Dual positive voltage supply from CT transformer help

[GTechno13]

I have a project using a 10V/5V dual output transformer. The secondary can be wired in parallel (5V) or series (10V) and series may be center tapped. The datasheet is here. My project requires regulated 3.3VDC, unregulated 5-6VDC, and unregulated of at least 8VDC. I connected the transformer in series and connected a (full wave) bridge across the outer connections (both coils in series) and connected another bridge to the center and one of the outside connections (across one coil). I connected the "minus" output of each bridge together to form my common ground. The ac input to each bridge was measured to be 13VAC and 6.5VAC respectively under almost no load conditions which is just fine. My problem is I measure 15.4VDC out of the center tapped bridge with a large smoothing cap (470u) and only a ld1117v33 regulator (which is under no load) as a load. The series bridge gives almost 18VDC which is correct. According to (6.5/.7)-1V I should see about 8VDC out of the problem bridge.
Could this be attributed to incorrect connection of the center tap bridge? Does dot convention come into play? Also, as I write this I wonder if oscillations in the regulator could be confusing my cheap meter. I have a large input capacitor and only a small (.1u) output cap on the regulator but I do measure a 3.3V output as it should.

Looking at my PCB and the datasheet I have the center taped bridge connected to pins 8 and (6/7) of the transformer.

Any help would be greatly appreciated.

Thanks,
GTech

Transformer is the 310 model. 5V/10V

 

[awt]

can you post the circuit diagram here to make your expression clearer?

 

[GTechno13]

Figured everyone would want that but I usually go straight to PCB layout from my head without schematics. I fixed the datasheet link and labeled the pins accordingly. The dots are at pins 5 and 7.
**broken link removed**

Top node I get 17.8VDC (good)
Middle node I get 15.4VDC (what!?!???)
Bottom = GND

 

[crutschow]

You can not connect the transformer that way to get two different voltages. You should have simulated the circuit before you built it.

The easy way to do what you want is to connect the windings in series to a bridge rectifier with the negative side of the bridge as your common (ground) as you have already done. That gives you your 17.8VDC.

1/2 of that DC voltage (about 9V with an added filter cap) will now appear at the center tap with no additional rectifiers needed. Magic? Nope. The configuration acts as a full-wave bridge for the high voltage with two of the bridge diodes that are connected to ground doing double-duty and acting as a center-tap 2-diode full-wave rectifier for the center tap output. (The circuit appears a little odd since the 2-diodes go to ground instead of being connected in the hot side as is usual, but it works the same).

 

[carbonzit]

A little tangent, if you don't mind:

Looking at full-wave rectifier circuits, it seems to me that the 2-diode full-wave rectifier is better than the bridge rectifier. It requires only half the parts and only gives one diode voltage drop instead of two.

So why is the bridge rectifier so much more popular? OK, because it doesn't require a center-tapped transformer.

But isn't that a trivial consideration? It costs pretty much nothing to bring out another tap from a transformer.

Up to now I've almost automatically gone the bridge rectifier route because it's "simpler". But now I'm wondering if it wouldn't be better to bet on the 2-diode horse. (Of course, there are times when one must choose one or the other because of grounding considerations.)

What do you think, Carl?

 

[GTechno13]

Hmmm sorry, I seem to have a hard time simulating AC circuits. I'm surprised that what I did didn't work as the input potential to each bridge indicated it would but I suppose a oscope would have indicated otherwise. If I understand your post correctly, below is what should work?
**broken link removed**
Lo and Behold! That DOES work! (according to simulations). Interesting that both waveforms appeared sinusoidal and did not appear like a regular rectified waveform dipping to 0. Maybe I messed something up in simulations. If this was correct this circuit would require much less smoothing.

Luckily this will only require one jumper to fix on my PCB. Thanks tons crutschow!

 

[The Electrician]

A little tangent, if you don't mind:

Looking at full-wave rectifier circuits, it seems to me that the 2-diode full-wave rectifier is better than the bridge rectifier. It requires only half the parts and only gives one diode voltage drop instead of two.

So why is the bridge rectifier so much more popular? OK, because it doesn't require a center-tapped transformer.

But isn't that a trivial consideration? It costs pretty much nothing to bring out another tap from a transformer.

The 2-diode arrangement with a center-tapped winding has the disadvantage that current flows in only one half the winding at a time. Alternate half sine current pulses are drawn from each half and the half of the winding that isn't supplying current is taking up space in the transformer and doing nothing to supply current. This means that a center tapped 2-diode circuit supplies less current for a given size transformer.

 

[carbonzit]

So a 2-diode rectifier gives less current, while a bridge rectifier has a greater voltage drop.

I guess it's up to the designer to choose their poison ...

 

[The Electrician]

Usually, copper is more expensive than silicon, so the bridge is the way to go if the designer has the ability to choose a transformer for a design that will go into production.

When a transformer is used for a one-off, hobby, application, other considerations may take precedence, such as maybe the experimenter only has two diodes of a suitable rating!

 

[crutschow]

Hmmm sorry, I seem to have a hard time simulating AC circuits. I'm surprised that what I did didn't work as the input potential to each bridge indicated it would but I suppose a oscope would have indicated otherwise. If I understand your post correctly, below is what should work?

Lo and Behold! That DOES work! (according to simulations). Interesting that both waveforms appeared sinusoidal and did not appear like a regular rectified waveform dipping to 0. Maybe I messed something up in simulations. If this was correct this circuit would require much less smoothing.

The reason your circuit doesn't work is because you would need an isolated ground for each bridge output. There is interaction between the two bridges when you try to use a common ground which causes what you observed.

With no filter caps, both outputs should look like a full-wave rectified sine wave. With filters the outputs should be DC.

Edit: Here's a sim of the circuit with no filters.