1. ALICE Muon Tracking Upgrade EDR Answers
Low Voltage Power Supply System
October 18, 2017

2. Present / Upgrade Present FEE (MANU: 64 ch)
Upgrade FEE (DualSAMPA: 64 ch)
3 LV lines : -2.5, + 2.5, 3.3 V
Power consumption: ~ 4 mW/ch on each line
Current max on one LV line (1 LVPS channel): 21 A
LVPS: WIENER PL512
8 V max per channel
25 A max per channel
200 W max per channlel
12 independent channels
LV lines grouped by 3  1 LVPS supplies 4 groups
2 LV lines: 1.2 V analog, 1.2 V digital
Power consumption: ~7 mW/ch on digital and 0.4 mW/ch on analog
 currents higher by a factor 3-4
Present Read-out board: CROCUS crate
Upgrade Read-out board: SOLAR crate
1 LV lines : 3.3 V
5 read-out boards in 1 CROCUS crate
Current max on one LV line (1 LVPS channel) for 1 CROCUS: 13 A
LVPS: WIENER PL512
12 independent channels
1 LV line : 5 V
6 SOLAR boards in 1 SOLAR crate
Current max on one LV line (1 LVPS channel) for 1 SOLAR crate: < 12 A
LVPS: WIENER PL512
12 independent channels

3. Quadrants of stations 1 & 2

4. Slats of stations 5

5. Upgrade scenario Slats (stations 3,4,5) SOLAR boards
We cannot afford to change all LVPS for cost reasons
Quadrants (stations 1 & 2)
1 LV group (= 2 LVPS grouped channels) supplies 1 quadrant cathode
With the DualSAMPA, 1 LV channel has a maximum current of 85 A !
Change PL512 by PL508 WIENER: 8 channels, 8V max per channel, 110 A max per channel… but 3.6 kW max for the whole LVPS
No DC/DC  need 8 new PL508 LVPS (+spares)
Slats (stations 3,4,5)
Need to use DC/DC coneverters which can stand 50 A  still to be found; tests in progress in Ithemba Laboratory (Rony Kuriakose)
The LV lines supplying 3 or 4 slats have to be splitted  2 new grouped lines needed per half-chamber
PL512 can handle 6 grouped lines for upgrade (compared to 4 in present)  2 new grouped lines “recovered”  need recabling inside each LVPS PL  need new grouped cables : (1 grouped cable = 2 LV + 1 GND 16 mm2 + 2 sense)
SOLAR boards
624 SOLAR boards installed in the cavern
6 SOLAR boards per SOLAR Crate
No need of DC/DC
Need 14 LVPS PL512: 7 INSIDE , 7 OUTSIDE
As we recover the 8 LVPS PL512 from quadrants + 2 from CROCUS:  need 4 new PL512 LVPS  need 84 cables (1 cable = 1 LV + 1 GND + 1 sense)  4 new racks (2 on each side): 1 rack with 4 LVPS, 1 with LVPS

6. Power dissipation and voltage drop in cables
Quadrants
Cables: ~30 m of 16 mm2 section between LVPS and LV filter boxes + ~3 m of 10 mm2 between LV filter boxes and detector
For 85 A current, the voltage drop is : 3.3 V
The maximum voltage needed is: 1.7 V (DualSAMPA 1.2 V + LDO) V= 5 V
The maximum power consumption would be : 85 A x 8 V x 8 ch = 3.2 kW for the whole LVPS PL508 (max foreseen 3.6 kW) (3.6 kW is if we had all the 8 channels corresponding to 7 mW/FEC channel; in fcat we have half with 7 mW/FEC ch and half with 4mW/FEC ch: kW  2.9 kW)
Slats
Cables: N (20 + 5) m of 16 mm2 section between LVPS and LV filter boxes with a current max of 50 A and 3 m of 2.5 mm2 with 25 A (split lines) between LV filter box and detector.
The voltage drop is: 1.9 V
The maximum voltage needed is: 1.7 V (DualSAMPA 1.2 V + LDO) V= 3.6 V at DC/DC level  the power is 3.6 V x 45 A max = 162W if we consider a DC/DC efficiency ~80%, the power becomes 202 W  Voltage at LVPS: 8 V ! Power consumption: 202x12 =2.5 kW

7. DC-DC Studies - Overview
Investigate and determine suitable commercial DC-DC converters that would meet the higher current requirements of the FEE for the upgrade.
Few apt DC-DC converters were chosen and ordered evaluation boards of each for testing their performance within magnetic field (~0.7T)
0.6V-3.6, 18A out LMZ A paralleled DCDC converter board from Texas Instruments
0.6V-1.8V, 50A LTM4650 DCDC converter board from Linear Technology
0.6V-5V, 50A ISL8272 DCDC converter board from Intersil
0.6V-2V, 70A TPSM846C23 converter board from Texas Instruments
Tests of these DCDC evaluation boards were done first at CERN DSF Magnet LAB and back here at iThemba LABS Magnet LAB
Spare Shielded Filter box from P2 was used for the DCDC evaluation boards.

