Tests-Protocol for Capes
Visual Inspection
Look for tombstones, shorts, missing components and parts with wrong rotation (diodes, ICs)
It’s beneficial to take macro-shots of the untouched PCBs for later reference as it is hard to determine error causes once the PCBs are fully assembled, encased and deployed.
Isolated Cape-PCB
apply external power to PCB directly (current limited 5V via lab supply)
initial consumption < 1 mA
Pull Up EN_Shepherd (P8-13) with 3V3
consumption on 5V rises to ~71 mA
Power-LED below recorder should light up
check generated voltages on board (marked at capacitors by silkscreen)
Name |
Expected Voltage |
Comment |
---|---|---|
L5V |
5.000 V |
Should be Spot On |
A5V |
5.000 V |
Should be Spot On (is L5V with pi-filter) |
L3V3 |
3.300 V |
Should be Spot On |
6V |
5.38 V |
[5.29; 5.47] V with 1% Res |
10V |
9.73 V |
[9.56; 9.90] V with 1% Res |
-6V |
-6 V |
[5.94; 6.06] V with 1% Res |
Cape-PCB on BeagleBone, powered
BB starts 390 mA on VDD-5V line, booted: 170-240 mA
sudden 66 mA increase on shepherd EN is no problem for BB
WD-Pins could be a problem - my current test-BB is sensitive for power-button and shuts down ⇾ use jumper?
P8-43 or 44 is sensitive for input - BB does not boot when shepherd connects with these pins
both are for boot-config, BUT
LA shows that both pins are high even before 3v3 gets to the pins, lasts 7.4s for a fresh BB
P8-45/46 stay low during boot an later on
p8-41/42 are HIGH for the same 7.5 s on boot
EN-Shepherd (P8-13) stays low during boot - perfect!
a fresh BB is sensitive for boot-pin during operation, it will shut down! start by triggering boot-pin again. Reset does nothing during power-off
boot is high even before power-up. then low 2.3s until 3v3 come to pin ⇾ shutdown command (short low), but then stays high during power-off
react to shutdown seem to be controlled by software (important for watchdog)
reset is low, changes to high when 3v3 come to pin, when 3v3 go on shutdown, the pin also changes to low
GPIO to target
GPIO 0 - 4 are working bidirectional
BATOK is always high (currently), pru-debug-answer shows it
uarx (target - right) shows 1.5V, TODO
urtx (target - legt) shows 3V, TODO
swdIO is inversed? low for 80ms instead of 10 ms high, or something is pulling
swdCLK is fine
Noise from Outside (v2.0r1)
BB 5V Lines (both) show cutting 1V transients every 23.6 ms, around 400 us long (quickshot 73/74/80) ⇾ due to diode between both 5V-Lines
entry-filtering is not doing much for these rails
6V has +120/-80 mV Spikes (qs77, 78)
5V and 6V are only used as intermediate voltage steps
10V +46/-30 mV
-6V +42/-30 mV
A5V +36/-28
36 us, +-10mV Spikes, 500 ns long (qs82, 84)
⇾ add a big external Cap on 5V
diode-connection between the two 5V-Rails could be the problem - there are no voltage-spikes over the diode, so the current seemed to be constant
without diode: Big Spikes are gone, 5V has now max -200 mV and other (qs85)
-6V & 10v & A5V are cleaner, delta 30mV (qs86-88)
a Cap, 1F 5V5, before the ferrites, does not improve the situation
Performance (v2.0r1):
0 to 5V Target A, with 1 kOhm Load, 75 us for 80%, ~100 us for 100% (QS92 & 94)
5V to 0V T-A, …., 75 us for 80%, 400 us for 100%?, qs93
Recorder is following, with 5V in, 1k pre-resistor, the op-amp switches from 0..5V with 20us period, shape of thin half sine
big voltage jumps take 1 to 1.7 ms, 0 to 5V, that’s also the period duration for no active load
Performance (v2.1r0)
Current PCB-Mods (v2.0r1): - P8-43/44 disconnected, messes with boot - P9-9/10 possible problem, unconnected for now - 2x 10k-PU from EMU/REC EN routed to 3V3 (easy) instead of 5V, ADCs still work - 2x 1k-PU for boot, reset pins, only on shep-pcb als external jumper - switched inputs of R13, Shunt of Recorder, 2 lines cut and rerouted - diode over reverse-pol-mosfet - 1k for LEDs ⇾ OK - 1 Level Translator fixed (single supply & lower threshold voltage ~ 900 mV) - (NO!) 5V_SYS switched over to 5V_VDD - 8 Ohm right before Shunt ⇾ stabilize
Current PCB-Mods (v2.1r0):
switched target-power-lines between OP-Amp and switch
1 mF / 16 V Cap on 5V PCB-Input ⇾ helps stability during shepherd-enable
change R15 (V_harv_sim), R8 (emu-rail_b) from 10k to 1k to bring lowpass to 160 kHz (6us) instead of 16 kHz (62us)
incomplete list, see “44_hw_performance_v2.1r0”