Skip to the content
2 layers68.6 × 53.3 mmMie1001 daq board, the printed circuit board in 3D

Mie1001 daq board

Chaospossum/mie1001-daq

MIE1001 Team 5: DAQ firmware and Arduino shield designs (preamp, ADS1115, MCP4725)

Size
68.6 × 53.3 mm
Copper layers
2
Thickness
1.6 mm
Parts
40
Nets
24
Pads
139
Tracks, vias
800, 56
Pours
8
Routed
Yes
KiCad
KiCad 10.0 (file 20260206)
Created
24 Sep 2026
Updated
20 Oct 2026

Overview

Mie1001 daq board is an open-source printed circuit board by Chaospossum, from the Chaospossum/mie1001-daq repository on GitHub. It is a 2-layer board of 68.6 × 53.3 mm with 40 parts on 24 nets, routed with 800 tracks and 56 vias.

It is built around the ADS1115IDGSR, MCP6002-I/SN and RC0603FR-071KL, with the RC0603FR-0710KL, CL21A106KAYNNNE and HVCB1206FKC100M among its other parts. Its parts include 17 resistors, 10 capacitors, 4 ICs, 4 connectors and 1 diode. It belongs with the audio and music and microcontroller boards boards of this library.

Main components

8

README

hardware/mie1001_daq_simple/README.md

MIE1001 — simple DAQ shield (rev E)

The beginner version of the DAQ board: an Arduino UNO R3 shield that the two breakout boards from class plug into (modular: both unplug). Fat tracks, the breakouts, connectors and through-hole parts are hand-soldered. Rev D adds a Faraday-cup preamp with 16 small SMD parts. Everything is ordered from Digi-Key (university procurement, see Ordering from Digi-Key) and hand-soldered (see Hand-soldering order); JLCPCB assembly is an optional alternative.

board, top board, bottom

text
SIGNAL IN (J1) -> R2 4k7 -> [CR1/CR2 clamp, C1 filter] -> ADS1115 breakout A0 -> I2C -> UNO -> USB -> laptop
UNO -> I2C -> SparkFun MCP4725 breakout OUT -> QPS OUT (J2), 0-5 V  (JP1, bridged: also -> ADS1115 A2 read-back)
TRIG IN (J3) -> R1 1k -> UNO D8 (R3 100k to GND)    ADS1115 ALRT -> UNO D2
Faraday cup -> J5 SMA -> R4 10k -> CR3 clamp to VREF -> R5 1k -> U3 LMP7721 preamp (R6 100M || C2 10p)
            -> R11/C6 1k/1u -> ADS1115 A1 (PRE_OUT)      VREF 2.5 V (R8/R9 100k/100k + C4 -> U4A MCP6002 -> R10 -> C5) -> A3
J4 = test header: 1 PRE_OUT, 2 A2 (QPS read-back while JP1 is bridged), 3 VREF, 4 GND
I2C: UNO SDA/SCL header pins AND A4/A5 (rev E)
CheckResult
ERC0 errors, 0 warnings
Guard check (check_guard.py, rev E)PASS: 0 items of any other net within 1.0 mm of the preamp input (COAX_IN, DET_IN, TIA_IN, TIA_REF), every input item has guard copper within 0.6 mm, and the bare-board flood fill from the input reaches only VREF and TIA_OUT on F.Cu and only VREF on B.Cu. The same script FAILS on the rev D board (7 leakage paths), so it catches the rev D defect
Independent critical reviewTwo reviews against the Adafruit, SparkFun, Arduino, Phoenix and TI source files and datasheets. Rev A: pinouts and UNO geometry correct, four fixes → rev B. Rev C: copper clean, no respin needed; one wrong BOM part number and several wrong claims in this file, all corrected below
Host boardArduino UNO R3 SMD (ATmega328P in a flat TQFP-32 package). Its headers are identical to the standard R3 (same positions, same pins, SDA/SCL next to AREF), so the shield fits unchanged
Still unverifiedThe height of the parts on the UNO under the shield (crystal, USB chip, capacitors). The SMD chip itself is only about 1.2 mm tall, so this is much less of a risk than with the socketed DIP chip, but look at your board before soldering
DRC incl. schematic parity0 errors, 0 unconnected, 0 parity issues (4 lib_footprint_mismatch warnings: ARD1 and J1–J3 are deliberately modified copies) — rev E
OrientationSame real-UNO geometry as the full board (rev E): flipped library footprint, keep-outs over the USB, DC jack and both ICSP headers
Rules0.5 mm tracks, ≥ 0.3 mm gaps, 0.8/0.4 mm vias: far above the Digi-Key PCB Builder (5 mil / 8 mil) and JLCPCB minimums

What changed in rev E (2026-09-25)

An independent review of rev D found a leakage path and a few smaller problems. No connector, socket or mounting hole moved (J1–J5, the two breakout sockets, H1–H3 are where they were in rev D). The router re-laid most of the signal tracks because the obstacles changed; that does not matter electrically, DRC and the guard check are clean.

