[Amps] When to stop.. caps for power supplies.

[Jim Thomson]

Any fool can overbuild a product. Good engineers know when to stop,
> and good engineers get paid good money to get it right.
> 73, Bill W6WRT

Agreed. We are paid to do good but not ridiculous engineering. Leave that for Hollywood. Or Star Wars (the project, not the movie). 

For serious high power (3 kW to Terrawatts), three phase service is the recommended design. With a simple 6 rectifier bridge, the ripple is down to 4.2% RMS, and with a 12 pulse Delta-Wye connected dual secondary transformer, 1%. 
The need for large capacitors is eliminated, and a simple LC or just C filter is enough. This eliminates the hazardous stored energy that takes a lot of expensive parts and design care to prevent damage to amplifiers and cables. 

73
John 
K5PRO

###  Some books  say 5%... some say 4.8% ripple  for a 6 rectifier bridge. [dahl sold a lot of em.. all using 6A10 diodes].    That's with NO  filter at all... VS  48-50% ripple  for single phase + no filter.  That's great IF  you have
208/360/480/600 vac   3 phase power available.  With any 3 phase, 6 x rectifier bridge, the caps get.."topped up"   360 x  per second  VS  120  for  single phase FWB... and only 60  for a doubler.   I only have 240/120 vac
single phase available.. and long drop line into the house.  Part  of the high C concept / experiment was to  eliminate ripple at 120hz + all harmonics, plus  improve dynamic regulation.  Sorta like a quasi brute force regulated B+ supply.
The big caps all reside in their own steel cab.   We found that  with a CORRECT type and sized B+ fuse PRECEDING the correct type and sized 50 ohm glitch R, that any short  from B+ to  chassis, grid, etc,  was not an issue, regardless  of whether 34 uf  was used... or 250 uf.   Fault current value is still the same.    7700 vdc / 50 ohm glitch = 154A fault current..... which will open off a 3A rated HV fuse in < 2 msecs.   All B+ wiring BEFORE the 50 ohm glitch R  IS a concern,
and precautions HAVE to be taken.  So is follow on energy. 

##  In the case, where multiple banks  of caps are in parallel,  individual HV fuses are used at the INPUT of each bank of caps.  Each bank gets it's own HV meter.  If the input hv fuse to any individual bank is removed, or goes open, the Hv meter  for that bank will drop to zero.  There is only one teflon hv wire, running from FWB.. over to B+ fuse, that feeds the 50 ohm glitch R.    There is only one real heavy duty teflon wire, running from the input of the same B+ fuse, b4 the glitch R.....that goes external, to the box full of caps.  This external teflon hv wire terminates  on the input buss, that is just cu strap running across the top of hv insulators. Fuse clipS sit right on top of the same strap.  NO hv connectors used anywhere,  it's all hardwired. 

## plate xfmr, contactors, FWB, + glitch R  {and it's fuse] , etc... reside in one box.    Caps reside in a 2nd box.    RF deck is in a 3rd box.   All  box's  have a B+ meter.   The box of caps, gets separate B+ meter's  for each bank of caps.  All 3 x box's  bonded together, via a star grnding system.   There is really little chance of the B+ accidently  being grnded out.


 The individual fuses  for the inputs to each bank,  alone won't limit current, but will  facilitate adding/deleting banks.   Everybody has been telling me for yrs, that anything over 40 uf is a waste of time, and you will soon reach the point of 'diminishing returns' .   Well I had the caps on hand, so proceeded with the experiment  to find the point of diminishing returns.      I haven't found it...yet.  

Later... Jim   VE7RF