[Rich] H12700 review

[Andrey]

Dear Marco,
Indeed all the questions are valid and necessary to understand in order 
to make a definite statement. We are working with Pavel to provide some 
information about his model and its motivations/assumptions.

To answer your questions about geographical dependence of SPE spectrum, 
we are going to apply small collimator to bring our beam spot size to 
smaller value in order to scan it more thoroughly, w/o overlapping.
*
Regarding independence of H12700 on HV:* I am glad you pointed it out. I 
didn't pay enough attention to this particular point, since gains 
themselves do not provide any criteria for MAPMTs comparison. The 
quantum efficiency doesn't depend on HV, and collection efficiency was 
smaller at any HV. It increased for H8500 at 1075V, but so should for 
H12700. You are right, the gains should increase as well. There are 
might be some problems with our HV setup, since slow control in CLAS was 
under some upgrade. Fortunately, it doesn't change final answer. 1000V 
is correct for sure, and collection efficiency is still smaller for 
H12700. Now if 1075V is in fact 1000V (HV slow control didn't work), it 
would bring collection efficiency for H12700 higher at 1075V. But I 
can't see any reason for it to be bigger than H8500 at 1075V. I will 
recheck this point though (thank you).

With that being said, it is extremely important that extraction of this 
efficiency relies heavily on the model framework. It leads to the your 
next question: the visible peak is indeed under the pedestal. These are 
the events, where photoelectrons were produced but weren't multiplied at 
all on first dynode due to the inefficiecy of electron multiplying 
system. And these events do heavily contribute to collection efficiency. 
*How Pavel's model determine how big is this peak?* From my 
understanding the main criteria is parameter myu - since it defines the 
relative number of 1PE events, 2PE events,... to the pedestal events. We 
could in fact remove this peak from 1PE, and include them into pedestal 
area, however, it would have a negative impact on 2PE, 3PE peaks, which 
would lead to the disagreement with data at higher amplitudes.

Also, just an assumption: As you can see from 1PE spectrum on page 11 
that H12700 events do not have apparent peak under pedestal, I believe 
that it might be an effect of Hamamatsu trying to optimize their dynode 
structure. Could it be that they succeeded to remove the areas where 
photoelectrons are not multiplied, but... these engineering solutions 
created large area on first dynode where electrons multiply with worse 
secondary emission ratio? Combine it with lower number of dynodes and we 
have smaller resultant collection efficiency. Meanwhile H8500 has areas 
of dynode without multiplication (peak under pedestal) and with 
"optimal" multiplication.
*These are clearly guesses,* that I am not sure we would be able to 
test, but it provides very brief insight how Pavel's model describes 
structures in H12700 spectrum.

In conclusion, I will add more slides with these corrections. 
Presentation will definitely be more complete with them.
Thank you,
Andrey.



On 08/20/2014 06:11 AM, Marco Contalbrigo wrote:
> Dear Andrey,
> thank you for this nice and interesting piece of work.
> We may discuss it this Friday at the RICH meeting, anyway
> I anticipate few comments.
>
> If this report is intended (also) for Hamamatsu I guess
> we should pay attention on completeness of information
> and motivation of the assumptions (model). This is
> important since your conclusions are somewhat in
> contradiction with Hamamatsu claim the new dynode
> structure is optimized for SPE.
>
> In this sense, I would say the report needs at
> minimum:
> - page number
> - a slide with the fitting formula/procedure
> - an explanation/motivation of Pavel's model
>
> but I think it calls also for - a reference SPE analysis where only 
> the center of
> the pixel is illuminated or a SPE analysis as a
> function of the laser spot within the pixel
> - a cross-talk analysis
>
> Indeed your conclusions heavily depends on the
> funny shape of the H12700 SPE signal. Thus one
> immediate question is how much this shape
> changes within the pixel and/or if it is
> affected by the cross-talk.
>
> I find odd that the H12700 gain is independent of HV
> (pag.8). Do you have an explanation for this ?
>
> On page 11, where the SPE signal is shown,
> are the dotted vertical lines the 5sigma cut ?
> How can be fit a (small, but evident for H8500) SPE
> peak under the pedestal ?
>
> Thnaks again, Marco.
>
> On Tue, 19 Aug 2014, Andrey Kim wrote:
>
>> Dear collaborators,
>> Please find H12700-H8500 comparison review attached.
>> "Short" summary: H12700 is still better than H8500 due to the higher 
>> photocathode
>> quantum efficiency. The nearly uniform illumination of pixel shows 
>> that SPE
>> spectrum is not as good as it was anticipated. It has more 
>> complicated structure
>> than H8500. The structure was described using Pavel Degtiarenko's 
>> model, and using
>> his model's fit results the collection efficiency was estimated for 
>> both MAPMT
>> types. The H8500 demonstrated higher collection efficiency than 
>> H12700 (unexpected
>> and rather saddening behaviour). Fortunately, H12700 has higher 
>> photocathode
>> quantum efficiency, and it was determinative factor in H12700 
>> superiority.
>> Please note that separation of quantum and collection efficiencies is 
>> possible
>> within model framework only. However, the global efficiency (decisive 
>> criteria) is
>> model independent as it is the fraction of events with signal over 
>> pedestal's 5
>> sigma (assuming, of course, that light source is stable and constant).
>> Any comments, suggestions or corrections are welcome!
>> Thank you!
>>
>>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/rich/attachments/20140821/338b84ab/attachment-0002.html>