Adapting a voltage doubler to the input of a PSU

[Avelino Sampaio]

Hi. I have an adjustable power supply that goes from zero to 10 V at its output, with a maximum current of 2 A. I want to make an adaptation that consists of doubling the input voltage (from 12 to 24 V). The voltage change will be done by switching a relay.

As you can see in the attached image, I am initially adding two switching contacts from a relay (RL1 and RL2). The guidance I would like to get is whether I could eliminate the RL1 switch and, if so, what the operating conditions of diodes D2 and D4 would be. I don’t know how to configure a simulator to check whether a large current would flow through D2 and D4 in the doubler configuration.

[wraper]

You already have 24V winding, I don't see why would you need voltage doubler to begin with.

[Avelino Sampaio]

Actually, my goal is not just to get a single 24 V output.

The transformer has two 12 V secondary windings, which I currently use to obtain a symmetric output of approximately +12 V and −12 V after rectification and filtering, with up to 2 A.  What I want is to keep this symmetric ±12 V / 2 A output and, from the same secondary, also have a symmetric output of approximately ±24 V / 1 A using a voltage doubler arrangement with a relay.  So I don’t just need 24 V; I need +24 V and −24 V with respect to 0 V, in addition to the existing +12 V and −12 V rails. That’s why I’m studying this doubler configuration with relay switching and checking the impact on the diodes (D2, D4) and on the transformer.

[wraper]

Then you don't need RL1, just short it. D2, D4 won't conduct in voltage doubling position of RL2 anyway.

[Avelino Sampaio]

Then you don't need RL1, just short it. D2, D4 won't conduct in voltage doubling position of RL2 anyway.

I decided to keep RL1 because, in the doubler mode, diodes D2 and D4 would basically be left doing nothing. By keeping RL1 in the circuit, I can make those two diodes work together with the others in the bridge, which gives a better current distribution between all of them.
I thought it made more sense to use the full bridge instead of leaving half of it “idle” when the circuit is running in doubler mode.

[wraper]

By keeping RL1 in the circuit, I can make those two diodes work together with the others in the bridge, which gives a better current distribution between all of them.

Diodes do not work well when paralleled. Chances are one of the diodes will pass 80-90% of the current due to Vf difference which only worsens when diode with lower Vf that passes more current heats up more and drops Vf further. You just add a needless part and additional failure point to the circuit. Look at ATX PSU input voltage switching example .

[Avelino Sampaio]

I understand the issue with diodes in parallel not sharing current properly, especially when they are separate components. But in the case of this D3SB60 bridge, I believe the situation might be a bit different, since the diode chips are all mounted in the same block, very close to each other and running at almost the same temperature. That leads me to think their behavior tends to be more similar, which should reduce the usual imbalance that happens with discrete diodes.   For that reason, I ended up choosing to make use of D2 and D4 as well. Since they’re already part of the same module, I assume it makes more sense to use the whole bridge instead of leaving two diodes unused in the doubler mode.

Still, I’ll take into account the point you raised about adding another component that could become a failure point. That’s definitely something to weigh in the decision.

[Kevin.D]

Here's a configuration I came up with and posted several years ago https://www.eevblog.com/forum/projects/preregulation-of-a-linear-bench-psu/msg2081389/#msg2081389 . I don't know if its what your looking for ? its single ended output pre regulator with four ranges so not symmetrical on third range when centre tap of caps is earthed.

regards

[max.wwwang]

But in the case of this D3SB60 bridge, I believe the situation might be a bit different, since the diode chips are all mounted in the same block, very close to each other and running at almost the same temperature. That leads me to think their behavior tends to be more similar, which should reduce the usual imbalance that happens with discrete diodes.

Since matching characteristics or curves are not essential for two diodes in a single bridge chip, the inference of matching diodes is flawed, so your hope of getting two this way is groundless and cannot be depended upon.