Cape v2.0 - PCB Changes
General rule for assembly-drawings
origin orientation
keepout
assembly notes (mech15)
designator
keepout
top designator (mech 2?)
Copper / Silk
L1 Pads / Via
Keepout
Top Overlay ⇾ print in Color
Implemented Changes v2.0r0
4 Layer! Planes for Sig, GND, A5V, (3V3)
decide Manufacturer, EC, Aisler, Betalayout
add design rules
add layer stackup
add default vias
divide in groups / rooms
optimize surroundings of ICs
change vias of pson50, dfn-10 (by lt3487 spec)
move lvlchangers to the left
change pads of pinheaders in inner layers
thermal pad of switch unused? yes, no word of use in datasheet
increase restring / holesize, sheph seems to have 0.15mm holes?, target 0.075 ring
Implemented Changes v2.0r1
Beaglebone
Emulator
DAC
previous: DAC8562SDGSR, mouser 595-DAC8562SDGSR
50 MHz SPI, 7-10 us Settling, 1-4 mV Zero-Scale-Error, 40 mA Short-Circuit-Current,
constraints: >= 16 bit, SMD, 2 CH, not maxim, v-range ~2.5…5.5 V, short settling time
this could also be low-res and slow for only bq-output-sim, but it would benefit to also tap in cap-voltage directly
https://www.mouser.de/Semiconductors/Data-Converter-ICs/Digital-to-Analog-Converters-DAC/_/N-4c44d?P=1z0w8k6Z1z0w2wwZ1z0w2wvZ1z0w2wtZ1z0z7ptZ1yz5pwlZ1yzmm10Z1yzml2aZ1yzmm18Z1yzmlprZ1yzmm0yZ1yzmm13Z1yzmlr9Z1yzmlh1Z1yzmlwtZ1yzmm16Z1yzmm0zZ1yyh4l4Z1z0zls6Z1yzxao2&Ns=Pricing%7c0
replacement: AD5663ARMZ-REEL7, mouser 584-AD5663ARMZ-R7
50 MHz SPI, 4 us Settling, Zero-Scale-Error<1mV, 30mA Shor-Circuit-Current, needs voltage reference, WATCH OUT - there are versions with midpoint-start
replacement: DAC8830, 1-CH, 50 MHz, 16bit, 10nV/sqrtHz, 1us Settling,
replacement: AD5545B, 2-CH, 50
OpAmp for V-BUF 2CH?
previous: OPA2388IDGKT, digikey 296-50277-2-ND
30-60 mA perm, 5 V/us, 7 nV / sqrtHz, 0.25 uV Offset,
constraints: opAmp, 3CH, supply ~ 3-5 V, Rail2Rail
https://www.mouser.de/Semiconductors/Integrated-Circuits-ICs/Amplifier-ICs/Operational-Amplifiers-Op-Amps/_/N-6j73m?P=1yzxao0Z1yzmm18Z1yzmm0xZ1yzmm13Z1yzmm14&Ns=Pricing|0
replacement: AD8606ARMZ-REEL, mouser 584-AD8606ARMZ-R
2CH, 80 mA, 5 V/us, 8 nV/sqrtHz, 20 uV Input Offset,
shunt-Resistor
nRF52 takes 9 mA @ 4dBm, 16 mA @ 8 dBm for ~ 200 us, rest is below 2 mA,
previous: 2 Ohm 1% ⇾ 16 mA => 32 mV, would mean 1% Voltage drop at 3V3, less would be better
current sensors are no alternative, too expensive, not enough resolution
replacement: 1 Ohm 0.1% ⇾ 1:1 mA:mA, 0603 or 1206
RT1206BRD071RL, mouser 603-RT1206BRD071RL
OP-Amp for Shunt
previous: AD8422BRMZ in combination with LM27762DSSR
constraints: 1 CH, > 2 MHz Gain-BW-Product, Supply ~ 2-5 V, >75 dB CMRR, Low input offset voltage
replacement: INA331AIDGKR, mouser 595-INA331AIDGKR, in combination with LM7705 (-0.23V) on V-, mouser 926-LM7705MMX/NOPB
ref: https://e2e.ti.com/support/amplifiers/f/14/t/700003
proper replacement: ad8429B
https://tools.analog.com/en/diamond/#difL=0&difR=0.05&difSl=0&gain=100&l=0&pr=AD8429&r=5&sl=0&tab=1&ty=2&vn=-8&vp=9&vr=0
https://training.ti.com/system/files/docs/1312%20-%20Noise%202%20-%20slides.pdf
ADC 2CH
previous: ADS8694TSSOP38 4 CH
18 bit, 4 CH, two V-Rails for A&D, 500 kSPS, 18 MHz SPI, variable LPF, 1175 ns Acq & 825 ns Conv.
constraints: 2CH, 18-24 Bit, SMD, >100 kSPS
replacement: ADS8691, 1CH 1 MSPS 8€, ADS8695 1CH 500kSPS 9€, ADS8699 1CH 100kSPS 6€, Acq 335/1000/5000ns, Conv 665/1000/5000ns
analog switch ⇾ is there a way to power the offline target? Switch up supplies
previous: TMUX1101DCK, 4 Ohm, 1 SPST SinglePole-SingleThrow
constraints: 2 Ch, legs, supply >= 5, rdson <= 500 mOhm,
replacement: NLAS4684MR2G, mouser 863-NLAS4684MR2G, 2CH, 300 mA Conti, 500 mOhm rds
Target-Port-IO (GPIO, SPI, I2C, UART, SWD/JTAG, BAT_OK PRU) ⇾ Q: is HS-GPIO enough? rest is userspace-logged
debug to target
voltage-level-translator,
previous1: TXB0304RUTR BiDir, autosense, min 3mA input drive current, 4 CH, >40 Mbps
previous2: SN74LV4T125PWR UniDir
constraints: 1 Mbps, high channelcount, autosensing, 2-5V, HighZ-Mode
replacement: NXS0108PWJ, mouser 771-NXS0108PWJ, 50 Mbps, BiDir, Autosense, open drain, 8 Bit, NXB-Version: 2mA input drive req.
⇾ WARNING: expected 18.01.2021, nxs0101 already in stock, nxs0102 in may
target-port ⇾ default pin-header, maybe smaller version of it
suppply for second target ⇾ 2. CH of DAC + Buffer
status-Leds
green 575nm, 0603, 60mcd 2V@20mA, 150060VS55040, mouser 710-150060VS55040
blue 470nm, 0603, 80 mcd 3.2V@60mA, 150060BS55040, mouser 710-150060BS55040
red 645nm, 0603, 70 mcd, 2V@20mA, 150060SS55040, mouser 710-150060SS55040
orange 605nm, 0603, 100 mcd, 2.2V@20mA,
LEDs for current active (and powered) Target
multipurpose nChannel MosFet
constraints: <50mOhm, smd, n-CHannel, VGS <=700mV
sot-323-3: DMN2058UW-7, mouser 621-DMN2058UW-7
i2c-storage, prev: CAT24C256WI-GT3
Cage
Part Properties:
price (for ten), manufacturer, manufacturer id, shop 1, shop 1 ID, …
special properties: max voltage, power, current, size / package, color, forward Voltage
extra information (i2c-adress, spi-speed, ) directly in schematic
power-recording-stage
DAC DAC80501ZDGSR
OPAmp OPA388ID, pin-compatible with LTC2050HV
nMOS SI2374DS, test with BSH103
ShuntOPAmp Ina190A1IDCKR
power in via vdd_5v (P5/6) ⇾ Test shows: BB does not power up via sys_5v
reboot / boot via Pin-Toggle (Shutdown via command), we should trigger both (RESn⇾PD,PWR⇾PD), Test shows: Reset works while PWR is in PD
add 256 GB USB-Stick
switch to smaller IC-Packages and 0402
order / add GPS
is the gps capable of alarm (wake up sys)
our 5V analog should be stabilized more! Add A5V with 2 Stage Bead, or real coil
add footprint for layer-windows
add footprint for shepherd-logo
give INA190 a negative supply (>1mV would be enough) on GND-pin, ref stays on common gnd, extra decouple
Debug-Pins with Ground
extend harvest-Port, add option to measure VSense, and output VCap (V_A of Emulator)
it would be wise to detach a5v even further from 5V, with a low-drop diode
EMI-guard SPI, currentlimit at pinheader, terminate at ICs, 33 Ohms close to cpu recommended (avoid reflections)
add alarm-feature, something SPI-programmable, that can act like a watchdog, with at least max 1-4h windows
check against shepherd v1.5
don’t shut down individual Emu / Rec - Parts (delete or just disable all at once) ⇾ done by Pwr-control
Harvester needs second channel ADC with very low input current, 1MOhm is too low
manual button with LED ⇾ connector S4B-ZR-SM4A-TF, P1 3V3, P2 LED ODrain, P3 SenseButton with PU, P4-6 GND
add ultra low noise LDO to A5V, and possibly a boost-converter upfront
find better level translator, less current (best if near 0)
reprocessed 11_concept.file
switched Ina190 for AD8421
added boost/Inverter for proper voltage rail
add target port (comparator-include?) System will be a nRF52840 and most likely a MSP430
try to make it compatible with breadboard / dev-Kit
is spy-by-wire physically compatible with swd ⇾ it is, TClock is uni-dir, TDIO is bi-dir
replace 100nF/16, 1uF/16, 10uF/16
BOM, more precise alternative - BB uses 32.768 kHz osci MC-306 (20 ppm, 8x3.8mm) or similar, package says 327A5M
alternative: 5 ppm, 12.5pF, 50 kOhm, https://www.mouser.de/ProductDetail/Citizen-FineDevice/CM200C32768HZFT?qs=rkhjVJ6%2F3ELrGt3qchcVtQ%3D%3D
BB also uses 24.576 MHz
check output limits of opax388 and DAC
compare lowNoise LDO to LM27762
750 kOhm 1%, 667-ERJ-2RKF7503X, 5 + 32
connect BB-Pins, 500 Ohm to input pins that could be driven from both sides
complete ERC
1uF/16V is still 0603, change to 0402, there are 34x (incl. Recorder)
redistribute capacitors
replace coil with smaller one, check recommended direction
add 1kR & 100R high precision for current measurement, EMU
order digikey (extBut, samtec), mouser, csv
add footprint for quality-control-panel
BB Pinheader Cape-Design Stays ⇾ possible alternaltive Producer is Samtech, design is now divided
add production-constraints
update BOM
v2r1 ordered
Implemented Changes v2.0r2
silkscreen - 10k array are 1k
3d-Step: shield transparent for better view below
silk. “P3” designator is below component
pads of 0402 bigger, 1.2x1.2 as min area for pads? paste 0.8x0.8? got some tombstones on current revision
half a pad distance to neighbour (keepout)
improve connection betwenn pads (esp. 0402)
feducial seems massive, shrink it?
SPI-CS with at least 33R to lower chance of interference
R for BatteryGood is wrong (handled by PRU), should it also be routed to userspace?
TC7WH-Footprint is to large
HC2G-comment (Name) is wrong / twisted
DAC pads could be longer, reach more under the IC
MP32 has no dot on package, just [ABC], lower left is pin1
NSR (or all diodes) could get a line on the 3d-file (current dot is confusing, because the actual package has line and dot)
the orange led is red! … bad UI for general blinking
describe gain of OP-Amp in schematic
100R, accuracy is missing in description / constraints
Silk for L9/L10
essential silk-numbers on headers can be bigger/bold, a bit more distance from header
rename rec to harvest, port on pcb more obvious
harvest dAC ch-b - drain power source OK?
improve marking on Voltages, maybe with an arrow
reverse diode of mosfet is too weak - already destroyed the reverse polarity protection
some popular Flag-Markers on backside for pinheader
Pin1 dot should be a shallow/sharp triangle, pointing in the right direction (less confusing in dense layouts)
transistor-footprint is not conform to datasheet, it is slighly bigger and just misses the pads?!?
u13 thermal to wide, reduce a bit to avoid shorts (u15 has same possible weakness)
rework als Pin-Descriptions (already mentioned)
round and divided (big) paste pads
add general power-led when shepherd is on? maybe on 6V line
switch to thinner stencil, bigger pads (paste is good for it)
header-row on target has paste, it should not
Font/Writing in copper is unreadable on target (slighly below spec)
add layer for manual / pick’and’place descriptions (m15, m?)
qr code is not readable, blurred, reduce size of “pixel”
warning for harvest V_sense ⇾ Voltage floats if not connected and will most likely show V-Max in this state
ADC nRST should not get A5V, only 3.3, also there is no need for a resistor-switch
Connector for external switch was copied from old schematic, but this one had inversed pin-numbering on connector, different from datasheet
Testpoint on RVS-Pin of
protect GND better around Pinheaders
U25A, Inputs are switched
diode between both 5V-lines should be two diodes that feed the watchdog
3V3 should also be switched - maybe even the 5V0 in, so the PU that hinder bootup are meaningless
vias came back only weak tainted (make it less severe if that happens), mainly for target pcb
more distance to gnd-plane (soldering is hard, even with thermals
TEST Watchdog, make compatible with BB
TEST GPIO to and from target
same orientation for transistors
switch p8-27 and 29 (batOK / GPIO4) for cleaner readout
switch p8-43/44 with p8-39/40 for proper boot
TEST boot with all pins except 3v3 & 5V
but make sure that shepherd-EN (p8-13) stays low during boot
switching main power to both targets shows, that the routes seem to have different current-readings for the same load! odd
1k for LEDs
check surroundings of mosfets, size has changed
mosfet of watchdog-wake could be tied to 3V3 with additional 1k resistor or diode (so that button stays usable)
100k PD for wd-pin (or keep 10k)
3V3 goes low on powerdown, so watchdog can enable (pull low wake) BB again,
polling has no effect during normal operation
TEST if 1k is enough to pull line low to enable ⇾ it is
add open source hardware logo?
Schematic Changes - Open
prepare calibration
ordered not enough 15uH Coils, need 30 more
check remainder of BOM for emu-only assembly
shepV1 had a user-space led, which is still there, same pin, but pru-controlled, was it the same in v1?
explain schematics,
Part Changes
Note: after Mouser-Order - NOW already ordered
DAC 100nF ⇾ 1uF
Boost 10uF ⇾ 10uF (inv)
Inv +47uF
Rec 2x 10k 0603 ⇾ 0402
All 34x 1uF/16V/0603 ⇾ 1uF/25V/0402
final Shield, cover & frame
all 4x 15uH/4mm ⇾ 15uH/3mm
parts of nRF-Target
Schematic Changes Postponed
internal calibration? with 2 switches and 1 calibration-linear-power-supply
OP-Amp, bias Subtractor: LMP7701MF, not needed now
sync to pps ⇾ external pcb
usb-socket is impossible to hand-solder right now
(maybe) add PU to watchdog outputs
separate PCBs for PPS-Source, Recorder, Emulator
Open Issues
only 5 diodes of type PMEG in order?
change 100 uF to 47u? one less component
Testpad should be square for GND, half-circle for Signal or similar
r3 (dac, emu)
proper naming for TP if there is space
TPs should have bigger hole, so probes stick
oscillating opAmp should be outside of cage
allow install of a big cap on A5V or 5V, 5v5 1F is cheap
add longer pinheader for p8/p9 23-46