More power supply

A couple of things to consider.. The wild leg will try to charge your capacitors to the wild leg X 1.414 So, the wild leg will try to charge the capacitors to 300 volts if you have 240 V three phase. The other legs will not try to charge the caps since they will be charged too high for them to reach.
If you are using 208 three phase, then it will try to charge them to 170 volts DC. 120 X 1.414
Start up, charging surge can be HUGE. Also, the diodes will only conduct for that VERY brief instant when the incoming voltage exceeds the capacitor voltage. Therefore the current that the supply draws will be in very narrow, very high current spikes, 120 time a second, per phase. This is what is known as Harmonic Distortion.
Adding a coil > Between < the diodes and the capacitors will radically reduce these current spikes. This is known as an input choke. This will reduce the Harmonic Distortion of the input power by a huge amount and reduce your output voltage somewhat. There once was a commercial radio transmitter that destroyed the power transformers and diodes on the power supply. I had them add an input choke, and it fixed it right up. The transformers and diodes no longer had to see those enormous current spikes. Without it, the current could Only flow when the voltage exceeded the charged voltage of the caps, and in a sine wave, that is a very short amount of time.
Also, if you have Two sets of capacitors, with a choke or resistor in between them, known as a Pi network, the ripple will be reduced by a HUGE ! amount.
When I was just a young lad, I gave up making a power supply with enough capacitance to eliminate the 60 hz hum for a telephone system. I could not buy enough capacitors to filter out the hum! Then I found out that if you add either a choke or a resistor in the Pi network, and Two capacitors, the ripple was completely eliminated. I ended up using a 100watt bulb as a variable resistor for the Pi network. I sure could have gone a lot farther if I had any kind of adult guideline back then.
Three ideas here.... 1. A Three phase rectifier will charge the caps to the Wild leg, and ignore the other two legs. ( If the output is not floating free of ground) ( If the output does float free, then the voltage above and below ground will be a really wild excursion (!!!) as it follows the three legs ab ove and below ground.
2. An input choke will reduce the current spikes across the rectifiers.
3. A Pi network will reduce the ripple by a HUGE amount and and use much smaller capacitors.

Three phase should need *way* less capacitance. Don't have the books around to look it up, but if you google 3-phase single phase rectifier ripple
You'll find lots of good stuff.

Are you sure you need any caps at all? Fullwave-rectified 3-phase never drops below 86.6% of peak, and ripple may not matter to servo amps as long as they have enough juice to drive their motors.

I don't know. The existing machine doesn't have any. I have eighteen 2400 MFD electrolytic caps just a gathering dust so there's no cost to me if I install some. just so the start surge is managed.
I don't think home made three phase quite meets that 86.6% reading as the L1-L3 runs line voltage but the L1-L2 and the L2-L3 run normally 20 volts to even 40 volts higher.
As I've said before, I'm a bit apprehensive about this whole DC power thing. that's why you're seeing all the dumb questions.
Say, how'd the ticker test go?
Karl

This may be part of why they used SCRs, they could set up the controller to slowly charge the caps. Yes, it will be a HUGE problem. You may destroy components, you may blow breakers. You almost certainly need the resistor and delay timer to charge the cap bank before closing the main relay. I have an 11 KW VFD that does it that way.
First, rectifying 240 V AC will give you 380 V DC. 100 A at 380 V DC is 38 KVA, or 51 Hp. Do you actually need that much?
Jon

The original power supply didn't use any significant amount of filter capacitors since it was running from three phase power and didn't have significant ripple to deal with. There is no need to make a huge problem where there is none, don't add excessive filter caps and you won't have excessive charging current. I'm quite sure his machine doesn't need anywhere near that much current to operate properly.

The existing power supply is 100 amp 400 volt. I'm sure that's just a surge at startup. Its a 20 horse spindle plus all four axis off the same supply. This is an 18,000 lb. machine. I would seriously doubt we'd ever draw 30 amps more than a fraction of a second in use. I did purchase an AMC 100 amp 400 volt servo amp for the spindle. Help me size that resistor and I'll install it on a delay timer.
Karl

Checked out OK. No shocks but we knew that. They said the pacemaker function acted less than 1% of the time but that was less than zero so probably a good thing that I had the implant.
Battery still shows 100%.
The device reforms the defib shock capacitor periodically. Oy din't know that! Good plan, electrolytic caps do need periodic reforming.
There are times when the presence of the device in a subdermal pocket near my clavicle is irritating. The device nurse said yes, that happens. Mary asked if it ever jiggles out of the subdermal pocket. The device nurse said she'd never heard of such an event.

Arrgh. I'm crossposting this to s.e.d, because it's like watching a bunch of lawnmower techs talk about how to soup up a NASCAR car.
Power supplies are almost trivial! Especially if you have real 3-phase.
So, would you be so kind as to restate the application, the need that needs to be met, the goal we're trying to accomplish? There are a bunch of people over at s.e.d who would be more than happy to jump in with electrical help - heck, they're almost as good at electronics as you guys are with metal! ;-)
Good Luck! Rich

Good news! :)

,,, With a long enough extension cord, he could run an electric commuter car. ;-)
Cheers! Rich

Ah ha! That was one piece of info I didn't know, a combined axis plus spindle power supply. Hmm, now the SCRs just don't make a whole lot of sense in that function.
The resistor for your direct connection relay? Well, 240 V RMS has a peak of about 340 V. You want, maybe a 50 amp worst-case charging current, so R=340/50 = 7 Ohms. Maybe make that 10 - 20 Ohms at 100 Watts. It just needs enough thermal mass to handle the pulse, as the current will drop very quickly as the caps charge.
Jon

Thanks Jon, I was trying to make this complicated. This is no big deal. I've got a call in to AMC for cap sizing.
One more query, I recently learned AMC amps need to be isolated so i just bought this tranformer: I'll put this in front of the rectifier/cap assembly.
My question, does a transformer like this have a large startup surge? If so I'll leave it on when ever the machine is powered. Otherwise, I'll drop power if Estop is hit. In other words, just a question of where to put the contactor for the power supply.
Karl

Some do, some don't. The problem is that the line phase at the moment the switch is turned on has a big effect on this surge. If the switch is turned on right when the voltage passes through zero, then the transformer core integrates flux for the full half-cycle of the mains. Normally, it starts with the reverse magnetization and de-integrates flux until the voltage hits peak, then starts integrating flux in that direction.
So, if power is turned on at the voltage zero crossing, the core gets twice the normal amount of flux, and saturates. The leakage inductance and winding resistance is then the only thing limiting current. The only way to know what the surge characteristics are are to turn it on a few times and listen for loud hums.
Jon