Star-fst-l mailing list

["Visser, Gerard" <gvisser AT indiana.edu>]

hi Zhenyu,
    All I am saying is that these plots do not indicate to me evidence of a 
problem. If the IST capacitance is "negligible enough" for instance it would 
look similar to FST unbonded. I wish we had such plots for an unbonded IST 
chip...
    The pattern of higher noise every four timebins for sure looks like 
something we don't want to see, but I think explanation for _that_ ties into 
some internal details of the APV chip that we have no choice but to live with.
   Anyway, I think we need to keep investigating, we agree. At the moment I 
would say I believe the FST noise looks a bit worse than IST simply because 
the capacitances are worse, but if we think the S/N ratio is still adequate I 
don't see a clear worry. Of course we should check everything that we can 
though.
  
       Gerard


p.s. At least the capacitance of IST to the backside contact / bias supply is 
<< than that of FST. This could be the capacitance that matters most for CMN. 
(I'd naively expect that capacitance to at least adjacent channels of same 
chip, does not generate CMN, or not significantly.)

________________________________________
From: Zhenyu Ye <yezhenyu2003 AT gmail.com>
Sent: Tuesday, August 11, 2020 7:14 PM
To: Visser, Gerard
Cc: Xu Sun; star-fst-l AT lists.bnl.gov
Subject: Re: [Star-fst-l] noise vs bias for worst channel and typical   
channel?

Hi Gerard,

In the other email, you argued that the CMN increases with input capacitance 
(I agree). But the argument you are making here, which is based on "CMN: IST 
with sensor ~ FST w/o sensor" and concluding there is no issue with FST, 
seems to neglect the fact that the input capacitance is different between IST 
with sensor and FST w/o sensor.

CMN:
FST with sensor >> FST w/o sensor ~ IST with sensor

CMN/Total
FST with sensor ~ FST w/o sensor >> IST with sensor

Zhenyu


> On Aug 11, 2020, at 5:59 PM, Gerard Visser <gvisser AT indiana.edu> wrote:
>
> hi Xu, Zhenyu,
>       I put the plot from IST that Xu sent the other day on a slide 
> (attached) together with the plot from FST03 with no detector that he sent 
> today, at the same scale. I think it looks fairly comparable, there is no 
> obvious reason I can see that these results indicate a problem.
>       I wish we had some understanding why the noise would be larger every 
> 4th timebin. I have no clue about that. It is especially odd since things 
> (readout, in particular) happen in the APV chip in groups of three 
> timebins, but I don't know of anything except some "black-box" internal 
> feature that happens in groups of four.
>       It is very likely related to some internal feature of the APV. That 
> doesn't mean we can conclude that it is not affected by some factor under 
> our control though, of course.
>       Sincerely,
>
>               Gerard
>
>
> On 8/11/2020 5:56 PM, Xu Sun wrote:
>> Hi Gerard,
>> As Zhenyu mentioned, please find the plot of noise before sensor mounting 
>> (APV chips only) of FST03 attached. Please ignore the APV 7 which shows an 
>> abnormal behavior, it is from the cable we used to take the noise run.
>> We see the large CMN with strong time-bin dependence with the noise from 
>> APV chips only. This time bin dependence is the same as noise with sensor 
>> mounted, therefore it is very likely from APV chips on FST.
>> Best,
>> Xu
>> On Tue, Aug 11, 2020 at 4:28 PM Zhenyu Ye <yezhenyu2003 AT gmail.com 
>> <mailto:yezhenyu2003 AT gmail.com>> wrote:
>>    Hi Gerard,
>>    The total capacitance of a readout strip for silicon strip detectors is
>>    given by the sum of the strip-backplane capacitance, and the inter-strip
>>    capacitance. The latter usually dominates.
>>    As we know, the APV noise is linearly dependent on input capacitance, 
>> i.e.,
>>    the total capacitance of the readout strip. For FST, we have seen that 
>> outer
>>    (larger R, Rstrip3 for the inner sensor or Rstrip7 for the outer sensor 
>> in
>>    Xu’s plots) strips have much smaller noises than the inner strips 
>> (Rstrip0
>>    for the inner sensor or Rstrip4 for the outer sensor), despite the fact 
>> that
>>    outer strips have larger area and thus large strip-backplane 
>> capacitance.
>>    This is consistent with the above statement that the interstrip 
>> capacitance
>>    dominates.
>>    The CMN, by definition, affects a group of channels in a coherent way. 
>> It is
>>    usually caused by a common electromagnetic pick-up, or noise on the 
>> supply
>>    voltage, etc. The fact that we see large CMN in FST, suggests to me 
>> that we
>>    should check FST inner cable/T-board/hybrid, such as grounding.
>>    P.S. Xu will send some plots on FST prototype modules before sensors 
>> were
>>    mounted. We see large CMN with strong time-bin dependence. I think this 
>> is
>>    consistent with my above assessment.
>>    Best,
>>    Zhenyu
>>     > On Aug 10, 2020, at 3:45 PM, Gerard Visser <gvisser AT indiana.edu
>>    <mailto:gvisser AT indiana.edu>> wrote:
>>     >
>>     > hi Xu, Zhenyu,
>>     >       Thanks; I didn't realize about this timebin dependence in the 
>> IST.
>>    By the way do we see any dependence of noise on the "cap id" (i.e. the
>>    "address" reported in the APV header)? I am assuming we are triggering 
>> at a
>>    low rate and asynchronous to the clock (ARC-II local clock) so we 
>> should be
>>    getting data in all values of capid. This could show some features (and
>>    perhaps can allow for some capid-dependent correction applied offline).
>>     >
>>     >       I make some estimate of the capacitances as follows: 
>> Neglecting the
>>    capacitance to neighbor pads, and all the capacitance of the routing 
>> line on
>>    the detector, let's consider only the infinite-parallel-plate 
>> capacitance in
>>    bulk of the detector:
>>     >
>>     > IST: Pad size is 594 x 6275 um, thickness 300 um, k=11.7  ==> C=1.3 
>> pF (I
>>    think this must certainly then be dominated by the other, neglected
>>    capacitances mentioned above).
>>     >
>>     > FST: worst case outer pad size is (about) 1087 x 28750 um. (Right? 
>> If you
>>    have more precise info please say.) Also 300 um thick. ==> C=10.8 pF. (+
>>    other again)
>>     >
>>     >       If you have some info about the gap in contact/metal between
>>    adjacent pads in the case of IST and FST I could try to roughly 
>> estimate the
>>    perimeter capacitance to neighbor pads.
>>     >       If you have some measured capacitance info or real calculated
>>    capacitances from detector design, of course we could better think 
>> about those.
>>     >       Anyway, my guess is that the capacitances in FST (outer at 
>> least)
>>    are probably 2-3x the capacitances in IST. This is probably responsible 
>> for
>>    the larger common mode noise and larger noise. If so unfortunately it
>>    probably means there is nothing that we can do about it.
>>     >       If we have an IST stave with a defective detector and wanted 
