[Polaris111688]

Just thought of a new power supply idea. I was wondering if this was a good idea for a rail splitter:

Drawn using Eagle's schematic program.

 

[Garbz]

Why do I get the feeling this won't work...

/Edit never mind. I was thinking of the wrong pinout for the LM 7910. This will work. Infact I've done it before in some project IIRC.

 

[nikongod]

i think the sds labs headphone amp uses a similar system (diferent chips perhaps) as its ground system.

 

[ATAT]

Would work...
IIRC I used 7805s in a similar config for a inertial guidance board, worked better than railsplitting a single 7810. So take that as you will.

 

[Polaris111688]

Heh, thanks guys.

 

[amb]

Quote:

Originally Posted by nikongod i think the sds labs headphone amp uses a similar system (diferent chips perhaps) as its ground system.

No it doesn't. The SDS Labs amp uses a straightforward LM317/LM337 split supply off of dual-secondary Amveco toroid and a bridge rectifier. No virtual ground there:

http://headwize.com/projects/showfil...stokes_prj.htm

As for polaris11688's idea here, what would guarantee an even split between the positive and negative sides? There may be an assumption that the 78xx and 79xx's internal circuitry are truly complementary when in fact they're not.

Also, since the ground pins of both regulators are tied together and now assume the task of sourcing and sinking return current from the load, is there enough current capacity through the ground pin for this role and would the virtual ground voltage be stable and not be affected by the load current? How much current is this PSU expected to deliver anyway? I have not experimented with this configuration but raise these questions as food for thought.

 

[Polaris111688]

Quote:

How much current is this PSU expected to deliver anyway?

I'm a little unclear as to which part of the circuit you're referring to. Are you referring to the +/- 12V that's going to the regulators or the +/- 10V that the load will draw from (after regulators)? If it's the latter, then the current draw depends on the load.

 

[amb]

Quote:

Originally Posted by Polaris111688 I'm a little unclear as to which part of the circuit you're referring to. Are you referring to the +/- 12V that's going to the regulators or the +/- 10V that the load will draw from (after regulators)? If it's the latter, then the current draw depends on the load.

I am referring to neither of those. Depending on the circuit that this PSU is supposed to power, the current it sources from the + rail may return to the - rail or to ground. Same for the - rail, the current it sinks from the load may be drawn from the + rail or from ground. If either of these are going to be ground, then you may have a problem. This is because the three-terminal regulators are not designed to source/sink much current through their ground pins. These ground pins are there only to establish a reference voltage for internal regulation purposes.

 

[Garbz]

Just a thought, what about a preceeding rail splitter of some kind?

 

[Polaris111688]

Quote:

Originally Posted by amb I am referring to neither of those. Depending on the circuit that this PSU is supposed to power, the current it sources from the + rail may return to the - rail or to ground. Same for the - rail, the current it sinks from the load may be drawn from the + rail or from ground. If either of these are going to be ground, then you may have a problem. This is because the three-terminal regulators are not designed to source/sink much current through their ground pins. These ground pins are there only to establish a reference voltage for internal regulation purposes.

Is it possible to buffer the ground to get a higher current output?

 

[amb]

Quote:

Originally Posted by Polaris111688 Is it possible to buffer the ground to get a higher current output?

Yes you could, but then it begs the question why you wouldn't just use a center tapped (or dual secondary) transformer and be done with it? If you're already using a positive and a negative regulator, you're only one capacitor away from a true split power supply if you use the right transformer. Adding a buffer merely increases the complexity...

Btw, below are two diagrams. "A" is the circuit you proposed (the negative regulator is redrawn to conform to the more usual convention despite the pin out difference to the positive regulator, and the transformer + rectifier sections are added to clarify what I mean). In A, the V+ output current may return to "G" and must be sunk into the ground ("COM") pin of the 79xx regulator in order to complete the loop back to the transformer. This is the problem area. The same thing occurs for the V- output current, which may need to be sourced from the ground pin of the 78xx regulator...

"B" is the true split power supply (where the regulators' ground pins are not called upon to source/sink current), because the ground is connected to the transformer where it completes the loop.

 

[Polaris111688]

You have a very good point there.