Amps
[Top] [All Lists]

[AMPS] series HV stacked rectifiers

To: <amps@contesting.com>
Subject: [AMPS] series HV stacked rectifiers
From: jtml@lanl.gov (John T. M. Lyles)
Date: Tue, 6 Jul 1999 11:49:30 -0600
Sometimes it helps to remind ourselves why HV stacks have RC compensation
(besides the obvious DC voltage division problem related to unequal leakage
and tempco in the diodes). And reading these also shows how the different
manufacturers feel about RC networks versus using avalanche diodes.


-----------------

Unitrodes 1986 Databook shows HV silicon Rectifiers and states:
"The KX/KXS silicon rectifier series is a unique concept for high current
HV applications. Matched junction characteristics and low stray capacitance
due to metallurgically bonded junctions eliminates the need for external
compensation networks."

In "HV Rectifer Tube Replacement with Silicon Doorbell Rectifier Modules",
bulletin N130, 1967, they state:
"Voltage equilization across individual diodes in each doorbell is provided
by the controlled avalanche behavior of the individual junctions and the
close matching of the PIV ratings of the diodes. The diodes behave like
zener diodes, with current limiting by those junctions operating at reverse
voltages less than their breakdown voltages."

"Transient energy during transition from conduction to blocking state is
safely handled with the high reverse pulse handling ability of the
junctions. Highest energy is supplied to part of the stack during reverse
recovery time. The stud and bushing that are molded into each module form a
shunting capacitor of 7 to 10 pF which further enhances this transient
energy handling capability. This capacitance shunts every group of (up to
30) series-connected junctions."

-----------------

Quoted from "HV Silicon Rectifier Stacks and Their Applications", Mullard
Technical Communications #110, March 1971 (Mullard, Philips, Amperex - were
all similar then):

"When diodes are used in series in a high voltage rectifier stack, there
are two main sources of transients. The diodes will encounter transients
from the AC supply, and also transients will be generated inside the stack
itself as the diodes start to conduct after the negative half-cycle of the
supply. The diodes will not begin to conduct at the same time, with the
result that, depending on the inductance of the circuit, very large reverse
recovery transient voltages can be applied to the diodes for very short
times. Avalanche diodes are able to absorb these commutation transients,
but non-avalanche diodes require equilizing resistors and capacitors to
ensure that the diode ratings are not exceeded. Thus the cost of a HV stack
can be reduced by using avalanche diodes since the ancillary components are
not required."

------------------

In "Protecting Silicon Rectifiers from Transient Overloads", Westinghouse
Electric Corp. Reprint from Electo-Technology, Nov. 1963:

"...current transfer between diodes occurs at relatively high voltage and
an abrupt spoke of reverse current flows. The current is limited only by
source impedance of the circuit. As the diode recovers its blocking
capacity this reverse current cuts off. The resultant fast drop in current
induces a voltage in stray circuit inductance and becomes a transient
source for other diodes in the rectifier.... In addition to the voltage
disturbance, radio noise of relatively high magnitude may be generated."

" In general, fast recovery diodes of the diffused junction type cause the
most severe commutation spikes and resultant disturbances. These problems
are usually less severe in the alloy-junction diodes. Methodes of
suppression include reducing the magnitude of the recovery spike,
minimizing distributed inductance, and most effective, protecting the
diodes with suitable suppression networks."

"High voltage rectifier assemblies offer transient-protection problems in
themselves. When rectifiers are connected in series, the reverse voltage
must be divided equally among the series units. Because the rectifiers may
have considerable difference in reverse resistance, voltage-dividing
resistors are commonly used. These should be chose so that their resistance
does not exceed one-half the minimum rectifier reverse resistance at rated
prv. This protection is satisfactory up to voltages of several KV. At
higher voltages, additional protection is needed in the form of capacitors
connected across each cell. These capacitors provide loading which masks
variations in the capacitances and recovery times of the diodes. It has
been the experience of Westinghouse design engineers that RC networks are
essential to reliable HV industrial assemblies, based on some 50 million
stack-hours of failure-free operation in the field."

-------------------

In "Power Semiconductor Handbook", Semikron International"

"Avalanche rectifier diodes can be used the same as normal rectifier diodes
in a variety of circuits but, due to their insensitivity to short term
overloads in the reverse direction, the use of overvoltage suppressors can
be avoided. In particular, avalanche rectifier diodes can be connected in
series without the use of static or dynamic voltage sharing networks."

For conventional diodes:
"The following guidelines for the protection of individual semiconductors
by RC networks are valid under the conditions that the circuit only gives
protection against excessive voltages arising from carrier storage
currents...."

"Making the assumption that about half the stored charge Qrr in the
semiconductor will result in the production of a transient voltage in the
circuit, the most suitable values for the capacitance and the damping
resistance are given by the formulae:

C =  Qrr/(V x 1.414), where Q is in uC, and V is the rms voltage

R = SQRT (Ls/C) where Ls is the total circuit inductance in uH (hard to get!)

Power loss in R is

P = 1.414 x V x Qrr x Freq

-------------------

In a 1980 edition of the Motorola Silicon Rectifier Manual (page 13-2):

"In the past, series units were compensated by the addition of a parallel
resistor and a capacitor to each cell that would dominate the diode
characteristics. In addition to degrading the rectifier characteristics,
the network adds bulk and cost. The only alternative is to carefully match
each cell going into a series string. This too can be expensive for the
user, since it involves considerable testing; for successful
implementation, five basic parameters must be matched: turn-on time,
reverse-recovery time, reverse breakdown, capacitance, and reverse leakage.
This overwhelming task can be done by the manufacturer relatively easily
and with a considerable cost advantage as will be seen."

-------------------

I hope all this helps you decide on your individual basis if you need to
build the networks. I would say if under a few kV, don't bother. If a 5 KV
stack must be built, you can build them or not, depending on the matching
of your diodes and whether they are avalanche junctions or not.

Its interesting that the present Motorola engineers have stated that they
do not control their diodes to a degree of matching needed to build series
HV stacks without RC compensation (I think I read this here last week).

73
K5PRO
John






























--
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


<Prev in Thread] Current Thread [Next in Thread>
  • [AMPS] series HV stacked rectifiers, John T. M. Lyles <=