Transcript A3016 test at CAEN

A3016 and MAO test at CAEN
Pigi, A. Boaino & CAEN
Test Schema
SY1527
3-phase
1. MAO 48 service from BC
2. EASY 3000 – Joint 48 Power & Service
48 service
3. Dummy LinkBoard load: R & C
4. Output power from A3016:
MAO
48 power
A301
6
load
–
–
48 service
•
EASY
2.5 Volts / 5.0 Amp
3.5 Volts / 10.0 Amp
MAO output current = 5 A
–
With a A3009 board more = 17 A
What powering OFF the MAO in the proper ways ?
• Powering OFF the MAO via SY1527 or using the interlock
placed on the front
What powering OFF the EASY in a “bad” way ?
• Powering OFF the EASY 3000 crate removing the power
or service connector from the back
What cutting the MAO 3-phase ?
• Shutting-down the MAO cutting 3-phase
• oscillation does not depend from
the MAO load (from 4 A to 18 A)
• sometimes 2 oscillations in few
msec
• oscillation disappear separating
the 48 power from the 48 service
• No other ways to reproduce it
What happen cutting MAO 3-phase ?
• MAO goes OFF in 2 steps: a first from 48 V to 22 V and a second
from 22 V to 0 V (slower)
• A3016 channels are switched OFF in 200 msec after 600 msec
from the MAO shut-down (software delay to be sure that was
not a voltage oscillation)
48 V
A3016 switched OFF
2.5 V
Step #1
Step #2
When and why the oscillation is generated cutting
the MAO 3-phase ?
• The oscillation on the A3016 output is generated when
the MOA output cross 22 Volts
• An oscillation has been seen when Vcc goes from 12 V to
6 V and this produce the spike.
Vcc = 12 V
Vmao = 22 V
6V
How prevent this spike during a MAO power-cut ?
• We can prevent the oscillation and spike separating the
48 service and power of the EASY crate as suggested by
the CAEN since long time.
• To do that we need to modifying the cable routing and
have a dedicated MAO under UPS to generate the 48
service.
• A faster solution is the put a “big” capacitor between the
48 power and service in order to give a couple of second
more to the service and let the EASY crate switching off
in the proper way.
• A capacitor of 33 mF + a protection diode are ok
Results with C and diode
service
power
1.5 sec
Results with C and diode
No spikes or oscillations
48 V
600 msec
2.5 V
Conclusion on the MAO
• MAO cannot reboot itself even in case of
“bad” shut-down.
– Need to clear the status alarm to switch back it.
• In case of SY1527 reboot the MAO and the LV
channels stay OFF until a “clear alarm”
• CAEN recommends to use 2 different lines for
the power and service of the EASY crates.
Conclusion on the A3016
• The LV board has been designed as a general purpose
board, for ALICE, ATLAS and CMS
– The output feedback circuit has not been optimized to supply
the link board (one crate has a large input capacitance)
• Let me remember you that Link Baord does not have any
regulator or protection on the power lines to reduce the
power dissipation.
• Oscillations have been detected during the switching ON of
the channels and have been partially cured via firmware 
we still need to work on that and on the 2 channels sync.
• An oscillation has been detected, during the last week test
at CAEN, on the A3016 channels in case of improper shutdown of the MAO (3-phase cut)
– Can this oscillation break the Link Board ?
– How can we can solve this problem and all the other possible
problems ?
Conclusion on the A3016
Long Term Solutions:
1. Designing a protection on the LB crate LV lines
2. Designing a “special” output feedback circuit of the A3016 for
the link board
Temporary Solutions:
• Working on the firmware in order to reduce problem during
the switching ON
• Synchronize the switching ON of the 2 independent channels
via firmware (now is via PVSS)
• Separating the 48 service and power of the EASY
– OR
• Introducing a big capacitor (33 mF) between the service and
power line of the EASY:
– Solution has been successfully tested at CAEN
– It does not have any side effect on the system
Action Items
1. Capacitor test at 904 and eventual installation
2. Definition of the synchronization requirements
and ramping-up sequence.
– two channels at same time ? How many steps ? How
long the sequence ?
3. Official request at CAEN of a new “firmware” to
develop and test asap and CMS validation.
4. Design of the LV lines protections