E U S O - B A L L O O N D E S I G N R E V I E W, 1 8 . 1 2 . 2 0 1 2 , C N E S T O U L O U S E

P. GorodetzkyAPC, ParisWork done with C. Blaksley at APCHV were conceived and produced by J. Szabeski and J. Karczmarczyk, Lodz, Poland

HV status and prototype results.Compliance with performance and technical requirements

HV Status 2

REQUIREMENTS:The High Voltage unit should be:

-Stable (HV voltage variations should have an effect on PMT output resolution negligible compared to PMT resolution).- remotely controlable- protected- protecting the PMTs to less than 100 µA output (Hamamatsu requirements)- allow dynamics from 0.01 photo-electron (pe) to 200 pe- the dynamics with the switches should be around 106 (ratio between a lightning and the background light). - PMT output should be linear with light flux falling on the PMT- If the light is larger than what is safe for PMT, the HV should switch in a mode allowing to continue measuring that strong light.-the switching time should be short and no longer than a GTU (2.5 µs)- size and weight should be small enough for Jem-Euso requirements (to hold in PDM system)- power should be less than 0.5 W per PDM for Jem-Euso and < 1 W for the balloon.

The Gate Time Unit (GTU) is 2.5 µs. The gain of the PMT is fixed (requirement of the ASIC) to 1.0 106. The light background of the balloon is about 1 pe / GTU / pix, that is 0.16 pC in 2.5 µs. Then the average current produced by the PMT (which has 64 pixels) is 64 x 64 nA = 4.1 µA.A strong light (lightning, city…) is limited in spatial extension, corresponding to some 8 pixels (a KI is 8 pixels). Then, we have 64 – 8 = 56 pixels with 64 nA each (3.6 µA), and 4 pixels with 200 pe, that is 102 µA. The total (106 µA) is close to 100 µA, and acceptable.As one CW will power 4 PMTs in parallel (an EC), the current to be delivered by the CW has to be slightly higher than110 µA (the strong light falls only on 8 pixels of the EC).

CDR – 18.12.2012

HV Status 3

REQUIREMENTS:

Comparison to the classic voltage divider:We can have 100 µA in the PMT, in a linear way, but only sometimes. This means that the current in the voltage divider must consume continuously at least 50 times 100 µA, that is 5 mA. At 1000 V, this is a power of 4.5 W per PMT.

For the balloon: 162 W, but, for Jem-Euso, whose balloon project is a patfinder:, it would be 22 kW. Exit the divider. We need something that provides to the dynodes exactly the current they need, getting rid of resistors. The best is a voltage multiplier, also known as a Cockroft-Walton (CW), taking into account the fact that the inter-dynode voltage is constant: here about 60 V. A big advantage of this scheme is that, the impedance is the lowest close to the anodes, where the current is the biggest, and at the cathode level , where we deal with individual electrons, the impedance is the highest.

IPMT

Idivider

IPMT and Idivider are in parallel

CDR – 18.12.2012

HV Status 4CDR – 18.12.2012

HV Status 5Command for switches Cockroft-Walton and switches

The prototype in the black box. It is connected to 4 PMTs in parallel, simulating an EC.The light source for background is a LED in a sphere equipped with a NIST photodiode.The light source for intense and localized light is here:

The switchesThe CW. Their size is 7 x 3 cm, hence 9 of them can be installed in one PDM.The switches command

CDR – 18.12.2012

HV Status 6

The CW gain is 1.2 times larger than the Hamamatsu divider gain.

Oscilloscope view of the 4 PMTS (anode 21 in each). All of them are illuminated uniformly on their full surface by the sphere at background level (1.2 pe/pix/GTU)The third one is also illuminated by the strong light on 8 pixels only at 150 pe/pix/GTU.

CDR – 18.12.2012

HV Status 7

Results for the CW

In abscissa, the number of photoelectrons per pixel in the PMT #3, the highly illuminated. Only 8 pixels receive the strong light.

In ordinate, the current in the last dynode D12, adding the 4 PMTs.

The linearity satisfies the requirement: the HvPS will support strong lights without loss of linearity.

