[Amps] Equalising resistors with HV diodes

[Ian White, G3SEK]

Radio WC6W wrote:
>
>> >    Without the R's, a string of unmatched and/or non-avalanche rated
>> > diodes might, or might not, exhibit a higher reverse breakdown
>> > capability
>>
>>   The avalanche characteristic is built-in at the time of epitaxy.
>
>   Are your pet diodes (1N5408) characterized in avalanche?
>

I researched this for the magazine column in 1997, including direct 
correspondence with the author of that part of the ARRL Handbook.

Here's the short version.

Rectifiers can be OK for hundreds or thousands of hours, and then 
suddenly fail. The main reason is a sudden spike of excess mains 
voltage, that happens to arrive near the moment when the AC mains 
voltage is also at its peak. This adds to the normal reverse voltage on 
(half of) your rectifier diodes.

In the first generation of silicon rectifiers, reverse breakdown was by 
surface arcing, which destroyed the diode and made it fail 
short-circuit. Another important factor was that diodes were expensive, 
so designers tended to use the minimum number in series. Then if one 
diode failed short, it would put significantly higher reverse voltage on 
the remaining diodes, so the whole string could easily unzip in a 
cascade failure. That is why designers used to take extreme precautions 
with equalizing resistors and capacitors.

But modern rectifier diodes don't have that mode of arc failure any 
more. You may not find it on the data sheet, but all modern silicon 
rectifier diodes are constructed so that excess reverse voltage causes a 
"zener-like" avalanche breakdown at a constant voltage, which is lower 
than the voltage required to cause a destructive arc. If the reverse 
current is limited by the other diodes in series that have *not* gone 
into breakdown, a brief reverse-avalanche event in an individual diode 
is usually survivable.

Diodes are now much cheaper than high-voltage resistors, so the most 
cost-effective design solution is to spend the money on extra diodes, so 
that the whole string has plenty of reserve PIV capability. For 
matching, I rely on branded diodes, all pulled off the same bandolier 
string - but plenty of 'em.

The K2AW rectifier packages do very much the same. They use typically 14 
of everybody's favorite 1N5408, potted in epoxy with no equalizing 
components. In spite of the much higher operating temperature than 
PCB-mounted diodes, the K2AW rectifiers have a very good reliability 
record - simply because they use lots of diodes.

If anyone is still worried, and still can't resist the urge to throw 
money at the problem, then spend it on a mains filter.



-- 
73 from Ian G3SEK         'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek