Balance resistors in series filtering

[azatplayer]

WHat determines the right value here?
Typically see either 100k or 220k used. SOmetimes 270k.
Im about to build a Naylor SD60, just love what ive heard of its tones, seems anyone that owns one, rates them highly.
Looking at the filtering vs the B+ they run, around 500V, and have two choices.
Go for spec which was using 300V caps in series first and second nodes, they used Illinois, i would go FnT.
OR use say, Nichicon or Panasonic radials. Small footprint and great spec.
Question is more for the preamp caps. Theres just no 500V small radials, 22uf value.
So, i m thinking using series in that end as well, pair of 47uf radials each stage, and use balance resistors.
Im thinking at this point, its only the preamp current, so figure on 1 watt balancers. But wondering how critical the resistor value is?
100K ok? Or should i go higher?

[azatplayer]

I did my express using small radials, also did my JTM50 using them. Oh and my AC15.. I guess i dig the look and theyre cheap and high spec and take up little space.

[VacuumVoodoo]

Look in datasheet for the caps you intend to use. Check specs for leakage current.
Calculate resistor so the current through it is at least 5x spec'd leakage (voltage across cap/leakage current).
Calculate resistor power rating to 3x power dissipated in it.
You may discover it's often not what is commonly found in amps.

[surfsup]

VV, I am looking at a datasheet for a 47uF/630V cap where leakage is given as = 3*SQRT(C*V) uA at 5 mins

I assume a 47uF means,

3sqrt(0.000047*630) = 0.5mA

assuming 500V at the node and two caps in series, 250v across each cap, 250/(0.5mA*5)=100k/1W resistor

What does the @ 5 minutes mean? Did I follow your advice correctly? TIA

[Structo]

Surf, I think you are a decimal point off on the 47uF. Micro meaning millionth.

It should be .00000047, right?

I know there is math to figure it out but if you look at existing amps with balance resistors, they are typically 220K-270K.

Not only do they balance the voltage between the two caps, they also act as bleeder resistors when you power down, draining the filter caps.

47000000

[roberto]

It should be .00000047, right?

No, this is 470nF. Surf is right.

[VacuumVoodoo]

VV, I am looking at a datasheet for a 47uF/630V cap where leakage is given as = 3*SQRT(C*V) uA at 5 mins

I assume a 47uF means,

3sqrt(0.000047*630) = 0.5mA

assuming 500V at the node and two caps in series, 250v across each cap, 250/(0.5mA*5)=100k/1W resistor

What does the @ 5 minutes mean? Did I follow your advice correctly? TIA

Use actual voltage that will be present on the cap, not it's rated working voltage in the first formula. In this case 250 instead of 630.
5mins means 5 minutes after application of voltage.
So, I(lk)=3sqrt(47e^-6*250)=325uA
This will give you R=~150k which means anything (within reason) up to 150k will be fine.
Power dissipation will ca 0.4W so at least 1W resistor, 2W if you want to be really conservative.

What capacitor brand/type is it?

[Structo]

It should be .00000047, right?

No, this is 470nF. Surf is right.

Why isn't 47uF equal to 47 millionths of a Farad?
10-6 is micro

47x10-6 or is it 47x10-7?

Now I'm confused

[roberto]

Because 1µ is .000001 then you have to multiply by 47 and it's .000047.

You wrote 47 after six zeros.

[dcribbs1412]

Question is more for the preamp caps. Theres just no 500V small radials, 22uf value.
So, i m thinking using series in that end as well, pair of 47uf radials each stage, and use balance resistors.

Cool idea
are you going use one long board for PS and preamp(like Naylor) or build a separate PS board?

Darin

[Firestorm]

Any special considerations when the series caps are used in an "economy" power supply (where the secondary center tap is connected between the two caps)? Marshall used no resistors; Fender sometimes used 100K, sometimes 39K.

[roberto]

Well, that way you have your PT's secondary windings that discharge those supply caps.

[Gaz]

Alex,

Thanks for sharing that formula - really useful, if even to check the traditional values commonly used with new caps.

Azatplayer, of course you don't need 500vdc caps after the PI, I'd just use inexpensive 22uf/450v radial Nichicons. I don't think the short time the amp is ever unloaded will hurt the caps.

And if you're going with the high oringinal B+, I'd stick with the stock Sovtek "5881"WXTs, which can handle the punishment. Same reason Soldano uses them.

Can't wait to see the amp!

[rdjones]

What does the @ 5 minutes mean? Did I follow your advice correctly? TIA

Some safety codes require that high voltage bleed off to a "safe" value (10% ?) within 5 min.

RedDog Steve

[surfsup]

Ok yea i get 154k...so can you explain the reason for the 5 times multiplier for the current?

Structo, just move the decimal over 6 places, so 47uF w the decimal after the 7 means 4 zeros...

EDIT rdjones, thx