[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: FEU84-3 / FADC Timing
Hi Matt,
I completely agree that it will be very useful to perform your tests
with the final system configuration, i.e., fADC250 on a VXS crate. We
could also include a F1TDC for checks. And the feedback from these tests
will provide us with valuable information for improving any features, in
both software and hardware.
I talked to Chris this morning and he has a VXS crate that you can
borrow along with a 12-bit fADC250 (8 channel). We also need to talk to
David Abbott regarding the drivers while you are here for the
collaboration meeting.
Regards,
Fernando
Matthew Shepherd wrote:
>
> Hi Fernando,
>
> By +- 4 ns noise, I just mean uncertainty in the clock
> synchronization. Of course this (or lack thereof) is getting designed
> into GlueX -- all electronics clocks will be synchronized. My point
> is that this is an absolutely essential feature!
>
> It is going to be very helpful to be able to migrate to the JLab FADC
> where we can have multiple synchronized channels. We should talk at
> the meeting about how best we can get one of these modules up and
> running at Indiana. There are only so many tricks we can play with
> one channel. I think very soon we will want start working with
> something that looks more like the real system so we can better
> explore channel-to-channel stability. This is where the noise
> discrimination and "bunch finding" power will come from the real FCAL
> anyway. It will be crucial to demonstrate that the multiple channels
> have a very tight and stable time variance when viewing a common light
> source.
>
> See you later this week,
>
> -Matt
>
>
> On Apr 29, 2008, at 8:32 PM, Fernando J. Barbosa wrote:
>> Hi Matt and Mihajlo,
>>
>> The results are indeed impressive! These also point to the stability
>> of the ADC.
>>
>> I don't understand what you mean by the +/- 4 ns noise. Do you mean
>> missing samples on the leading edge? In order to get the full
>> resolution from a 10-bit ADC running at 250 MSPS, the clock must be
>> stable to less than 2 ps (jitter). For 12-bit ADCs running at the
>> same 250 MSPS, the clock must be stable to less than 350 fs! For the
>> energy sum, these numbers can be considerably relaxed but for timing
>> from the fADC, we need to measure the timing resolution dependence on
>> clock stability. This would be a very interesting measurement that
>> could be checked by modulating the clock signal. In any case, we have
>> tested clock distribution circuits with optical fibers to be stable
>> to better than 2 ps and that's our present plan.
>>
>> I completely agree that there are quite a few tests left regarding
>> the timing stability. The PMT & CW stability tests will provide a
>> measure as to what level of control we need to have over the voltage
>> setting, temperature, etc. The temperature in the hall is expected to
>> be stable around 70 F but the characteristics of the cable must be
>> well understood so that temperature does not become a factor.
>> Eventually, I think it would also be useful to characterize the
>> timing algorithm for single shot resolution and if it changes over
>> time, perhaps in combination with a high resolution TDC. The F1TDC
>> has a resolution of 60 ps if needed.
>>
>> I look forward to your presentations next week during the
>> collaboration meeting. It may also be a good time to start discussing
>> the algorithm implementation on a FPGA.
>>
>> Thanks and regards,
>> Fernando
>>
>>
>>
>> Matthew Shepherd wrote:
>>>
>>> Hi Fernando,
>>>
>>> Mihajlo has been making nice progress with a single FADC card. We
>>> do this by dividing the pulse and send one copy into the card. The
>>> other part we bounce off the end of the long cable and also send it
>>> into the card. This leaves then two copies of the same pulse in a
>>> single readout buffer of the FADC. We can shift the clock phase
>>> between the samples by varying the length of the cable used to
>>> generate the reflection. We apply the time measurement algorithm to
>>> both pulses and look at the RMS of the difference between the pulses
>>> over many events (the mean of the difference gives the delay). For
>>> large pulses (near full scale on the 8 bit FADC) we are see typical
>>> RMS differences of order 150 ps or single channel timing resolutions
>>> around 100 ps. It is impressive (suprising?) that the timing
>>> resolution is this good. We are also investigating dependence on
>>> pulse height.
>>>
>>> One conclusion is already evident: the synchronized clock for all
>>> readout channels is absolutely essential. If we would had some +- 4
>>> ns noise on top of this it would completely wash out the sharp
>>> timing resolution we can get with the PMT FADC system.
>>>
>>> I think we also need to explore absolute time stability of a single
>>> tube. The cable lengths will certainly stay fixed, but changing the
>>> voltage may shift the absolute transit time within in a tube at the
>>> level that we are sensitive to in our calibrations.
>>>
>>> We look forward to hearing your input and plan to present some
>>> results at the collaboration meeting next week.
>>>
>>> -Matt
>>>
>>>
>>> On Apr 15, 2008, at 7:45 AM, Fernando J. Barbosa wrote:
>>>> Hi Matt,
>>>>
>>>> We have clock distribution boards that output multiple (five) NIM
>>>> replicas and were designed for providing synchronized clocks to
>>>> multiple fADCs on a small setup. These were recently designed by
>>>> Gerard and Kim and are available with 250 MHz clocks. These sources
>>>> include the power supply.
>>>>
>>>> If these are of interest to you, we can ship you one now.
>>>>
>>>> Regards,
>>>> Fernando
>>>
>>
>
begin:vcard
fn:Fernando J. Barbosa
n:Barbosa;Fernando J.
org:Jefferson Lab
adr:Suite #10, 12B3;;12000 Jefferson Ave.;Newport News;VA;23606;USA
tel;work:757-269-7433
version:2.1
end:vcard