# Design of a switchable dual voltage power supply

My project would benefit from power supply with two modes selected by a switch. The 'normal' mode is full voltage (±13.5V) at full current (~5A), the 'high' mode is double voltage (±27V) at half current (~1.8A, see note 2). I came up with pretty basic doubler/rectifier circuit with the circuit ground either being connected to a transformer winding or being disconnected from it.

simulate this circuit – Schematic created using CircuitLab

According to my simulations the switch would need to handle quite high peak currents, up to 60A with a 27V 5A transformer. The question is, how should I design the switch? At currents this high relays tend to get rare and expensive. Also, automotive relays for instance are usualy spec'ed for 14VDC so there is a bit of uncertainity there too. The switching would need to occur a few times in an hour at most, but I'm trying to design for longevity anyway.

Is there a better way to design such a supply? Without buying another transformer that is.

Note 1. Ideally I would use a proper transformer with center-tapped or dual windings. However, this is not really an option, transformers get kinda expensive and hard to find where I live. So I have to compensate for that with schematic tricks. I also don't mind extra capacitance in the filter, capacitors are easier to get.

Note 2. I understand that at high mode there will be less than half of maximum current due to increased peak current and thus heating.

• Good first post, Eugene. A few pointers: (1) Double-click the diodes and delete the 1N4148 part numbers. They're only 100 mA, I think. (2) Add in a couple of nodes where the output of this circuit is. It's not clear at the moment. (3) Minor error on C2 "10 mF F". (4) If you've done any calculations for voltage droop between half-cycles with C1 and C2 at 10 mF then add them in. +1. Jun 10, 2018 at 14:04
• do you need the central voltage pin as an output? it can unbalance your caps in "normal mode" Jun 17, 2018 at 23:02

To reduce ripple and peak current, at the expense of reduced voltage, you can add an inductor to the positive and negative legs to make a choke-input filter. This will also reduce the chance of destroying the diodes from the inrush current on closing the switch.

The chokes can be optimized to reduce ripple, in which case output would be closer to RMS than peak voltage, or a smaller choke would increase the output and ripple, yet might keep peak current tolerable.

Incidentally, choke-input filters reduce transformer heating because of lower peak current, so average power output does not suffer, though the output voltage is less.

First, i should say that some parts are not clear like switching requirements i.e. the frequency and response time specs. Have you ever consider using electronic switches, IGBT, FET?

If yes maybe you can consider using soft switching techniques to reduce inrush current.

Another option is using using NTC or PTC to limit inrush current. This is a commonly used trick in the situations that are similar to yours. Please take a look at here: https://www.ametherm.com/inrush-current/ptc-thermistors-for-inrush-current-limiting

• The are perhaps beneficial comments, but a posting in the answer for is required to actually answer the question, and this does not. Jun 17, 2018 at 21:37

If you don't need the 0V output at all, your circuit is mostly fine. I'd just add 100k resistors in parallel with the capacitors to ensure they won't get too unbalanced (in extreme cases one can go negative, and then bad things can happen). Besides that, you just have to be sure not to destroy your relay and diodes when switching occurs. That may warrant a series resistor somewhere.

If you do need that 0V output though, I think your circuit will need to get a lot more complex.