+/-40V SMPS virtual ground

[j0no89]

Hey, I am currently designing a switchmode power supply with +/-40V @ 2A based on the TOP258YN. So, getting on with my question, in the reference designs for a split rail supply they use a transformer with a centre-tapped secondary winding, but the only transformer i can find on digi-key (where im ordering all my other parts) that is capable of putting out this much power and has an appropriate turns ratio has no centre tap on the secondary (Wurth Electronics 750311269).

To overcome this problem i found a 'virtual ground' circuit on http://tangentsoft.net/elec/vgrounds.html. So, my question is, what is the actual potential of the common rail in this virtual ground circuit w.r.t the mains input earth. I dont really understand how the virtual ground works so im not sure if it is a true split rail supply or is it just creating a reference halfway between the 2 rails and saying that the 2 rails are +/-40V w.r.t that reference (which would be 40V w.r.t earth). Mainly i just want to know if this common rail is safe to connect to the mains earth reference.

Thanks in advance for any help, if its unclear what im asking just let me know and ill try and be more clear!

 

[duke37]

1. I would never connect the mains to the center rail either directly or via a capacitor. You could connect it to ground.
2. You have plenty of windings to give a +/- supply.
3. Diodes in parallel is not good. One could hog all the current.
4. Your PNP transistor is upside down.
5. As you have drawn it, there is no way of setting the center rail to the center.
6. As drawn, you only need one diode to generate V+ to V-, the transformer may saturate.

 

[Zork]

This circuit waste two diodes, doesn't make much sense to use two diodes on a half wave rectifier, without the center terminal of the transformer, the four diodes are doing the job of only one.

I can see three windings on the transformer, it will be much easier and efficient if you use the 8 and 12 terminals on the same way of the first schematic.

The virtual ground wouldn't be applied to this level of power, a lot of power will be used to heat the two transistors, this procedure is useful on small signal problems, where symmetrical voltage supplies aren't avaiable.

 

[KrisBlueNZ]

So, my question is, what is the actual potential of the common rail in this virtual ground circuit w.r.t the mains input earth.

The transformer should provide isolation between the primary-side circuitry and the secondary-side circuitry. So you can connect mains earth to the virtual ground rail if you want to.

I dont really understand how the virtual ground works so im not sure if it is a true split rail supply or is it just creating a reference halfway between the 2 rails and saying that the 2 rails are +/-40V w.r.t that reference (which would be 40V w.r.t earth). Mainly i just want to know if this common rail is safe to connect to the mains earth reference.

The latter description is right. The virtual ground generator operates from an 80V supply and generates an output that is half-way between the two rails. This then becomes your 0V point for the power supply's output, and the positive and negative rails of the 80V supply become +40V and -40V relative to 0V.

There are several problems with this approach though.

An active supply splitter like this is not normally used at such high power levels. Consider what will happen if you connect a 2A load between the +40V output and the 0V rail. 2A of current will flow through the PNP transistor and there will be 40V across that transistor. This means that transistor will dissipate 80 watts continuously, and will need a big heatsink - preferably fan- or water-cooled. It is just not practical. Zork also mentioned this issue.

There are also errors in your supply splitter diagram - the PNP is drawn upside down, as duke37 already pointed out, and there are other errors, including an unneeded double diode. Also I would use a single diode with a higher rating, instead of two diodes in parallel. Diodes have a negative temperature coefficient of forward voltage, so they don't share current when connected in parallel. One gets warmer than the other, which causes it to take more current, which makes it warmer, which makes it take more current, and so on.

My suggestion would be to rewind the secondary of the transformer. Get a transformer that's a bit bigger than you need, unwind the secondary wire, match the gauge and rewind the secondary with a centre tap.

Alternatively, it looks like that transformer already has three separate secondaries. If it's rated for at least 50% more power than you need, you could ignore one winding, and connect the other two in series, to form a centre-tapped secondary.

If you post again, could you please post the full schematics, rather than just part of them, or post links to them, and identify them. I assume the first one is from an app note for the switching regulator you're using, and the second one is your design? This isn't clear from your post.