VTT ASSEMBLY 5CHARACTERISATION

KATE DOONAN & RICHARD BATES

2012

SETUPROIC is 200um thick, no special thin device treatment during the re-flow process

VTT Assembly 5 was operated:

• in the dark

• at -100V and

• at room temperature (20C)

INITIAL TESTS• Analog and Digital scans looked good

• Noise occupancy scan and subsequent mask revealed noisy pixels in the 2 right hand columns which were to be masked

TUNING DONE WITH THRESHOLD TARGET OF 3200ELECTRONS

Prim list was:

IF TUNE

GDAC (note in STControl version 3.0 need to run Fast_GDAC)

TDAC

FDAC

TDAC

Repeating the FDCA, TDAC loop three times improves the uniformity of the Vth 50% point.

AFTER 1ST TDAC/FDAC/TDAC TUNE CYCLE

THRESHOLD SCAN NOISE -

Very non-Gaussian threshold – misses target value of 3200electrons

Mean = 130.7Sigma = 7.0

Mean = 3314Sigma = 349.8

AFTER 2ND TDAC/FDAC/TDAC TUNE CYCLE

THRESHOLD NOISE

Mean = 3194Sigma = 26.77

Mean = 134Sigma = 6.54

Significant improvement in threshold distribution

AFTER 3RD TDAC/FDAC/TDAC CYCLE

THRESHOLD NOISE

Still a slight tail to lower thresholds for several columnsSlight improvement in the target but probably not worth the extra tune cycle

Mean = 132.9Sigma = 6.36

Mean = 3200Sigma = 26.91

HUNTING FOR NON-BONDED PIXELS• Perform threshold scans

with and without Detector HV

• With HV on, sensor is depleted and there is low noise on all pixels

• With HV off, disconnected pixels will exhibit just the noise of the bare FE (around 120e)

• Small bump can be seen around 120e

• Zoomed in noise plot

• There are a number of pixel which appear to be unbonded

• They are in the top corner sections of the assembly

• Number of un-bonded pixels is 630.

• 630/26,400 = 0.0238

• 2.4% is too high

• Data to be fed back to VTT to understand how to over come bow problem

DISCONNECTED PIXELS

CROSSTALK SCAN: MERGED BUMPS• With high voltage off, crosstalk

is seen due to the bulk of the sensor being undepleted

• With high voltage on, any crosstalk is due to merged bumps

• This happens at columns 0, 78 and 79 which are masked out

• No merged pixels are visible

AMERICIUM-241 SOURCE SCAN• Columns 0, 78 and 79 masked

• Full sensor illumination achieved with 12million events

• Strange pattern seen at corners identified in previous scans

• Placed source over each corner to investigate

CORNER BEHAVIOUR• Scan of 12million events at

top right hand corner shows a dark section where full illumination is expected

• Scan of 4million events at top left hand corner shows similar effects

CORNER BEHAVIOUR

CONCLUSIONThere is some form of disconnection at the corners of the sensor

• They do not perform as well in the threshold noise scan

• Analysis on going• They show much less illumination in the source scan although the

source is placed directly over these corners

• Analysis on going• 2% disconnected bumps at the corners probably due to ROIC bow

during the reflow process