The AM Forum

I wonder if someone can set me straight on the reason for a current limiting resistor in the B+ of a HV supply?

For example:

I currently have a 100W 10 ohm power resistor in series with my 5KV supply.  At first I thought it would help to limit current in case of a direct short to ground. But doing the calculations, it seems senseless...

At 5KV /  10 ohms = 500 amps. (I= E/R)    500 amps wud be the theoretical current of a direct short, assuming the supply could deliver that much current.  (2.5 megawatts!) Since the supply would sag and blow a breaker way before hitting even 2 amps, the resistor seems like a waste of time.  It wud help a tiny bit, but not much.

In addition, it's a power hog -  when it is normally passing 1A of current, this is 10 watts of wasted power. (I2 * R)

Now if I increase this resistor to 100 ohms, this is STILL 50 amps of limited current during a direct short...  still fantasy.  (250 kw) It would eat up 100 watts of power when drawing 1A.

In the case of using a 1K resistor, this  would limit it to 10A, (50kw) but the normal supply regulation would be terrible, as a result.

What am I missing here? It seems like a big waste of power and effort for nothing.  In this situation, I would think the power transformer's internal resistance  and house wiring is about the only MAJOR thing slowing the current flow during a shorted condition.


** Update:  OR, is this resistor meant to be like a fuse, say a few watts rating and blow apart when the shorted current hits a certain point??
If so, I shouldn't be using big power resistors in there like in the primary for a step start circuit, rather small-sized ones..

** Maybe I'd be better off just sticking with my fine wire fuse. This consists of a few strands of the fine braid wire from RG-213 coax - bridged across two stand off insulators in the HV lead. When a short occurs, it vaporizes into a fast arc.  (The stand-offs are spaced about 2" apart)


73,
Tom, K1JJ

It is called the Rocky Point resistor. The caps in your power supply will deliver a lot of current if shorted. Also a good idea to limit the wire size off the HV supply. It is used to protect the tube. (grids I suspect)

Under short circuit conditions, a capacitor can deliver a nearly infinite amount of current for a short period of time.  Without any limiting to the peak current, a cap can destroy itself or other components, especially if you get into a rat-a-tat repetitive arc condition.  This current has little to do with the average current the supply is capable of delivering under normal loads. (At least this is the reason we put resistors in series with the HV caps for the HV power supplies we make at my work.)

Think about a xenon flash circuit.  The charging circuit is very low current, but once the capacitor is charged, it can deliver a tremendous amount of current to the flash tube for a short period of time.

So why not fuse it with a small fusing wire? Caps will only discharge partially by the time the wire fizzles.

Thanks Guys -

I had just updated the post above and then saw the responses. Look like I was partially wrong and partially right... :-)

Right now there's a fine wire from the diode stack to the caps to protect the diodes.  It has worked so far and haven't lost a stack in years. But based on what you've said, I will add another fine wire "fuse" after the HV caps to the rig. This will protect the cap and rig/tubes somewhat.


I'm gonna install the new fuse now. (2" spaced stand-offs with 2 strands of fine wire from coax braid as the fuse in the HV lead to the rig)



BTW, Frank, is the, "Rocky Point Resistor"  supposed to blow up like a fuse?

T

Tom,
I would leave the resistor in and add the fuse. AH yes the time the cat shorted the PDM supply....Fuses blow slower than grid wires

OK will leave the 10 ohm resistor in for the cap protection and add a fine wire fuse to the rig.

TNX lots!!

T

Well, it needs to be large enough to handle the steady state dissipation so it can't be too small.  Also I would think it would be preferable to replace a fuse than a resistor, so you would want it to limit the peak current but not be so small that it would be the first thing to fail under a short.  Also 1A may be the normal steady state current, but what is the worst case steady state current it might see?

OK, John -

So please clarify...  is the 10 ohm, 100w doing anything if I keep it in series in addition to adding a fine wire fuse, based on the peak currents I calculated during shorts? Or does the fuse pretty much take it's place and do a better job on the output?

Or as Frank suggested, use both the resistor and fuse on the output?  The resistor hogs power, so I just assume to get rid of it if I can.

