K5-517 Module

1st Setup

K5-517, K5.1 hybrid with 1 dead channel per chip, production chips.
Valencia Test Box.
Grounding at VME: Shield - 10nF - VME ground.
Grounding at Support Card: AGND - DGND - Shield - Box.
Kapton on cooling block (isolated module).
Hyb. Temp. = 38慢.

Strobe Delay Scan
The Strobe Delay Scan did not show a tendency along channel number but showed some activity when the 4fC pulse is not present.

Strobe Delay Scan.

Then we played with grounding schemes configurations to see if this behaviour changes, the worse results was when the shield cable was not connected at neither end:

Grounding at VME: Shield - not connected.
Grounding at Support Card: AGND - Box (DGND and Shield not connected).
Kapton on cooling block (isolated module).
Hyb. Temp. = 38慢.
Strobe Delay Scan.

The best result, having Kapton on the cooling block, was for:

Grounding at VME: Shield - 10nF - VDD return (DGND).
Grounding at Support Card: AGND - DGND - Shield - Box.
Kapton on cooling block (isolated module).
Hyb. Temp. = 38慢.
Strobe Delay Scan.

Then, with this configuration, we removed the Kapton on the cooling block and the strobe delay was clean:

Grounding at VME: Shield - 10nF - VDD return (DGND).
Grounding at Support Card: AGND - DGND - Shield - Box.
Nothing on cooling block (connected module).
Hyb. Temp. = 32慢.
Strobe Delay Scan.

These results were repetible. The next day we could still see the same behaviour. Putting and taking off the Kapton we have extra activity or not. To keep in mind, the strobe delay is done in EDGE mode with the Threshold at 2fC. So, this extra activity when there is not a pulse is bigger than 2fC. It is infrequent, some channels fire once over 1000 events. With this configuration we characterised the module:

Three Point Gain (untrimmed)
The gain and noise versus channel number is flat. The mean input noise value for K5-517 is ~1520 e-.

Three point gain.

Trimming
The trimming gives 13 channels masked.

Trimming for K5-517.

Response Curve (trimmed)
The gain and noise versus channel number is flat. The mean input noise value for K5-517 is ~1520 e-.

Response Curve for K5-517.

Noise Occupancy (trimmed)
Occupancy @ 1fC is 1.8e-5.

Noise Occupancy for K5-517.

Timewalk (trimmed)
Timewalk is ~13ns.

Timewalk for K5-517.

During the Timewalk Scan, Strobe Delay Scans are made, in EDGE mode, with Threshold at 1fC and different input charges. We could see in the Strobe Delay Plots that there was some extra activity. However, when the Threshold is set at 2fC in a normal Strobe Delay Scan we don't see the extra activity.

Is this compatible with the noise occupancy scan?. It can be, several noise hits per event over 1000 events and 1536 channels can give the figure that we see in the noise scan (1.8e-5). But, we have never seen before this extra activity.

Is EDGE mode more sensitive to noise than LEVEL mode?. For comparison we have made a noise scan in LEVEL mode and inmediately later one in EDGE: