Re: Some more preamp questions

["Fernando J. Barbosa" <barbosa@jlab.org>]

Hi Mitch,

I hope you also enjoyed your holidays and Happy New Year!

I will address some of the items from the previous email:

2) I have modified the pinout per your suggestions. I have attached a 
pdf  that shows the pinout, footprint and signal polarities. Let me know 
if all this is correct. I would also appreciated if you could let me 
know the signal excursions for both input and output pins.

3) Let me clarify my observations on page 23. For the high amplitude 
signal (mag=300), the differential amplitudes [peak-baseline] (from the 
graph) are 950mV @ VCC=2.4V (450mV+500mV), 890mV @ VCC=2.5V 
(430mV+460mV) and 840mV @ VCC=2.6V (415mV+425mV). Therefore, the 
amplitude variation with the power supply is 
[(950mV-840mV)/(2.6V-2.4V)]=55%. For a 1% amplitude variation due to VCC 
instability, I would expect the VCC to be stable to better than 18 mV, 
say 10mV. My point is that the VCC at the chips must be well regulated 
as the chip's single-ended and differential amplitude characteristics 
are very sensitive to VCC variations. There are a few regulators on the 
market with these level of precision that we can use.

5) It sounds like a good plan. I agree that the comparator function 
would make this preamp much more generic but time is short to 
incorporate this into the first prototype run in February 2007. 
Regarding the funding, perhaps Elke, Elton and Daniel can comment on 
future budgets and plans. I will bring this up at our meeting tomorrow.

Thanks for your reply and best regards,

Fernando

mitch@hep.upenn.edu wrote:
> Hi Fernando,
>             hope you're enjoying your holidays.  I thought I might
> try to get a head start on the new year and try to answer your note.
>
> 1) Costs are in line with my expectations.
> 2) How about adding:  lead 25, 56 = GND
>    Also we may opt for simple shape control so 22, 23 would be shape control
>    contingency if we get time to add it in.
> 3) I am not seeing the 50-60% variations on page 23. It looks like we
>    should clarify this page and perhaps come up with a table.  One thing
>    that may help is to say that the DC offset changes with Vdd.  I assume
>    that Vdd will be  fixed during a run and that since you will be
>    digitizing the offset values, only difference values  (peak - baseline)
>    will be used to calculate the signal amplitude.  My guess on the
>    difference values over the 2.4 - 2.6V range is less than 10%.
>    While it doesnt appear to be an issue here keep two things in mind when
>    considering amplitudes.
>     1) Milind is considering only  the charge collected in the first 12ns
>        here ( in some cases 7ns) This is ~20% of the total ionization
>        charge. The rest of the charge, collected over the several micro
>        seconds it takes for the +ions to reach the cathode is cancelled.
>     2) The impulse charges we are using represent a worst case charge
>        deposition, most of the time charge should be distributed over longer
>        time periods with a much less determined shape.
>
> 4) Unless there is some differential non linearity between the ranges shown
>    it seems clear that the difference in amplitude represents a difference
>    in slope.  We use a 55C junction temperature for normal simulations.
>    Milind changes this value in spice to do the temperature sensitivity runs.
>    Note that the linearity at 55C is quite good up to 400fC for an
>    impulse input.
> 5) Yes we would only consider multi purpose outputs with the existing outputs.
>    This is not hard and I would very much like to add a comparator function.
>    Probably the best thing to do though is to get this prototype out of the
>    way and see how close it comes to meeting your analog performance
>    requirements.  I am hoping to get a February MOSIS submission.  When that
>    is out the door we can think about the comparator design. We will need to
>    discuss funding for that though. I think we have just about dried up our
>    first allocation. ( If it has in fact made it here.)
>
>
>                                                            Mitch
>
>
>
>
>
>
>
>
> Quoting "Fernando J. Barbosa" <barbosa@jlab.org>:
>
>   
>> Hi Mitch,
>>
>> GAS - GlueX Amplifier Shaper sounds like a good part name. I will add a
>> number so people don't confuse its name with a real gas. What about
>> GAS-1? You are the author!
>>
>> 1. COST - I have been contacting Sandy Connolly at I2A and she sends
>> regards to you. She provided me with information on the LQFP-64 and
>> QFN-64. The thermal resistance for the QFN-64 is on average about half
>> as that of the LQFP-64 but may not be a factor (soon I will go over the
>> thermal simulations). The packaging costs are the same. Sandy's cost
>> estimate for the production packaging, which includes wafer dicing,
>> handling, wire bonding, etc, is $2,560 for 4K parts and $250 for the
>> bond diagram generation if needed. She also quoted me $1,500 for the 40
>> part prototype run.
>>
>> The total production cost is $84,000 for 4K DIE from TSMC plus $2,560
>> for packaging 4K DIE plus $250 for the bond diagram generation or $86,810.
>>
>> 2. PINOUT -  I am including a pdf file that shows the preliminary pinout
>> for the GAS-1 and following your recommendations from an earlier email.
>> Any changes?
>>
>> 3. Referring to Millind's Powerpoint document dated 11/11/2006, page 23,
>> the plot shows the differential output response for two input amplitudes
>> and at 2.4V, 2.5V and 2.6V supply voltages. The amplitude variation with
>> supply (PSRR) at low amplitudes is about 5-20% and at higher amplitudes
>> it is 50-60%. For this variation to be much less than the non-linearity,
>> say 1% from chip-to-chip, the power supply at the chips must be +2.5V
>> +/- 10 mV. This necessitates the use of linear regulators on the preamp
>> cards which is not a problem. How does the linearity change
>> (sensitivity) with the supply?
>>
>> 4. Page 24 shows the differential amplitude variation with temperature.
>> I believe this is for a 50 degree C range. For the 100, 200, 300 input
>> amplitudes the variations are 10.2%, 3.5% and 9.9%, respectively. As
>> above, how does the linearity change with temperature? What were the
>> junction temperatures for your simulations?
>>
>> 5. Page 29 suggests pin selectable comparators. There have been some
>> questions concerning the use of discriminators/TDCs on the FDC anodes.
>> We don't have enough pins on the 64-lead package for accommodating both
>> differential analog and discriminator signals simultaneously and I
>> believe we don't need both types of outputs simultaneously. We also need
>> to keep the package size in check (~10mm x10mm) as there is limited
>> space. If it is possible to add the discriminators in a transparent
>> fashion and use the same output pins as the analog outputs , we would
>> need to enable/disable the discriminators circuitry. How difficult would
>> it be to add this extra feature? The cost would also go up with the
>> additional Si area.
>>
>> Thanks and best regards,
>>
>> Fernando
>>
>>
>>
>>     
>
>
>

gas-1.pdf

begin:vcard
fn:Fernando J. Barbosa
n:Barbosa;Fernando J.
org:Jefferson Lab
adr;dom:;;12000 Jefferson Ave.;Newport News;VA;23606
tel;work:757-269-7433
version:2.1
end:vcard