Electrical safety - verifying zero energy in large capacitors

[Chris Carlen]

Greetings:

We have numerous flashlamp pumped lasers with one or more 60-80uF 2kV
capacitors inside. They operate at about 1.5kV peak. We are possibly
moving toward a requirement that the caps must be grounded and strapped
before doing work on the power supplies (including non-electrical work
such as changing water filters). This will be extremely cumbersome
since the caps were intentionally located in a difficult to access
location in the power supply by the manufacturer.

Additionally, this grounding work would have to be performed with
arc-flash clothing/face protection to the voltage/energy levels
expected, and appropriately rated insulating gloves. Yet the use of
this equipment makes the task even more difficult, and potentially more
prone to error.

The manufacturer's circuitry includes bleeder resistors for the caps.

I propose that we use SHV or similar, rated HV connectors connectors to
provide a safe direct monitoring port to the capacitor terminals. Then
by simply plugging in a high-voltage probe + DMM, we can verify that the
caps are at zero volts.

After that, one can proceed with work without concern. Such a rapid
means of verifying a safely discharged condition is essential for we
frequently need to iterate many times to troubleshoot or perform mods to
the lasers.

There is a concern though about a direct wire to the caps. If anyone
managed to short the SHV connector while the cap was charged, this would
create a large arc-flash hazard. Originally I had proposed using a
resistive voltage divider to monitor the caps, and argued that one could
actually watch the RC decay on a scope. Barring any step
discontinuities in the resulting trace or large deviation in time
constant from that expected, one could then conclude with a high degree
of condifence that zero energy was verified. I also provided detailed
analysis of the failure possibilities of the divider in this scheme.

The SHV connector idea was a simplification of the divider idea,
eliminating the failure modes and the fact that one cannot verify after
zero energy is read, that the divider hadn't suddenly broken (though I
argue that you DO know that, by looking at the scope trace).

Alternately, several parallel resistors could be placed in series with
the SHV connector, to limit current, while reducing the probability of
resistor failure causing a faulty reading to

1 / ((probability of a single failure)^(the number of resistors))

which would be pretty close to zero for 3-4 resistors.

This is all too logical and our safety folks want to hear none of it.
Even regarding the directly wired monitoring port they say "but there's
no way to know that the wire didn't break the moment after you measured
normal cap voltage, and now measure zero."

Would you EEs think it is sufficiently safe to verify zero energy via
such a monitor connector (with or without intervening signal
conditioning circuitry) wired to the caps, and perform work without
visibly grounding and strapping the cap terminals?


--
Good day!

________________________________________
Christopher R. Carlen
Principal Laser&Electronics Technologist
Sandia National Laboratories CA USA
[email protected]
NOTE, delete texts: "RemoveThis" and
"BOGUS" from email address to reply.

 

[Baron]

Greetings:

We have numerous flashlamp pumped lasers with one or more 60-80uF 2kV
capacitors inside. They operate at about 1.5kV peak. We are possibly
moving toward a requirement that the caps must be grounded and
strapped before doing work on the power supplies (including
non-electrical work such as changing water filters). This will be
extremely cumbersome since the caps were intentionally located in a
difficult to access location in the power supply by the manufacturer.

Additionally, this grounding work would have to be performed with
arc-flash clothing/face protection to the voltage/energy levels
expected, and appropriately rated insulating gloves. Yet the use of
this equipment makes the task even more difficult, and potentially
more prone to error.

The manufacturer's circuitry includes bleeder resistors for the caps.

I think you should discuss your thoughts with the manufacturers !

I propose that we use SHV or similar, rated HV connectors connectors
to provide a safe direct monitoring port to the capacitor terminals.
Then by simply plugging in a high-voltage probe + DMM, we can verify
that the caps are at zero volts.

After that, one can proceed with work without concern. Such a rapid
means of verifying a safely discharged condition is essential for we
frequently need to iterate many times to troubleshoot or perform mods
to the lasers.

