Bleeder Resistors

[Ed/KB1HYS]

I am building a 2000V  0.300A Plate supply for a class C RF deck. 
Choke input, 6Henries, with an 8uF cap on the output (usual configuration).
The math says this should be good for around 1.5% ripple.

I would like to use 10 (ea)  10kOHM @ 10Watt resistors in series as a bleeder.
This gives me 0.020A bleeder current which is a little low, but I can tap them for other usefull voltages in 200 volt increments. ( I want to get 400 volts for the screen).

So I know this sounds simple, but since I'll only be dissipating 40 watts through the bleeder, and I have 10 10 wat resistors so this should work right?
Or did I miss something that will make it go POOF? 

[w8khk]

Hello Ed,

Ripple percentage is 100 divided by inductance times capacitance, in your case  100 / 4*8 = 2.083%, or 4.166 volts, sufficient filtering for the class c final.  The bleeder string itself does not appear to be a problem, but several other items should be evaluated before continuing.  This assumes the power supply is designed for 2kv average output under load.

First, the choke value vs the bleeder resistance should be evaluated to be sure the output voltage will not soar if the supply is unloaded.  This may not be an issue if the supply is always loaded, in the case of an AM final where the power supply is switched on only during transmissions.  In that case, the bleeder string current is not critical.  To avoid voltage soaring, critical choke inductance L in henries should be at least equal to load resistance divided by 1000.  Figuring in reverse, for a 100K bleeder, a 100 henry choke would be required to maintain average output.  This is unrealistic for a practical supply.  Alternatively, a 6K load with your 6 henry choke would meet the requirement.  Assuming a 6K load, at 2KV, equates to 300MA load.  This is 600 watts, independent of the bleeder.  (Note: This 6K load is your class c final, not a bleeder resistor.  But as long as your filter capacitor is sized for sufficient voltage, the final current could run less than 300 ma. with no problem, other than screen tap voltage variation from your 400 volt target value.)

A smaller load would still allow the choke input supply to rise above average, but not to peak voltage of 2828 volts (2000 x 1.414).  Filter capacitor should be sized above 3000 volts to allow for voltage rise to peak, with some margin for surges.

The other issue relates to the screen current required from the tapped bleeder.  If the current is very small, and regulation is not critical, then the tapped bleeder string may be acceptable.  I would suggest recalculating the voltage at the 400 volt tap, taking into consideration the reduction in voltage based on the screen current at that point.  Small values of screen current could significantly reduce the tap point voltage, considering the bleeder string is drawing only 20ma.  Worst case, if you should happen to short the 400 volt point to ground, the remainder of the bleeder string should be safe, as 80 k at 2kv would increase the bleeder current to 25ma, for a dissipation of 50 watts spread across 80 watts of resistors.  This is safe, assuming the resistors are well ventilated.

In summary, be sure that the filter capacitor can handle the peak plus surge voltage, and evaluate screen voltage and regulation requirements.

[Ed/KB1HYS]

For simplicities sake I wanted to use a dropping resistor from the plate supply to get screen volts.  Making it part of the bleeder seemed logical.  The handbooks recomend that the bleeder current be ~ 10% of the total current so I should have had .030A through the bleeder to meet the requirements, but I rounded off the resistor values to something standard and make the bleeder/voltage divider easy to build. 
I handn't considered the effect of the screen current on the bleeder.
 I had thought that in keeping the screen voltage a product of the plate voltage that variations in the plate voltage would result in an "adjustment" in the screen voltage that would keep the bias level constant. I assumed (wrongly?) that they were a simple linear relationship.
  The tetrode (4-250) I'm using has a max screen current of 60mA, so I thought all would be well if I limited the screen current to always be less than that value.

I did base the filter design on that used by the Globe King, whos final power is similar to mine.