T

a fuse is very slow to open.....

It's all about timing.  A resistor is infinitely fast at limiting the current, while a fuse does not limit current at all, but opens after some time delay that is a function of the amount of overload and the characteristics of the fuse element.  I think that's why Frank suggested to use both.

Got it.

Well, I guess because a capacitor could deliver almost infinite current, maybe the resistor limiting it to even 500 A is a good thing.  I'll add both.

Just wanted to understand the reasons why, before proceeding.

Thanks again for the fast answers, guys!

T

my reading of it was that you want a limit to the absolute peak joule limit that the filters can deliver. very small slice of time.

Yep, you got it concisely there Derb. Those caps turned out to be the bad MoFo to watch out for.  I just rewired the supply to give a cap rating of 140 uf at 10KV for lots of MaxHeadroom.  That's a lot of jingling jewels.

That's why I'm running scared with these precautions... :-)

T

Jingling joules fer sure there, Vu.  A mushroom cloud from Marlboro when those caps let go.

Rich

I put in some NOS Sprague "blue devils" in the Gonsets. wirewounds.  Never use those "cement block" resistors - if you get a discharge they will blow sandy looking crap into every nook and cranny in your chassis.

I wonder if there is modeling software out there to take a given voltage, current, and supply joule capacity that would tell you the optimum resistor size and wattage to limit the timeframe of the peak discharge the best? 

Isn't there a resistance and inductance associated with filter caps that would have a small-but-limiting effect on current upon initiation of the dead short? Wouldn't this keep the current down until the fuse blew?

The resistor is recommended by Eimac, RCA, and ? if I remember correctly. A fuse wire is too slow to save a grid wire inside a tube in case of an arcover. Remember that you have to limit the energy deposited to 5-10 J if you want the small wire grids to remain intact. So, to do this, the arc will have maybe 50 volts across it, then the remaining instantaneous voltage is your power supply - 50, for example, 2350 volts. This is across 10 ohms, leaving 235 instantaneous amps,  until the cap is drained and the breaker kicks or blows a fuse. Besides, if you try and fuse 235 amps at that voltage, you get a plasma across the fuse, and it will continue to burn there until your grid wire is toast first. Without the resistor, take 2350 and divide by .1 or whatever the internal wiring impedance is. You will see that you might get thousands of amps for a some microseconds. 30 AWG wire melts with about 1500 amps through it, for a few hundred uS.
I have been testing this very thing at work few weeks ago, with a big crowbar circuit. My tube requires that a 29 AWG wire be dropped across the 225 uF power of capacitors, charged to 32kV DC, and not melt. I divert the fault current to a parallel device, in this case a large mercury ignitron. I have about 1.5 ohms in series, with lots of cooling air. The wire survives beautifully. So my grid wires will survive if the tube decides to burp, outgas, from Rocky Point effect. In smaller plate supplies, a series R is used without the shunt crowbar device sometimes, as ham equipment usually does.

Interesting thread.

Say, a WW resistor also acts as an inductor, even if only a few micro-Henries.  That will affect the leading and trailing edges of a squarewave, in effect, soften your very sudden short's high freq. components in addition to instantaneously limiting the peak surge.

The old designers were not only very technically smart, they were intuitively very canny. ;D

Yup E=L di/dt

The WW resistor acts as an inductor and limits the di/dt. To an infinite step in current, the resistor looks like an infinite impedance.

A fuse is not nearly as safe as you think- Look at the curve on a small standard fuse. Also, look at the interupting current rating. A fuse that can interupt a large current pulse at 5 KV is very expensive.

A 5 amp fuse does not "blow" at 5 amps. Most simple fuses can carry twice the rating for many  seconds.

A small 250 volt fuse will do nothing but explode into an arc at 5 KV.

Proper fusing is a complicated process and involves knowing the time damage curve on what you are protecting, the available fault current, and the voltage across the fuse at the time of short circuit.

Good info.

I like the idea of the wire wound and the di/dt  opposing effect. I'm using two large 100w 5 ohm wire wounds in series, so that will help.