There is a concern though about a direct wire to the caps. If anyone
managed to short the SHV connector while the cap was charged, this
would create a large arc-flash hazard. Originally I had proposed
using a resistive voltage divider to monitor the caps, and argued that
one could actually watch the RC decay on a scope. Barring any step
discontinuities in the resulting trace or large deviation in time
constant from that expected, one could then conclude with a high
degree of condifence that zero energy was verified. I also provided
detailed analysis of the failure possibilities of the divider in this
scheme.

The SHV connector idea was a simplification of the divider idea,
eliminating the failure modes and the fact that one cannot verify
after zero energy is read, that the divider hadn't suddenly broken
(though I argue that you DO know that, by looking at the scope trace).

Alternately, several parallel resistors could be placed in series with
the SHV connector, to limit current, while reducing the probability of
resistor failure causing a faulty reading to

1 / ((probability of a single failure)^(the number of resistors))

which would be pretty close to zero for 3-4 resistors.

This is all too logical and our safety folks want to hear none of it.
Even regarding the directly wired monitoring port they say "but
there's no way to know that the wire didn't break the moment after you
measured normal cap voltage, and now measure zero."

Would you EEs think it is sufficiently safe to verify zero energy via
such a monitor connector (with or without intervening signal
conditioning circuitry) wired to the caps, and perform work without
visibly grounding and strapping the cap terminals?

Doesn't this equipment have safety interlocks that disconnect the supply
and ground the caps if the case/door is opened ?

 

[Mike Harrison]

Greetings:

We have numerous flashlamp pumped lasers with one or more 60-80uF 2kV
capacitors inside. They operate at about 1.5kV peak. We are possibly
moving toward a requirement that the caps must be grounded and strapped
before doing work on the power supplies (including non-electrical work
such as changing water filters). This will be extremely cumbersome
since the caps were intentionally located in a difficult to access
location in the power supply by the manufacturer.

Additionally, this grounding work would have to be performed with
arc-flash clothing/face protection to the voltage/energy levels
expected, and appropriately rated insulating gloves. Yet the use of
this equipment makes the task even more difficult, and potentially more
prone to error.

The manufacturer's circuitry includes bleeder resistors for the caps.

So what's wrong with " allow x minutes before opening"?

I propose that we use SHV or similar, rated HV connectors connectors to
provide a safe direct monitoring port to the capacitor terminals. Then
by simply plugging in a high-voltage probe + DMM, we can verify that the
caps are at zero volts.

There is a concern though about a direct wire to the caps. If anyone
managed to short the SHV connector while the cap was charged, this would
create a large arc-flash hazard.

So add a large value series resistor to the socket.
Part of the test procedure would be to verify that the resistance is present.
i.e.
1) check that voltage across terminals is 0
2) check that resistance between terminals is < x

 

[Don Bowey]

I think you should discuss your thoughts with the manufacturers !


Doesn't this equipment have safety interlocks that disconnect the supply
and ground the caps if the case/door is opened ?

I once picked myself off the floor, about ten feet from the workbench where
I had been working on a photoflash unit, wondering what the hell just
happened.

What happened was the safety interlock, which should have grounded the caps
when I removed the unit from its case, didn't work, and in a potent rebuke
it taught me a lesson.

I've never since trusted an interlock, and I recommend that nobody should.
Nothing beats the insurance of manually discharging HV caps.

 

[Eeyore]

There is a concern though about a direct wire to the caps. If anyone
managed to short the SHV connector while the cap was charged, this would
create a large arc-flash hazard.

Fit a resistor inline with the 'live' connection to the SHV connector. It
won't interfer with the voltage measurement but will prevent anything being
shorted. It could perhaps even work as a low value bleeder to discharge the
cap fairly rapidly in the event the connector is intentionally shorted.

Graham

 

[[email protected]]

This is all too logical and our safety folks want to hear none of it.

Even regarding the directly wired monitoring port they say "but there's
no way to know that the wire didn't break the moment after you measured
normal cap voltage, and now measure zero."

I think there will always be some voltage over the capacitor. So a break can
be detected as a true zero voltage.
One could also send a small AC signal and measure it's attenuation.

 

[linnix]

We have numerous flashlamp pumped lasers with one or more 60-80uF 2kV
capacitors inside. They operate at about 1.5kV peak. We are possibly
moving toward a requirement that the caps must be grounded and strapped
before doing work on the power supplies (including non-electrical work
such as changing water filters). This will be extremely cumbersome
since the caps were intentionally located in a difficult to access
location in the power supply by the manufacturer.
...
This is all too logical and our safety folks want to hear none of it.
Even regarding the directly wired monitoring port they say "but there's
no way to know that the wire didn't break the moment after you measured
normal cap voltage, and now measure zero."

Would you EEs think it is sufficiently safe to verify zero energy via
such a monitor connector (with or without intervening signal
conditioning circuitry) wired to the caps, and perform work without
visibly grounding and strapping the cap terminals?

Build your HV monitor with a divider network and micro. Have the
micro monitor the HV continuously and activate/verify the decay prior
to service work. If your safety committee raises any doubt (I believe
this is the proper word here, as opposed to many questions elsewhere),
tell them to check with your programming committee.

 

[Chris Carlen]

So what's wrong with " allow x minutes before opening"?

It is an unacceptable risk that the bleeder resistor/circuit may have
malfunctioned.

So add a large value series resistor to the socket.
Part of the test procedure would be to verify that the resistance is present.
i.e.
1) check that voltage across terminals is 0
2) check that resistance between terminals is < x

Yes, I suppose that works, since you'd be measuring the monitor
connector resistor in series with the bleeder resistor.



--
Good day!

____________________________________
CRC
[email protected]
NOTE, delete texts: "REMOVETHIS" and
"BOGUS" from email address to reply.

 

[Chris Carlen]

Build your HV monitor with a divider network and micro. Have the
micro monitor the HV continuously and activate/verify the decay prior
to service work. If your safety committee raises any doubt (I believe
this is the proper word here, as opposed to many questions elsewhere),
tell them to check with your programming committee.

Egad. The mentality here is that if you can't see 2/0 gauge bare wire
strapped across the capacitors, then it's not sure to be discharged.

I think this is overkill, and that it is possible for a carefully
thought-out measurement procedure to both ensure safety and make it much
more expedient to verify zero energy.

But I wouldn't trust my safety to a micro in this case. Plus, they
would never understand it even if it were a ultra-high reliability type
of system.



--
Good day!

____________________________________
CRC
[email protected]
NOTE, delete texts: "REMOVETHIS" and
"BOGUS" from email address to reply.

 

[Frithiof Andreas Jensen]

Would you EEs think it is sufficiently safe to verify zero energy via
such a monitor connector (with or without intervening signal
conditioning circuitry) wired to the caps, and perform work without
visibly grounding and strapping the cap terminals?

No - Eventually someone will use a broken probe, read the meter wrong or
forget about it.

Instead, I would use a HV relay for each capacitor (or -bank if the
connections are sufficiently solid), with a low-resistance high-power
resistor in series with it (maybe one of those resistors made from discs -
"Ceramcarb"??).

The relays are powered by a safety circuit that runs through a series of
switches on all access doors and hatches. When the loop current is
interrupted, by turning off the supply or opening a door, the relays will
drop and dump the energy in the capacitors. The access doors are naturally
only released by using the captive safety interlock key that locks the main
breaker open ;-).

Then one will hang an earthing stick off the capacitors. *Then* one will
work on/near them.

If the capacitors are removed, one will place a short across the terminals
and the housing because in my painful experience those suckers charge back
up again due to charges migrated into the insulation. As long as the
capacitors are in the equipment, the earthing rod should be sufficient.

For simple things like water filters one can, I.M.O., avoid the full
earthing stick e.t.c. by having the "non-electrical" parts in separate
compartments monitored by the safety loop.

So the supplies are Off & Grounded but not hardwired to earth.

Yeah: It *does* take time getting the breaker out, getting the key and
getting access. So one must compensate by making the tuning easier, remote
or superfluous! When I worked with high voltage we used to say that we did
not want to do this work the rest of our lives ;-)