[AMPS] Vacuum relay and HV Fault Protection
Vincent Fedele
vince.fedele@vsttech.com
Fri, 02 May 1997 09:20:05 -0400
Ian:
My understanding of defibrillators from the application point of view is
that there are many similarities and differences. For example, believe
it or not, the energy level (at the highest settings) is similar to the
B+ energy levels, (C*V^2)/2 Joules of a typical linear amplifier
(150-250 Joules). Differences are that the Defib works only with stored
energy in a capacitor, which ensures a known, limited amount of energy
is available. The P/S is disconnected after charge is reached, and the
energy is dumped (ALL CLEAR!) . If your intent is fault protection,
etc..., there are several lessons we can learn from that application.
Most B(+) faults result in a HV to Chassis short. Examples are: Anode
to Grid (GG) short inside tube, RF Plate Choke to Chassis, Plate
blocking capacitor flashover through antenna RF choke to chassis, B(+)
bypass short. Since the voltage across the filter capacitor can not be
changed instantaneously with respect to time, the neg side of the filter
will try to spike neg the full supply voltage. Fault current must be
directed around fragile components, and the stored energy absorbed.
In theory, the best application to employ during a B(+) to chassis fault
might include the following list of desirables:
1. Limit the peak current with 10-50 ohms of glitch resistance in the
[fault] path to avoid damage to the tube structure(s) and other
expensive components;
2. Prevent the fault current path to B(-) from raising the B(-) to
levels that would wipe out metering and biasing components;
3. Detect the fault, and immediately remove AC Mains power within 1/2
cycle so that the energy to be absorbed or diverted is limited to that
stored in the filter cap +plus 1/2 cycle (max) of [fullwave] charging;
4. Provide an energy absorbing device to dump capacitor into while
maintaining 1-3, above.
I don't think interrupting the B(+) on fault detection is a good idea.
It has been tried, usually unsuccessfully, because once the fault
occurs, most relays (even vacuum relays not intended for this purpose)
will arc and breakdown internally to the frame (i.e., chassis) if
current is interrupted at that level. This occurs because of inductances
in the fault path which force a very high (many tens of KV) EMF. You can
not change the current through an inductor instantaneously [with respect
to time]. Relays capable of interrupting power at this level are very
large and expensive, and are used, for example, in vacuum interrupters
of hi-tension power lines, etc.
Most modern amplifiers only employ protection items 1) and 2), above.
This is better than nothing and often protects the tube and metering,
but is not adequate protection for several reasons. First, the AC mains
are left ON, with only the mains fuse (or breakers) to trip--if they
trip at all. This allows the B(+) to remain ON, and eventually,
something in the path WILL FUSE OPEN. This results in destruction of
components in the fault path, sometimes the glitch resistor in the B(+),
along with other components often not detectable (i.e., the biasing
zener) until several more faults occur when trying to re-start after the
initial failure. Sometimes the component that fuses also arcs, and if
that’s the glitch resistor, little current limiting will occur across
the arc. In most cases, even if the mains were shut down, the glitch
resistor commonly used is not large enough to absorb the stored energy,
nor high enough in value to provide the desired limiting. Usually a 10
ohm,, 10w vitreous enamel resistor is used, limiting the current to
3,500/10 = 350 amperes! Eimac recommends the combination of glitch
resistor and B(-) diodes because most amps have neither, and the cost is
next to nothing-while offering some (minimal) protection for the tube.
The 1994 edition of Wm. Orr ‘s "The Radio Handbook" has a very well
executed design of a commercial grade 8877 amplifier by Jim Garland,
W8ZR. In that design, an attempt was made to break B(+) when a fault was
detected (high B+ current), but the relay itself broke down, and flashed
over internally. The updated design (not yet published), which adds QSK,
and 160m also employs a different method of dealing with the B+ faults,
the B+ relay is no longer used.
Protecting with all 4 items, above, is actually easy, and inexpensive,
and reduces damage to the faulty component only, in nearly all cases,
and can be implemented as follows:
Items 1. and 4. Limit Peak Fault Current and Absorb Stored Energy
The glitch resistor is a 50 ohm, 75 watt non-inductive resistor in the
B(-) lead to the filter capacitor - not in the B(+) lead. This limits
peak current to 80 amperes, and is large enough to absorb the energy
stored in the capacitor plus several charge cycles of AC mains.
Items 2. and 3. Protect Metering/Bias Circuits and Detect HV Fault
The typical glitch diode string of 3 or 4 1N5408 diodes is at the
cathode-return side of the 50 ohm resistor, and keeps the B(-) path
within 6 Vdc of chassis gnd. The filter end of the resistor will spike
to neg 4kv, so must be on the rectifier bd, and isolated as a HV
component. A resistor divider across the 50 ohm resistor can now be used
to sense fault current, and set to fire an OPTO Isolator LED (7.5KV
isolation) at 6-10 amperes. The sense resistor is usually about
4Kohm/5Watt to the (-) filter cap in series with a 100ohm/3w to B(-).
The LED sense is across the 100 ohm. These values may need to be
adjusted for your application.
Finally, the phototransistor side of the OPTO isolator may now be used
as a control signal to shut down the AC Mains. If a Solid State Relay
(SSR) is used for the AC Power Mains, the control voltage may be shut
down immediately upon fault current detection (or at the next
zero-crossing). If a mains contactor relay is used, the control signal
from the OPTO may be used to remove AC from the Mains contactor coil.
This may be done with a small (less than $8.00) SSR from CRYDOM via
DigiKey.
Testing can be done at the bench with a low voltage p/s connected to the
100ohm sense resistor. Just calculate the voltage across the string at
5-6 amperes trip current, and you’re done. A small integrating capacitor
across the OPTO LED may be used to eliminate false detects on peaks.
Now, you have the following:
1. Limited peak current;
2. Fast fault-current detect and shut down of AC mains, limiting total
energy;
3. B(-) protection for metering/bias and other circuitry;
4. Energy absorbed, no shrapnel to clean up.
Good building, think safety when working with HV.
Vince Fedele
WA2PKE
*******************************************************
Ian White, G3SEK wrote:
>
> W8JI, KN6BI and AG6K wrote:
> >>ratings and switching speed of the type RFE-26N1070 vacuum relay?
> >
> >Ian, That number isn't in my catalog. Are you sure it is correct?/
>
> Sorry everybody, that should have been RFIE- - - Damned astigmatism!
> ^
> Apparently they're used in defibrillators. I was thinking of making a
> short-circuit tester for the 3kV supply, which is kind of the
> opposite...
>
> 73 from Ian G3SEK
>
> --
> FAQ on WWW: http://www.contesting.com/ampfaq.html
> Submissions: amps@contesting.com
> Administrative requests: amps-REQUEST@contesting.com
> Problems: owner-amps@contesting.com
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com