W8JI wrote:
>> The need of equalizing resistors and spike caps due to diode manufacturing
>> variations pasted almost two decades ago. Old habits are slow to die.
>
>Interesting. What manufacturing process suddenly changed or
>improved??
>
>I've heard that mentioned several places, does anyone know the
>source of that information?
For my monthly magazine column, I checked that information with Ken
Stewart, W3VVN, who wrote the Power Supplies chapter for the current
ARRL Handbooks. Ken is a power supply designer for Grumman, and I asked
him to expand on the relatively few words in the Handbook.
In older types of silicon rectifier diodes, excessive reverse voltage
would cause a destructive arc, leaving the failed diode in a low-
resistance state. This failure would impose more reverse voltage on the
remaining diodes in the string, so the whole string would probably fail
in cascade. Also high-PIV diodes were expensive, so manufacturers and
amateurs tried to use the minimum possible number of diodes, which
increased the risk of cascade failure.
On the other hand, physically large (high-voltage) resistors and high-
voltage ceramic caps were cheap in those days, so that's why the habit
set in. That habit seems to have "over-run" while the technology and
economics have changed.
W3VVN explained that modern rectifier diodes are designed not to fail by
the destructive arc mechanism. At a lower reverse voltage than is
required to cause an arc, all modern rectifier diodes go into a
controlled avalanche breakdown, and start to conduct rather like a zener
diode. Thus the meaning of the "PIV" rating has changed: it is no longer
the voltage above which the diode may arc and be destroyed, but the
voltage above which it may go into avalanche mode - but still won't
necessarily fail.
If one diode in a series-connected string happens to avalanche, the
current through it is limited by the leakage resistance of the other
diodes that are reverse-blocking in the normal way. If the avalanche
current is not excessive, the diode will not be harmed. Also, the diode
that has avalanched will continue to maintain a high reverse voltage
drop, so it does not dump its entire share of the total reverse voltage
onto the other diodes.
The economics have also changed. PIV is now cheap, but high-voltage Rs
and Cs probably cost as much as the diodes. Thus it's possible to build-
in plenty of PIV (at least 2x the nominal) and not use the Rs and Cs.
There are still problems with mixed-lot diodes having unequal reverse
leakage at high temperatures. In a 1-2A HV supply it's obviously a high-
risk strategy to use mixed lots of 1N4007s of unknown origin - they are
being pushed to the limits of their current rating, and you can bet that
a cheapskate manufacturer won't use enough of them either. The ideal
solution is to use strings of diodes that are matched in a high-
temperature-reverse-bias test. On the other hand, it seems pretty safe
to use branded 1N5408s (3A rated) from a reputable supplier, all off the
same reel, and with enough diodes to give a good margin on PIV without
help from any Rs and Cs. There are also several types of 6A-rated diodes
available now, and prices are coming down. The reason for using diodes
with high current ratings in a 1-2A supply is that the internal heat-
sinking is better than in lower-current diodes, so the chip temperature
is lower and reverse breakdown is less likely.
Also, we now have better ways of dealing with mains spikes than
connecting a C across each diode. Mains filters have never been cheaper
than they are today. In Europe it's practically impossible to market any
large power supply without a mains filter to ensure compliance with
interference regulations - that is, to prevent interference FROM the
power supply going back into the mains - but the filter also protects
the PSU components from mains-borne transients.
MOVs are another good development. A few months ago I did an Internet
trawl for stories about MOV failures, and they all seem to have been in
situations where the MOV took the full force of a lightning-induced
surge. Longer-term, it does seem that MOVs that have been subject to
repeated minor hits may suffer a reduction in breakdown voltage, in
which case they can go into regular conduction on the peaks of the mains
cycle. This is obviously worse if the MOV was selected with too low a
voltage rating and is also at the low end of the production spread - but
those are the only ones you hear about! If they are intelligently rated
and located INboard of a mains filter, MOVs seem pretty reliable and
provide good protection to the rest of the power supply.
No individual builds enough power supplies to collect reliable
statistics on failures. The only reliable sources are commercial PSU
designers such as W3VVN, which is why I took good care to check the
background before going into print.
Further comments would be appreciated, of course!
73 from Ian G3SEK Editor, 'The VHF/UHF DX Book'
'In Practice' columnist for RadCom (RSGB)
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|