The Card Bypass Capacitors ------------------------------ Original Rev. 27-JUNE-1996 Latest Rev. 27-JUNE-1996 This file is the gathering place for our understanding of what bypass capacitors to use on THE Card. Ceramic Chip Capacitors: ------------------------ We need a 1206 Z5U Ceramic Chip Capacitor. We use ?? per Foundation Module card. We will use KEMET part number: C 1206 C 104 Z 5 UAC This is a 0.1 uFd 50 Volt 1206 Z5U ceramic capacitor. KEMET part number: C 1206 C 104 M 5 UAC +- 20% tolerance KEMET part number: C 1206 C 104 Z 5 UAC +80% -20% tolerance Newark Stock No. 89F5966 for the +-20% tolerance About 4 cents each in reel of 4000. Tantalum "B" and "C" Case Capacitors: ------------------------------------- We need a "B" Case Tantalum Capacitor: at 10 V standard range largest value is 6.8 uFd ESR is 3.5 Ohm at 10 V extended range largest value is 15 uFd ESR is 3.5 Ohm at 16 V standard range largest value is 4.7 uFd ESR is 3.5 Ohm at 16 V extended range largest value is 10 uFd ESR is 3.5 Ohm We need a "C" Case Tantalum Capacitor: at 10 V standard range largest value is 15 uFd ESR is 1.8 Ohm at 10 V extended range largest value is 47 uFd ESR is 1.2 Ohm at 16 V standard range largest value is 10 uFd ESR is 1.8 Ohm at 16 V extended range largest value is 33 uFd ESR is 1.2 Ohm 10 Volt vs 16 Volt and standard range vs extended range: The failure rate of the 10 Volt Tantalum Capacitor running at 5 Volts at 40 deg C will be about an order of magnitude higher than a 16 Volt Tantalum Capacitor running under the same conditions. In the Run I equipment we used: 10 Volt, 16 Volt, and 35 Volt hermetic tantalum capacitors. We had no failures (expect for units that were damaged during installation. In L1 Cal Trig it was mostly 33 uFd 10 Volt parts. We have 320 CTFE's x 34 Tantalums each plus 40 CHTCR's x 3 Tantalums each plus 220 CAT's x 3 Tantalums each. This is about 12,000 Tantalum Capacitors in L1 Cal. Trig. with no known failures. The Run II frameworks will have about 120 THE Cards with ? Tantalums each for a total of ? Tantalums in the Run II Frameworks. Estimate the useful amount of capacitance. Let's say that we want 16 msec hold up from the on board capacitors. Each FPGA will draw about 1 Amp and run from 3.3 Volts so it looks like 3 Ohms. So from an RC of 16 msec we need 5,000 uFd of capacitors. Well we are not going to get this. The biggest we can hope for is 7 "B" case and 7 "C" case per FPGA. With a "B" case of 15 uFd and a "C" case of 47 uFd (i.e. the biggest 10 Volt extended range parts) this gives a total of 7 x (15 uFd + 47 uFd) = 434 uFd. So at best the tantalum capacitors local to an FPGA will provide about 1.3 mill sec of hold up time. Examples of Tantalum Capacitors on 5 Volt systems: IBM/PC 10 uFd 25 Volt "D" Case 1990 IBM/PC 10 uFd 16 Volt thru hole 1988 IBM/PC 10 uFd 16 Volt thru hole 1996 IBM/PC 10 uFd 25 Volt "D" Case 1991 IBM/PC 10 uFd 16 Volt thru hole 1992 IBM/PC 10 uFd 25 Volt "D" Case 1992 3" hard disk drive 10 uFd 16 Volt "C" Case 1992 3" hard disk drive 47 uFd 10 Volt "D" Case 1992 Pier ERPB 10 uFd 16 Volt "C" Case 1993 NONE of these tantalum capacitors used the extended range: 10 uFd 25 Volt "D" Case 47 uFd 10 Volt "D" Case 10 uFd 16 Volt "C" Case ALL of these are the biggest capacitance per the case size in the standard range. 8 of the 9 examples used 16 Volt or higher capacitors in the 5 Volt application. Is the absolute lack of use of extended range tantalum capacitros (even in the compact 3" hard disk) just a matter of price, i.e. premium price for extended range ? I cann't tell because Newark lists only the standard range parts from KEMET but this in itself is a hint ! So visit the hold up time question again: Tantalum FPGA "B" and "C" Hold Up Selection Based On: Capacitors Picked msec -------------------------------- ----------------- ------- Biggest Extended Range 10 Volt 15 uFd 47 uFd 1.3 Biggest Standard Range 10 Volt 6.8 uFd 15 uFd 0.5 Biggest Standard Range 16 Volt 4.7 uFd 10 uFd 0.3 OK, Final Decision, Use 10 Volt Biggest Standard Range parts. This is based only on the fact that most of the THE Card tantalums will be running from 3.3 Volts and not from 5 Volts. So the parts that we order are: "B" Case 6.8 uFd 10 Volt KEMET part no. T491B685M010AS 20% T491B685K010AS 10% "C" Case 15 uFd 10 Volt KEMET part no. T491C156M010AS 20% T491C156K010AS 10% These are going to cost about 25 cents each in quantity. The cost increase for 10% tolerance parts compared to the 20% tolerance parts is about 6%. The "B" Case 6.8 uFd 10 Volt KEMET part no. T491B685K010AS 10% is Newark Stock No. 89F5039 about 22 cents each in reel of 2000. The "C" Case 15 uFd 10 Volt KEMET part no. T491C156K010AS 10% is Newark Stock No. 89F5044 about 29 cents each in reel of 500. Aluminum Electrolytic Capacitors: --------------------------------- The diameter of the aluminum electrolytic capacitors can be up to 0.4" The lead holes on the pcb are spaced 1.7". So the maximum case length can be something like 1.4" to 1.5". What voltage do we go for: 10 V or 16 V. I have no idea about the failure rate or the relative failure rate: Look at some various possibilities: Type uFd Volts Dia. x Length Temp. Qualt Cost ------------------ --------------- --------------- ----- ----- ---- Sprague TE series 250 uFd 16 V 0.385" x 1.286" 105 C Inds 2.56 Sprague TVA series 1000 uFd 16 V 0.375" x 1.500" 85 C Com 2.76 Mallory TT series 1000 uFd 16 V 0.394" x 1.062" 85 C HQ .75 Mallory TT series 1500 uFd 16 V 0.512" x 1.260" 85 C HQ .90 Mallory SKA series 1000 uFd 16 V 0.394" x 1.063" 85 C Com .64 Mallory SKA series 2200 uFd 16 V 0.512" x 1.260" 85 C Com .86 United Chem SME 1000 uFd 16 V 0.393" x 0.984" 85 C Com 1.10 United Chem SME 2200 uFd 16 V 0.492" x 1.181" 85 C Com 1.63 CDE 201 series 2200 uFd 16 V 0.512" x 1.240" 105 C Inds 2.44 OK, the final decision is that 0.5" diameter capacitor is an OK size. So let's pick the CDE 201 series Axial capacitor. This is Cornell Dubilier Part No. 201I222P016XX Newark Stock No. 46F6951