What are good resistor values (1KV rated) for balancing HT rectifier diodes?

UsableThought

Member
Messages
1,650
Another silly PT/HT/rectifier question -

I'm building a fully fused HT rectifier for a singled-ended tube output stage. The center-tapped PT is a little elderly, having come from an old tube tape recorder. It tends toward high voltage rather than substantial current; thus, even with my bucking the mains voltage down to 105VAC or so, it still puts out about 300VAC on each of the secondaries prior to the SS rectifier. For the rectifier (remember this is two-phase, not bridge) I've got two UF4007 diodes on each secondary. I need two diodes and not just one on each leg of the secondary to get past the 1K Vrrm rating for a single diode (these have the same rating as a 1N4007). And from what I have read, it's best in such a case to balance the voltage drop across the diodes on each leg. Apparently the balancing can be done by adding either 1KV-rated ceramic caps or 1K-rated resistors in parallel.

In this case, because these are UF4007s, or "fast diodes," I don't want to go with small caps; I have been told by folks such as @trobbins that snubber caps don't go well with fast diodes & in fact would potentially create rather than reduce diode noise. So it seems like balancing resistors would be a better choice. My question is, what value resistors should I look for, assuming I can locate 1K rated resistors? I haven't found that info anywhere yet, though I'm still looking. I'd also be interested in learning the rationale behind the choice of value, since I'm sure there is one (I just don't know what it is yet).

I have no interest in debating fast diodes, so those who hate 'em please refrain! I'm looking only for informed answers to my question.
 
Last edited:

00JETT

Member
Messages
565
Are you sure you need them? Thousands of amps have been designed with diodes in series without balancing resistors. I have used UF4007s before and agree you don't want the snubber caps over them, but you really shouldn't need the resistors either.
 

UsableThought

Member
Messages
1,650
Are you sure you need them? Thousands of amps have been designed with diodes in series without balancing resistors. I have used UF4007s before and agree you don't want the snubber caps over them, but you really shouldn't need the resistors either.
I trust Merlin Blencowe pretty well, so if he suggests I need them, I'm inclined to go with his recommendation. I'm putting a lot of effort into this & a simple addition of resistors seems worth it to avoid shorting a diode & perhaps sustaining a whole lot more damage after that. Both secondaries are fused prior to the diodes, but even so it seems worth taking the caution.

This is the article where he mentions why you'd want them: http://www.valvewizard.co.uk/bridge.html
And here is a quote:

The diodes [in a two-phase rectifier] must also have a Reverse Repetative Maximum (Vrrm) rating that exceeds the peak-to-peak AC voltage (measured from one end to the cente tap), twice the value needed for a bridge rectifier. This is equal to 2.8 × Vrms. A 1N4007 is rated for 1000V. This corresponds to an AC voltage of 1000V/2.8 = 357Vrms. Knock off 10% to allow for variation in mains voltage, plus another 10% for transformer regulation and we are left with about 290Vrms. In other words, we shoudln't use the 1N4007 with anything more than a 290-0-290V transformer.

What if the transformer voltage is higher than this? The best option is to buy diodes with higher voltage ratings, but they're not nearly as common as the 1N4007. A classic alternative is to use two or more diode in series, so they share the burden. However, we must ensure that the voltage is shared (at least roughly) equally. This can be done by adding a 10nF to 100nF capacitor in parallel with each diode. High-value resistors could alternatively be used, but it is a lot easier to find 1kV-rated ceramic capacitors than 1kV-rated resistors.
I have seen frequent discussion of this & similar topics over on Music-Electronics-Forum, and am going to go search old threads over there. But if someone here knows about resistors rather than snubber caps, please share!

EDIT: I just saw he suggests that the best option is to get diodes with a higher rating - I can look into that also.
 

pdf64

Member
Messages
7,376
The method I've seen for balancing resistors on stacked ecaps is to look up the likely leakage current (of the diodes in this case), and use a resistor value that, given your reverse voltage, will pass about 10 times that (or more), ie so that the resistor current should dominate the reverse diode current, and so achieve roughly equal voltage across them.
But as you've only got 300-0-300Vac, and it's a 'best practice' rather than 'essential' thing, I think it would be reasonable not to bother.

Also perhaps consider whether it is feasible for the mains to rise a further 10% above its current level, which is already so high you need to wire a bucker permanently in place.

If you do go for it, consider the voltage rating of the resistors (limiting element voltage or similar); you may need 2 or more in series to keep within their rating.
 
Last edited:

UsableThought

Member
Messages
1,650
But as you've only got 300-0-300Vac, and it's a 'best practice' rather than 'essential' thing, I think it would be reasonable not to bother.
Yeah, I don't know. I guess since Merlin is writing a book he doesn't want to overtly suggest practices to his readers that might sometimes result in a "Boom." On the other hand he's usually pretty good at signaling nuances and here his delimiter for when to add caps or resistors is 290-0-290 - he's not waffling or hedging or shrugging verbally. I don't have the experience to know, myself.

Thanks for the method - I will try that out. If there is no downside to adding properly rated resistors (or several as needed) then it seems sensible to do so. Over on M-E-F.com, they have long and fervid discussions of this sort of thing - how to protect transformers, the impossibility of protecting transformers, gruesome failure scenarios, etc.
 
Last edited:

00JETT

Member
Messages
565
Yeah I remember reading that valve wizard article before, your diodes should be rated high enough to not worry about the resistors is all I thought. Best of luck
 

UsableThought

Member
Messages
1,650
Solution for a non problem? ;)
I dunno. One expert says something, another expert says something else; and here I am stuck in the middle. Merlin B. is generally helpful but is obviously not the mouth of God.

Meanwhile I'm waiting for those who are anti-balancing to explain why, other than "Well, they seem to do it that way & it doesn't blow up, so, uh . . . "

Which leads to the question, what might lead the voltage drops between series diodes to differ in the first place - a quirk of manufacture in the diodes themselves, circuit behavior, or something else? And correspondingly, aside from the "It didn't blow up yet so it's OK" reasoning, would there be a more detailed reasoning that tolerances or circuit behavior make it very unlikely the voltage drops will differ significantly?

There's got to be SOME logic to it. We are talking physics and industrial manufacture after all.
 

darkfenriz

Member
Messages
207
Because . . . ?
Try proving the opposite.
Assume you have 1010V reverse drop on 2 series diodes, but not equal 505V+505V distribution but the worst case of 1000V on first and 10V on the other.
So what?
Pretty much nothing happens, first in its zener range, i.e. starts reverse conducting (if it could conduct) but the other is still blocking. Both happy and not subject to any power loss as there's zero current flow through the series.
Worst case scenario results in a non-issue.
 

UsableThought

Member
Messages
1,650
Assume you have 1010V reverse drop on 2 series diodes, but not equal 505V+505V distribution but the worst case of 1000V on first and 10V on the other. So what? Pretty much nothing happens, first in its zener range, i.e. starts reverse conducting but the other is still blocking.
PICARD: Anything yet, Mr. Worf?
WORF: Sir, we are picking up a transmission from the planet, but it appears to be in an unknown language. Not in any of our databases.

Sorry - just kidding. Thanks for the detailed response. I will re-read carefully until my slow brain starts processing it!
 
Last edited:

Blue Strat

Member
Messages
30,182
I dunno. One expert says something, another expert says something else; and here I am stuck in the middle. Merlin B. is generally helpful but is obviously not the mouth of God.

Meanwhile I'm waiting for those who are anti-balancing to explain why, other than "Well, they seem to do it that way & it doesn't blow up, so, uh . . . "

Which leads to the question, what might lead the voltage drops between series diodes to differ in the first place - a quirk of manufacture in the diodes themselves, circuit behavior, or something else? And correspondingly, aside from the "It didn't blow up yet so it's OK" reasoning, would there be a more detailed reasoning that tolerances or circuit behavior make it very unlikely the voltage drops will differ significantly?

There's got to be SOME logic to it. We are talking physics and industrial manufacture after all.
I think it's compelling evidence that it's virtually never seen in commercial designs. If that's not compelling enough I don't know what to say.
 

UsableThought

Member
Messages
1,650
I think it's compelling evidence that it's virtually never seen in commercial designs. If that's not compelling enough I don't know what to say.
Yes, that is exactly the "it didn't blow up yet" approach. On the one hand, that is the way a lot of bad s___ happens in this world. On the other hand, as an approach it does seem to have worked pretty well for pushing tubes beyond their data sheet limits in some categories - from what I read that became a common practice under Leo Fender? And it relates to what pdf64 was saying about theoretical best practices vs. essentials - that is something I can grasp.

Anyway you know all about these commercial designs, so to you it seems obvious. Not to me; I am new to this & learn slowly. One of the first things I did was look at my Laney Cub 10 schematic, but it was no help since the rectifier output is a puny 300VDC. I will look for other schematics where the HT out of the res cap is bumping up against the margin as mine is - thanks for the indirect tip.
 
Last edited:

ironman28

Gold Supporting Member
Messages
590
Anti-Balancing...?
Most commercial MI designs are traditionally budget driven due to the large number of units produced. Small scale builders and DIYers can afford to build to higher standards so I would use the higher voltage diodes. There is nothing wrong with designing to "adequate" vs a bullet proof standard. There is a sense of satisfaction in knowing you have built something to a higher standard though.
 

UsableThought

Member
Messages
1,650
Also perhaps consider whether it is feasible for the mains to rise a further 10% above its current level, which is already so high you need to wire a bucker permanently in place.
I guess not? It's 121VAC at the wall socket where I live and apparently in many places in the U.S. . . . going up 10 percent would be 134VAC at the wall socket. Doesn't seem likely.
 

Kyle B

Member
Messages
5,074
The method I've seen for balancing resistors on stacked ecaps is to look up the likely leakage current (of the diodes in this case), and use a resistor value that, given your reverse voltage, will pass about 10 times that (or more), ie so that the resistor current should dominate the reverse diode current, and so achieve roughly equal voltage across them.
But as you've only got 300-0-300Vac, and it's a 'best practice' rather than 'essential' thing, I think it would be reasonable not to bother.

Also perhaps consider whether it is feasible for the mains to rise a further 10% above its current level, which is already so high you need to wire a bucker permanently in place.

If you do go for it, consider the voltage rating of the resistors (limiting element voltage or similar); you may need 2 or more in series to keep within their rating.
I agree, this is the way to go about it. Check out the leakage current of each diode, then pick a resistor that will draw enough current to completely swamp the effect.

The 1N4001-7 datasheet, on page 3, shows the reverse current for a "hot" diode to be around 100uA.
https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0ahUKEwi1-dOO9_zLAhVI0GMKHY63DXgQFgg9MAA&url=http://www.diodes.com/_files/datasheets/ds28002.pdf&usg=AFQjCNGNkYePUumtp96okqcBKscYMaoK0A&sig2=MgnlH55iburJGDWqWDDqLw

So to swamp this effect, you want around 100uA * 10 = 1mA flowing in each resistor.

If the voltage the resistor(s) would see is 1000V, then R=V/I=1000/0.001=1MEG. You would be using 2 resistors, so a couple 500k resistors (one across each diode) would do the trick. Don't forget power P=I^2*R --> 0.001*0.001*500,000= 1/2W. So 1W resistors would reasonably be the "right" size (probably can use 1/2W since the resistors would only be seeing 1000V when the diodes are reversed biased, but this is for a military-grade bulletproof build, right? :) ). Most 1W resistors can withstand the voltages we're talking about, but that's another thing to verify from their datasheets. eg: http://www.farnell.com/datasheets/1716725.pdf

Caveat: This is just EE theory discussion, which I love ;) It really is probably overkill for guitar amps. But WTH, it's cheap and won't hurt to put 'em in!
 




Trending Topics

Top