Determining HP-23 capacitor replacement equalizing resistors

[w5rkl]

Morning everyone,

I’ve been reading Stu’s response to a question about capacitor equalizing resistors.

The original capacitors in my HP-23 have a leakage current that is too high. I've decided to replace the caps with Sprague Large Can 36DX 160ufd 450VDC screw terminal capacitors in series for a total capacitance of 80ufd at 900VDC.

The HV winding provides approximately 282VAC rms, 398.748VAC peak. In a voltage doubling circuit the peak voltage would be 797.496.

The total capacitance of two 160ufd series connected capacitors is 80ufd (C1 x C2 / C1 + C2) or simply 160 / 2.

I needed to compute the voltage across each capacitor so I used the formula "total capacitance / individual capacitor value x applied voltage, i.e. 80 / 160 x 797.496 = 398.748 volts across each capacitor.

The capacitor specification sheet does not provide a leakage current formula or a leakage current value at a voltage and temperature rating. What I did find by looking at similar capacitors was a leakage current formula of I (leakage current) = .006 x sqrt C (capacitance) x V (supply voltage). I know this may not be the correct formula but, I used it to get an approximate idea of what the capacitor’s leakage current would be (see the following).

A.  160ufd x 797.496 = 127599.36 (C x V)

B.  Sqrt of 127599.36 = 357.2105 (sqrt of answer "A")

C. .006 x 357.2105 = 2.1432ma approximate capacitor leakage current (.006 x answer "B")

D. 797.496 / 2.1432ma = 372.105 (R = E / I)

At this stage I figured "4" 200K 3 watt resistors, 2 across each capacitor for a total resistance of 100K across each capacitor and a total series resistance of 200K at 3 watts. Am I correct or are my calculations totally wrong?

73
Mike
W5RKL

[w4bfs]

100K for each cap sounds very reasonable ... I always check the voltage drop across a series stack of caps for each capacitor ... this gives a real idee of how the voltage is dividing both by leakage and equalizing resistors ...what you have sounds conservative and should be long lasting ... 73 ...John

[AB2EZ]

Mike

Sounds like an excellent solution to me 

Three very minor comments:

I looked at the HP23 schematic that is posted on the BAMA web site.

In the posted schematic, each half of the voltage doubler uses 125uF in parallel with 50k ohms (a pair of 100k ohm resistors in parallel). The capacitors in the HP23 will bleed down with a time constant of 125 uF x 50 k ohms = 6.25 seconds. Therefore, the voltage will decay to 5% of its full value in 18.75 seconds

In your new design, the capacitors in each half of the voltage doubler will bleed down with a time constant of 160 uF x 100k ohms = 16 seconds. Therefore, it will take 48 seconds for the voltage to decay to 5% of its full value.

So... if you are going to work on the supply or anything connected to the supply...you need to make sure that you wait 48 seconds, instead of 19 seconds, before you use your insulated / grounded safety discharge stick to short the HV output to ground.

Each 3 watt resistor will dissipate up to 3 watts, so your total bleeder design is good for 12 watts.

Each 200k ohm resistor will have to dissipate approximately 400 volts x 400 volts / 200,000 ohms = 0.8 watts ... so your 3 watt resistors should be very comfortable.

Since this is a voltage doubler design, the output voltage will split evenly across the two capacitors even if the leakage resistance of one or both of the capacitors is not swamped by your bleeder resistors. In effect, each capacitor is being charged independently by the transformer on each AC cycle.

Best regards
Stu

[w5rkl]

Thanks John and Stu for your comments.

The CGC capacitors are installed and appear to be working quite well. The caps are running quite cool. The no load HV measures 788VDC. I haven't measured full output voltage yet, have to remove the rear panel on the final cage on the HW-101 and measure at the base of the final's RFC. I replaced quite a few resistors in the voltage divider circuits for LV and bias which has greatly improved stability. Normal stability is 100Hz after 30 minutes warm up and 100Hz per hour after warm up. The 101 doesn't appear to be drifting at all, at lease during the initial receive test.

Still have some additional test to do tomorrow but so far everything appears to be working quite well.

Again, thanks John and Stu.

73
Mike
W5RKL

[w5rkl]

Further testing wasn't needed with the supply. After finding additional information it was clear the supply is providing the correct output voltage even though it is less then the manual states it should be. Variations in the different transformers Heathkit used in these supplies plus differences in line voltage can contribute to different HV AC voltages. The supply is working very well, not heating up like it did with the small snap-in capacitors, and I'm only noticing a slight decrease in output power when using the supply. The slight reduction in output power is not due to an alignment issue and it sure won't be noticed at the receiving end.

Thanks again to all who responded, I greatly appreciate your comments and suggestions. Now I have to figure out what to use in order to get the coffee and doughnut gunk off the calculator buttons. hmmm..DeoXit and a wire brush, yeah, that's it, "get'r done".

73
Mike
W5RKL