- In 2012, we discovered that some issues related to output diodes used in output modules of LHC13kA-18V converters happen more frequently.
- Some systematic measurements when repairing (changing diodes) showed strange high leakage current on the non-destroyed reamining diodes, and that, even on modules which were not faulty. (When a sub-converter made of 5 modules fails, procedure is to change the whole sub-converter, even when trouble is identified on a given module).
- Moreover, some additionnal measurements made on a known healthy converter (never failed from start of operation) showed that some diodes were starting a slow degradation, visible on Irr.
- What is visible then is a higher than expected leakage current while testing the 135V diode at only 30V in reverse.
- All data below are available to try to identify the source of the problem.
- Diode being used is 249NQ135.pdf
- Steps to be followed:
- Identification of a possible common mode from measurements and observation.
- Identification of trouble spread over the modules and diodes: are all diodes touched, from slightly to severely, or are some not damaged at all?
- Possible explanation to evaluate:
- Too high torque on diodes (copper link to transformer leg)?
- Could explain the fact that problem appears randomly on all diodes, and moreover, would explain that some seem still brand new.
- Would only be credible if current sharing is completly not working, since LHC13kA converters haven't operate at very high current (not more than 50% from 2007).
→ Estimation of torque on all diodes before measuring them in reverse.
- Too high Junction T°C operation?
- Strange since 20.5kA is running in a lot more severe conditions, and no failure of this kind up to now. Water T°C difference?
- Would only be credible if current sharing is completly not working, since LHC13kA converters haven't operate at very high current (not more than 50% from 2007).
→ Check of current sharing between diodes whatever the operating mode is.
- Over-voltage across the diodes?
- Efficiency of the snubber not sufficient, or overvoltage conditions higher when operating in 1/2 bridge conditions.
- Would explain why LHC20.5kA and LHC13kA are different.
→ Check of over-voltage across the diodes whatever the operating mode is.
→ Check of snubber state (R-C value, frequency plot), or typical oscillation period (1st sinus full period) being measured.
- Thermo-mechanical constraints?
- Power Transformer is really hot, when baseplate of the diode is quite cool. Delta T inducing mechanical stress?
- Why LHC20.5kA and LHC13kA are different if so, when LHC20.5kA operates at higher output current, then high delta T°C?
→
- Operating cycles being stressfull?
- Would be coherent vs LHC20.5kA which are in DC mode, and doesn't suffer from this trouble (Irr not measured up to now).
- Cycles numbers are nevertheless not very high, and very smooth and slow. Say 1 cycles every 10 hours, going smoothly (some A/s) from 0A to 7kA max.
- Cycles are completly different in the VxI area, and perhaps the 1/2 to full bridge operation is a lot more used in LHC13kA.
→ Determination of when the 1/2 bridge mode is active on superconductive magnet, and compute timber data.
- Environmental conditions?
- Humidity level?.
- Would explain why LHC20.5kA and LHC13kA are different.
→ .
- 3 output modules were removed from the machine, converter: RPHE.UA47.RQF.A45. This converter almost never encountered any failure, and is then considered as a reliable and healthy one.
- Initial position in the converter are:
- HCRMOC_ : TT000082 → SUB4-Pos3
- HCRMOC_ : TT000081 → SUB4-Pos2
- HCRMOC_ : TT000080 → SUB4-Pos1
lhc3.25kA-Inverse-current-rectifier .xls
lhc3.25kA-Inverse-current-rectifier .xls
Physical position of the diodes in a output module
Electrical Position of the diodes in a output module and position of modules in a converter
- Diodes were analysed considering 2 blocks, and results are displayed, being compared to brand new diodes:
- Low-in-specif Irr (<10uA)
- high Irr (> 10uA)
Comparison of diodes in between, and "good ones in range" vs brand new diodes.
- Result is:
- Some diodes seem completely "new", when other ones are critical.
- Some diodes experienced then "a constraint" the other don't.
- Several modules were removed from machine operation (several years being operating a medium output current). Some were faulty, some not. All diodes (2 were dead in short circuit) were measured at Vrr 30V to get Irr. Several observations:
- 7 modules are split: 4 faulty ones, or coming from faulty converter. 3 are coming from healthy one, even if degradated slighlty (see above).
- Mean value over different modules considering the diode position is not really relevant, even if displayed below.
- Diodes measured with still low Irr (like brand new diode) are of interest since representing locations were diodes are "less damaged".
- All in one, on 7 modules, and considering a limit of 10uA @ Vrr=30V as a treshold:
- on 3 "healthy" modules:
11/24=45% diodes with Irr > 10uA
- on 4 "suspect or damaged" modules:
25/24=78% diodes with Irr > 10uA
Overview of Irr @ Vrr=30V of several modules .xls
- All data available are displayed below
Overview of Irr @ Vrr=30V on all modules available.xls
- A measured diode with relatively high leakage current is tested vs Vrr. While diode was still acting "like a healthy one at low voltage", it goes dead (short circuit) at around 60V (130V rated).
lhc3.25kA-Inverse-current-rectifier .xls
- Irr @Vrr = 30V was measured on 24 brand new diodes. Results are summarized on 2 plots and these additionnal numerical data
- Average Irr@ Vrr:30V=5.25uA
- Standard deviation: STD(Irr)@ Vrr:30V=0.4uA
lhc3.25kA-Inverse-current-rectifier .xls
|