# Design for a balanced (positive and negative) HV DC regulator

I'm trying to design a high voltage DC regulator for driving a PZT that will provide both +200 and -200 V outputs referenced to ground. The input is a rectified AC waveform that will provide UK mains voltage (230 V ac r.m.s.). Since it's driving a capacitive load only, the current requirement is small (less than 15 mA per rail).

I've got a few ideas for how to do this, but the problem is that they are based on positive voltage regulator designs. I'd like to build something that works for both positive and negative rails of the input.

The most promising idea I have is to take a good design for a positive regulator and convert it into a negative one, and pair them up with a common ground. I am thinking of basing it on the "Maida" style floating regulator. Pete Millett has provided an updated version of this design that uses a high voltage MOSFET and LM317, freely available on his website:

The rectified, filtered mains is input to the MOSFET to perform coarse regulation and fine regulation is performed by the LM317 which only sees a fraction of the total input voltage due to the zener diode D1. The design shown is only for a positive voltage, but I would like both positive and negative varieties. I imagine that it would be possible to adapt this circuit to instead work for negative voltages simply by using a P-type instead of an N-type MOSFET and an LM337 instead of an LM317.

Can anyone think of a reason why the above idea might not work? In particular, I wonder if any difference in the time it takes for each rail to come up to full output (given tolerances of capacitors, differences in attached load impedance, etc.) would have any damaging effect on the sensitive components such as the MOSFET. Will the suppressor diodes in parallel to the MOSFET and LM317 prevent such problems (e.g. by blocking a reverse current)?

I'm happy to hear any other ideas. The output doesn't need to be regulated heavily so perhaps there's a simple approach I'm missing.

• Just to be clear: your W1 and W2 input nodes are coming from a bridge rectified 230VAC mains line? Just the mains line through a bridge? – jonk Aug 6 '16 at 1:02
• @jonk yes, that input will be the rectified AC mains. I guess I would put some smoothing capacitors to ground to get rid of some ripple. – Sean Aug 6 '16 at 7:16
• Is there a reason you aren't concerned about galvanic isolation? (Or the original authors weren't?) – jonk Aug 6 '16 at 7:18
• You say "doesn't need to be regulated heavily." I'm not familiar with what a PZT is (piezo thing?) How regulated does it really need to be? What is X in $200V\pm X$? And is 15mA really all you need? – jonk Aug 6 '16 at 7:37
• @jonk: sorry, I should have mentioned there would also be a transformer before the input to provide isolation. The PZT is a capacitive load (around 1uF). It won't draw much current at all, so 15 mA is already more than enough. I guess X would be a volt or less, but I'm not so concerned about that right now - I would like to know whether it's feasible to group the above design with a negative voltage equivalent sharing the same ground. – Sean Aug 6 '16 at 8:02