to update the virtual grounds page you have, is that in the later parts you refer to the Pimeta/PPA ground channel designs with a buffer in the schematic without mentioning that it is NOT a BUF634. Also in the upper parts where you do specify BUF634 you might mention that the LMH6321 might be a better option. This is what put me into that question mode...

Btw, you also link to Arn Roatcap's page with positive/negative voltage regulators in a kind of cross connection. That page is also on HeadFi's forum. How would you think that design compares to the others?