For "fuses,"  I use two stand-off ceramic insulators spaced about 2" apart. I bridge them with 2 fine wires from the braid of RG-213. I've used this for year between the diode stack and the filter capacitors to protect the diodes.   I used to lose a diode stack about once a year due to foul ups on my part.  But since adding the wire fuse, I've not lost a stack in ten years.

The fuse disintegrates and I see some black residue on the standoffs showing their was an arc. But the diodes survive every time.  This may not apply for the capacitor bank when discharging into the rig, as has been discussed.  But for now, I have an additional 10 ohms of wire wound resistance and another 2" wire fuse in line to the rig.    At 140 uf / 10KV rating, (at 5kv) those caps can hold some nasty potential.

Later -

T

I always wondered what YupE meant.  Now it's clear, although I've been spelling it wrong all these years.

Not just potential (V), but more importantly, a stored energy of 1750 J (1/2 CV^2)

By contrast, a 1.2 KW, 5kV supply using high frequency inverter technology we would manufacture at my work, has a stored energy at the output of about 1.2 J.

That's some high voltage and you have some Joules.

Mike WU2D

Yup .... big stored energy and arc over can mean big BOOM and shrapnel ....duck and cover...beefus

Yeah, I often wondered if these caps can actually blow up like a grenade and what the damage wud be.  1750 Joules sounds big, but I have no practical reference to gauge it. How many HP = a joule or is there some other real whirl comparison analogy?  ;D

I'm using one 70 ufd @ 10KV cap that is about 24" tall and weighs maybe 90 pounds. In parallel is an additional pair of  35 ufd @ 10kv. The 35uf caps are about half the size of the 70uf ones.  = 140uf at 10kv.


I wonder what could happen to make them explode? Perhaps a typical internal arc over failure - or maybe a direct short that caused it to overheat inside in less than a second?

I've not heard of it happening much in big BC transmitters.  Of course, these are high quality HV caps vs: electrolytic types that are always puking out their guts...

T

Another term for joule is Watt Seconds so 1750 watts for 1 second, 17,500 watts for 1. seconds then there is 175,000 watts for 10 ms. etc It all depends on the ESL / ESR of the cap and series reisitance. So you can quickly see how 10 ohms can help limit current.
When we do lightning testing a Waveform 5 pulse can be as high as 2000 volts at 2000 amps. Storing that kind of energy in a bank of caps will put fear into you. 

Frank, I think you meant 17,500 watts for 0.1 seconds  :P

One calorie is equivalent to exactly 4.184 J.  A calorie is the amount of heat required to raise the temperature of 1 g of water at 14.5°C to 15.5°C.

My college chemistry professor used to say:  "We use calories here...Joules are too expensive."

yup

Interesting.

After doing some reading, I see the quantity of watts is directly related to the speed at which it is delivered. It's like the force= mass X acceleration formula. Here's a link showing some every day examples:

http://home.earthlink.net/~jimlux/energies.htm

Stated in watts, notice that 1 pound of high explosives can put out 240 gigawatts of power cuz some explode in 4 microseconds. Whereas, a cup of gasoline explodes in milliseconds, thus spreading it over time and much lower watts delivered.

A nuclear bomb explodes in 1 microsecond and goes off scale in the example... :-)

It's also interesting that a 9V battery has a total capacity of 16,000 joules.

 

T

Hey yo! Watts are the product of energy and time (i.e. one watt is one joule per second) not what your Dosey meter sez on your Texas Star.   ;)

How is the cat?

He sed Dosy....

Don'tcha know us big strappuhz measure in pidgeons now? :)

--Shane

Used to be that a standard size capacitor for high energy research was a 3 kJ can. These were about 2 1/2 feet high, 8 inches wide, and 15-16 inches deep. I am at home, so these are from memory. They are 45 kV DC rating, and about 5 uF. Thats about 5 kJ. We put 50 in parallel. Failure mode is an internal pack shorting, as GE didn't make 45 kV individual caps, but stacked series of 10-15 kV capacitors inside the can. When they failed, (we lost quite a few in 2001) the metal can would peel back like a banana, and the can buldged. The insulator was broken off, and flames lasted for about 30 minutes. Not to mention the mess of all the oil that is squirted all over the vault.
Shrapnel for sure. We have steel capacitor rooms to contain  it.

