Author Topic: No drawbacks? Split rail PSU from two voltage regulators and virtual ground  (Read 507 times)

0 Members and 1 Guest are viewing this topic.

Offline igor78Topic starter

  • Contributor
  • Posts: 21
  • Country: us
I recently learned that one can do a surprisingly simple split rail PSU, using a virtual ground approach.

Single rail DC goes in, split rail goes out.
https://tangentsoft.com/elec/vgrounds.html
https://www.goldpt.com/virtual_ground_circuit.html



My ltspice simulation fails, but I've built it, and it works. I'm in the state of - "This is the best thing since sliced bread. There must be drawbacks, but I don't see large ones.

I found that capacitors on the output C2 and C3 must be of the same capacitance. Otherwise, regulators fight for the virtual ground. Input voltage can't go below the sum of rated voltages of U1 and U2, otherwise, the fight for the virtual ground resumes.

If there are no drawbacks, why is this design not super popular for split power supplies? It seems to handle load imbalance quite well.
I can't simulate it, so not sure of the frequency response. Is it the problem why it's being avoided?

Compare it with this circuit

It's absolutely same, but instead of BJT transistors voltage regulators are used.


 

Offline Benta

  • Super Contributor
  • ***
  • Posts: 5817
  • Country: de
but I've built it, and it works.
I highly doubt that.
Is this some kind of "AI" joke?
 

Offline igor78Topic starter

  • Contributor
  • Posts: 21
  • Country: us
It does work. And works rather well.

I'm slightly worried about load mismatch, but it seems stable. 100 Ohm + 10K Ohm loads is mismatched loads, right?

I was worried about heat - but no issue there as well.
Let's assume that there is only one load on the positive side of PSU. From V+ to virtual ground. Say the current is 1A.  Then the 78xx is idle, almost no heat.79xx is burning it's voltage drop, say 1.5V * 1A = 1.5W
1W will be burned by the 1 Ohm resistor. Wasted power 2.5W

 

Offline Benta

  • Super Contributor
  • ***
  • Posts: 5817
  • Country: de
Well, let's do the maths.
Say your LT1084V is set to 24 V output.
The 7805 will give a +5 V output.
The 7905 will give a 24-5 = +19 V output.
Now we connect the +5 V output to the +19 V output throught two resistors.
I don't think the 1-ohm resistors nor the regulators will work well under these conditions. And the current flow between the two regulators is outside all data sheet specs.


 

Offline igor78Topic starter

  • Contributor
  • Posts: 21
  • Country: us
I think you misunderstood the circuit.
It's meant to split the rail, not necessarily evenly, but split it.

>Say your LT1084V is set to 24 V output.

if you have 24v, you are expected to use 7812 and 7912.
1 Ohm resistors are "ballast"
 

Offline Kim Christensen

  • Super Contributor
  • ***
  • Posts: 1247
  • Country: ca
I found that capacitors on the output C2 and C3 must be of the same capacitance. Otherwise, regulators fight for the virtual ground. Input voltage can't go below the sum of rated voltages of U1 and U2, otherwise, the fight for the virtual ground resumes.
If there are no drawbacks, why is this design not super popular for split power supplies? It seems to handle load imbalance quite well.

Well, you just listed a couple of drawbacks... Plus the impedance of the virtual ground will be 0.5ohms, while pretty good, is not great for an active circuit.
You could simply use a power Op-Amp, or maybe even a class AB audio amp IC, to create the virtual ground with fewer components than the competing circuit shown.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf