• About PSAS
• Get Involved!
• Schedule
• Projects
• Presentations
• Mailing lists
• Contact us

• Teams
  • Airframe
  • Avionics
  • Communications
  • Ground
  • Payload
  • Propulsion
  • Rocket Science
  • Safety
  • Software
  • Uncertainty
  • Regulation
  • PSU Student Club

Design notes for the AV3 node "Generic Front End" (GFE) circuit and PCB.

Purpose

AV3 stands for "Avionics System 3", our fourth generation of avionics systems for amateur rockets (it's zero based, of course). "Generic Front End" means that this circuit, and most of the layout, is replicated across the ~8 nodes found in the avionics system. The GFE contains three sections:

• Switching Power Supply (SPS)
• Highly available Power Supply (HAP)
• Microcontroller (UC)

History

LV0, LV1 had (woefully underpowered) monolithic flight computers (FC). LV2 had an FC with many CAN-based Microchip PIC18 nodes. This revision, now called "AV" instead of "LV" in order to separate it from launch vehicle, is now based on USB and CAN, and uses the NXP LPC2368 ARM7TDMI microcontroller.

Original work on this board was done using the USB-only NXP LPC2148 by the PSU ECE Capstone 2006 team and can be found here. This board was significantly revised by adding the CAN and USB-based NXP LPC2368 by the Capstone 2009 team and can be found here.

GFE Requirements

• General
  • Should have parts cost <150$ in small quantities
  • Should be (,20,25)cm^2
  • Should be (,,12)mm thick
  • Must have (2,,4) board layers
• Interfaces (connectors)
  • MUST use a polarized and locking connector
  • Should be able to be non-locking during development
  • MUST be rated at (100,500,) mating cycles
  • MUST have (2,4,) high reliability, high current (2,5,) A rated contacts for power
  • Must have (2,4,) high reliability contacts for USB data
  • Must have (2,4,) high reliability contacts for back-channel bus
  • Should minimize board area.
  • Debug connector MUST contain all signals to program and debug GFE.
• PFE
  • MUST have OVP at (,20,) V
  • MUST have UVLO at (,10,) V
  • MUST have reverse voltage protection
  • MUST have OCP at (0,2.5,) A
  • MUST filter EMI such that other systems are not affected (e.g., GPS)
  • Must filter ESD (human body model)
• SPS
  • Vin (10,14.4,20) V (Voltage supplied by APS node)
  • Vout = (3.3,,5) V (selectable) (requires changes in component values)
  • Imax = (0.030,1,2) A (requires changes in component values)
  • > 70 % efficiency (prolong Battery life)
  • Circuit must be fault tolerant/resistant/robust
  • MUST be EMI compatible with onboard communication systems (e.g., crystal oscillator locked for noise shaping)
  • Should have high Frequency Switching (> 1 MHZ) (low noise production)
  • Low Quiescent Current Draw (prolong battery life)
  • Should do soft starts
  • Should be able to shut down on external signal.
• HAP (battery)
  • MUST use single Li Ion Polymer cell
  • MUST have a seamless transition from externally powered to battery-powered
  • MUST meet standard lithium charge/discharge safety requirements (thermal, voltage, current, time, fuse)
  • MUST be able to charge at (,1C,C/2) rates at internal air temperature of 50 deg C.
  • MUST indicate charging status (on/off)
  • MUST store enough power to run generic node for (1,4,) hours (Pyro requirement)
  • MUST be able to measure battery voltage
  • MUST be able to have external charge control (from microcontroller)
  • MUST be able to be bypassed for boards not requiring HAP functionality
  • MAY indicate some kind of charge % (blink rate, color, LED bar graph, etc)
• HAP (supply)
  • Vin (3.3,,5) V (Voltage supplied by SPS node)
  • Vout = (3.3,,5) V (selectable) (requires changes in component values)
  • Imax = (0.030,1) A (requires changes in component values)
  • MUST be EMI compatible with onboard communication systems (e.g., crystal oscillator locked for noise shaping)
• UC
  • MUST use a ARM-based microcontroller
  • MUST have USB and back-channel (CAN) peripherals
  • MUST have (32,64,) K RAM
  • MUST use 12 MHz crystal (for USB) and PLL to highest Freq.
• UCM
  • Should use UC crystal for SPS and HAP clocks

Design Notes

Left to do:

• Clean up silkscreen
• Switch to 2A capability (for FPGA)

Use Cases

• We expect GFE users to rip out breadboard, pin-to-hole array, and the HAP as necesary.
• Users chose between the SPS and HAP at the 'SPS/3.3V' and 'HAP/3.3V' polygon interfaces (see table below).
• The microcontroller requires 3.3V so you can't get out of that.

Vout Requirements	What to do
3.3 V, battery back up	Leave the circuit as is.
3.3 V, no battery	Rip out the entire HAP circuit and connect 3.3V plane via 'SPS/3.3V'.
3.3V, 5 V, no battery	Leave SPS at 5V, bypass the battery circuit, leave HAP at 3.3V output.
3.3V, 5 V, battery	Leave the circuit as is but attempt to get "hack" 5V supply off HAP IC to work.

Component Values

Most of these entries come directly from the Capstone 2006 and 2009 component entries.

The component numbers are as follows (see RocketNames):

Section	Numbers
SPS	2000 - 2099
HAP	2100 - 2199
UC	2200 - 2299

Each components should have three lines:

• Manufacturer and part info.
• Distributor info
• Purpose
• Design notes

SPS (FRONT END)

J2000 Rocketbus connector

• JST Connector 16PS-JED 1.25 mm pitch side entry-type (right angle) 1.0 A plug
• Available only through JST website.
• Purpose: Connects to rocketbus (power, CAN, USB, and AUX).
• Notes: Gang two pins together to get 2 A max current flow. FIXME: this maximum has no margins compared to the desired 2A at Vbatmin = 10 V. Mates to the 16R-JED.

R2000 chassis ground isolation resistor

• 100K 0603 resistor
• Digi-Key
• Purpose: Provides a DC path from the SPS ground to chassis ground. See page 29 of the CAN Node Switch Mode Power Supply (SPS) (200) section in the Component Design for LV2 Power Electronics (Except Main Battery) engineering design notes.
• Notes: Arbitrarily choose 100 kOhm. Value not very important; should be relatively high resistance so there's not a ground loop path.

C2000 chassis ground bypass capacitor

• 100nF 0603 >= 50V capacitor
• Purpose: Used to bleed high frequency signals from the ground shield.
• Notes: Value not important. Should be high voltage >= 50 V to deal with ESD, battery voltage, etc.

L2000 input choke

• Coiltronics CMS1-7-R CMS-SERIES Common Mode Inductors 41.5µH 2.6Amax 19.00 mOhm
• Digi-Key 513-1113-1-ND
• Purpose: Common mode choke (balanced inductor) used as an EMI filter between the power bus and the SPS.
• Notes: The capstone 2006 value of 100uH was chosen through a trial and error process from the previous LV2 SPS design. See page 13 of the CAN Node Switch Mode Power Supply (SPS) (200) section in the Component Design for LV2 Power Electronics (Except Main Battery) engineering design notes. However, we've upped the current and and the old model is not available, so a new 41.5 uH value that runs at 2.6 A was chosen.

F2000 fuse

• TYCO Electronics, 1206SFF200F/63-2 2000 mA, 63 V, 1206, Fast Acting Short Time Lag
• Digi-Key 1206SFF200F/63CT-ND
• Purpose: Protects SPS from short cicuirts on the SPS side, and reverse currents through the TVS2000.
• Notes: The specified maximum SPS output current is 2 A; given a 10 V input into the SPS and a 5V output at 80% efficiency, that's 1.25 A input current. We chose a fuse rated at 2 A, which is a safety factor of 160%. The opening time for the fuse according to its datasheet is 50 ms at a current of 8 A, or 20 s at 4 A, and at least 4 hours at 2A. This is OK, since the fuse isn't meant to protect from slight overcurrents (that's U2000's job). The resistance is 50 mOhm, so that's a 200 mW power dissapation at full capacity; that's a lot of power, so it might be worth it to slash that by half by switching to the 1206SFF400F/32 (other components, need to be relooked at, specifically TVS2000).

C2001 bulk charge filter capacitor

• Kemet T491X226K035AT 22 uF 35 V Tantalum T491 Series, X (7343-43) package
• Digi-Key 399-3816-1-ND
• Purpose: Acts as a noise filter between the power bus and SPS and serves as a local energy storage resevoir.
• Notes: Value chosen relatively arbitrarily. Tantalum capacitor was chosen due to their low ESR at a higher capacitance. Would have liked to have the T491D226(1)035A(2) which is 2 mm shorter, or the T491X226(1)050A(2) which is 50V, but Digi-Key doesn't stock either.

C2002 filter capacitor

• 330 nF 0603 >= 50V capacitor (TDK C1608X7R1H334M)
• Digi-Key 445-5951-1-ND
• Purpose: filter out higher frequency noise than the 22uF C2001 can
• Notes: Value chosen arbitrarily.

C2011 filter capacitor

• 10 nF 0603 >= 50V capacitor (Murata GRM188R71H103KA01D)
• Digikey 490-1512-1-ND
• Purpose: filter out higher frequency noise than the 330nF cap (C2002)
• Notes: Value chosen arbitrarily.

TVS2000 transient and over voltage supressor

• Diodes Inc SMBJ20A-13-F TVS unidirectional 600W 20V SMB package
• DigiKey SMBJ20A-FDICT-ND.
• Purpose: A transient voltage suppressor (TVS), this "zener like" diode protects the SPS (specifically U2002) in the event of an overvoltage at the input or a reverse battery.
• Notes: In both cases (over voltage and reverse voltage), the TVS protects the SPS by clamping the voltage and conducting enough currrent to blow fuse F2000. Thus we need a big beefy TVS to have the fuse blow first (thus the SMB). It should have a breakdown voltage of > Vbatmax which is about 19 V and a current carrying capacity greater than the fuse rated current. It should have a fast response time and be unidirectional.

Q2001 UVL/OVP/OVC switch

• Vishay/Siliconix SI7309DN-T1-GE3 P-Channel -60 Vds 8 A MOSFET in a 'PowerPAK 1212-8' package.
• Digi-Key SI7309DN-T1-GE3CT-ND
• Purpose: This is the external PMOSFET of U2000 which will turn off for any of 4 events: under voltage,SPS output over voltage, SPS power down is asserted, or the over current trip is exceeded. U2000 uses the RDS(on) of the saturated Q2001 as a current sense resistor (along with R2003).
• Notes: This MOSFET should be able to handle large Vds and have a low Rds(on). The one chosen has a 60 Vds and the Rds(on) is ~ 115 mOhms at Vgs = -10V. Since U2000 measures the voltage drop across this uncertain Rds(on), we add another resistor (R2003) to make that value less sensitive to Rds(on). At 8 A this MOSFET is grosslly over-rated, but there's no problem with that.

R2003 shunt resistor

• 0.033 ohm 0805 1% (e.g. Vishay/Dale WSL0805R0330FEA)
• Digi-Key WSLA-.033CT-ND
• Purpose: This resistor dominates the circuit-breaker resistor's (Rcb) value. It is in series with the drain (hence RDS(on) of Q2001 to make up Rcb. The voltage drop across it is used to detect an overcurrent event given that it is greater than 300 mV. See Q2001 and U2000.
• Notes: Since the typical value of RDS(on) of Q2001 is 0.115 ohm and Rcb was chosen to be 0.15 ohm, R2252 = Rcb - RDS(on) = 0.035 ohm. The closet standard value was 0.033 ohm. Given this value of Rcb and the circuit-breaker trip threshold voltage of 300 mV, the maximum SPS current which can be drawn before an overcurrent event is Imax = 300 mV / (0.115 ohm + 0.033 ohm) = 2.02 mA. Although this is slightly over 2A, the reality is that Q2001 will have a higher Rds(on) in practice, which will bring down the trip current. How much will have to be measured. At 2A, this resistor should dissipate 132 mW.

U2000 circuit breaker chip

• Maxim MAX5902AAEUT +72 V, SOT-23, Simple Swapper How-Swap Controller.
• Available through Maxim only.
• Purpose: This auto-restart hot-swap controller IC serves three purposes: (1) circuit breaker, (2) under-voltage lockout (UVLO) protection, and (3) SPS over-voltage protection (OVP).
• Notes: The version we chose had a input voltage range of +9 V to +72 V, a 300 mV circuit-breaker threshold voltage, limited inrush current ("soft start") and was an automatic retry circuit-breaker. The 300 mV drop voltage was chosen to minimize power dissipation. The automatic retry circuit breaker version was chosen so that the SPS to be able to recover from a fault condition without external help. It also had a built-in thermal shutdown and active low power good (!PGOOD) indicator output pin. There are four events which will cause Q2001 to turn off: (1) if there is undervoltage at the input, (2) if there is overcurrent, (3) if the die temperature exceeds +125 C and (4) the ON/!OFF pin 6 is forced low for at least 10 ms.
  • Soft-start: Upon power up U2000 keeps Q2001 off and barring the trigger events (1) and (2), then it gradually turns Q2001 on to saturation at a 9V/ms (measured at the source).
  • OCP: Rcb, which is the series combination of Q2001's RDS(on) and R2003 between the two pins 1 and 2 of U2000, generates a voltage when current flows through them. When this voltage exceeds 300mA, U2000 triggers an over current protection (OCP) event. U2000 waits for 150 ms, then tries again.
  • Restart: If any one of the 4 trigger events occurs U2000 will turn Q2001 off, de-assert !PGOOD, and reinitiate the start sequence given that the trigger event(s) disappears during the 150 ms period, if not the 150 ms period will repeat. There are two typical turn off times regarding Q2001: 10 ms and 4 us. If there is an ON/!OFF or UVLO trigger event, they need to exist for 10 ms before U2000 turns Q2001 off, which will take an unspecified amount of time. If there is an overcurrent or temperature trigger event, then Q2001 is turned off in 4 us. If the trigger events disappear after Q2001 turns off within 150 ms, then the normal start sequence is reinitiated.
  • !PGOOD does not require a pullup resistor: this pin is open drain, and the LPC can provide a pull-up resistor on it's GPIO pin.
  • UVLO and HAP testing: The UVLO threshold was specified as 9 V. See R2001 and R2002. This meant that when a trigger event other than a UVLO condition happens, U2000 would turn Q2001 off in 4 us and would reinitiate the start sequence after a trigger event free 150 ms time period. We needed a way for the LPC microcontroller to turn off the SPS in order to test the HAP, Q2000 can pull down ON/!OFF to ground.
  • U2000 has a 1 mA to 2 mA standby current, which seems very high. We believe that U2000 will draw the most current from the power bus when the SPS is in a standby/shutdown mode.
• NOTE: According to the MAX5902 datasheet there are multiple package options for the different versions of the MAX5902: a TDFN and SOT23 package. All SOT23 packages have the specification that, "This device is constructed using a unique set of packaging techniques that impose a limit on the thermal profile the device can be exposed to during board level solder attach and rework. This limit permits only the use of solder profiles recommended in the industry standard specification, JEDEC 020A, paragraph 7.6, Table 3 for IR/VPR and convection reflow. Preheating is required. Hand or wave soldering is not allowed.", which may or may not present a problem.

R2001 UVLO resistor divider of U2000

• 61.9 kohm, 0603, 1%, 1/8 W, (Rohm, MCR10EZHF6192)
• Purpose: R2001 is part of the UVLO voltage divider for U2000.
• Notes: UVLO was specified to be 9 V. See page 8 and Figure 3 in the MAX5902 datasheet. Letting R2002 = 10.0 kohm and using the typical value of Von/!off = 1.26 V, the UVLO formula from page 8 in the datasheet is R2001 = R2002 * ((VUVLO / (Von/!off)) - 1) = 61.4 kohm. The closet standard value was 61.9 kohm.

R2002 UVLO resistor divider of U2000.

• 10.0 kohm, 0805, 1%, 1/8 W (Rohm, MCR10EZHF1002)
• Purpose: R2002 is part of the UVLO resistor divider of U2000.
• Notes: R2002 was chosen to be 10.0 kohm based on page 8 and Figure 3 in the MAX5902 datasheet.

Q2000 SPS off command from LPC

• Logic level N-channel mosfet, 100V, SOT23, (Zetex BSS123TA)
• Digikey BSS123CT-ND
• Purpose: This N channel logic level MOSFET allows the LPC to activate the shutoff of U2000 (circuit-breaker) resulting in a turning off power to the SPS. This is particularly useful for testing the HAP. With no HAP, this MOSFET may be not placed.

R2012 Pulldown resistor for Q2000

• 10.0 kohm, 0805, 1%, 1/8 W (Rohm, MCR10EZHF1002)
• Purpose: Pull down Q2000 so it doesn't turn on when the LPC first starts up.
• Notes: The crazy LPC designers turned on pull-up resistors in the LPC's GPIO after power-on reset. How crazy messed up is that?! Without this resistor, the LPC would turn off the SPS supply as it powered up.

R2004 over-voltage protection buffer

• 1 kohm, 0603, 1%, 1/8 W (Rohm, MCR10EZHF4702)
• Purpose: R2004 along with Q2002 will "emulate" an overcurrent trigger event as seen by U2000 when an SPS output overvoltage triggers U2001.
• Notes: When the SPS output rises above a certain threshold, the output of U2001 goes high, turning Q2002 on. When this happens it pulls pin 2 of U2000 very close to ground, emulating overcurrent trigger. R2004 limits the extra current pulled through Q2002 when it is turned on. In normal SPS operation, the voltage drop across Rcb will be less than 300 mV and R2004 is connected to the high impedance pin 2 of U2000. Note that although "no current" should flow, some current does indeed flow, so values of 10K and above will not work (found out through experimentationon the FC carrier). 1K seems to be a reliable value that works. Although dropping 20V across 1K to ground burns 400 mW (3.2x the 125 mW rating), R2004 works discontinusously because of U2000's 150 ms timeout after a few ms of being on, so the PWM nature of the signal during an OVP event should be small.

Q2002 over-voltage protection switch

• Logic level N-channel mosfet, 100V, SOT23, (Zetex BSS123TA)
• Digikey BSS123CT-ND
• Purpose: This general purpose N channel logic level MOSFET allows the U2001 comparator to emulate a over-current trigger in U2000. It does this during an SPS overvoltage event - whenever the SPS is 10% above it's rated voltage, U2001 shuts down U2000.

U2001 Over-voltage protection comparator

• Texas Instruments TLV3012AIDBVT nanopower Push-Pull Output Comparator with Voltage Reference, 1.8 V < Vin < 5.5 V, SOT-23-6
• Digi-Key 296-16830-2-ND
• Purpose: This comparator compares the SPS output voltage (see R2005, R2006) to its internal reference voltage (1.242 V), and triggers an SPS overvoltage trigger event if the output is higher than it should be.
• Notes: It is powered by a "backup supply" consisting of D2001 and C2003 so that it can shut off the SPS supply for long periods of time. Also it has a pseudo low-pass filter consisting of C2004 and its output (pin 1) with the input being pin IN+ (pin 3). See D2001, C2003 and C2004 respectively. We wanted a low power push-pull output comparator to get rail to rail output swing (approximately 200 mV to 3.1 V) and have reasonable switching and rise/fall times, on the order of several microseconds and nanoseconds respectively. We tied the IN- pin (pin 4) to the internal reference voltage REF pin (pin 5) which will be compared to the divided SPS output voltage at its IN+ pin (pin 3). See R2005, R2006 and C2004. The "undervoltage" protection is actually provided by D2001 and C2003 where U2001 will remain powered for a specified amount of time if the +3.3 V SPS output rail drops. See D2001 and C2003.

C2003 Backup power for U2001

• 3.3 uF, 10 V, 0603, X5R (TDK C1608X5R1A335K)
• Digi-Key 445-5168-1-ND
• Purpose: Along with D2001, C2003 forms a "backup" power supply for U2001. C2003 is charged to the SPS voltage less a forward diode drop (see D2001) from the SPS output rail under normal operating conditions.
• Notes: U2001 draws a constant 2.8 uA supply current so D2001 is always trickle charging C2003, therefore the voltage across C2003 will be VfD2001 = 0.37 V (this is worst case Vf) less than 3.3 V or 5V (depending on what the SPS is set to). We specified U2001 to have power long enough to turn Q2002 back off while U2000 is turning back on again. Under normal operating conditions (no UVLO event) and assuming that there are no long or catastrophic transients, U2001 should always be on. This design minimizes any unknown states at the comparator output resulting from U2001 turning off, then on again, etc. Upon initial power up we assume that the output of U2001 will be logic low keeping Q2002 off to prevent a false overcurrent event for U2000 which may prevent the SPS from working as U2000 will never turn Q2001 on and will just cycle. This seems unlikely because upon initial power up, U2001 has no power and cannot output logic high. However as the +3.3 V SPS output is being brought up the output of the comparator is undefined which is not good being directly connected to the gate of Q2002 but we still think that Q2002 will remain off or will rapidly switch off if it is ever on after the transients. We calculated the needed amount of charge, Qt, C2003 would have to store such that U2001 would have power for at least 0.5 s (our specified amount of time U2001 should have power during these events) given that the SPS output voltage dropped by 1 V from which we calculated C2003's capacitance. C2003 = [((Qt * 1.2) + (Iq * tp))] / Vp, where Qt = [((Input Capacitance) * Vt) + ((Reverse Transfer Capacitance) * Vin)], Iq = 2.8 uA, ts = 0.5 s, Vp = 1 V Qt is the sum of products of the input capacitance of Q2002 times the maximum threshold voltage plus the reverse transfer capacitance of Q2002 times the maximum Vds swing, namely Vin. Iq is the supply current of U2001, tp is the amount of time we want U2001 to have power, Vp is the amount of voltage the SPS output drops and the 1.2 term is a fudge factor because Qt is dependent on some other factors not explicitly shown. tp was specified to be 0.5 s, this time is the time C2003 can supply power to U2001 which is longer than the propagation delays mentioned above including some margin just in case any trigger events do not go away and U2000 has to go through another 150 ms start sequence. If the trigger events remain longer than the 0.5 s, then U2001 turns off and the whole SPS will go through another initial power up sequence. Vp = 1 V, i.e. the input voltage V+ (pin 6) to U2001 can drop to about 3.0 V - 1 V = 2 V. Iq = 2.8 uA, see page 3 in the TLV3012 datasheet. Qt = (85 pF * 2 V) + (15 pF * 20V) = 470 pC, see page 3 in the BSS123 datasheet. C2003 = 1.40056 uF we decided to chose a 2.7 uF cap to give us a little more tp due to any unknown delays and the like we did not consider. However, 2.7 uF is not currently available, we chose 3.3 uF which was.

D2001 Backup power for U2001

• >= 30 V, 100 mA, < 350, 0.35 mV Vf @ 10mA (Comchip Technology CDBU0130L)
• Digi-Key 641-1282-1-ND
• Purpose: D2001 and C2003 form U2001's power supply. This diode prevents C2003 from discharging anywhere but to the V+ supply pin of U2001.
• Notes: Schottky diodes were chosen for their low forward voltage and fast switching and reverse recovery times. In response to a overvoltage event at the +3.3 V SPS output, U2001 will output a logic high and turn Q2002 on which will cause an overcurrent event at U2000 which will in response turn Q2001 off thus circuit breaking the bus rail from the SPS and the +3.3 V SPS output voltage rail will go to zero. D2001 has a low forward voltage and U2001 has an input voltage supply range of 1.8 V to 5.0 V so when the SPS voltage is being brought up C2003 is being charged through D2001 leaving the voltage of SPS minus the Vf of D2001 at pin 6 (V+) of U2001.

R2005 SPS OVP divider network 1 of 2

• SPS = 5V: 34.0 Kohm, 0603, 1% (Rohm, MCR10EZHF1822)
• SPS = 3.3V: 18.2 kOhm, 0603, 1%
• Purpose: R2005 along with R2006 form a voltage divider with respect to the +3.3 V SPS output rail. When there is an overvoltage at the +3.3 V SPS output the voltage at the IN- pin (pin 4) of U2001 will be greater than the internal reference voltage of U2001 (typically 1.242 V) and will result in the comparator in U2001 outputing a logic high value. When the SPS output is below a certain threshold the input voltage (pin 4) to U2001 is less than the internal reference voltage and the comparator's output is a logic low.
  • Notes: We arbitrarily choose +10% overvoltage (3.3 -> 3.63 V and 5 -> 5.5 V) for the OVP cutoff. For 3.3V, Since we have been using several 10.0 kohm resistors we specified R2006 to be 10.0 kohm. Therefore using the overvoltage trigger event value to be 3.6 V and the compared voltage to be 1.242 V we solved for R2005: 1.242 V = (3.6 V * R2006)/ (R2006 + R2005) and solving for R2005 = 18.985 kohms. The closet conservative standard value was 18.7 kohm. For 5V, that comes out to be 34.283 kohms which is 34.0 Kohm.

R2006 SPS OVP divider networkh 2 of 2

• 10.0 kohm, 0603, 1%, 1/8 W (Rohm, MCR10EZHF1002)
• Purpose: R2006 along with R2005 form a voltage divider with respect to the +3.3 V SPS output rail. When there is an overvoltage at the +3.3 V SPS output the voltage at the IN- pin (pin 4) of U2001 will be greater than the internal reference voltage of U2001 (typically 1.242 V) and will result in the comparator in U2001 outputing a logic high value. When the SPS output is below a certain threshold the input voltage (pin 4) to U2001 is less than the internal reference voltage and the comparator's output is a logic low.
• We arbitrarily R2006 = 10.0 kohm for power savings and simplicity. See R2005.

C2004 LPF cap on SPS OVP divider

• 15 nF, > 25 V, 0603, X7R,
• Digi-Key 445-5190-1-ND
• Purpose: C2004 is used as a low-pass filter to node IN+ (pin 3) of U2001 and as positive feedback to make the comparator switch faster and to make sure that once the comparator is switching it completes the transistion. Under normal SPS operation, when U2001 is keeping Q2002 off, the OUTPUT (pin 1) of U2001 is at zero volts thus the cap is acting like a low-pass filter, the node connected to pin 3 of U2001 is the input to the filter. If there are transients at the +3.3 V SPS output this node will also experience proportional transients. If the magnitude of these transients are great enough (but fast in duration) then U2001's comparator switches, which is undesirable so we want true ovevoltage events to trigger the comparator. C2004 will remove most of these false event transients. When there is a true overvoltage event the comparator starts to switch. If there is another transient (false event) where the magnitude of the voltage goes below the threshold the comparator could possibly try to switch back. We want the comparator to avoid reacting to false transients. C2004 prevents this because as the comparator is rising its output voltage, C2004 raises the voltage on pin 3 of U2001 thus reinforcing the comparator to keep on raising its output voltage. This is positive feedback. Also C2004 decreases the rise time of the comparator. Basically if there is something weird going on at the SPS output, i.e. it is oscillating between 0 V and 3.3 V, C2004 will help to make sure that U2001 turns Q2002 off, eventually turning off Q2001 which will give a 150 ms time period for the weird things to go away, given that U2001 does not shutdown during these transients. See C2003. We wanted C2004 to filter transients which lasted less than 100 us therefore we need to find the output resistance C2004 sees under a transient (or AC) condition. We used the zero-time coefficient technique to solve for the resistance and eventual capacitance. Under a transient condition the SPS output and comparator output are grounded (DC voltage) and removing C2004 the resistance it sees is the parallel combination of R2005 and R2006. R2005 || R2006 = 6.46 kohm, with an RC = 100 us we can solve for C = 15.4 nF. The closest standard value is 0.015 uF. The voltage rating needs to be greater than 20 V just for safe measure.

SPS

U2002 buck regulator

• Linear LT3972 "33V, 3.5A, 2.4MHz Step-Down Switching Regulator with 75μA Quiescent Current"
• Digi-Key LT3972EDD#PBF-ND
• Purpose: Main buck switching power supply chip for the SPS (hey, that's redudant! :)
• Notes: The Linear LT3972 is the heart of the SPS regulator. It can withstand Vin up to 62V which minimizes front end protection circuitry and it can be synchronized to the 1.5MHz clock coming from the UC. The main function it performs is dropping the 10-20VDC coming in from the battery down to 5 V. For nodes without the HAP, the SPS can be reconfigured to run at 3.3VDC directly removing the need for the entire HAP section. The layout for this chip is pretty much directly from the datasheet - they had a pretty sweet design compared to what we came up with. Middle pad has ground via; good for low impedance, bad since it'll suck some solder paste out.
• Maximum Switch frequency and duty cycle calculations
  • F(switch max) = (Vd + Vout) / (t(on min)(Vd + Vin - Vsw)), Vd=0.5V,Vout=3.3V,t(on min)=150nS,Vin=(10,14.2,20)V,Vsw=0.5@max load
  • F(sw max)= 3.66MHz for Vin=10V, 2.58MHz for Vin=14.2V, 1.83MHz for Vin=20V
  • Min Duty Cycle = 22.5% = F(sw)*t(on min)
  • Max Duty Cycle = 77.5% = 1 - (F(sw) * t(on min))
  • These all appear fine as they are all well above the 1.5MHz we are planning to sync at.
• Input Voltage range (at 1.5MHz operation)
  • Vin(max) = 36.66V = [(Vout+Vd) / (f(sw) * t(on min))] - Vd + Vsw, Vd=0.5V (catch diode drop), Vout=3.3V, t(on min)=100nS, Vsw=0.5V
  • Vin(min) = 6.47V= [(Vout+Vd) / [1 - (f(sw) * t(off min))]] - Vd + Vsw

C2005 Input Capacitor Cin1

• 10uF , 50 V, 10% , Ceramic , 1210 , X5R (Taiyo Yuden UMK325BJ106KM-T)
• Digi-Key 587-2247-1-ND
• Purpose: Local bulk charge storage for U2002.
• Notes: Using a physically large 10uF further away from U2002 is fine, but then add a physically small 1uF cap in very close.

C2006 Input Capacitor Cin2

• 1uF , 50 V, 10% , Ceramic , 0805 , X7R (TK C1608X7R1E105K)
• Digi-Key 490-4736-1-ND
• Purpose: Local low impedance charge storage for U2002
• Notes: Xc = -j.01 ohms = 1/jwc , w = 2pi * 1.5MHz, c=10uF So long as Xc is small at the switching frequency this cap will work, any cap in the range is fine, but the smaller they are the closer they need to be to the IC and the catch diode in order to effectively handle the EMI from the switching. Watch out for Ringing during hot plugging (although U2000 should soft start) - the cap needs to be able to close to the full 62V that the LT3972 can during faults and plug in ringing. Use type X5R/X7R, NOT Y5V (warning per data-sheet). Place as close to the IC as possible.

R2007 soft start resistor @ LT3972 'RUN/SS' pin

• 500k ohm 0603 1%
• Purpose: Works in combination with the C2007 soft start capacitor (see C2007).
• Notes: See C2007.

C2007 soft start capacitor

• 100nF, >= 50V, 0603 (TDK C1608X7R1H104K)
• Digi-Key 445-1314-1-ND
• Purpose: Works in combination with R2007 to set the soft-start time.
• Notes: Choose a large RC time constant to minimize voltage overshoot and reduce Imax for the switch at startup. Choose resistor to supply 20uA when RUN/SS is at 2.5V. Want (3.75k, 585k, 875k) ohms for Vin = (10,14.2,20). 500k worked fine in LT-Spice and is the value chosen. Datasheet shows 220nF, but doesn't work in LT-Spice simulation. 100nF did, so we choose that value.

R2008 frequency compensation resistor @ LT3972 'Vc' pin

• 15k ohm 0603 1%
• Purpose: Set the SPS frequency compensation. See C2008.
• Notes: Change to 19K ohm 0603 1% for Vout = 3.3 V. Works in combination with the C2008 compensation capacitor (see below).

C2008 frequency compensation capacitor @ LT3972 'Vc' pin

• 680 pF 0603 >= 50 V (TDK C1608X7R1H681K)
• Digi-Key 445-5077-1-ND
• Purpose: Set the SPS frequency compensation. See R2008.
• Notes: Directly from datasheet. May need to be tuned insitu. See LT1375 for longer rightup on tuning these values.

R2009 frequency set resistor @ LT3972 'RT' pin

• 26.7k ohm 0603 1%
• Purpose: Sets the default startup frequency when the LPC is off.
• Notes: As per datasheet, set to 20% less than the 1.5 MHz synchronization frequency, which is 1.2 MHz, which becomes 26.7k Ohms.

C2009 Boost and Bias Capacitor

• 0.22 uF, 25V, 0603 10% (TDK C1608X7R1E224K)
• Digikey 445-5191-1-ND
• Purpose: Used to boost the supply to fully turn on the internal switch in U2002.
• Notes: This is set at 0.22uF per figure 5A in the data sheet. So long as the input voltage remains above 7.5V, the boost/bias considerations should be fine.

D2002 catch diode

• Diodes-Inc PDS340-13 "3A SCHOTTKY BARRIER RECTIFIER PowerDI 5"
• Digi-Key PDS340DICT-ND
• Purpose: SPS buck supply diode.
• Notes Using a Schottky diode, faster is better for efficiency, 40Vmax is fine since the OVP will shutdown the IC above 35V. Vreverse = 40Volts , Iave = 3.0Amps , 6.6mm x 3.6mm footprint. Id(avg) = (1.875,,2.14)A = Iout(Vin-Vout)/Vin where Iout=2.5A(max), Vin = (10,,35)V, Vout=5V. Alternative: On-Semi MBRA340T3: Vreverse = 40Volts , Iave = 3.0Amps , 6mm x 3mm footprint. Note apparent junction temperature difference and minimum buy compared to PDS340.

L2001 SPS buck Inductor

• Coiltronics DR74-6R8-R, 6.8uH, Irms = 2.60Amps , Isat = 3.67Amps , DCR = 0.0418Ohms , Shielded , 8mm x 8mm x 4.25mm
• Digi-Key 513-1536-1-ND
• Purpose: Main inductor for the buck SPS.
• Need 5.9uH min. Per data-sheet, start with di/dt for L and use dI= 1A = 0.4(Ioutmax) , Ioutmax=2.5A. Ipeak must be lower than Ilim (the switch current limit). I(L-peak)= 3A = Ioutmax + dI/2 , Ioutmax = 2.5A, dI = 1A. This is good as Ilim ranges from (5.5A to 4.5A) and is always greater than 3A. Check ==> Ioutmax = Ilim-dI/2 = 4A , Ilim(worst case) = 4.5A, dI = 1A, So Ioutmax is at worst 4A, resulting in a value that is still above 3A, this is good. Largest dI occurs at max Vin, choose L using the following formula: L=[(Vout+Vd)/(FswdI)][1-((Vout+Vd)/Vinmax)], Vout=5V, Vd=0.4V, Fsw=1.5MHz, dI=1A, VinMax = 20V (possibly 30V during failures?). L=5.2uH for Vin = 20V, or L=5.9uH for Vin = 30V. Larger values are OK as they will only make dI smaller resulting in improved system responses. RMS rating of L must be greater than Imax = 2.5A. Saturation current should be about 30% higher, Isat = 3.25A, but want Isat above 5A during startup, shorts, Over Voltages, and other failures. To keep efficiency high, DCR should be kept below 0.1 ohm. To minimize EMI, use a shielded toroid with a ferrite core. Again, larger L is good since they should have better DCR values, smaller ripple current (dI), and an easier time keeping out of discontinuous mode, but the physical size will be a limiting factor. THus choose > 5.9 uH = 6.8 uH.

R2011 Feedback resistor R2 for U2002

• 100k ohms, 0603, 1%
• Purpose: Feedback voltage to U2002.
• Notes: This component value was chosen from the reference design. It needs to be big enough to draw very little current through the feedback loop and still feed the appropriate voltages to the feedback error amp and the power good error amp.

R2010 Feedback resistor R1 for U2002

• 532k ohms, 0603, 1%
• Purpose: Feedback voltage to U2002.
• Notes: R1=R2((Vout/0.79V)-1 , where R2=100kohms, Vout=5Volts (chosen as Vin for the Charger of the HAP). 532.911k ohms, closest = 532k ohms.

C2010 Output Capacitor for U2002

• 22uF , 25 V, 20%, Ceramic, 1210 package, X5R (Taiyo Yuden TMK325BJ226MM-T)
• Digi-Key 587-2086-1-ND
• Purpose: Filter the output of U2002.
• Notes: 13.3uF minimum: Cout = 100/(Vout*f(sw)), Vout=5V, f(sw)=1.5MHz. Ensure that ESR is as low as possible to maximize efficiency (0.05 ohms or less). Look for High performance electrolytic or Tantalum caps. Try to use type X5R or X7R. Alternate Capacitor = Digikey Part Number = 445-3945-1-ND , TDK = C3225X7R1C226K , 22uF , 16Volt, 10% , Ceramic , 3.2mm x 2.5mm , X7R. Alternate Capacitor = Digikey Part Number = 478-4594-1-ND , AVX = 1210YC226MAT2A , 22uF , 16Volt, 20% , Ceramic , 3.2mm x 2.5mm , X7R.

HAP (BATTERY)

U2100 HAP Battery charger & power control

• Linear LTC4085 USB Power Manager with Ideal Diode Controller and Li-Ion Charger
• Digi-Key LTC4085EDE#PBF-ND
• Purpose: Charge the HAP battery and provide power path control from the SPS to the HAP.
• Notes: The LTC4085 is a linear charger that has the capability to control 2 external P-MOS devices while charging the battery. During normal operation power is supplied by the SPS and there is no need for the battery. While power is supplied from the SPS The !ACPR! signal from the charger (AC power Present, though in our case the power is DC) will enable power-flow through Q2100 to the HAP Output Regulator (U2003) bypassing the LTC4085. During loss of SPS power, the !ACPR! signal goes away blocking the reverse flow of power from the battery toward the SPS. Anytime the load draws the output voltage down, the ideal diode controller in the LTC4085 will begin to feed power from the battery to the load. This is done both through an internal ideal diode between the BAT and OUT pins, as well as by controlling the gate of Q2101 and using it in parallel with the internal diode. Paralleling the internal diode allows lower resistance sourcing of the battery power to the load. The !CHRG pin is grounded to indicate the battery is charging when the charge current threshhold is passed, which is 5000V/Rprog, or 50mA. Note that the SUSP pin is not connected, because with wall power, SUSP doesn't do anything. HPWR is connected to the LPC so that the LPC can control fast (100% charge) vs slow (20%) charge rates.

Q2100 External power ideal diode

• Vishay SI7123DN-T1-GE3 P-channel MOSFET 20V 25A PowerPAK 1212-8
• Digi-Key SI7123DN-T1-GE3CT-ND
• Purpose: Allow external power to reach HAP battery and HAP supply. This MOSFET is used as an ideal diode by U2100.
• Notes: FIXME: better design info. The current MOSFET is WAY over designed, but that's just fine. Rds(on) = 0.011 ohms, and at 25 A, it's overkill. Yay, overkill!

R2102 Q2100 pull up

• 1K ohm, 0603
• Purpose: Make sure Q2100 is off by default by pulling up the gate.
• Notes: Although 1K seems low, the LTC4085 datasheet recommends it. Huh.

R2100 PROG resistor

• DEPENDS ON BATTERY: 100k ohm, 0603
• Purpose: Sets the default charging current to the battery.
• Notes: Ichrg(A)=50,000V/Rprog. Rprog = 100k ohms sets charge currentto 500mA which should be ~ 1 C for the Lipo battery chosen. There's a test pad at the PROG pin that can be monitored by a high impedance ADC to measure the charging current. Note that this is not connected to the LPC ADC's by default, even though it should be: this error was caught to late to change. Also, Rprog affects the !CHRG pin: it's grounded to indicate the battery is charging when the charge current threshhold is passed, which is 5000V/Rprog, or in this case, 50mA.

R2101 CLPROG resistor

• 660 ohm, 0603
• Purpose: None, really. This is only used when the LTC4085 is powered from USB, which it never is in this circuit.
• Notes: Icl = 1000 V / Rclprog, but the current limit is never used because we have "wall power". Just in case - especially if Q2100 fails - we'll set it to the max recommended which is 1.5 A which comes to 660 ohms. This resistor may be able to be a "NP" (no place), but just in case, we should place it.

C2100 Charge Timer Capacitor

• DEPENDS ON BATTERY: 66 nF, 0603
• Purpose: ALong with Rprog, this cap sets the Lipo battery charge time out.
• Notes: Ttimer(hours)=(CtimerRprog3hours)/(0.1uF*100k). Arbitrarily setting the timer cutoff to to 2 hours gives 66 nF

Q2101 Battery power ideal diode

• Vishay SI7123DN-T1-GE3 P-channel MOSFET 20V 25A PowerPAK 1212-8
• Digi-Key SI7123DN-T1-GE3CT-ND
• Purpose: Allow battery power to reach HAP supply. This MOSFET is used as an ideal diode by U2100. Note that it is NOT where the charging current comes through; charging current comes from the "BAT" pin on the LTC4085.
• Notes: FIXME: better design info. The current MOSFET is WAY over designed, but that's just fine. Rds(on) = 0.011 ohms, and at 25 A, it's overkill. Yay, overkill!

RT2100 Battery temperature thermistor

• 100K ohm 5% NTC in 0805 (Vishay/Dale NTHS0805N17N1003JE)
• Digi-Key 541-1140-1-ND
• Purpose: provides temperature measurement of the battery. Note that this thermister can be placed off-board using J2100.
• Notes: This thermistor is an 0805 package with a 100k ohm value at 25C. Rntc is 100k @25C and Rcold is 3.363 @0C, Rhot is 0.3507 @50C (3.363 and .3507 are from the conversion table for the vishay thermistor R2214 <http://www.vishay.com/docs/33011/convtabs.pdf>)

R2103 Thermistor Rnom

• 121k ohm, 1%, 0805
• Purpose: This resistor forms a voltage divider circuit with the thermistor RT2100 and its delta modifying resistor R2104 which results in a voltage dilivered to the NTC pin that represents the current temperature of the battery.
• Notes: As we want the thermistor to use a minimum amount of power in this design, we are using a 100k thermistor. The resistor value is then calculated by: Rnom=((Rcold-Rhot)/(2.815-.4086))*Rntc (where Rntc is 100k @25C and Rcold is 3.363 @0C, Rhot is .3507 @50C (3.363 and .3507 are from the conversion table for the vishay thermistor R2214 <http://www.vishay.com/docs/33011/convtabs.pdf>) resulting in a value of 125k, using the nearest standard 1% resistor results in Rnom = 121k ohms.

R2104 Thermistor Rdelta

• 15k ohm, 1%, 0805
• Purpose: This resistor forms a voltage divider circuit with the thermistor RT2100 and bias resistor R2103 which results in a voltage delivered to the NTC pin that represents the current temperature of the battery.
• Notes: This particular resistor is in series with the thermistor and widens the temperature delta of the thermistor to set 50C as the the T-hot trip point. The resistor value is calculated by: Rdelta=([(.04086/(2.815-.4086))(Rcold-Rhot)]-Rhot)Rntc (where Rntc is 100k @25C and Rcold is 3.363 @0C, Rhot is .3507 @50C (3.363 and .3507 are from the conversion table for the vishay thermistor R2214 <http://www.vishay.com/docs/33011/convtabs.pdf>) resulting in a value of 16k, using the nearest standard 1% resistor results in Rnom = 15k ohms.

LED2100 Red charge LED

• Red LED, 0603, 2 mA, Vf =1.8V (OSRAM Semiconductor: LS L29K-G1J2-1-0-2-R18-Z)
• Digi-Key: 475-2506-1-ND
• Notes: When the battery charge current is above 50mA this red LED is lit indicating the battery is charging. We don't want it to draw a lot of power, so it has been chosen as a 2mA part in an 0603 package. While not bright, it should be useable. The current is set by resistor R2105 to be 2mA when the battery is charging (Vout-Vfwd)/Iset = (5V-1.8V)/2mA = 1.6k ohms.

R2105 Charge LED resistor

• 1.6k ohms. Ohms, 0603
• Purpose: Set the LED current.
• Notes: See LED2100 discussion.

R2112 PORT1.10 protection resistor

• 10K ohm, 0603
• Purpose: protect LPC inputs from 5V 'PROTECTEDDCPOWER' supply.
• Notes: It's probably true that the 5V - Vf(LED2100) - (V drop from R2105) is probably no problem for the LPC, but just in case, we added R2112 to reduce any over 3.3V power to non-lethal levels.

B2100 HAP battery

• 860mAh Lithium polymer battery
• Sparkfun.com PRT-00341
• Purpose: provide backup HAP power
• Notes: FIXME: How did we arrive at 860 mAh?

HAP (SUPPLY)

U2102 TPS63000 buck-boot supply

• TI TPS63000DRC (FIXME: The TPS63020 is the chip we REALLY want (4A!) but it's not available. Clearly, we should switch to this chip in v2)
• Digi-Key 296-19641-1-ND
• Purpose: Take either the 2.0 - 4.2 V HAP battery or the 5V SPS output and turn it into 3.3V. This clearly requires a buck-boost switching supply.
• Notes: U2102 takes the voltage supplied by the battery or charger in the HAP and DC-DC converts it to the required voltages for the remainder of the circuit. In most cases this is 3.3V, though with small modifications this voltage can be set to anywhere from 2.5 to 5.5 Volts. The circuit is synchronized to the 1.5MHz clock that is stepped down from the system clock by connection to the PS/Sync pin resulting in constant frequency operation that should not interfere in audio bands. The TPS63000 is noteworthy in that it acts as either a Buck regulator, or a boost regulator and not as an inefficient buck-boost regulator. It manages this by only activating two of it's four internal switches at a given time. The TPS63000 changes automatically from buck to boost operation as required on a per cycle basis.

C2101 Bulk input capacitor

• 10 uF, >= 10 V, 0805 (Murata GRM21BR61A106KE19L)
• Digi-Key 490-1709-1-ND
• Purpose: input cap for HAP switcher.
• Notes: Datasheet recommends at least 4.7uF bypass cap from the out pin of the LTC4085 to ground. This capacitor holds up the output voltage when the battery is initially switched in. It's ALSO part the right value for the TPS63000. But we already have 10 uF in stock, so we'll just use that. See also C2105, C2108.

C2102 bypass cap

• 100nF, X5R ceramic, 0603
• Purpose: Bypass paranoia
• Notes: We had the room, it makes sense to through in small capacitance when you can. This could be a 10 nF capacitor... maybe should be?

U2102 HAP shutdown latch

• NXP 74LVC1G74DC,125 single D flipflop with set/reset in a VSSOP8
• Digi-Key 568-4494-1-ND
• Purpose: Lets the LPC power down the HAP output and keep it off until external power is restored.
• Notes: We need a way for the node to shut itself off when the external power is disconnected, else the node will run until the battery is killed every time. We can't just let the LPC run the U2102 enable line, since as it powers down, it'll reset, and bring the enable line high due to internal pull-ups (crazy NXP engineers!). So what we have to do is latch the shutdown command. We use a D flipflop that's powered by the battery, which is OK because it takes Iq = 500 + 10 = 510 uA. When connected to external power, the SPS is running, and thus the !ACPR line on the LTC4085 is asserted low which asserts the !SD (or !PRE in the schematic) in U21202. When !PRE is asserted low, it forced Q to be high, which asserts the TPS6300 enable, which turns on the HAP circuit. No matter what the LPC does with to the clock input, asserting !PRE will keep the HAP on. Once the SPS is shutdown, then !ACPR is no longer asserted, so !PRE goes high (is deasserted) via R2102. At this point, when the LPC brings the PORT1.14/HAPDISABLE high, it will clock in the 'D' which is tied to ground. This will bring Q low, which brings EN low on the TPS63000, which shuts of the HAP. The HAP will power up once external power is applied via !ACPR and !PRE to Q to EN, like before. The 74LVC1G74DC has the right voltage supply range (1.65 V to 5.5 V), it has nicely tolerant inputs for high voltage ranges, and will even clock on the LPC's 3.3V GPIO when the HAP is being run on battery( ~4.2 V). U2102 needs no bypass caps, since it's connected to a couple of them that bypass the PROTECTEDDC_PWR line.

R2107 U2102 clock pull down

• 1k ohm, 0603
• Purpose: Agressively keeps the clock line pulled down.
• Notes: The low impedance is to make sure any weirdness on reset from the LPC doesn't affect the clock pin. Theoretically, a low pass filter or a one-shot would be a good idea here. But for now, just a stiff resistor is fine.

R2108 Startup delay resistor (?)

• 100 Ohms, 0603
• Purpose: Becuase the datasheet had it? WTH is this resistor for? It looks like a soft start, or least a "delayed start" after power up.
• Notes: From the datasheet, for no explanable reason. In combination with C2103 this resistor probably creates a time constant that forces the controller to wait for a while after power is applied. TODO: Find out what this is for, and how these values were calculated.

C2103 Startup delay capacitor (?)

• 100 nF, 0603
• Purpose: See R2108.
• Notes: See R2108.

R2109 U2101 feedback resistor divider 1 of 2

• 1M ohm resistor for 3.3V operation. 200k ohm for 5V operations
• Purpose: Set feedback voltage for regulation. Also see R2110
• Notes: Calculation for 3.3V: R2((Vout/Vfb)-1) = 1.12M ohm, given Vout = 3.3V, Vfb = 500mV, R2110 = 200kohms. For 5V operation: R2((Vout/Vfb)-1) = 1.8M ohm, given Vout = 5.0V, Vfb = 500mV, R2110 = 200kohms. CAUTION! Ensure that L2100 is capable of safe operation at 5Vdc before making this change.

R2110 U2101 Feedback resistor divider 2 of 2

• 200k Ohms, 0603
• Purpose: Set feedback voltage for regulation. Also see R2109.
• Datasheet rcommends keeping this part in the range of 200k ohms. No good reason to change this, though the efficiency could be slightly better if a larger value is used. Keep Feedback divider current at or above 1uA.

C2104 Feed-forward capacitor

• For 3.3V: 2pF, X5R ceramic
• Purpose: help voltage feedback
• Notes:C2014(3.3V)= 1.96pF = Feed-forward capacitor = 2.2uS/R1. For 5.0V, go with 1.2pF, X7R ceramic (5V= 1.22pF = Feedforward capacitor = 2.2uS/R1)

L2100 HAP output inductor

• 2.2uH, 1.44 A, 72m ohm, (Taiyo Yuden NR4018T2R2M)
• Digikey 587-1669-1-ND
  • Recommended value: 2.2uH, 1.75A Irms, 2.26A Isat
  • L2202(suggested) = 2.2uH, this is the inductor value suggested in the datasheet for 3.3V operation.
  • L2202(min) = 1.57uH, this is the larger of 1.57uH = (Vout(Vinmax-Vout)/(Vinmaxf0.3A) or 1.34uH = (Vin_min(Vout-Vinmin))/(Vinminf0.3A), where Vout=3.3V, Vinmin=2.5V, Vinmax=4.2V, f=1.5MHz.
  • Imax(3.3V) = 1.74A, Isat= 2.26A = Imax+30%

C2105 HAP supply bulk output cap (1/2)

• 10 uF, >= 10 V, 0805 (Murata GRM21BR61A106KE19L)
• Digi-Key 490-1709-1-ND
• Purpose: SPS bulk output capacitance.
• Notes: TI datasheet says 22 uF, and break it up into two caps. That's brilliant, so that's what we did. Cout=5L(uF/uH), L=2.2uH. In combination with L2100, this capacitor acts as the output filter. Datasheet recommends small ceramic cap as close to Vout and Pgnd pins as possible.

C2108 HAP supply bulk output cap (2/2)

• 10 uF, >= 10 V, 0805 (Murata GRM21BR61A106KE19L)
• Digi-Key 490-1709-1-ND
• Notes: TI datasheet says 22 uF, and break it up into two caps. That's brilliant, so that's what we did. See C2105.

HAP (MISC)

LED2101 Amber 3.3V on LED

• Amber LED, 0603, 2 mA, Vf =1.8V (OSRAM Semiconductor: LY L29K-H1K2-26-Z)
• Digi-Key: 475-1196-1-ND
• Notes: Amber LED when 3.3V power is on. The current is set by resistor R2111 to be 2mA when the battery is charging (Vout-Vfwd)/Iset = (3.3V-1.8V)/2mA = 750 ohms.

R2111 Charge LED resistor

• 740 ohms, 0603
• Purpose: Set the LED current to 2mA.
• Notes: See LED2101 discussion.

D2100 HAP "secondary" Buck (Catch) Schottky Diode

• 30 V, 1.5 A, 4 ns, New MiniPower 2P, (Panasonic - SSG, MA2Q70500L)
• Digi-Key MA2Q70500LCT-ND
• Purpose: D2100 along with C2106 and L2100 form the "crazy Tim second buck switching power supply" which will be may never be used, but if it is, it'll be regulated down to 5 V possibly with a low-dropout (LDO) linear voltage regulator. Also the general understanding was that this secondary buck will power specific parts like 5 V ADCs on certain nodes (like the IMU) and we expect that this diode's rated specs are more than enough.
• Notes: Using the formula on page 9 in the LT1767 datasheet we calculated the average DC current that D21-- should be able to handle. Id,avg = Io (Vin - Vout) / Vin, where Io is the secondary output current of the SPS, Vin is the voltage at the node which is between L200a and L200b and Vout is the SPS secondary output voltage. We assumed that Vin would be switching somewhere between 1.6 V and 7 V. With Vout = 5 V and using the worst case Io = 1 A and Vin = 7 V values Id,avg = 286 mA (again this diode is overrated). We knew that not every node would need a secondary 5 V supply but even the ones that did, the added current should not cause overcurrent events. Hopefully. Due to the nature of the new TPS63000, this is unlikely to function. In LV2B this was connected to a simple buck supply and could have worked. In LV2C, it is unlikely to raise the voltage from 3.3 to 5 volts, but it has been left in as it can be depopulated if it turns out to be nonfunctional.

C2106 HAP "secondary" buck filter cap (1 of 2)

• ???
• Purpose: This cap along with D2100, L2100, and C2107 form the second buck switching voltage power supply which will be eventually regulated down to 5 V possibly with a low-dropout (LDO) linear voltage regulator. These are application specific caps whose values are mostly independent from the SPS design.
  • Specifications/ Calculations: The only difference between C2252 and C2252a are the packages and that only one of them will actually be on the PCB. Since we do not know any details about the actual application specific circuitry each SPS will power from an SPS design point of view, we chose to use both a 0805 and 1206 package. We chose two packages because we moved all relevant parts to the 0805 package from 1206 as in the LV2 SPS design and in case a specific application node needs a more beefy cap a 1206 package cap can be used. The lay out of the parts will not be side by side as suggested in the schematic but are offset and superimposed on top of each other on the same side of the PCB. Because only one cap will be used we offset the pads such that they are not directly on top of each other and either package can be placed down thus saving space. The values are TBD. Due to the nature of the new TPS63000, this is unlikely to function. In LV2B this was connected to a simple buck supply and could have worked. In LV2C, it is unlikely to raise the voltage from 3.3 to 5 volts, but it has been left in as it can be depopulated if it turns out to be nonfunctional.

C2107 HAP "secondary" buck filter cap (2 of 2)

• ???
• Purpose: See C2106.
• Notes: See C2106.

UC

LED2200 RGB LED

• RGB LED in a PLCC-6 (Osram LRTB G6TG-TU7+VV7+ST7-IB)
• Digi-Key 475-1319-1-ND
• Notes: Used in other projects, seems OK. And everything needs a RGB LED these days.

Q2200, Q2201 RGB LED driver

• Logic level N-channel mosfet, 100V, SOT23, (Zetex BSS123TA)
• Digikey BSS123CT-ND
• Purpose: Buffers the green and blue LEDs (LPC probably won't drive, so we use a mosfet).
• Notes: Any old N channel MOSFET will work.

R2205, R2206, R2207

• 1k ohm, 0603
• Purpose: current limits for RGB LED.
• Notes: Values probably need to be tuned to get the right light balance.

U2203 Clock divider for SPS and HAP

• SN74LV163A
• Digi-Key 296-13936-1-ND
• Purpose: Takes LPC 12 MHz clock and divides it down by 8 to get 1.5 MHz clock for SPS and HAP.
• Notes: Simply takes bit 3 of 4 bit counter. TODO: replace with D flipflops to get cool spread-phase clock.

J2204 debug connector

• JST Connector 16R-JED 1.25 mm pitch 16 pin receptacle (female)
• Notes: Wanted a high density debug connector with hundeds of cycles. Female chosen so that you can't plug the debug connector into the rocketbus connector by accident. Mates to the 16PS-JED or the 16P-JED.

TVS2200, TVS2201 Surge protectors

• TI SN65220DBVR sigle USB port TVS in SOT-23-6
• Digi-Key 296-9694-1-ND
• Purpose: Transient Voltage Supressor (TVS) for USB and UART lines.
• Notes: Used in other projects, seems to work OK. Specifically designed for USB 1.1.

L2200, L2201 "T" EMI filter

• Panasonic EXC-CET471U EMI filter, 470PF, 50V
• Digi-Key P9829CT-ND
• Purpose: Filter out incoming and outgoing EMI on USB lines. May need to be bypassed if they introduce too much filtering on USB lines.

U2200 LPC2368

• NXP LPC2368FBD100,551
• Digi-Key 568-3997-ND
• Purpose: It's the uC!
• Notes: TODO

X2201 12 MHz crystal

• Abracon ABM7-12.000MHZ-D-2-Y-T 12 MHz 18 pF 20 ppm crystal
• Digi-Key 535-9836-1-ND
• Purpose: Main 12 MHz crystal for the LPC2368. Also is divided down and clocks the SPS and HAP switchers.
• Notes: TODO

X2200 32.768 KHz crystal

• Citizen CM415-32.768KEZF-UT 32.768 KHz 12.5 pF 10 ppm
• Digi-Key 300-8750-1-ND
• Purpose: Real time clock (RTC) crystal. Can be used for low power mode.
• Notes: TODO

U2201 CAN driver

• TI SN65HVD235D CAN controller
• Mouser 595-SN65HVD235D
• Purpose: drive the differential CAN bus from the single ended UC
• Notes: High on the AB ("autobaud") pin disconnects the transmitter.

D2201, D2202 CAN disconnect "OR"ing diodes

• >= 30 V, 100 mA, < 350, 0.35 mV Vf @ 10mA (Comchip Technology CDBU0130L), 0603
• Digi-Key 641-1282-1-ND
• Purpose: Allows the stuck dominant stuck bit circuit and the LPC to both disconnect the CAN transciever.
• Notes: TODO

R2203 CAN disconnect pulldown

• 10K ohm, 0603
• Purpose: Pull down the AB ("autobaud") pin to enable the CAN transceiver (U2201).
• Notes: Arbitrarily chosen. It may want to be a lot higher resistance, so that the ~ 60 K ohm of the LPC pull-up resistors can disconnect the transceiver. That said, if there was a problem with a dominant bit, the stuck dominant bit timer would get it. So 10K is probably fine.

U2202 Triple Schmit inverters for CAN stuck-dominant bit

• TI SN74LVC3G14DCUR Triple inverter in US-8
• Digi-Key 296-13006-1-ND
• Purpose: Flexibly implement stuck-dominant-bit time-out circuit for CAN bus.
• Notes: When the CAN transmit line goes low, it should only go low for a short time, like 7 CAN bit times (7 us). If it gets stuck low (a dominant bit), it could bring down the whole CAN bus. To prevent this, we disconnect the CAN transceiver U2201 from the bus using it's "Autobaud" feature ("AB") which disconnects the driver in the chip from the CAN bus. The Schmitt trigger U2202C charges C2200 through R2202 when there's a dominant bit (a '0'). After some time constant set by C220,R2202 and the voltage thresholds of U2202, it triggers U2202B, which gets inverted by U2202A, which then asserts AB and disconnects the driver from the CAN bus. Note that the LPC can also disconnect the U2202 using PORT0.4. On the CAN bus, 6 dominant bits in a row is considered an active error flag since every 5 bits of the same type in a row requires an opposite bit to be "bit stuffed" in there (this helps recover the implicit clock in the CAN bus). at 1 Mbps, you should never see more than 6 us of dominant bits. Being conservative, you can give a nice safety margin by a factor of 10 and say if you ever see 60 us worth of dominant bits, something is very wrong. One tau, or 63%, of 3.3 V is 2.08 V. Vih for the inverter is (1.3,,2.2) V. So choosing the more conservative 1.3 V, and calling that about 63%, means we want a time constant of 60 us / 63% = 95 us so we'll call it 100 us since that's easy. We choose the capacitance to be small so we don't get big current spikes when we discharge it, so call it 10 nF. That gives us a resistance of 10K, which is handy. For the record, 100 us is about 2.5 minimum CAN frames, so we could be more conservative and go even lower. But we'll stick with this for now, and see how it works.

D2200 Stuck dominant bit diode

• >= 30 V, 100 mA, < 350, 0.35 mV Vf @ 10mA (Comchip Technology CDBU0130L), 0603
• Digi-Key 641-1282-1-ND
• Purpose: Provides a quick "reset" of the stuck dominant bit RC timer if there's a recessive bit.
• Notes: TODO

R2202 Stuck dominant bit timer resistor

• 10k ohms, 0603
• Purpose: Charges C2200.
• Notes: See U2202 discussion.

C2200 Stuck dominant bit timer capacitor

• 10nF, 0603
• Purpose: Stores charge for timer.
• Notes: See U2202 discussion.

Design log

2011-03-14

Fabbed the boards and got them back! Gerbers that were fabbed were in git commit 5f6bbb58075a9bea24e6dde172e43951f1043f0d. Issues so far:

• Critical
  • TBD
• Major
  • TCK probably requires a high value pull down resistor; see page 661 in the User's Manual. There should room next to J2204 to place an 0603 pull-down.
  • UNCOMFIRMED: PORT1.18/GPIOUSBHIGHSPEEDSELECT should have gone to a P ch MOSFET to pull that line high. See Olimex LPC2378 board.
  • F2000 ought to be a 2.75 or higher fuse, given the 2A circuit breaker threshold.
• Minor
  • Redid the schematic of the rocketbus connector after building board. No real change, mostly to match pin #s which were flipped before.
  • D2001/D2200/D2201/D2203 have a different value than shown on the schematic.
  • Some signals have an extra '!' which I removed: '!signalname!'