Re: your note on timing

[Gerard Visser <gvisser@indiana.edu>]

Hi Alex,
	I think Paul was referring to 
http://www.iucf.indiana.edu/~gvisser/scibath/fee_note.pdf.
	_However_, please note that that is in an entirely different context 
(PMT pulses from a scintillation detector), and the 100ps resolution 
figure (factor of 1000 better than sample rate) was for test pulse data, 
i.e., no deviation in the waveform shape. And also we won't shape the 
waveform as described in that note (ringing waveform). That's great for 
timing if the pulse is known to be narrow, but here it's not so applicable.
	I personally believe that Curtis's figure of factor of 3 better time 
resolution than the raw sample rate is too low / too conservative, but 
probably only a little bit so. I think that 2ns timing from 100 MSPS 
data should be achievable, i.e., factor of 5. Perhaps 1ns...
	A drift distance measurement will of course have other errors even if 
the leading edge timing was measured absolutely without error. Therefore 
I continue to argue that the proper question is whether FADC timing can 
make the drift distance measurement within xx um, not whether the FADC 
timing agrees with the leading edge timing within 1ns - especially since 
the leading edge timing in practice is not perfect. Is it possible to 
measure with the FDC prototype the drift distance error in the case of 
FADC timing method and in the case of discriminator/TDC timing method? I 
think it is those two numbers which should be compared to judge the two 
methods.
	Calibration and stability issues are of course a concern, but probably 
no more so than in the case of discriminator/TDC readout. And without 
recording the waveform it is hard to see how one would monitor or 
correct for some of the sources of error, e.g., baseline shifts, 
amplitude effects, ...
	As with Paul, I think (if it is possible) that a final decision on this 
be deferred until Simon has attempted to make timing measurements with 
intentional waveform shaping optimized for timing. [And preferably with 
the real preamp; it will have lower noise than the CLAS preamps.] I 
don't think that has been explored yet. He has been working with the 
CLAS preamps and the post-amp boxes given to him; timing resolution 
versus pulse shaping parameters should be studied.
	I have received the reciever/shaper boards this past monday and have 
one nearly ready to be used for initial tests with the FDC prototype. 
I'll have to get it to JLab 1st week of january. I think we should also 
plan for me to come out for a week and work together with Simon on some 
pulse shaping and timing measurements.

	Gerard


Alex Dzierba wrote:
> Hi Gerard
> 
> Could you send your note on timing - the one Paul referred to?  I 
> believe the
> Private Documents area of the GlueX portal is currently down,
> 
> Thanks
> Alex

-- here is the other message you refer to:
Gerard has been able to get something on the order of 100s of ps with  a 
10 MHz fADC - I believe he has written a note on his technique.

I wouldn't make the final decision on this until trying his new  shaper 
boards with the Struck ADCs.

Paul


Begin forwarded message:

 > From: Alex Dzierba <dzierba@indiana.edu>
 > Date: December 21, 2006 10:07:00 AM EST
 > To: "Daniel S. Carman" <carman@jlab.org>, ptsmith@indiana.edu
 > Cc: staylor@jlab.org
 > Subject: Re: FDC timing and FADC's
 >
 > Daniel
 >
 > Thanks for your quick response to Paul's questioning of the
 > assumptions about achieving 1 ns timing from 100 MHz FADC's
 >
 > I'm forwarding this to Paul - we need to get his response and
 > maybe we can further define what is needed to come a resolution.
 >
 > Alex
 >
 > At 9:53 AM -0500 12/21/06, Daniel S. Carman wrote:
 >
 >> Alex,
 >>
 >> With regard to timing resolution from FADCs, a usual rule of
 >> thumb is that resolution of 1/3 of the sampling time is achieveable.
 >> So, for a 100MHz clock rate, this amounts to about 3 ns.  Our
 >> current studies with the 100MHz FADC in our test setup have yielded
 >> a time resolution of more than 4 ns.  We are still working to
 >> improve this.  Thus I think that 1 ns is totally out of the
 >> question.  There is an important difference between what may be
 >> theoretically possible if all the stars are in alignment, and
 >> what actually is possible given real signal and noise levels in
 >> the readout.  Note that my rule of thumb above came from discussions
 >> with Curtis who has more practical experience than I do with FADCs.
 >>
 >> Another important issue that I have no experience with is how
 >> hard it will be to maintain a good timing calibration for the
 >> FADCs and how strongly the calibration parameters will change with
 >> time.  Also tied in with this question is the fact that the
 >> pulses coming from the chamber have a fairly broad range of
 >> shapes, rise times, and amplitudes, and I worry about the systematic
 >> deviation of time resolution and t_o resolution as a function of
 >> pulse shape, rise time, and amplitude.
 >>
 >>
 >>                 Regards,
 >>
 >>                     Daniel
 >>
 >>
 >> ********************************************************************* *
 >> *                                                                     *
 >> * Dr. Daniel S. Carman                   e-mail :  carman@jlab.org 
   * * Staff Scientist                         office : (757)-269-5586     *
 >> * Jefferson Laboratory                   web: www.jlab.org/ ~carman 
   *  * 
     *
 >> ********************************************************************* *
 >
 >
 >
 > --
 > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 > Alex R. Dzierba
 > Chancellor's Professor of Physics (Emeritus)
 > Department of Physics / Indiana U / Bloomington IN 47405 /  812-855-9421
 > JLab Visiting Fellow
 > Jefferson Lab / 12000 Jefferson Ave / Newport News, VA 23606 / 
757-269-7577
 > Home Phone: 812-825-4063  Cell:  812-327-1881  Fax: 866-541-1263
 > http://dustbunny.physics.indiana.edu/~dzierba/
 > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~