Re: BCAL timing

[Blake Leverington <leverinb@uregina.ca>]

Hi David,

I did a quick fit to your corrected resolution_#deltat vs Ephoton plot I.
    TF1 *f1 = new TF1("f1","sqrt(([0]*1/sqrt(x/1000))^2 + ([1])^2)  
",150,650); 
The result is in the plot with the energy dependent term being 152ps and 
the constant term is 53 ps. This is for the sigma of (tn8+ts8)/2.0 - 
(tn7+ts7)/2.0 so the resolution is sqrt(2) smaller than that as I 
understand it for reading one end.

-Blake

David Lawrence wrote:
>
> Hi Blake,
>
>    To minimize the contribution of the timewalk in tn8 to the tn7-tn8 
> time difference, I cut on n8>1000. Likewise, for the timewalk 
> functions of the other tubes, I cut on the "reference" channel's ADC 
> value being large. Granted, this will not remove it completely, but it 
> does reduce it down to a level that is not bigger than the intrinsic 
> spread of the timing distribution of the PMT. Of course, the procedure 
> could be iterated using the previous iteration's timewalk functions to 
> correct the reference channel's time. But again, since the first 
> iteration had so little influence from the timewalk of the reference 
> channel, I don't believe you will see any difference in the final 
> resolution.
>
>    You are right. The distributions I produced will have the intrinsic 
> resolution of both cells and so are sqrt(2) larger than the intrinsic 
> resolution of a single cell. In principle, one mat be able to use the 
> timing and energy info from multiple cells to improve the resolution 
> even further. That has to be thought about how to do it correctly.
>
>    After talking with Elton and Simon about this current technique, 
> out biggest concern is that using the 2 cells from the BCAL may be 
> canceling some systematic part of the error giving a misleading-ly 
> lower resolution than we would otherwise be able to obtain. I think 
> this depends on how valid it is to assume cell 7 and cell 8 are 
> independent detectors.
>
> Regards,
> -David
>
> Blake Leverington wrote:
>> Hi David,
>>
>> I had done something like this previously except for the last step 
>> (ts8+tn8)/2 - (ts7+tn7)/2 but I'm wondering how good the walk 
>> corrections are when you plot say n7 vs tn7-tn8. There is a walk in 
>> tn8 as well that does not completely correlate with n7 though they 
>> appear to be corrected nicely. This is my only concern, otherwise 
>> everything looks good from what I see. I guess this a naive question 
>> but since you are taking the difference of 2 sums of the PMT's, does 
>> this still give you the intrinsic resolution or is it 2 cells folded 
>> together? I suppose I can check this myself too.
>>
>> I'll use your code to see what I see in the other cells.
>> Thanks!
>>
>> -Blake
>>
>> David Lawrence wrote:
>>>
>>> Hi Blake,
>>>
>>>    I have just added a page to the wiki under BCAL analysis and 
>>> today's date. The URL is:
>>>
>>> http://www.jlab.org/Hall-D/software/wiki/index.php/BCal_Beam_Test_Plots%2C_April_4%2C_2007 
>>>
>>>
>>> It shows some plots and describes *a* method of extracting the 
>>> timing resolution from the BCAL without using the tagger timing 
>>> (we're still working on that). I would urge you to take a look and 
>>> think about how valid the method is. I also uploaded the ROOT macros 
>>> used to make the plots. They are linked at the bottom of the page. 
>>> You should also take a look and make sure I didn't make any mistakes!
>>>
>>> Let me know what you think.
>>> Regards,
>>> -David