8. DC-DC Studies – Tests under Mag Field
LMZ A tested at CERN DSF Magnet LAB 0.5T):
Tests were done with input supply of 5V, DC load of minimum 18A, magnet with 0.5T field.
With no field, efficiency is seen to be better when board is perpendicular to field and same with increasing B field.
Output ripple like efficiency is better when board position is perpendicular to field at 0.5T field.
Output Ripple 7.6mV
Lemo, 90 deg, 0.5T
Output Ripple 11.6mV
Lemo, 0deg, 0.5T

9. DC-DC Studies – Tests under Mag Field
LMZ A tested at CERN DSF Magnet LAB 0.5T):
Testing the LMZ31710 with shielded integrated inductor at 18A was a starting point to understand effects of magnetic field on these converters.
The LMZ A evaluation board performed satisfactory both with and without shielding but at 18A opposed to 45A requirement of increased current of new FEE– we would have to place 5 LMZ31710 modules in parallel to achieve output of 45A.
This could mean the board size for DCDC would be greater and design of DCDC board to meet space constraints for Filter box will be crucial.
In addition, with the voltage drop of 3m length from LV filter to slats ~1.9V, total DCDC output voltage required =3.6V – this will be fine as output voltage for LMZ31710 is 3.6V.

10. DC-DC Studies – Tests under Mag Field
LTM A DCDC board tested at iThemba LABS Magnet LAB 0.7T):
Tests were done with input supply of 5V, DC load of minimum 50A, magnet with 0.7T field.
Tests without shielded filter – DCDC board failed within few minutes - temperature of DCDC module exceeded safe operating conditions due to interference of high B field – no results able to be taken.
Unlike the LMZ31710, LTM4650 integrated inductor not shielded - added cause to failure
Another note about the LTM4650, the maximum output voltage is at 1.8V – with the volt drop consideration across 3m length to slats of 1.9V to achieve 1.7V at the end of 3m, this DCDC board would not work. This is the same case with the 70A TPSM846C23 from Texas instruments with output voltage max of 2V which still not tested

11. DC-DC Studies – Tests under Mag Field
ISL A DCDC board tested at iThemba LABS Magnet LAB 0.7T):
Tests were done with input supply of 6V, DC load of minimum 50A, magnet with 0.7T field.
Shielded integrated inductor with module. Output voltage can be adjusted from 0.6V to 5V.
Initial tests without shielded filter – like the LTM4650, DCDC board failed within few minutes - temperature of DCDC module exceeded limit due to high B field.
Shielded LV Filter box from P2 modified and used to test ISL8272 –
at 20A DCDC board ran without fail with output ripple seen below at less than 5mV and efficiency 79.4%
At 45A DCDC board fails even within shielded filter box enclosure after 10 minutes

12. DC-DC Studies – Summary
The ISL A evaluation board performed satisfactory under but fails at higher current outputs of 45A as per specifications of current requirement for new FEE.
Shielding seems to be crucial when dealing with commercial DCDC modules with or without shielded integrated inductors – shielding of LV Filter box not sufficient for DCDC to sustain under magnetic field of 0.7T at higher currents.
ISL8272 at high current of 45A lasted longer with shielding enclosure even with failure – hence looking into even better shielding material might be best option going forward – order of Mumetal placed to retest with this board to evaluate best options going further.
In addition, with the voltage drop of 3m length from LV filter to slats ~1.9V (total DCDC output voltage required =3.6V) – this will be fine as output voltage for ISL8272 is 5V and this DCDC converter see to be ideal to be implemented if best shielding is established.
Also in parallel, contact with Synqor is maintained to see their solution for DCDC converters with shielding – their 50A DCDC converters are not tested for magnetic field conditions but they suggest shielded enclosure with their DCDC converters with 3 layer silicon steel, each layer 3mm thick and Dimensions 70mm X 70mm – to follow up and see feasibility for the LV system both on budget and upgrade constraints and maybe have a sample unit for testing.

13. Scenario: Checks to be done
DC/DC: other invesigations and tests
LVPS: Contact established with WIENER; no real answer yet.
Cables: Discussion with Arturo t-Tauro for the cables : information has been sent; discussion tomorrow.