The power consumed (28 V and 3.3 V) is 70 mW, so 630 mW for the balloon with requirements < 1 W. The requirements of Jem-Euso: about 500 mW for a PDM.

CDR – 18.12.2012

HV Status 8

Gain of one of the PMTs versus the control voltage (DAC acting on the multiplier frequency)

Voltage to apply to a resistive divider (Caen volts) to have the same gain than with the CW, versus the CW control voltage.The gain can be remotely varied from 0.8 106 to 3.2 106. The requirement is to be around 1.0 106.

CDR – 18.12.2012

HV Status 9

The basic idea behind the switches is to vary the cathode voltage while keeping all other voltages on dynodes constant. Then, the electrostatics between the cathode and dynode 1 are modified, and the collection efficiency is diminished. The real gain of the PMT is kept at its original value (around 106), so that the tube is always working in single photoelectron mode, whatever the light intensity (no pile-up at the anodes). On the ordinate, the word "gain" is improper, but it is easier to understand.The dynamics allowable are > 106, which corresponds to the requirement.

The switches: how they work

The switches are actuated by the KI (integrator part of the front-end ASIC) through a simple algorithm in the PDM board FPGA.

CDR – 18.12.2012

HV Status 10

Timing properties of the switches

When we test the switching, we apply squares pulses (1.2 s at high gain and 1.2 s at low gain) to the 4 PMTs of the EC. The light is DC.In the next scope pictures, the anode pulses are shown in violet, and the command in green.

Vcw = 2.005 V (G = 1.4 106), switching from 900 V (G = 1.4 106) to 739 V (G = 1.4 104) and vice-versa

CDR – 18.12.2012

HV Status 11

Vcw = 2.005 V (G = 1.4 106), switching down from 900 V (G = 1.4 106) to 739 V (G = 1.4 104)

Very fast, very important

Same pictures when going from 739 v to 250 V

CDR – 18.12.2012

12

Vcw = 2.005 V (G = 1.4 106), switching up from 739 V (G = 1.4 104) to 900 V (G = 1.4 106)

Here we pulse also the light, that is the light is back to the background level when we switch the gain up. Actually, the bckg level has been increased by 10 (ID12 = 160 µA instead of 16 µA).This is close to what happens in reality.

Now the recovery time is < 4 µs

Same pictures when going from 739 v to 250 V

Bckg multiplied by 10, and no strong lightHigh gain (900 V)

Bckg multiplied by 10 and strong light (100 times Bckg on.Low gain (739 V)

Conclusion: good PMT protection AND no dead timeCDR – 18.12.2012 HV Status

HV Status 13

We illuminate with 20000 pe/pix/GTU

"Gain" is 1.4 104 with the switch: we see pulses at the amplifier output of the ASIC, hence the KI (who is a "time over threshold" integrator in the ASIC), works.

We illuminate with 20000 pe/pix/GTU

Gain is 1.4 104 without the switch:We reduce the voltage of all dynodes proportionally: Only pile-up, -8mV DC level, no pulses and the KI does not work.

Difference between switches and conventional gain reduction:

1.0 µs

1.0 µs

CDR – 18.12.2012

HV Status 14

Conclusion on switches

a) They work as intended Gains are OK Going down in gain in less than 2. 5µs Going up is slower (4 µs), but perfectly OK

b) Dynamics: at 900 V, we have from 0 to 200 pe (per pix and per GTU) emitted by Kat 739 V, we have from 150 to 20000 peat 250 V, we have from 1.5 104 to 2 106 peat 0 V, we have from 1.5 106 to 3.4 106 pe

total dynamics > 3 106

CDR – 18.12.2012

HV Status 15

Where are the HVPS

2 potted HVPS boxes(6 CW in one, 3 CW + connection to HK in the second).

Cables go here

CDR – 18.12.2012

HV Status 16

Cables between potted HV boxes and potted HV board under PMTs

9 x 14 cables

Tyco-Raychem TE: 44A0312-24-9Rated at 2000 VDiameter: 1.45 mm

Connectors: 2 x PD-9, from Mac8, phi = 0.45 mm

PTFE tube: 1.6 x 3.0 mm x 3 cm

1 2

3

CDR – 18.12.2012