Cape v2.4 - PCB Changes

Implemented Changes

  • 74LVC2T45GS has too small pads ⇾ prone to errors (very hard to see, but shorts under IC in all cases)

  • drc-rule: Force proper Fanout with Neck-Down (<=100%) ⇾ EC seems to extend solder mask expansion on its own

  • drc-rule: increase solder mask sliver (Gap) >= 0.2 mm

  • drc_rule: solder mask expansion default 0.04mm (was .06) ⇾ with manual override for fine-pitch footprints (min. 0.005 for 0.35mm-Pitch)

  • drc-rule: silk from pad distance >= 0.08 mm

  • paste mask pad fill ~ 66 - 80 % (30% reduction), thermal pads ~ 50-60% (45% reduction) for 100um Stencil

  • paste mask minimum dimensions 0.26 * 0.30 mm (T3 Paste with 25-45 um Balls), only exception: 0.35mm-Pitch part and small BGA with round d=0.25 Cutout

  • move ~1/3 of paste to not under IC

  • silk more symmetrical, cleaner and helpful

  • rework footprint-lib accordingly

  • feducials can go, are on outer frame

  • reverse order of diode & shunt in harvester ?? No, seems fine currently

  • rotate harvest port in schematic to reflect board layout

  • add usb-c to pcb and ditch BB-pwered mode?

  • emulator FB-Resistor-Switch can be removed

  • ADC RVS-Pin not needed, remove TPs

  • revisit datasheets for lower noise suggestions

  • A5V needs 1mF, +10V gets 1x more 47 uF, -6V gets 100 uF

  • remove coils for 6V, 5V, 10V, -6V

  • bring sense- / FB-line directly to target-por ⇾ NetTie

  • testpoints don’t need gnd - its all around

  • hrv-sense directly at pin, netsplit, also FB-lead

  • EMU, replace opa388 with opa189 for main-line

  • raise 6V to 6.2V (from 5.4), 578k + 100 k (or 680k)

  • Pin1 on Headers not clear ⇾ direction 1 2 … put mark directly under pin1, in direction of pin1&2

  • 74LVC2T45GS

    • dir is referenced to VCCA ⇾ switch side with GND

    • QFN-Pinout is wrong! https://4donline.ihs.com/images/VipMasterIC/IC/NEXP/NEXP-S-A0002881467/NEXP-S-A0002881253-1.pdf?hkey=6D3A4C79FDBF58556ACFDE234799DDF0

  • harvester

    • ADC-IN: R22 low but not 0, C140 to 1 - *10nF, R16 & TP6 same

    • DAC-OUT: R27 to 33R - 100R, C36 to 1 - *10nF

    • Drain: R20 lower ⇾ faster response to nonlinearity (diode-voltage from PU to PD)

    • VSense: remove R18 1k

    • ShuntBuffer C35 can be 10 - 100nF

    • update to latest profilings

  • 5V to BB before the Inductor? Yes

  • 1uF should be X7R (not X5R), check others

  • 1mF to 6V and A5V

  • Sense-Resistors with lower PPM/K - Value, higher precision

  • emu

    • OpAmp FB 2k/1nF is 20% better than 1k/1nF (current, smu), 5k also improves on that (+10%), but might be too slow

    • ADCIn-LPF 33R, 10nF is a good compromise, 4% better tan 100R

    • Shunt-Buffer 10 nF is 10-20 % worse than 100nF unlimited but similar in limited space, 570nF is 3-5x worse than both

  • External LED-Button-Connector S4B-ZR-SM4A-TF ⇾ Top Entry type: B4B-ZR-SM4-TF

  • order new parts:

    • usb-connector, 2x 5.1k R,

    • more Opa189

    • HRV 3x 10nF, 2x 33R, 1x 1nF

    • 100R (1HRV

    • EMU 1x 33R, 3x 10nF,

    • 1x 680k (6V)

    • 39x 1uF X7R

    • 1x 100R 1% 100PPM

    • 50x 100nF 25 V

  • [ADC can use 2x more 10uF on ref-pins] ⇾ NO, skip this one, had min to no effect

  • add >16 V Cap to BOM, or 2x ~10V ⇾ wurth, see orderlist ⇾ lifetime,

  • new 100nF to +10 - 6V directly, 2x

  • more pads for Caps on backside

  • big 0402 caps near device ⇾ don’t bother with 100nF or smaller ⇾ NO, skip this one, ESR / impedance is better on smaller values (same package)

  • change 0402 footprint, bring pads closer together

  • ref-input for InAmp AD8421 (voltage divider + op1177)

  • emu, use free opa388 for reference voltage offset, 5mV (60uV input offset * 50 + 400uV output offset) ⇾ 33R || 10k + Cap

  • 10nF <should be NP0, but this seems expensive

  • level-translators need to reach 1MHz, 1kOhm is limiting to ~200kHz, 2x 400 Ohm is more fitting

  • correct op-fb

  • 10uF should be X7R, but X5R has now 16V, X7R will be <6V? (ADC-Bypass) ⇾ Skip

Changes in Layout

  • LATER:

  • try V-FB without C ⇾ same for Emu-OpAmp, tune emulator similar to harvester

    • https://e2e.ti.com/blogs_/b/analogwire/posts/do-it-yourself-three-ways-to-stabilize-op-amp-capacitive-loads

    • just using R_ISO would have the lowest noise ⇾ but unstable

    • R8 5k is slower for current changes

    • ⇾ best would be to increase shunt-resistor back to 10 Ohm BUT maybe this is no problem after the OpAmp-Swap

  • remove 10R by just using 33R?

  • switch back to 10 R shunt with 1k1 Ohm RG ?

  • add option to supply Opa2388 with noisy 6 V

New Components

  • SPDT Switches Pi5… & the other Test ICs (DG612, DG2501)

  • 470 Ohm 1%

  • 2x 0R & 100nF

  • new NP0 10nF

  • Kai, https://www.mouser.de/ProductDetail/Nordic-Semiconductor/nRF-PPK2?qs=sGAEpiMZZMv0NwlthflBi%2FwrKI1rLznmCjMIzu8H1xs%3D

  • replace corrected BOM-parts