You don't want this to happen at home!

Ken W2DTC uses a "glitch" HV fuse.

http://w2dtc.com/w2dtc-hv-power-supply/2005-0304-ps-hv-fuse.jpg

The idea sounds good to me.

Al VTP

Photos attached of what happens when a 5 kJ capacitor ruptures...Note that the photo of the bank of caps, is only half of them. Identical wall full on the right of the camera. You can see the series wirewound resistor beside one cap. The resistor has no wire (the emperor has no clothes!). The mechanical force of the magnetic field caused the wire to bunch up and rip off the core, despite being epoxied there.

One more capacitor rupture. This one on top - where flames lept out.

Just beautiful!  Makes me wonder why we do what we do :)

--Shane

Those are pretty revealing pics, John!  First I've seen of blown HV caps like that.

My biggest cap looks close to that size.

Do you know what caused the flameout?  ie, What happened in the circuit to produce that failure?  - or did the cap fail as a normal bell curve percentage of using a large quantity of caps? (random)

T

I'd vote for Lentz's law inside the caps ... rapid energy discharge from a capacitor can form large mechanical moments and internal stress ... somewhat similar to Derb's recent oriental food experience... caps used in this form of testing are generally large for their rated capacity / voltage ... I have seen large thick cables 'jump' with this ...73 ...John

140 uF @ 10 kV? Why so much? Do you really need all that?

Hi Thom,

I did have the caps lying around doing nothin... ;D

Well, first, the voltage rating (5kv extra overhead with a 5kv supply) is always a good thang.

The capacity was needed cuz my power transformer doesn't have the greatest regulation. The core size is probably too small for the job. If I throw a full carrier, the HV sags too much when I had only 35 ufd.  But now it sags a lot less with 140uf -  but the real important thing is now I see NO fluctuation on big AM voice peaks. The dynamic regulation has improved markedly with the X4 capacitance increase.  Same thing on ssb, so less distortion.

Some day I'll get a better HV transformer, but for now this will do.

BTW, Chuck has about 200 ufd in his HV supply, using a quality Peter Dahl transformer, so I'm in good company, caw mawn.

T

Yes.  Dynamic regulation is a good thing, when you can get it.

Hence the reason most don't use a doubler in AM service.


--Shane

The caps failed about a year after their full warranty ran out. (10 years of operation, ran half time, so about 20 years old). It is a complicated situation that caused it. They are charged by a three phase rectitifier, and run 88 kV total, or 44 kV per cap, series parallel arrangement with grading resistors. The discharging current is not to zero, we have a few kV of sawtooth ripple from it. They are discharged at anywhere from 20 to 120 pulses per second. To save money, it was decided to discharge them with less pulses, but not evenly spaced out. (don't ask, too complicated to explain further). So here is a train of pulses, spaced in a stuttering rate of discharge. I measured the ripple, it was higher, and the FFT of the ripple current was much higher frequency. GE rates their caps with a certain amount of harmonics in the ripple current. We probably exceeded that with old caps. They begin to heat from dielectric loss. Eventually one pack shorted. They kept going, until the second one shorted inside, then it was boom. And all the other capacitors nearby discharged their energy into the short, as the5 ohm wirewound resistors in series with each just turned into plasma and shorted out for a moment.

 

Gotcha.

I need to pull the HV supply on the Junkyard Dawg and shove in this 15 uF cap I've got laying around for the same reason. I blew (well, flashed over, anyway) a 10 uF cap ages ago and just pulled it. I think another 10 has opened up and left me with 4 uF grand total, because I'm starting to get noticable sags on modulation peaks that weren't there when I pulled the shorted 10.

FWIW, the power supply in my avatar is 22.5 uF @ 10 kV, about 7 or 8 H, and a 25 kVA pole pig to deliver 2.5 to 3 amps at about 6400 volts. You might want to try that. Sounds about right for your needs.  ;)


Yeah, but that's Chuck. Chuck doesn't do anything small. I delivered the tube he's currently running (or would be if he'd ever get on the air, at least on 160). I used to swing by a few times a year when I was working down there. That is quite a few pounds per square inch of rig.

