HV Cap PCB

[W7SOE]

Hi all,
   I am going to do a PCB for a HV cap for my ART-13 power supply.  (picture attached)

I thought it would be nice to have a neon lamp on the front panel, this way I can see when there is HV present, monitor the bleed resistor health, etc.

The PCB has a current limiting resistor (~1 Meg) for the lamp.  Would it be considered safe to bring HV to a panel lamp holder like this?

Thanks

Rich

[steve_qix]

I suppose you could stick a zener across the lamp, in case it fails (open).. This would limit the HV (dropped across the resistor) to the zener voltage - and of course if the lamp fails open AND the resistor shorts, you're really protected 

But, for all practical purposes, it's probably fine     I've certainly done this with panel meters reading high voltages through a dropping resistor.

Regards,

Steve

[W7SOE]

Perhaps I would be better of with an actual panel meter....  The PS will have a variac, it would be nice to know the voltage.


Rich

[W1AEX]

Somewhere along the way in my adventures with high voltage, someone suggested that it was wiser/safer to use a string of resistors (example: 5 x 200k rather than a single 1 meg) to add more isolation from the supply to the front panel metering, or in this case neon indicator. Seems like a good idea. Someone might comment if it really is valid.

Looks like a neatly done project.

Rob W1AEX

[W7SOE]

OK, changed my mind about the neon bulb, I will use a meter.  I have added a resistor string to the PCB..

Rich

[The Slab Bacon]

Rich,
I think you may have hit the point of way overkill?? The pcb is a nice layout, but......
Why so many capacitors? If you are using 450v caps 4 or 5 of them would give you more than a comfortable cushion @ 1500v. The same with the resistors, I agree with using a stack instead of 1 single, but just about all common resistors whether they are carbon or metal film are good for 500v each. Dont beat up on yourself by making it harder than you have to. 1500v isnt anything to sneeze at, but it requires a lot less design cushion than say, 4000v. Remember to always try to keep it a simple as you can.
 
                                                       the Slab Bacon

[KA1ZGC]

What Frank said, and this too:

If you're using electrolytics, you don't want to undervolt them too much. With most electrolytics, the electrolyte doesn't form until there's a certain threshold voltage applied. Until then, you have what looks like a dead short (or darn close to one).

If you string too many caps together in series, you'll bring up the collective electrolyte-forming threshold dangerously close to your operating voltage.

Undervoltage is not quite as dangerous as overvoltage, but can still ruin your day.

20-25% safety margin is plenty. They lowball the voltage ratings on a lot of modern caps, anyway (to avoid liability issues).

FWIW.

--Thom
Keep Away One Zorched Germanium Capacitor

[w8khk]

Another thing I learned when series connecting electrolytics - To equalize the voltage distribution, use carbon comp or film resistors in parallel with each capacitor, and size these resistors conservatively.  They will outlast the capacitors.  Separately, use a string of wirewound resistors as necessary for the bleeder.  DO NOT connect the individual wirewound resistors in parallel with the electrolytics. 

Three advantages using this method:

1) If a wirewound resistor zorches, you will lose the main bleeder, but you will not unbalance the voltage distribution on the electrolytics.

2) Heat from the bleeder string is not radiated directly into the capacitors.  They live longer if they run cooler.

3) If you lose the bleeder string, the equalizing resistors will still bleed the caps, albeit a lot more slowly.  The meter will indicate this situation.

[steve_qix]

Another thing I learned when series connecting electrolytics - To equalize the voltage distribution, use carbon comp or film resistors in parallel with each capacitor, and size these resistors conservatively.  They will outlast the capacitors.  Separately, use a string of wirewound resistors as necessary for the bleeder.  DO NOT connect the individual wirewound resistors in parallel with the electrolytics. 

This is a fantastic recommendation !!  Also, do run as close to about 80% of the rated voltage on the electrolytics as you can.  They will "form" much better, and achieve their full value.

[W7SOE]

Another thing I learned when series connecting electrolytics - To equalize the voltage distribution, use carbon comp or film resistors in parallel with each capacitor, and size these resistors conservatively.  They will outlast the capacitors.  Separately, use a string of wirewound resistors as necessary for the bleeder.  DO NOT connect the individual wirewound resistors in parallel with the electrolytics. 

Three advantages using this method:

1) If a wirewound resistor zorches, you will lose the main bleeder, but you will not unbalance the voltage distribution on the electrolytics.

2) Heat from the bleeder string is not radiated directly into the capacitors.  They live longer if they run cooler.

3) If you lose the bleeder string, the equalizing resistors will still bleed the caps, albeit a lot more slowly.  The meter will indicate this situation.

So you are saying to run an individual resistor in parallel with each cap and then one large bleeder across the whole mess (Like a tradional bleeder)?

Iwas trying to make a "general purpose" cap board, that is why there are so many.  Any un-needed caps can simply be jumpered.

I happen to have a bag of 220uF 250V caps, that is why I have nine on there.  I could lower the number which would lower the voltage rating and raise the capacitance.  I am not familiar with modern caps "forming", is this a high voltage phenomenon?

Rich

[KA1ZGC]

I am not familiar with modern caps "forming", is this a high voltage phenomenon?

Nope, it's an electrically-induced chemical phenomenon that makes an electrolytic cap work in the first place.

It's not a question of old vs. new, or high- vs. low-voltage, it's how electrolytics work. The electrolyte does not form without voltage present.

[WA1GFZ]

Most resistors are only rated for a few hundred volts unless you use large wire wounds. Why not put the neon across the cap at the bottom of the stack. I prefer a volt meter with a string of series resistor. You can monitor the discharge time to see if anything is going away.
Look at the possible line transient multiplied in the transformer and quickly learn you want plenty of voltage margin. you may not fully form it but you also won't deform it on a line transient.

[w8khk]

So you are saying to run an individual resistor in parallel with each cap and then one large bleeder across the whole mess (Like a tradional bleeder)?

Exactly.  Using the individual carbon composition or metal film resistors in parallel with each capacitor allows you to equalize the voltage across each capacitor during the forming process, as well as the subsequent operation.  Choose a resistor which will have sufficient current to equalize the voltage based on the leakage current.  (I typically use a 100K 2 watt resistor across each 450 volt cap in a series string.) 

Since wire wound resistors are less reliable, they should not be used to equalize the voltage on each of the electrolytics; instead they are better suited to the bleeder function, dissipating the required power to stabilize the output voltage, especially in the case of choke input power supplies.  In summary, the equalizing resistors and the bleeder resistors are totally separate in function.  The failure of a bleeder in a separate series string will NOT cause unequal distribution of voltage on the series capacitor bank.

Iwas trying to make a "general purpose" cap board, that is why there are so many.  Any un-needed caps can simply be jumpered.

Is the general purpose board to be moved from one project to another?  Or just reduce the design time and board artwork for multiple projects?  A general purpose board could be populated with the required quantity of components to meet the voltage needs of the project.  If fully populated, a portion of the string could be used when the max voltage is not present.  If moved to a different project, be sure to form the unused caps in the bank to assure equal voltage distribution when moving it to a different project.

I happen to have a bag of 220uF 250V caps, that is why I have nine on there.  I could lower the number which would lower the voltage rating and raise the capacitance.  I am not familiar with modern caps "forming", is this a high voltage phenomenon?

Rich

All electrolytics need to be "formed", whether ancient or modern.  Capacitors that are used frequently form quickly, and no manual precautions need to be taken; those that have been stored for a long time form more slowly.  This is why it is advisable to form NOS caps using a series resistance and limited voltage supply prior to installation in a circuit, thus limiting the leakage current during the initial forming process.  The following link provides a rather detailed explanation of the function of the electrolytic cap, including the forming process:

http://electrochem.cwru.edu/ed/encycl/art-c04-electr-cap.htm

Regarding the need to build with enough voltage tolerance for surges and transients, this is a factor that needs to be balanced with component cost and properly forming the capacitors; 80% of working voltage is a good rule of thumb.  Other methods of surge protection at the primary of the HV transformer, as well as some series resistance added between the transformer secondary and the rectifier are also prudent.  If lightning strikes, I would much rather lose a small series resistor instead of a capacitor bank.

[WA1GFZ]

Sorry Rich I'll take a wire wound over a film any day for tuff resistor in a power supply application. A properly derated WW will outlast your children.
Carbon comps increase in value as time goes on.

[N3DRB The Derb]

the most important thing about equalizing resistors is that they be matched pretty closely in value so you don't have any hot spots across the caps in the chain. you want the voltage to climb up and down the caps in equal measures across the entire string from B+ to b- ( or gnd as the case may be)

the gonset amps use 5 100uf @ 450 vdc in series with 2w 100K rx's across each cap for 20 uf total. When they went to the 572 B tubes, they added another 100K 2watter across each cap making 4 watt 50 K's across each cap. voltage 1500++ in each case. They're all carbon, but these days i'd use 1% or 2% MOF and oversize the current rating by 100%.

I really like the idea of a safety bleeder not in the supply string from B+ to B- or gnd, something like a 2W 2 meg..... by the time you get the case off, it will have done it's job regardless of what kind of fireworks happened in the rest of the power supply. 

[W7SOE]

the most important thing about equalizing resistors is that they be matched pretty closely in value so you don't have any hot spots across the caps in the chain. you want the voltage to climb up and down the caps in equal measures across the entire string from B+ to b- ( or gnd as the case may be)

the gonset amps use 5 100uf @ 450 vdc in series with 2w 100K rx's across each cap for 20 uf total. When they went to the 572 B tubes, they added another 100K 2watter across each cap making 4 watt 50 K's across each cap. voltage 1500++ in each case. They're all carbon, but these days i'd use 1% or 2% MOF and oversize the current rating by 100%.

I really like the idea of a safety bleeder not in the supply string from B+ to B- or gnd, something like a 2W 2 meg..... by the time you get the case off, it will have done it's job regardless of what kind of fireworks happened in the rest of the power supply. 

Good idea!  I found these:

http://www.ohmite.com/catalog/pdf/v_slimmox.pdf


2Meg 2 watt, rated to 15KV.

rich