Ground loop elimination in parity experiments.

 

While performing parity experiments it is vital to prevent electronic cross-talk that could feed real-time helicity information to the experimenter.  Ground loops could transmit this signal and must be eliminated.

 

When we developed the original Pockels Cell driver for CEBAF, parity experiments were not at the front of our mind.  We were a bit careless with the layout of our electronics.  This is evidenced by the following schematic layout:

 

float Pockel-Model

 

The errors should be obvious, but I will point them out.

  1. The DAC’s that program the 4000V DC-DC converters have a common ground reference with the VME crate in the injector service building.
  2. The 24VDC bulk supply for the DC-DC converters has a ground reference in the injector service building.
  3. The Helicity generator chassis was in the same rack upstairs, sharing power with the VME crate electronics.
  4. Not shown on the drawing—A 1000:1 high voltage probe would monitor the voltage on the pockels cell as a safety precaution to warn the halls if the flipping stopped.  The signal was buffered and sent into two ADC channels.  Fundamentally it was a good idea, but only if it is implemented in such a way as to not pass real-time helicity information.  It shared a common ground, so it was a suspect for passing information.

These grounds were all in common with a distant ground on the Pockels cell and 4000V dc-dc converters.  Every time the cell would switch, the grounds could all take a small jump as they bleed down the current that passed through the high voltage switches.

 

 

These errors were corrected in 2004 by isolating the system grounds to a common ground in the tunnel as shown below.

 

 

float Pockel-Model1

 

All system commons are designated by the blue lines.  All commons join earth ground at a single point.  The helicity generator is located in a floating rack and the helicity information is fed on fiber to the high voltage switch.  The DAC programming voltages are optically isolated, so there is no possibility of transferring helicity information back up to the injector service building.

 

The remote voltage monitoring has been eliminated for now.  The halls tell us that their data would tell them almost immediately if there was an unexpected loss of helicity flipping.

 

2007 plans

In the summer of 2007 the wiring to the laser room will be cleaned up.  All information for programming and readback of signals related to helicity components (Pockels Cell and IA’s) will pass through a new isolator card.  This card uses the ISO-124 chip to isolate signals from -10V to +10V with a bandwidth of up to 50kHz.  The card we have designed will isolate 8 channels from the service building down to the laser room, and 4 channels from the laser room back up to the service building.  We will have 3 of these cards in place.  We will need to ensure that any signal that is placed on a channel from tunnel to service building is heavily filtered with at least a 100ms time constant filter to ensure there is no real time helicity information being passed.

 

FAE8FD6C

 

January 2008:

The helicity system is configured as shown in the figure below:

 

All circuits seeing real time helicity are floating with respect to earth ground.  The option remains to connect a single point ground if requested by the researchers.  Thus far we have had no complaints about unintended helicity pickup.

 

April 2010:

A “quad-IA” system has been developed that will permit small corrections to the IA cell for Hall C on a quartet basis.  The system is configured as shown in the figure below:

 

 

 

 

 

 

BACK