HV power supplies, how much smoothing capacitance is TOO much capacitance?

[Chris Wilson]

I am building an HV power supply for a 3CX3000A7 valved (tubed) linear amp a pal and I are converting. The supply will be circa 5kV at 1.8 amp. I am hoping to use an array of electrolytics with close tolerance bleeder resistors across each one for smoothing after a full wave rectifier. I am getting conflicting advice from two obviously very clued up guys. I am in no position to judge who is right or wrong, or which way to go. Maybe I should toss a coin

One guy says having a vast amount of stored energy in the bank is asking for trouble if something goes wrong, and the extra capacitance won't aid the smoothing any significant amount. The other says having a load of capacitance enhances the smoothing and makes for a better amp used for SSB or CW as it helps keep the HV from changing up and down as the loads change.

Let's put some figures on this

Guy A: 15 off 470uF  450 volt electrolytics in series, with 50k ohm 5 watt bleeders across each cap. Total capacitance circa 31uF, rated voltage 6750 volts


Guy B: 16 off 2300uF  450 volt electrolytics in series with 100k ohm 3 watt MOF resistors across each cap plus a 6A10 diode across each cap in case a rectifier diode block fails.
That's circa 143uF capacitance and a rated voltage of 7200 volts. And a lot (about 2000J?) of stored energy.....

Both put their case well, what do I do? Thanks!

[Smokey]

If you put a good quality fast acting high energy fuse in there, it would probably keep things from blowing up if the cap bank dumped into something.  I've never used these, but they look really cool.
http://www1.cooperbussmann.com/pdf/abe14678-ac07-4a08-8813-faa12ccda82e.pdf
Just a thought. 

What voltage ripple is acceptable to you?  Since the simple version of the voltage ripple equation, ripped off Wikipedia, is just V_ripple-pp = I/(2fC), more capacitance means less ripple, but the lower capacitance might get the job done.

The resistors across series caps like that are more balancing resistors than bleeders. 
http://www.illinoiscapacitor.com/pdf/Papers/voltage_balancing_resistors.pdf

One more question... Why tubes?

[PA0PBZ]

One more question... Why tubes?

Because there is no easy solid state way to go from 400 Watts to 6 KWatts? 

[Chris Wilson]

I am building the supply as a "generic" one that will have enough oomph for any forseeable project. It may be going to the USA with my pal when his UK convalescence is over, or I may end up keeping it. here in England. I have a supply up to about 80 amps (main fuse in distribution panel is 100 amps. I have pulled over 75 amps with a TIG welder from it, the line voltage drooped a bit though

[Topband RF]

Do you mean a 3CX3000A7 ?  I couldn't find a data sheet on a 3CX300A7 anywhere. If your looking for robust HV designs 1-2.5 amp check some of the ARRL handbooks. I personally wouldn't use that tube, as replacement costs are abt 1500.00 USD. A 3CX1500A7 (8877) will put out full legal power for any licensing class in any country and a whole lot more. Thier cheaper, easy to find medical pulls and can provide 2 kw out with less than 4kv and less than 100w drive. I have seen good condition 8877's available from solid sources from 250-400 USD. The difference from 2kw rf out put to 4kw rf out put on the receiving end of an rf signal is minimal. They also require low drive which is an added benifit, because the drive stage (most common HF transmitters) can be dialed down from their max output rating. You may also consider oil filled HV caps or a pair of them, I had a 4000v oil filled cap @ one point for an 8877 Project that never got off the ground. I think I paid 60 dollars for it if i recall. Good luck with your project.. and you may have heard this 1000x before.. please work safe. A friend of mine was killed working on a 4000v 1.5 amp supply about 2 years ago, it was a huge loss to me and many others.

[Chris Wilson]

We have the base for the amp already, a Henry 2000D RF generator, it's a fairly well documented base for HF band linear conversion. So the tube is already set in stone. Typo on the first post, will amend should be 3CX3000A7  Oil filled are very had to find in suitable values here in the UK, and as they are so big and heavy they are costly to ship. I saw these in Scandinavia, but the ad says pulse caps, which, as far as I know, are far from ideal for ripple regulation, yet the ad goes on to say they are ideal for linear PSU's....  Cost is far greater than an electrolytic array, too. I think practicalities of supply dictate I stay with electrolytics. A further question is what's a suitable unclad board material to build on at these voltages? Thanks for the warning re the dangers of high voltages at these sorts of current level, and with charged capacitors, I intend to show it great respect and take great care though, I can assure you

[mianchen]

Well actually per ring mains it's about 30A per ring mains, so if everything is shut off 7200kW is possible
But how are you going to achieve 9000kW huh? 3 phase?

6kW with solid state? No problem. 2 of these
http://item.taobao.com/item.htm?spm=a1z10.3.17.28.ae7fb7&id=19348548600&

he's doing rf stuff not audio amps

[Psi]

I'm pretty sure the guy who's saying  147uF is doing so to lower the ESR

I'd probably go with the 31uf option, but i'd put some 942C 2kV 0.15uF Cornell Dubilier pulse capacitors in series to get 8kv or so and then put that in parallel with the main electrolytic cap bank.

That would give your electrolytic cap bank some awesome peak current, since the 942C cornell caps have ~5mohm ESR and can supply crazy current continuously without failure.

We use 942C in tesla coils, they're awesome for high current low esr.
 The 940C is easier/cheaper to get and ok for your application though.

[T4P]

Oh, sorry didn't notice the SSB/CW written

[Topband RF]

We have the base for the amp already, a Henry 2000D RF generator, it's a fairly well documented base for HF band linear conversion. So the tube is already set in stone. Typo on the first post, will amend should be 3CX3000A7  Oil filled are very had to find in suitable values here in the UK, and as they are so big and heavy they are costly to ship.

Check this out - http://www.ebay.com/itm/NOS-DUBILIER-OIL-FILLED-CAPACITOR-B-1440-3300-5000-VOLTS-0-5mF-FEA-XK-TUBE-AMP-/170895432604?pt=Vintage_Electronics_R2&hash=item27ca295b9c

46 bux to your door...Its listed as NOS there was a guy in Russia selling HV NOS oil filled caps.. I'll see if I can find out if he has em.

Good luck on ur amp.. if you get it together..please post picks

[Topband RF]

Here is another one..
http://www.nd2x.net/ur4ll.html#caps1
100microFarad, 5KV For Europe only; at between 13 and 14 kg, it's too heavy (expensive) to ship outside Eu.

[NiHaoMike]

Use a switching supply to allow the use of smaller caps.