>> to do
>>    further tests, we could bond some APV input pads to test capacitors of 
>> value
>>    similar to the FST detector and see how that looks. I think this could 
>> be a
>>    significant effort though.
>>     >       Really the only question that we must answer is whether the 
>> PPB,
>>    purple cable, T-board, and hybrids are working well together to deliver
>>    clean supply voltages and clock/trigger signals to the APV chips (and 
>> clean
>>    bias to the sensors). This is probably so, I think all your noise plots 
>> look
>>    reasonable, but we could try to check more directly with low noise 
>> probing
>>    of the supply voltages on one of the prototype hybrids. We probably 
>> would
>>    have done this already if it weren't for the virus situation.
>>     >       Sincerely,
>>     >
>>     >               Gerard
>>     >
>>     >
>>     > p.s. I suppose that, at least for lower trigger rates <3 kHz or so, 
>> we
>>    should consider to setup to read 4 timebins and ignore 0 in offline, use
>>    only 1-3. If 3 timebins is all that we really need, that is. What do 
>> you think?
>>     >
>>     >
>>     > On 8/10/2020 2:29 PM, Xu Sun wrote:
>>     >> Hi Gerard,
>>     >> Sorry for the late reply. Please find the FST & IST noise study in 
>> the
>>    attached file.
>>     >> I see a similar behaviour for IST with a much smaller magnitude.
>>     >> Best,
>>     >> Xu
>>     >> On Tue, Aug 4, 2020 at 1:12 PM Gerard Visser <gvisser AT indiana.edu
>>    <mailto:gvisser AT indiana.edu> <mailto:gvisser AT indiana.edu
>>    <mailto:gvisser AT indiana.edu>>> wrote:
>>     >>    Hi Xu, Zhenyu,
>>     >>             Do we see the timebin-dependece of noise as you show 
>> here
>>     >>
>>    
>> https://drupal.star.bnl.gov/STAR/event/2020/05/04/star-forward-silicon-tracker-meeting/prototype-module-assembly-and-test
>>     >>    in the IST data too?
>>     >>             Thanks,
>>     >>                     Gerard
>>     >>    p.s. And, if possible to answer, there is also the question 
>> whether
>>    this was
>>     >>    seen in the IST installed in STAR/HFT? I don't remember hearing 
>> about
>>    it before.
>>     >>    On 8/4/2020 12:39 PM, Zhenyu Ye wrote:
>>     >>     > Hi Gerard,
>>     >>     >
>>     >>     >> On Aug 4, 2020, at 11:19 AM, Gerard Visser 
>> <gvisser AT indiana.edu
>>    <mailto:gvisser AT indiana.edu>
>>     >>    <mailto:gvisser AT indiana.edu <mailto:gvisser AT indiana.edu>>> wrote:
>>     >>     >>
>>     >>     >> hi Zhenyu,
>>     >>     >>      That timebon dependence sounds definitely odd. Are you 
>> sure?
>>    Do we
>>     >>    see that in IST too, only more mildly? I wasn't aware of this.
>>     >>     > ‘
>>     >>     > Please take a look at Xu’s presentation in FST meeting on May 
>> 4
>>    (FST) and
>>     >>    11 (FST and IST):
>>     >>     >
>>     >>     >
>>     >>
>>    
>> https://drupal.star.bnl.gov/STAR/event/2020/05/04/star-forward-silicon-tracker-meeting/prototype-module-assembly-and-test
>>     >>     >
>>     >>
>>    
>> https://drupal.star.bnl.gov/STAR/event/2020/05/11/star-forward-silicon-tracker-meeting/prototype-module-assembly-and-test
>>     >>     >
>>     >>     >>      The capacitance of FST is much larger than IST, I think.
>>    This may
>>     >>    certainly be relevant. For sure it is relevant to CMN.
>>     >>     >>      Anyway, I agree we should investigate these noise 
>> issues, I
>>    do not
>>     >>    like to ignore them. However on the other hand it _may_ be an 
>> inherent
>>     >>    property of APV chips.
>>     >>     >
>>     >>     >>      - Gerard
>>     >>     >>
>>     >>     >>
>>     >>     >>
>>     >>     >> On 8/4/2020 12:12 PM, Zhenyu Ye wrote:
>>     >>     >>> Hi Gerard,
>>     >>     >>>  From the plots that Xu sent, there is a clear pattern 
>> where the
>>     >>    channels showing enhanced noise level after mounting the sensors,
>>    also show
>>     >>    enhanced noise level before mounting the sensors.
>>     >>     >>> An independent topic, we see that the CMN in FST is much 
>> higher than
>>     >>    IST, and show a strong time-bin dependence, i.e., when we read 
>> in 9 time
>>     >>    bins, the 1st, 5th and 9th time bins have much higher CMN than 
>> the other
>>     >>    time bins. I don’t feel comfortable to ignore it w/o knowing the
>>    cause, as
>>     >>    it may get worse in the real experiment.
>>     >>     >>> Best,
>>     >>     >>> Zhenyu
>>     >>    _______________________________________________
>>     >>    Star-fst-l mailing list
>>     >> Star-fst-l AT lists.bnl.gov <mailto:Star-fst-l AT lists.bnl.gov>
>>    <mailto:Star-fst-l AT lists.bnl.gov <mailto:Star-fst-l AT lists.bnl.gov>>
>>     >> https://lists.bnl.gov/mailman/listinfo/star-fst-l
>>     > _______________________________________________
>>     > Star-fst-l mailing list
>>     > Star-fst-l AT lists.bnl.gov <mailto:Star-fst-l AT lists.bnl.gov>
>>     > https://lists.bnl.gov/mailman/listinfo/star-fst-l
> <cmn_noise_comparison.pdf>_______________________________________________
> Star-fst-l mailing list
> Star-fst-l AT lists.bnl.gov
> https://lists.bnl.gov/mailman/listinfo/star-fst-l