Pixel Alive test results: inconsistency in Pixel Alive and Address decoding errors.

[messineo]

Dear all we have identified a problem in "pixel alive" test: the detected pixel alive problems and address deconding errors are compatibile in the histograms and are not in the log summary. We attach here the log file and histograms for a ROC test. Test is performed with 150V bias. Alberto

Log File starts here

16:48:33.725] INFO: ###################################################################### [16:48:33.725] INFO: PixTestAlive::doTest() [16:48:33.725] INFO: ###################################################################### [16:48:33.725] INFO: ---------------------------------------------------------------------- [16:48:33.725] INFO: PixTestAlive::aliveTest() ntrig = 10, vcal = 200 (ctrlreg = 0) [16:48:33.725] INFO: ---------------------------------------------------------------------- [16:48:34.504] INFO: Test took 778ms. [16:48:34.590] INFO: PixTestAlive::aliveTest() done [16:48:34.590] INFO: number of dead pixels (per ROC): 791 [16:48:34.704] INFO: ---------------------------------------------------------------------- [16:48:34.705] INFO: PixTestAlive::maskTest() ntrig = 10, vcal = 200 (ctrlreg = 0) [16:48:34.705] INFO: ---------------------------------------------------------------------- [16:48:35.432] INFO: Test took 727ms. [16:48:35.515] INFO: PixTestAlive::maskTest() done [16:48:35.515] INFO: number of mask-defect pixels (per ROC): 0 [16:48:35.643] INFO: ---------------------------------------------------------------------- [16:48:35.643] INFO: PixTestAlive::addressDecodingTest() ntrig = 10, vcal = 200 (ctrlreg = 0) [16:48:35.644] INFO: ---------------------------------------------------------------------- [16:48:36.421] INFO: Test took 777ms. [16:48:36.509] INFO: PixTestAlive::addressDecodingTest() done [16:48:36.509] INFO: number of address-decoding pixels (per ROC): 0 [16:48:36.584] INFO: PixTestAlive::doTest() done

[messineo]

I forget to add the information: FW version is 3.1 and SW version 3.3.

[ursl]

Hi Alberto,

I am not sure I understand correctly your point. Do you argue that a dead pixel should also be counted as an address decoding error?

At the moment, only the non-dead pixels with address-decoding pixels are counted. I think this is a feature.

Cheers, --U.

PS: Both address-decoding errors and dead pixels are displayed as "0" in the map. Iirc there was a discussion on this in the past, and people thought it having a three-state (alive, dead, address-decoding problem) display was too confusing.

On Wed, Sep 24, 2014 at 12:45 PM, messineo notifications@github.com wrote:

I forget to add the information: FW version is 3.1 and SW version 3.3.

— Reply to this email directly or view it on GitHub https://github.com/psi46/pxar/issues/221#issuecomment-56653105.

[messineo]

Dear Urs, the problem is the following (I will try to summarize my understanding) Both 2D histograms (Addres decoding and Pixel alive attached in the issue report) show the same failures. In the log file we have listed “791" pixel alive test dead pixels, and “0” address decoding errors. First point : I suspect that the problem of pixel address decoding errors triggered the pixel alive dead, it is true ? If this is the case, and it is relevant to know, the address decoding error should not be 0 in the log file. Anyway the 2D histogram of address decoding errors should be correctly summarized in the log, so if failures are "0" the map should be all red.

Best regards Alberto

[ursl]

Hi Alberto,

the problem is the following (I will try to summarize my understanding) Both 2D histograms (Addres decoding and Pixel alive attached in the issue report) show the same failures. In the log file we have listed “791" pixel alive test dead pixels, and “0” address decoding errors.

I would think that you have 791 dead pixels. Or your setup is not optimal, such that for those pixels (they are in a contiguous region) the signal is recorded in a neighboring WBC. You could play with ctrlreg -> 4, VCAL, VthrComp, and caldel, to check whether the pixels are really dead or not.

First point : I suspect that the problem of pixel address decoding errors triggered the pixel alive dead, it is true ? If this is the case, and it is relevant to know, the address decoding error should not be 0 in the log file.

Not sure about this. If the pixel gives no response for your setup, then my initial guess is that this is not from address decoding problems. You could run with -v DEBUG, this will print out explicitly all pixels with address decoding problems. They are characterized by having a pixel readout value < 0. For display purposes this negative value is reset to 0.

Anyway the 2D histogram of address decoding errors should be correctly summarized in the log, so if failures are "0" the map should be all red.

The histogram is simply a visualization of pixels with problems. Non-zero usually means something good.

We can change the logic of the display, no problem. Let's discuss this on Friday in the workshop. We'll have to keep moreWeb in the loop, however, as they look at this (and take out the overlap with dead pixels). I have no (strong) opinion on what is displayed in which form.

Cheers, --U.