From: Raybould, Keith [keith@mbari.org]
Sent: Tuesday, June 16, 2009 1:47 PM
To: MARS Operations and Maintenance Discussion List; Mellinger, Ed
Subject: [marsoandm] MARS MV converter status

Attachments: MV-LV_telemetry_22_10_version2 1_21_11_08.pdf; 29E_HV00_801_AAA_5_n_01.doc; Mars MVC_16_06_2009.ppt

 

 

Here is a quick summary of my visit to Alcatel last week.  Follow up actions, or info I will send on when I receive it from Alcatel, are in blue.  

 

Attached are the MV telemetry spec, MARS MVC test results and MARS MV converter test specification.  More information and documentation to follow,

 

Keith

 

Current status of our MV converter

 

  1.. All of the MARS MV sub-converter stacks (6) have been fully tested individually and have passed all of the acceptance specifications.  These tests are performed with 1.7kV across each stack. 
  2.. The MARS converter (P4) has been fully assembled but has not been tested at 10kV.  Alcatel redesigned the mid-ring during the converter redesign (P4’s) but made the mid-ring in a single piece to fit our MV housing and then only last week preparing for my visit, realized that the larger diameter ring will not fit into their test vessel. They indicated that after sub-converter stack testing they have never experienced any failures with the full assembled converter, so they are confident that there will be no problems with our MV converter.  However, I asked for full system testing before I authorize final payment so they are looking at options.  The favored option right now is to redesign the center ring so that it is fabricated as two pieces. The smaller ring will allow the converter to be installed in the test housing for the full power testing, and then after being removed from the housing, the extension can be added so that it will fit into our housing 
  3.. After the new rings have been designed and the 10kV tests completed, I have asked Alcatel to ship our converter and their test housing assembly to MBARI so that the tests can be repeated here.  They will send someone over to assist us with these tests.  After the 10kV testing has been successfully completed at MBARI, the test housing will be shipped back to Alcatel. Modifications to the rings may take approx 2/3 months so we will not see the MV converter till Sept- Oct ’09.  The test housing has integrated feedthroughs in bulkhead for monitoring converter performance during high power testing.  Very nice unit. 
  4.. I have asked Alcatel to send us recommended storage conditions for our MV converter. 
  5.. (side note: During the fluorinert fill they do not continue vacuum pumping on the housing; they pump down the housing, valve off the vacuum pump, open valve to the fluorinert chamber.  Complete the fill with a hand pump between the fluorinert tank and the housing.  I have asked them to send details and photos? 
  6.. Earlier I sent around the SPI specification for the telemetry system (MV I, V, pressure and 2 temps).  I asked Alcatel to send the block diagrams, schematics and sensor data sheets so that we can evaluate how to interface to the new sensors 
  7.. We had a lot of discussion on whether we should continue to leave the P2A converter in the MARS node or whether it would be advisable to replace it with the P4 converter. The general feeling was that we could continue to leave the P2A deployed for much longer (1-3 years) but we should first check whether any of the 48 converters have failed.  Each sub-converter stack contains 8 converters, and three converters can fail before a sub-converter stack fails.  If we loose one sub-converter stack, the converter will continue to operate (with reduced output I) with the remaining 5 sub-converter stacks.  If two converters in any of the 5 remaining sub-converter stacks fail, then we will be left with 4 sub-converter stacks.  Regulation output with only 4 sub-converter stacks will be significantly reduced.  Any further converter failures and the entire MV converter will shut down.  If we send Alcatel Voltronics output V, I, cable resistance/length, and 375V and 48V currents they will determine from efficiency estimates whether any of the converters have failed (Ken/Craig coordinate getting readings, send to me). 
  8.. We also need to send Alcatel our projected science loads over the next few years (Steve coordinate, send to me) 
  9.. Florinert removal from the housings after power testing is now not regarded as a major problem and they have much simpler processes in place.  One of the earlier model converters P2C experienced catastrophic failure when they tested the converter with a direct short across the 10kv input.  All of the converters fried in a spectacular fashion.  They cautiously removed a sample of the fluorinert and sent it to 3M for mass spectrometer testing.  3M could not detect any level of HF or PFIB.  All of the testing for HF and PFIB at 3m has failed to detect any presence of these chemicals.  The process Alcatel follow now prior to emptying fluorinert from a housing after testing is as follows: 
    1.. Move housing into open well ventilated space (outdoors) 
    2.. PPG used by personnel include a flow hood respirometer and gloves (I have details of the hood and will order 2 units).  No other protective gear is needed 
    3.. Drain sample into container (polyethylene not glass) 
    4.. Send sample to 3M for mass spec analysis (Alcatel are sending me info on contact in US who will be able to perform these tests.  We will need to ask MBARI safety to advise on shipping sample to 3M for testing) 
    5.. When results come back negative continue with draining housing 
  10.. I have copies of all of the mass spec tests performed at 3M 
  11..  To prevent the catastrophic failure caused by a direct short across the input to the MV converter (cable being severed), Alcatel have included a series filter into the -10kV line.  Direct shorting tests have shown that the P4 versions survive this direct short test. 
  12.. I have asked Alcatel to quote for a spare P4 converter.  If it looks reasonable, Steve could request the cost in the next MARS O&M budget 
 

Significant changes between P2A and P4 converters:

 

  1.. Wiring flexes are installed between sub-converter stages instead of individual wiring, much more reliable 
  2.. 50kHz PWM increased to 100kHz, more stable operation, lower ripple 
  3.. Converter stable powering straight into a 700 micro-farad capacitive load (P2C showed instability at ~40 microfarads or less with no resistive loads) 
  4.. Converter does not need no load resistors in MV housing to be stable 
  5.. Converter can survive direct short across input lines 
  6.. Converter has lower undershoot and overshoot on step load changes 
  7.. Telemetry board and sensors included with converter (current, voltage, pressure, 2 temp sensors)