Transcript High Voltage power supplies HV Switches - Euso

High Voltage power supplies
HV Switches
Design:
J. Szabelski, J. Karcmarczyk, Lodz, Poland
Tests:
P. Gorodetzky, C. Blaskley, S. Selmane, APC, Paris
For the JEM-EUSO collaboration
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
1
MAPMT polarization
Current flowing in PMT loads
the voltage divider. For JemEuso, and balloon, with this
design, we have 0.1 W / EC (4
PMTs) and for 1233 ECs, we
have 123 W. Too much.
Also, impedance to the different
dynodes is given by the voltage
divider resistors: impedance is
high, then dynamics are low.
Solution: Cockroft-Walton (CW): a
succession of voltage multipliers.
Works because most voltages between
adjacent dynodes are equal (60 - 65 V
for Hamamatsu M64)
Also, impedances are low at the first
dynode, raising when going towards K.
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
2
Cockroft-Walton
Remote commands:
End of phase A meeting
- total ON-OFF
- variation of the "60 V", to adjust the gains between 1 and 5 106
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
3
Tests at APC
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
4
Cockroft-Walton tests on one PMT
1 nW on sphere NIST = 3000 ph / sec (1000 pe / s) per pixel
Total power for 137 PDM (Jem-Euso) if one CW per PDM = 6 W. It is 50 W if one
CW per EC. Safety factors for Jem-Euso are 1/137 or 1/1233. For the balloon, they
are 1/1 or 1/9. So it is here mandatory to have one HV per EC.
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
5
HV Switches
LIGHTNINGS AND EXTREMELY BRIGHT SIGNALS
A system of HV switches is foreseen in JEM-EUSO to
reduce the PMT current when the PMTs are subject to very
bright signals. A lightning is more than 106 times more
brilliant than the UV background.
An example is described in the following.
The balloon could give us the opportunity to check the
functioning of this switch system. In principle the current
level to apply the switch can be arbitrary decided so that it
could be tested even in presence of city lights, if the balloon
will not be exposed to lightning.
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
6
Original idea by M. Sato
-640
-900 V
-860 V
End of phase A meeting
-640 V
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
7
#7
Switch organigram
The light amount is measured by the KI: charge integrators
located in the LAL ASIC digitizing the charge of 8 anodes
every GTU (2.5 µs). This is counted in the FPGA of the
PDM board, in which this organigram is written. The
dynamics of the KI are a factor of 100, from 2.5 pC to
250 pC. So, 4 switches are enough to go from a gain of 106
to a gain of 100.
However, changing the voltage of the cathode only can
change the gain such as
illustrated in the adjacent
switch logic.
The PDM board cannot give more than 4 lines of signals to
the switches. We decided that the full PDM would switch
gain at the same time. Then 4 lines (one per gain) are
enough.
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
8
The Lodz switches
Power measured at APC was about 6 W. However, most of it was for the
GTU generator which will be the PDM board in the final version. So, the
power due to the switches only, and not their command, is virtually 0.
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
9
Switching in one PMT
Going from HIGH to LOW
Going from LOW to HIGH
(one square is 4 µs)
One square is 100 µs
The overshoot (not dangerous in that
direction) has since been corrected.
Only of two days ago do we have at APC 4 M64 MAPMTs to make an EC, and the corresponding CW and
switches (the prototype at the right size). Tests will be done immediately after this meeting.
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
10
Simulations of strong light problems
(M. Bertaina)
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
11
Development plan
Many prototypes have been realized in Lodz, and their tests at APC showed they worked perfectly. We are
building actually a qualify model with the right technology and dimensions to be tested in February 2012 at
APC. Then we start to make a flight model.
The difference between them is that the FM will be potted.
It will require optimization in the cabling in order to ensure that all remnants of high frequency are
eliminated. The FM will be tested at APC at 3 mbar in a 1 m3 vessel in order to mitigate the risk of Corona
discharge at PMT-HV with damage to FEE.
The HV cabling is the object of a special study which is now finished and needs to be tested with the flight
model (it cannot be made with the prototypes because they are independent of the EC-boards) Once the
design is completed, and the prototypes OK, the Flight Model will be produced (by the industry, in Poland)
=> assembly and potting.
These FM elements will be delivered to APC for the acceptance and validation tests which will include
tests in an opaque vessel set at 3 mbar with every component (a mock-up of an EC, that is 4 PMTs, the
HV cables and the HV box) potted inside, and everything working.
There will be at least 2 tests at 3 mbar: one for the HV box to deliver something that does not spark and
one for the EC-boards with the MAPMT soldered and powered. Here, we are dealing only with the test
related to the HV box.
Integration inside the PDM will follow. Then the PDM could require a test at 3 mbar, however, each
component apt to spark will have been previously tested.
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
12
Risk analysis
End of phase A meeting
CNES (Toulouse) 2/02/2012
Ph. Gorodetzky APC - Paris Diderot
13