Cape v2.2 - PCB Changes
Implemented Changes
INFO for v2.1r0:
first 4 Target-Boards has green LED1 (down) and red LED2 (up)
Vanilla Shepherd v2.1 Cape stopped BB from booting and even putting it on while running failed (p8 41-44)
removed watchdogpins (pwrOn&reset) allowed to put cape on while running (still not while booting)
L3V-TestPosition is bad, easy to produce shorts with neighbour
update expected voltage levels in schematic
label for pwr-led
AB-label of target bigger
P7/P8 seems wrong name for big pinheaders
testpoints without paste, and small hole in copper to lock a probe ⇾ just less paste
is there a way for cage-selfalignment? ⇾ negative solder mask exp
remove layerwindows and make number bigger
describe external powering better in schematic
pin1 markings should be bigger
emulationBug! - Critical
solutionA - MinimalEffort: Swap FbB- with RailA-Trace
FbB ⇾ rotate R1/10k clockwise and attach it to middle pad
RailA ⇾ cut Trace between R9 and Via
connect enamel from R9 to (now) free Pad of R1 ⇾ this routes RailA to Switch
connect enamel from TP1-Pad to rotated R1 ⇾ this routes FbB to Switch
solutionB - clean: rewire output of AnalogSwitch
silk: vCap is now V_SimBuf
schem: recorder gain is now just 10
more Metal for cooling the recorder-Path
Rail-LED is twisted! ⇾ No its not
add 2x23 to BOM SSQ-123-03-G-D , digikey SAM1196-23-ND
VHarv lowpass is 160 kHz not 16 kHz as shown
change all lowpasses to 160 kHz
emulator should go back to 1 Ohm Shunt, 100 Ohms for InAmp for 100:1 Amp ⇾ now 2R & 1:50
Switch Ext-Pwr Pins (instincts are strong for edge-pin to be GND) and don’t forget Silkscreens!
EEPROM needs to be always powered (BB 3V3)
reduce V_IO_BUF OPAmp-Resistor to 10 Ohms
a little bit bigger 0402-pads, they get loose quiet fast
level-change performance is still bad! maybe add 1k back?
add back 1 kOhm as series resistor for LSF0801. maybe a bit lower because edges are a bit slow (1 us)
put a note in schematic for WD
BB_START has 5V Level when BB is on, gets pulled to 3V3 when WD does its routine because schematic uses BB3V on it. but that seems ok!
get target A/B/1/2 straight. it is target 1/2 from now on! ⇾ Skip it
C over Shunt is sometimes contraproductive (100 nF || 2 Ohm produce 1 MHz oscillation without any load but 100 nF buffer, 1uF/0uF is fine)
provide more help with switching to external pwr, switch? backside would be good
switch + jumper?
add 74HC4066 and OPA189 to consumer-list, build a low-pwr Overview
optimize shepherd-EN (unlink rec/emu?`, better buffering?)
hw may be glitching BB - caps are getting big, voltage drop also ⇾ critical!
maybe add lowpass to en-pin of regulators or limit power
TEST: run unittests 2-5x, often bb hangs itself somewhere between 47…80%
scope shows:
AC: 600mV dip on shepherd-enable, 3x 200mV stages, lowest point 600 us after start, then fast recovery to -300mV, slow after (quickshot 1-2)
DC: 800 - 1100 mV Dip, but the first one (after a break of some minutes) is more severe
another hang: converters did not start as planned, only 6V is up, L3V3 is at 2V, the others well below that
mitigations: 1mF/16V Cap on 5VBB reduces voltage drops to ~ 200mV (max 300), in a 5 ms Windows (quickshot 4)
Add Cap
run through hw_performance_v2.1r0 for final cross-check
allow to turn off ADCs (not that important for EMU-only)
maybe add 2nd Switch for PRU-Ports, or lower resistors to <= 100 Ohm (speed-improvement)
around switch-IC, avoid solder bridges per design even more ⇾ space out vias, reduce solder mask expansion
Additional Parts
Note: numbers for one Unit, refs are Mouser-Numbers
2 Ohm precision shunt: 603-RT1206CRD072RL
2 Ohm shunt: 667-ERJ-2GEJ2R0X
more 1k 0402 (2xRec, 2xEmu, 10x LvlChg)
210R InAmp-Gain-Setter: 603-RT0402BRD07210RL
1mF, 6V3, 647-UVR0J102MPD1TD (D8mm,L11mm,3.5mm Spacing)
analog switch, 2x 771-HC4066BQ115
4x 963-NR3015T1R5N
LSF dev-kit
Open Issues
feducials could be removed - panel has plenty
more pinnumbers on big headers
test changed recorder
is harvest-LED working? YES, but only with higher currents
var1: bind to vsim to allow more refined feedback
var2: 3x 2.0V Leds from adc-input to -6V (with resistor, should always light up)
var3: v-to-i converter
switch to TI Version of LSF0801 ⇾ out of stock till feb.22
optimize LSF for 1 MBaud (see
WDog is still not triggering nSTART (pulse too short)
HC2G04 still has wrong comment
wrongly linked in schematic
C53 L3V is part of recorder, but should be open domain
C48 is recorder, but should be emu
C24 is open domain, but should be recoder
has DAC also pin to disable?