Quote:

wakibaki said:

Thanks, KT88, but what took you so long? We could have been here about 60 posts back...
 

 

 
w

Click to expand...

This isn't exactly what I meant. I meant really build it like a class-AB buffer. You should add a resistor from D3 to the positive supply and another one from D4 to the negative supply. This will keep the diodes slightly conducting at all times. It can also be done with BJTs instead of diodes:

But in this case its not really important since what's driving this point is an op-amp so no need for the extra current gain of the first transistors.
 
You can now also get rid of D1-D2. The bases of the output transistors only have 2 (forward biased) diode drops between them, so the B-E voltages will be well within the allowed voltage range.
 
Here's the output for this exact circuit with 2.2K for the resistors to the left, and 0.1R resistors at the output. The op-amp used for this simulation is the good old opa134, it can only work with a +-18V supply, so I've used that value and not a +-20V like in the previous sim. Everything is wrapped in feedback of the opamp. Capacitors are just 1uF (due to the very low output impedance there very little need in the capacitance, even lower value could be used) and load amplitude is exactly the same as before:

The feedback drops the output resistance much lower than before.
You can also try that with the LT1097, but it doesn't perform as good in this case (yet still much better than the previous circuit did under the same conditions).
In this case since the feedback to the op-amp is coming from the output directly, using huge caps might hurt the stability. The output resistance of the circuit and the load capacitance add another pole in the loop. It must be kept high enough to maintain stability. This turns out not to be an issue since the very low output resistance is so low there's no real need for the capacitance anyway (and the results back-this up), so just use a small cap and you're fine. It's possible to have large capacitance at the output but than the feedback must be slightly modified to have an alternative path at high frequencies (using a small cap from op-amp out to inverting input + resistor between the inverting input and the vGND node), but it will actually only hurt performance in most cases and requires extra parts/larger caps.
 
BTW, the reason I didn't post this circuit before is very simple, this is what everybody else do. tangent has an IC buffer in the pimeta, and a discrete (and better performing) buffer in the PPA. So this is simply a simpler implementation of the same thing that everybody else use. The reason I liked the way this thread started is that the circuit didn't use the straight forward way, it had a very strange look to it. Finding new and different ways to do things is always more fun than doing it the "right" way

Quote:

sonic wonder said:

Sour grapes - porque so sour? How very childish some of these comments are! We apparently allow unsupervised, spoiled kids to post on this site, but something should be done about that...
 
The load itself (not shown here) demands more than enough quiescent current for the regulators to be happy..
 
Um, I like testing for sonic qualities - actually listening to circuits - not just saving a few pennies with minimal parts counts or whipping out a theoretical "best circuit". Sketches and "computer simulations" only don't cut it completely. Know what I mean Sparkie and Wobblie?

Click to expand...

 
At least some of us have the good grace not to edit our posts retrospectively to remove our mistakes in an effort to make ourselves look better.
 
If you don't like the site you have the option of posting elsewhere.
 
w

Quote:

kt88 said:

 tangent has an IC buffer in the pimeta, and a discrete (and better performing) buffer in the PPA. So this is simply a simpler implementation of the same thing that everybody else use.

Click to expand...

 
Yes, but both these options will cost more in quiescent current, and simpler is not without its advantages. I've never been an advocate of virtual grounds and I drew up this circuit in the middle of the night, and have had little opportunity to think about improving it, having been obliged to defend it versus a much worse performing circuit in part due to your interventions.
 
w

As you said already, lets leave it at that. I'm sure whoever is reading this thread will able to decide for themselves.
 

Wakibaki,
 
I removed only what might have been confusing. This thread overall has had a lot of activity while we were working out the "self adjusting" virtual ground using the LM317s/LM337s. Furthermore, as you may have noticed, I integrated the final results into the first post of the thread, again for the sake of new readers of the thread - not to "cover up" anything about my posts - jeeze!
 
There is plenty of room for multiple circuits by many people who have worked with virtual grounds - or who wish to develop new circuits. That's OK with you, right?
 
Live and let live, especially recommended in public forums.
 

so where am i suppose to connect the adjust pin of lm336?

Quote:

curiousmuffin said:

so where am i suppose to connect the adjust pin of lm336?

Click to expand...

 
You don't. We don't need to use this pin in this design. The ADJ pin is there for an external circuit that would compensate for a change in current. Seeing the large margins of error in this design, the 3 mV per 1 mA change falls way bellow the acceptable margin. 

so my build works, and works well. i was first drawn into this design for the simplicity of parts, powering my desktop cmoy. it first ran a passive divider which sounded horrible. but goldpoint's circuit is good enough for me to keep the amp on the desk, thanks goldpoint!

Thanks curiousmuffin!
 
Yep, I concur. This circuit works well enough - and SOUNDS better than anything else I've tried, well almost...
 
For the "ultimate in sound quality", I do like putting another regulator out in front. The LD1085 3A adjustable regulators seem to sound really good in audio circuits wherein I've compared them to other voltage regulators. Here's that - what I actually use in my headphone amps at this time: The very BEST sounding virtual ground as far as I can find or develop.
 
 
Note that using this extra voltage regulator "in front" is likely beneficial for me because I power it all from a DC adapter. The LD1085 regulates my incoming DC and must be "cleaning that up".  (But if you are using a battery to power the virtual ground, I'm not sure this extra voltage regulator would really do anything for you - in that case it might be only a waste of power.)

Is this circuit really appropriate for battery-powered use?  There are going to be huge I-V losses through the linear regulators when the ground is sourcing current.

Don't mention the I-V losses. I did, but I think I got away with it.
 
w

From my point of view, this whole design was nothing more than a mental exercise. When considering efficiency, price and footprint, there really is no reason to use this circuit over an op-amp-based circuit. 

Hi,
 
Thank you for your thread.
My source is 2 x 2 Lipo = 14.8 V (16.8 V when full charged). 
 
1/ As I don't want to have a very accurate output voltage, my question is can I replace the 2.5 V voltage reference by 1 2 ou 4 diodes (basic or Schottky) ?
 
2/ Is it possible to use transistor like 2N3055 or FET to handle more current like 5 Amperes ?
 
Thanks in advance.

by 5 A current most people become worried about efficiency - the linear techniques in this thread essentially double your "ground" current drain
 
usually it becomes cost effective to use a switching circuit for ~ 80-90% efficiency
 
and if you have 2 batteries why not just use the middle point
 
if current draw is heavily unequal then again switching converters could balance the battery load

Hello,
 
And what about this :
http://electronicdesign.com/power/hex-buffer-mosfets-build-high-power-lossless-virtual-ground