[Collins] 30S-1

[Dr. Gerald N. Johnson]

The effect of capacitance and load on capacitive filtered power supplies 
has been converted to simple charts. Proceedings of the IRE July 1943, 
by Otto H Schade, title "Analysis of Rectifier Operation." They have 
been reproduced in some ARRL handbooks, like the 1988 in my lap shows a 
voltage doubler in figure 20 of the power supply chapter and both half 
and full wave rectifiers in figure 13 of that chapter. Most charge the 
capacitors to nearly peak voltage if the rectifier side resistance is 
small and at some load point the voltage drops back to about DC = RMS. 
On the curves that point is defined by the product of the load 
resistance and the filter capacitance. So a bigger capacitor keeps the 
output up on the peak for a larger load current.

The problem with going to an infinite capacitor is that then the 
charging current pulses are infinitely narrow with infinite amplitude 
peaks. OK the transformer and rectifier impedance do limit the current, 
but still the pulses are short and the amplitude tall which upsets 
voltage regulators on generators when running on standby power and 
registers as a poor power factor on any power supply because of the 
extreme harmonic content to the current. The world is leaning towards 
demanding high power factor power supplies (already in Europe, even 
small ones) to drastically reduce the distortion of the load waveform. 
Unfortunately few ham supplies are high power factor, as are few 
computer supplies though they do exist with a switching regulator after 
the high power factor input section. And so use a half pound transformer 
for half a KW rather than a 100 pound transformer.

On 9/20/2010 5:09 PM, Carl wrote:
> A lot of capacity in screen and plate supplies has been the subject of
> much discussion in the amp world these days. Almost new industrial caps
> are selling for pennies on the dollar in surplus shops and fleabay. I
> picked up a couple of 2400uF 105C CDE caps from a swap forum for $4 each
> a few years ago and recently a bunch of 1200uF 105C 500V CDE for a bit
> under $2 each. Less than 2 year old date codes from unused industrial
> products.
>
> They should make for an extremely low impedence screen supply and do
> away with complicated and expensive regulators. The ESR is extremely low.
>
> About 10 days ago I hooked up a string of the 1200uF to a 2x 3CX800A7
> amp. With the stock 220uF caps the no load to full load HV sag was 362V.
> With the 470uF I replaced them with the sag was 250V, and 56V with the
> 1200uF. PEP average was higher and I suspect the IMD was improved. The
> amp had a quality step start from the factory.
>
> Some builders are using a string of 2400 to 5600uF on BIG tubes with
> handles and 5-8KV. That calculates to some serious energy storage
> requiring extreme safety provisions.

I did a paper on that protection for the amplifiers for the 2010 Central 
States VHF conference and in my presentation I referred to the 250 KW 
Collins 821A-1 a lot. There it was important to protect the tubes and 
vacuum variables from arcs dumping the stored energy in the filter 
capacitors and the modulation reactor into the arc destroying the parts. 
The standard test was to see if a 1X2 rectifier that would arc at the 
standard PA plate voltage would survive. Generally the first test took 
the filament but the glass survived. Usually by the 4th or 5th test the 
envelope shattered. That article is on line at 
http://www.geraldj.networkiowa.com/papers/CSVHF2010/protection.pdf. My 
power point with more 821A-1 pictures is there too, but it won't stand 
along.
>
> The only other concern I have with these caps is the effect on the tube
> longevity by increasing the average power considerably.

My concern is also the dumping of energy from the power supply filters 
into a stray arc in the tube or tuning capacitor leading to severe 
damage. The arc will do more to the tube. One might need to include a 
series switch or a crowbar to dump that energy outside the PA tube else 
the tube will likely not survive. Point in my paper is that we aren't 
using surplus 807s or 829Bs that we bought for $2 a case, we are using 
$100 transistors and $500 or more tubes and we can protect them but we 
have to work at it. There was a paper in the most recent proceedings of 
the International EME conference about that topic too. I have submitted 
(very late) my concepts paper as a possibility for 2010 MUD but haven't 
heard back from the editor.

I did 6 papers for CSVHF 2010 that are in that same directory.
>
> Carl
> KM1H
>
73, Jerry, K0CQ, Technical Adviser to the Collins Radio Association.