CT transformer secondary wiring for two regulators

[MarkF]

Sorry I saw another thread discussing this but can't find it.

I have two power supply PCBs that I want to drive with one transformer. There is a 12V transformer with a CT secondary, a 5VDC PCB and a 12VDC PCB that I want to wire. I've seen two different secondary wiring configurations. One of which I think is just plain wrong.

Is this doable or do I really need two transformers?

[MarkF]

After staring at those two wiring diagrams, I realized that they are the same. Just drawn different.

[Ian.M]

If you had a somewhat higher voltage transformer, you could wire the AC terminals of both PCBs across the whole secondary, and for the 5V regulator, omit D2 and D4 and link the track their cathodes went to to the secondary C.T, to get half the DC voltage that the full winding feeding a bridge would give.   However 6V AC is only 7.4V peak and by the time you've lost another half a volt or so drop across D1 and D3, you'll have under 7V on C1 at the peak of the ripple.  The trough will be well below that depending on the load current and as MC78xx series regulators need 2V headroom, you've got problems.

It will probably work if the 5V rail is very lightly loaded as the transformer regulation factor will be in your favour, but, unless the transformer has a grossly excessive VA rating, it will drop out of regulation and start passing ripple long before you reach the regulator's max current rating.

[MarkF]

Will the diodes connected to DC GND of one PCB carry all the return current instead of being shared between both bridges?

Ian: I saw in another thread what I think you're talking about. Where the second supply only used half of the bridge. I can't find that wiring setup and I'm not sure how it was connected. By doing what you suggest, wouldn't I be reducing my diode drop since I don't have D2 and D4?

[Ian.M]

It would be a good idea to link the grounds of the two regulator PCBs to avoid a DC offset, and of course paralleling diodes is not the best of ideas, as you certainly wont get their full rating, but if you must use those PCBs, minimally modified, its the best option.   A better idea would be to strip both boards of diodes, replacing those on the 12V PCB with 3A Schottky diodes for less voltage drop - you'll probably need to drill out the holes in the pads and mount them vertically to get them to fit, and on the 5V board, replacing D2 and D3 with wire links to convert it to DC input, from the C.T. and the 12V Gnd respectively.

A center tapped secondary, two diode full wave rectifier has half the voltage drop of a full bridge, at the expense of much less efficient transformer usage.  However even with Schottky diodes its likely to be marginal feeding a MC7805 from a 12V CT secondary.  If you substitute a LDO, you *MAY* be able to get enough headroom (depending on the load current and the value of C)1 for a reasonable output current.

[MarkF]

I'm confused.
Is this the connections you suggest?

I would like to get 1A from both outputs. I have 6800uF capacitors for C1 on both boards. Overkill for sure.

It's looking like I would be better off making room for two transformers.

[mariush]

You should really consider simply using a switching regulator for the 5v output, or one of the ready made 5v linear regulator replacement products.
You'll have too much losses in the bridge rectifier, and dropping from 14-16v to 12v, and from around 7v to 5v and so on..

There are chips these days that are super easy to use and cheap and don't require a lot of extra parts. Sure, they'll be slightly nosier than a LDO but there wouldn't be so many losses.

Here's an example, the boring LM2576: https://www.digikey.com/product-detail/en/microchip-technology/LM2576-5.0WT/576-1518-5-ND/1027686

Datasheet is full of information and even schematics and tells you how to choose the components and so on. But for such a low frequency switching regulator, you can use plain cheap schottky diodes, cheap through hole inductor, electrolytic capacitors...  see http://www.onsemi.com/pub/Collateral/LM2576-D.PDF

[Ian.M]

@MarkF,
Yes. That diagram looks correct. The minimum VA rating for the transformer would be 27.5VA (assuming 12V secondary) - a higher VA rating would be desirable, as would a higher secondary voltage to avoid loss of regulation at full load if the mains input is at its lower tolerance limit.   6800uF caps, assuming worst case of 20% under nominal value, and 60Hz supply gives 1.53V ripple at full load.  That means the 12V output will be fine with a MC7812, however the 5V output will certainly brown out unless you either use an 8V transformer to feed it,  or use Schottky diodes in the bridge and a good LDO.  Increasing C1 would also help give you a little more margin.

Whether or not you take Mariush's advice on a switching 5V regulator is dependent on whether or not you need a low noise 5V rail, and if you care about RFI.   If for example you want to receive radio time signals, a switching regulator is likely to compromise your noise floor.

[MarkF]

Whether or not you take Mariush's advice on a switching 5V regulator is dependent on whether or not you need a low noise 5V rail, and if you care about RFI.   If for example you want to receive radio time signals, a switching regulator is likely to compromise your noise floor.

I wanted to stay with linear regulators for that exact reason. I'm looking for very clean outputs.