I miss the days of you, him, and several other now-scarce tall ships getting on and peckermatcing.  :(


Yikes! I wouldn't want to be standing within a hundred feet of that. Your ears would probably ring for a month!

Were these PCB caps, or did you have giant rooster-tails of mineral oil simultaneously boiling and burning?

--Thom
Keep Away One Zorched Grid Capacitor

Any of you guys get to see IEEE Spectrum  3/09 check out the 2.4 million volt Marx generator at LTI in Pittsfield. The have set it off 4 times for me.
Strap a 10 spark. There is a big cap about every 18 inches of height on the charge bank. The charge supply is 100,000 volts. The last two they took pictures for me.

No one was in the building when our caps ruptured. So don't know if they spouted like an oil well or not. But oil was all over every inch of the room, it takes about 12 hours with 3 people or more to clean it, a very disgusting job. Ruins clothing. Not PCB oil, these are vintage 1980s GE capacitors.

I want to work where you work.

The only time I ever blew up an oil cap was once when I dead shorted a 10 mfd 4 kv oil capacitor I was using in a 2500 volt supply, to discharge it.  Nothing big happened; it just sparked as expected.  But next time, the power supply output voltage was way low, and I finally figured out that the capacitor was completely open-circuited.  I pried the top off the case, and saw that the wire leading from the guts of the capacitor up to the terminals was small, only about #20 solid copper, spot welded to the bottom side of the terminal.  The wire had melted in two right at one of the welds.  I probably could have repaired it, but it wasn't worth the trouble , with oil all over the place and no easy way to re-seal the can.  After that, I always use a hefty resistor in series with a shorting stick.

Perhaps a good design would be to make a shorting stick with two tips, one a dead short and the other through the resistor.  You would first touch the resistor to discharge the capacitor, and then do the dead short just to make sure the cap is fully discharged.

I'd bet I could get quite a bang out of those 400 mfd 2500 volt caps I picked up.  The case is made of 1/8" steel.  The fellow who gave them to me said they were used for "energy storage" for some research project at Vanderbilt.

Is there any real difference between "energy storage" caps and standard oil caps??

I recently picked up a pair of 35mfd @ 4KV peak "energy storage" caps at a local fester for a very reasonable price. I had planned to keep them for spare filter caps for my 4X1 rig. (3400v plate supply)

They are not much, if any bigger than normal oil caps of that value. I just assumed that they are just standard oil caps made a little more robust to handle the repetetive charge / discharge cycles.

                                                              The Slab Bacon

John,

With the price of CHEAP IP based cameras today, do you guys have any video monitoring of the rooms now?  Would seem like a good idea, if only for "damage review"...  As well as some DAMN fine youtube video footage :)

--Shane

Slab,

The energy storage caps (from what I've been told / experienced) are for charging and QUICK discharging.  Whereas standard electrolytics are NOT designed for quick discharge.

Photoflash caps are storage type.

I've used them in smaller power supplies, and they work...  Bought some 450 or 550 volt types off EBay, and they worked FB in a sweep tube amplifier, no noticeable difference (other than larger PEP capabilities, the power supply went from about 45 uF to 150 uf or so of total Cfilter)..

I have heard from others that they didn't work so well, and eventually went south.  So, long term, they might not like being charged all the time (used in power supply work)...  But, my experience has shown them to work fine business.

--Shane

Shane,
         that is pretty mouch what i have found so far. It should be ok to use storage caps as filters, but not to use filters for storage. as they wont hold up to the heavy charge / discharge cycles. Other than that, no one has been able to give me any hard answers as to what the difference is.
 
                                                            the Slab Bacon

I find it surprising that so few commercial ham amps include this surge suppressor resistor as it would save a lot of expensive items. The NCL-2000 of 1963 is the first that I am aware of and I was fortunate enough to be an engineering tech on that project and learned more than a few things.

