Non-polarized start motor capacitor in DC

[jc40]

Hi, I am planning to use a non-polarized, aluminum electrolytic AC motor start capacitor to filter a full-wave rectified DC voltage line. I am constructing an DC motor drive to be used with a 1/8 HP Leeson PM 90 VDC motor, and the problem is that I do not need only to vary the velocity, but also to vary the direction (forward-reverse). So I designed a H-bridge using IR2104 drivers and IRF740 mosfets. One of the problems is to get a 90 VDC clean power supply, but since I do not plan to use a transformer, because of size, I am using the following scheme:

127 VAC line -> Quadrac based dimmer -> full wave rectifier -> smoothing capacitor

The dimmer is adjusted so I could get the necessary 90 VDC at the final of the chain. So the pulsating DC output is 120 hertz, but my non-polarized capacitor is only specified for 50-60 hertz. Could really this be a problem? Should I need to use a polarized type? The capacitor is very rugged so I am starting to think that it is optimized for its function only (starting AC motors) and nothing else.

Also, by the way, does anyone knows how much current demand the IR2104? Could be low enough so I could use the same 90 VDC with a string of zeners to obtain an extra voltage of +12 VDC?

Thank you very much for any advice.

[Al_The_Man]

First the capacitor will be fine on 120hz ripple, it was designed for optimum power factor/for capacitive reactance on 60hz AC, you will be using DC ripple.
But I question the method of control, it would probably be much more economical to build a 4 quadrant SCR drive, this would be connected across the incoming supply without any need to reduce the voltage or build a power supply.
Either way, if you need 12v for logic supply, why not use a simple supply like a wall-wart type etc, if necessary add a 12v regulator.
Al.

[jc40]

Thank you for your reply.

The first though I had was to use some SCR's setup and avoid mosfets. However, I was wondering if two back-to-back SCRs were enough to change direction of the motor. The whole setup would be like a parallel-SCR's based dimmer, but then other problems arise like undesired turn on of both SCRs damaging my DC motor (I believe that the motor does not "understand" it like dynamic breaking). I was doing a search for 4 quadrant SCR drives, and I am surprised that I did not found a single DIY scheme, only manufactured commercial units. Also, in some texts explain about 6 or 8 SCRs, so I am more confused. Of course I am trying (if possible) to avoid the IR2104 route, and as you point the SCRs would be easier.