Do PIV's add?

[Radio Fixer]

On another Forum saw two 1N4007 being used in series and implied that this doubles the PIV.  My gut feeling is that it doesn't?

Does two 1mm air gaps have the same breakdown voltage as one of 2mm, my feeling is not.

What do members think  Gary

[boater sam]

Unsure, but there are many circuits with semiconductor diodes in series, usually silicon with capacitors in parallel with each diode.

What about valve diodes in series? Isn't a metal rectifier diodes in series? Then PIVs must add.

[ppppenguin]

If you put 2 perfect identical diodes in series the PIV would double. But diodes are neither perfect nor identical. The voltage will distribute between them depending on the leakge current. This is minuscule for most silicon diodes but can't be ignored. The less leaky diode will take more of the voltage which in turn will cause it to leak more. I don't know if this process reaches a safe equilibrium or not. As a conservative rule of thumb I'd take the PIV of 2 diodes in series as 50% higher than a single one.

Of course you can fit high value resistors across each diode to swamp the leakage. This may or may not be acceptable, especially as you'd be using high voltage rated parts such as VR37 series. Easier and cheaper to stick a third 1N4007 in series to be sure.

Not sure what happens as the diodes switch off. This is a relatively slow process in diodes like 1N4007 and might cause poor voltage sharing on a transient basis. No idea if this matters.

Air gaps are a totally different thing, especially as the breakdown voltage is ill defined anyway.

[Refugee]

I was under the impression that normal diodes act like zena diodes at some voltage well above the published PIV.
You see stacks of diodes everywhere including in TV triplers and those finned smelling sticks found in 1950s tellies where the earlier ones had a valve rectifier.

[Mark Hennessy]

This comes up from time to time, and the question of balancing resistors usually comes up. Same with capacitors in series.

In most cases (of diodes and capacitors), resistors aren't needed.

It is just like putting two zener diodes in series. The voltages add up. Hence yes, the PIV doubles.

If the voltage across one reaches the breakdown voltage, then nothing happens beyond the voltage across it staying reasonably constant. The other diode is still an open-circuit, passing nothing but a tiny leakage current. As that current really is tiny (perhaps 100uA at worst for a 1N400x, and more like 0.1uA at room temperature), no power to speak of is dissipated in the diode that is acting like a Zener, so nothing bad can happen to it. Obviously, when the total voltage is such that both diodes can break down, then that gets "interesting", but that would be the same for a single diode of double the voltage rating.

Unlike Zener diodes, the breakdown voltage is not tightly specified. Generally, it is very conservatively specified for the 1N400x types that I've measured in the past. Most of the lower-voltage parts are "binned" after manufacturer, so it's quite feasible to find a 1N4001 that meets the 1N4004 spec.

The point about transient behaviour is an interesting one. Very much "it depends" - careful testing would be required. At 50Hz, in a low-energy situation, it's probably not a big deal.

Think back to those early metal rectifiers - the type that were full of stacked discs. Did you ever see balancing resistors in there?

http://www.vintage-radio.net/forum/showp...p?p=510861&postcount=17

[ppppenguin]

Strictly speaking, high voltage diodes break down by the avalanche effect, not the zener effect, but the result is the same, they become approximately constant voltage devices.

As for early stacked disc rectifiers, I reckon they were leaky enough to balance without any problems.

[Mark Hennessy]

I decided not to get into Zener-vs-avalanche in that post, so was quite careful with my choice of words; while I drew parallels with "Zener" diodes, at no point do I state that Zener action was taking place.

[Mike Watterson]

In most cases (of diodes and capacitors), resistors aren't needed.

For selenium rectifiers it's true.
Capacitors need derated by a factor related to tolerance spread. Electrolytics especially even for only two, need resistors as leakage can be variable. Capacitors and silicon rectifiers ought to have high value resistors, especially if more than two parts are in series.

It depends on the level of risk / reliability. In the 1970s Hawker Harrier head up displays were failing because the multiple ceramic discs in the EHT capacitor package were not matched enough, nor derated in working volts due to tolerance spread.

[Mark Hennessy]

Capacitors self-stabilise in a very similar way to diodes, although obviously the internal mechanisms are different.

Naturally, we can please ourselves in our own projects, but in a mass-production scenario, balancing resistors rarely make it past the bean counters. Capacitor manufacturers usually recommend fitting them, but that's as much about covering themselves as anything else.

Personally, sometimes I do; sometimes I don't. If feasible, I'd rather just fit a single component properly specified to do the job. You have to ask if balancing resistors will do more harm than good - either because of increased leakage, or because of failure (some folk are quick to forget that many types of film resistors can fail at well below their voltage or power rating). Every scenario is different. But in most cases, they aren't needed.

[Mike Watterson]

Indeed every design is different, how you fudge not having the correct working part can depend on many things. It's sad that purely finance considerations can remove parts costing less than 1/2 cent in volume. Even for neons I put two series resistors (one is marginal rating), and especially for series resistors of my HV capacitor leakage tester (they don't actually break down but voltage is nearly twice rating if the capacitor on test is leaky)