The purpose of the vitreous enamel WW is to limit the surge current until a safety device opens. It is not there to be sacrificial and if it fails it wasnt big enough. I use 25 Ohms @ 50W in the typical 1200-1500W out ham amp and all have survived a bout of tube flatulence.

Carl
KM1H

Thanks for all the great info on capacitor failures, guys!

Don, I like your idea of having TWO paths on the shorting stick. One through a resistor for the first pop and the next a direct short to be drain the cap completely.  This eases the strain on the caps by using a current limiting resistor first - then they are drained completely using the direct short second. I would then leave the direc short on while working on the rig.

Last night I was working on my QRO 40M dedicated amp and it shorted the HV supply. The limiting resistors did their job and the wire fuses evaporated... :-)     I took out the 20KV  JJ homebrew hi-pot tester and found the problem was the plate coupling cap. It was mounted too close to a grounded bracket and arced through itself.  The evidence was only a TINY spot on the cap - almost impossible to find without the hi-pot tester.  The tester uses a Variac controlled neon sign xfmr that goes thru a 1/2 wave rectifier stack, into a .1 20kv cap...then thru about 20megs, then thru a  meter.

BTW, I have used  "PhotoFlash" caps (55 uf @ 5kv types) in many of HV supplies and love 'em.... and still do.  Never had a failure yet over 20+ years. 


T 

If you're talking about an independent unit for discharging loose capcitors, maybe. If you're talking about your rig's Jesus Stick, absolutely not!

A transmitter's shorting stick should be a shorting stick, period. Nothing more, nothing less. Straight path to chassis ground. You do not want to add any parts which may fail. That shorting stick is your last stop before The Great Beyond. It's not just your life at stake, either. Someone else may wind up with your rig after you die, and they'll need to depend on that shorting stick, too.

If the choice is blow a capacitor or die, hell... I'd rather blow a cap.

Now, there's a way to do what Don described within a power supply. Harris did just that in the power supply in my avatar (FM-25K).

Here's how they did it:

Mounted on one wall of the cabinet are two electrically-identical solenoids. Their field coils are tied to the 120V cabinet interlock. These solenoids are arranged vertically such that they form normally-closed switches (they have 1" dia. brass slugs on the end, as do the stationary contacts below).

Of course, "normally closed" means closed when power is cut or interlock is open. The rest of the time (from Filament On to Filament Off + 2 minutes), these two solenoids are held open.

The stationary contacts are perfectly level with respect to one another. One contact leads to one side of the cap bank through a 500 ohm (IIRC) carborundum resistor, the other stationary contact is a straight shot to the same spot.

The (non-stationary) solenoid contacts are both tied directly to the opposite side of the capacitor bank.

All seems pretty obvious at this point, right? But, you're probably asking yourself where they reliably derive the time delay between the resistive short and the dead short?

It's so simple it's brilliant: they simply mounted the dead-short solenoid coil 1" higher than the resistive-short solenoid coil! So open the interlock (or lose the mains), both solenoids drop at equal velocity, the resistive-short contact gets there first, followed by the dead-short contact a few tenths of a second later. As long as you pay your gravity bill, this step-stop configuration doesn't have too many things that can go wrong with it!

We've dumped the mains on this transmitter with the plate on at 11 kilowatts output. No bangs, no flashes, not even a slight "snap". Just the "chuh-chunk" of the solenoids falling, and the HV meter instantly falls to zero.

I have close to 100 pictures of this supply, not sure I took any of the step-stop. I can certainly take one the next time that supply is opened up (probably next week, since we need to take it off the air for some finishing touches very soon).

I like the "What Would The Egyptians, Mayans, Incans, or Aztecs Do" approach to these things. You're far better off putting your life in the hands of the laws of physics than in the hands of human engineering!

...for what any of that's worth.

--Thom
Kilovolts Alter One Zorched Gate Crasher

The typical storage cap is not intended to remain at its rated voltage for very long and might break down if done that way. 75% for continuous filter service is what I was told by a laser technician, but with so many kinds of caps and as many educated opinions, only your datasheet vendor knows for sure.