Pelletron HV stability status

1
Pelletron HV stability(status)

• Recycler Meeting
• March 19, 2008
• A. Shemyakin

2
Problem description

• Sometimes, the Pelletron voltage starts to
  oscillate
• Also, the average level of the HV noise seems to
  be higher

20-hrs plot of HV as read by R GVMVLT Data from
D44, 1100 9-Mar-08 to 700 10-Mar-08
Good
Present norm
Bad
3
Impact on operation

• 1-hr RMS of RGVMVLT (1 Hz data from D44).
• 20 hours starting 11AM 9-Mar-08 left plot,
  16-Mar-08 right plot
• Bad/Good 2
• Number from the good past 0.25 kV
• Operational impact at present level
• Seems to be negligible when as on the right plot
• Could be noticeable if bad time happens during
  the shot
• Bad if the energy calibration is done in the time
  of a strong voltage deviation

4
Possible culprits

• Greg and Howie has shown that the problem starts
  with GVM reporting a non-existing HV perturbation

Location of the culprit

• Corona current follows the GVM signal therefore,
  the reason for the erroneous signal could be
• Mechanical problem with the Generation VoltMeter
  (GVM)
• First stage of the GVM preamplifier
• Electron flow to GVM plates (corona from the
  terminal)

5
Investigation corona (3)
electrons

• Comparison of 1-hr RMS of RGVMVLT at 4.33 MV
  (16-Mar-08, curve 1) and 3.0 MV (18-Mar-08, point
  2)
• Decrease of the voltage should dramatically
  decrease the level of corona currents
• No significant difference in the noise level
• In both cases, the software loop was off

1
2
6
Investigation GVM preamplifier (2)

• Options to address the issue (Greg and Kermit)
• Replace the chip with another (better) one
• Tap to the GVM signal
• 2-3 hrs MI-31 access
• In parallel, Greg prepares a new box that will
  include CPO into the regulation circuitry in
  addition to the GVM with a crossover frequency
  around 0.3 Hz
• Should significantly suppress HV jumps in 1 Hz
  range if they are really coming from the GVM

7
Plan for the GVM

• Kermit reports that the GVM noise has changed in
  the recent months
• Change in the rotation frequency may change the
  readback
• A predictable way to repair GVM is to replace the
  entire assembly
• Would be assembled and tuned at NEC
• Replacement can be done in a shift or so
• Kermit and Ron have put a PO
• 12 k
• Because of a large sum and furloughs, it still
  has not been placed
• Delivery time is between 6 weeks and 4 months
  from placing the order
• Emergency plan replace moving parts
• Not obvious that disassembling and reassembling
  can be done without a damage and changing the
  balance
• Kermit has all parts

8
Summary

• The Pelletron HV is somewhat unstable
• Most of the time, it doesnt create noticeable
  impact on operation
• However, common sense tells that once started,
  the problem will become worse
• Incorrect reporting by GVM is the most likely
  reason
• So far, we cant ruled out any of 3 possible
  explanations
• GVM mechanics, electronics, corona
• Plan to test electronics this week if there is an
  opportunity
• PO goes through administration
• An emergency plan is ready
• Not working readbacks complicate analysis
• Replacement parts are on site (Kermit)