The pix shown are from a pocket LM-6171-based 3-channel headamp from an old (2006??) project. (the photo with two amps shows another similar-sized project, but is not 3-channel).
Virtual ground headamps were quite popular about 12-13 years ago, esp. on head-fi diy forums.
Note that non-use of dual-voltage (e.g. two 9v batts) was not the main criteria.
Yes, NWAVguy did write that lengthy article. Which is interesting but quite verbose.
Virtual ground headamps were quite popular about 12-13 years ago, esp. on head-fi diy forums.
Note that non-use of dual-voltage (e.g. two 9v batts) was not the main criteria.
Yes, NWAVguy did write that lengthy article. Which is interesting but quite verbose.


yep,On the normal 2-ways amp,the channel separation problem get even worse when you use a low impendence headphone.That`s why the virtual ground design should be a must-own in a professional headphone AMP.
Nowadays people went to BTL(balanced drive) instant of 3-ways.More and more hi-end headphones have balanced cable in stock,not to mention about the 4-pole 2.5mm and 4.4mm jack.I haven`t use SE for a while, back to the days I don`t have a balanced system,I was wondering"one day I will try a good virtual ground amp"too.
Nowadays people went to BTL(balanced drive) instant of 3-ways.More and more hi-end headphones have balanced cable in stock,not to mention about the 4-pole 2.5mm and 4.4mm jack.I haven`t use SE for a while, back to the days I don`t have a balanced system,I was wondering"one day I will try a good virtual ground amp"too.
yes, instead of a passive rail splitter, the active ground channel amp is driven at a point VCC/2. there are many reasons not to do this. it places the impedance of the active ground channel in series with the L/R channels. thus it is higher impedance than a well laid out GND. balanced amp THD into 32Ω will only be higher if output impedance is not low enough, nor drive current ability high enough. it is by no means a rule of thumb, just an aspect that needs proper attention in the design stage. Noise is generally higher then SE, as the +/- amps are in series in a balanced amp
The maximum linear output current is say 80mA peak, at 95mA happens clipping.
The virtual gnd has to deliver the sum, so 40mApeak this means with 32 ohm phones you can only deliver 25.6mW rms
At higher impedances soon is reached the maximum voltage swing which is +/-2.5V with a 9V supply. 70 ohm will get 44.6mW and higher Impedances even less.
Also the dissipation of the virtual gnd IC is double.
Without the virtual gnd and caps at the output almost 100mW at 32 ohm can be drawn.
The virtual gnd has to deliver the sum, so 40mApeak this means with 32 ohm phones you can only deliver 25.6mW rms
At higher impedances soon is reached the maximum voltage swing which is +/-2.5V with a 9V supply. 70 ohm will get 44.6mW and higher Impedances even less.
Also the dissipation of the virtual gnd IC is double.
Without the virtual gnd and caps at the output almost 100mW at 32 ohm can be drawn.
Last edited: