I used 4400μf for 2.5HP.
Max.

I have seen some designs like this which uses a load resistor of 100K and a TVS diode in parallel with the capacitor. Is that relevant?

 

Usually bleed resistors are much lower than that, personally never found the need for them.
Max,

 

Usually bleed resistors are much lower than that, personally never found the need for them.
Max,

What about the TVS diode? I heard it's for load protection. But haven't seen them so frequently in this kind of design.

 

The diode is probably there to clamp high voltage tranzients that can damage the circuit. Inductive loads, like motors, can be very noisy and generate a lot of spikes when switched. And a bleed resistor to discharge the power supply when it is turned off is a good idea although not always fitted. Add an LED in series with the bleed resistor for a power indicator? It is always good practive when working on these sort of things to assume the caps are charged even when unplugged. Measuring the volts on the caps is an idea before diving in. The screwdriver across the caps can be impressive but not really ideal.

 

The screwdriver across the caps can be impressive but not really ideal.

Ive' had to re-grind the ends of more than.........
Max.

 

The TVS in the schematic appears to be a 1.5KE200A which is far too small to do anything of value and has a conduction voltage threshold too low to be safe to use for 120 VAC (120 at 10% above nominal is 185 V peak; the TVS has a standoff voltage of 171 V, which is barely enough for 120 VAC actual line voltage). Typically electrolytic capacitors have a fairly generous short-term overvoltage rating and can probably handle dumped energy better than the TVS. If there is an expectation of fast "spikes" that need to be absorbed, it is probably better to put a reasonably large film capacitor across the electrolytics. I have never use a TVS in such a position nor seen one recommended. A TVS could fail short-circuit and protect the electrolytics from destructive voltage, but there is really no mechanism for applying destructive voltage in that circuit.

TVS devices can be useful in some snubber circuits where they are essentially used as zener diodes to help dissipate some spike energy. If there are spikes to be suppressed, it is probably much better use of a TVS to put it near the generator of the spikes.

A discharge resistor can be a good thing, but generally isn't strictly necessary. Where a discharge resistor is essential is where "X" capacitors are used right across the incoming AC line, since they present a shock hazard if someone pulls the plug and gets a finger across the blades.

~~~
I used to have a screwdriver with a corner of the blade "detached" - and embedded in the glass lens of my spectacles.

Someone I used to work with told of someone he used to work with who dropped a wrench across the terminals of a submarine battery. He simply waited until it melted in two, then went to get another wrench. Someone else I knew told of someone who got his metal watch band across the terminals of the battery in a diesel-electric railroad locomotive - with painful results.

 

I'm selecting the transistor now. I got away with a MOSFET instead of an IGBT. Do you think that paralleling MOSFETs is okay for this application? I wonder it because I've seen many MOSFETs nowadays that claim to stand high currents.