  • Guard closed under U3. In rev D the guard pour stopped halfway under the LMP7721, and TIA_OUT ran from the feedback parts under the chip body to pin 4. The copper-free gap alongside it was a bare-board path from the input to GND (pin 3) and +5V (pin 6). Now a VREF bar runs under the chip from pin 2 to pin 7 (both N/C pins, on the guard), so the input pins 1 and 8 sit in a closed guard pocket, and TIA_OUT leaves the guard on the bottom layer only: via next to R6/C2, under the chip, via below the bar, then to pin 4 and R11. On the top layer the input is now completely surrounded by guard copper; the only other net it faces is TIA_OUT across C2 and R6 themselves, which only adds in parallel with the feedback resistor (TI LMP7721 datasheet, 10.1). DET_IN was re-laid as two straight runs so the guard hugs it everywhere.
  • No solder mask on the guard (both layers). The mask openings are the guard pour shrunk by 0.10 mm and kept ≥ 0.50 mm from any pad and ≥ 0.40 mm from any track of another net, and off all silkscreen. So every other track, every gap next to the guard and the edge of the GND pour stay under mask: exposed guard copper is never closer than 0.4 mm (mask web) to exposed copper of another net, which keeps solder bridges unlikely. Care points: the SMA J5 ground pins sit next to the bare guard ring around the centre pin (≥ 0.5 mm mask web); don't flood them with solder, and check there is no bridge from a J5 ground pin to the ring before powering up.
  • check_guard.py is strict now. It fails (exit 1) on any unguarded input item (it used to only print them) and adds a bare-board flood fill from the input nets on each copper layer: reaching copper of any net other than VREF / TIA_OUT, or leaving the guard area, fails the build. check_guard.py .history/revD-final/mie1001_daq_simple.kicad_pcb shows the rev D failure.
  • Quieter VREF. R8/R9 are now 100k/100k (were 10k/10k). With C4 10 µF the divider's corner is 0.32 Hz instead of 3.2 Hz, so 50 Hz ripple on the USB 5 V reaches VREF attenuated ~160× instead of ~16×. The MCP6002's ~1 pA input bias current makes no measurable error through 50k.
  • A4/A5 also carry SDA/SCL. On a genuine UNO R3 they are the same signals as the SDA/SCL pins next to AREF; some clones don't have those two pins, and then the shield works through A4/A5. Consequence: never use A4/A5 as analog inputs with this shield fitted.
  • JP1, QPS read-back. A solder jumper, bridged by a copper strip from the factory, connects QPS OUT (MCP4725 OUT, J2) to ADS1115 A2 (J4 pin 2). The firmware can read the command voltage it actually set: ads.readADC_SingleEnded(2), with the ±6.144 V range (ads.setGain(GAIN_TWOTHIRDS)), because 0–5 V does not fit the ±4.096 V range. To use A2 for something else, cut the thin copper bridge between JP1's pads with a knife (check with a meter); a blob of solder closes it again. JP1 is next to J4, marked "JP1 / A2=QPS".
  • C5/C6 part (optional JLC route only). LCSC C15849 had no LCSC stock; now C5673 (Samsung CL10A105KA8NNNC, 1 µF 25 V X5R 0603, extended part: one extra JLC loading fee). There is no other basic 1 µF 0603 at JLC; JLC's own library still listed C15849 in stock on 2026-09-25, so it is a valid fallback.
  • Docs: J4 firmware note rewritten (pins 1 and 3 are op-amp outputs), wrong 8 SPS/mains claim corrected, comparison with the full board corrected, U3 "no solder mask" note is now true.

What changed in rev D (2026-09-24): the preamp

The Faraday cup gives a current of pA to nA. The ADS1115 measures voltage, so rev D adds a transimpedance amplifier (U3, LMP7721) that turns 1 nA into 0.1 V. Everything that was on rev C stays where it was; the preamp sits in the band under the digital header, where the text used to be (the text moved to the left edge and the back).

PartJob
J5 SMA jack (vertical, PTFE)Detector input. Use a BNC-female to SMA-male adapter for a BNC cable. A BNC did not fit anywhere on this board
U3 LMP7721 + R6 100 MΩThe preamp: Vout = VREF − I × 100 MΩ. Full scale ~25 nA; 1 ADC count = 1.25 pA at GAIN_ONE
C2 10 pFStops the preamp oscillating with the cable capacitance. Sets a 159 Hz bandwidth
R4 10k + CR3 BAV199 + R5 1kProtect U3 from a discharge on the cable. CR3 is tied to VREF, so it leaks ~0 into the input
VREF 2.5 V: R8/R9 100k/100k (rev E), C4 10 µF, U4A MCP6002, R10 100 Ω, C5 1 µFThe input sits at 2.5 V so the output can swing both ways on a single 5 V supply
R7 1kProtects U3's + input
R11 1k + C6 1 µF159 Hz anti-alias filter before ADS1115 A1
Guard ring (copper at VREF, both layers, no solder mask)Surrounds the whole input so no other copper can leak pA into it; closed under U3 between pins 2 and 7 (rev E)

Firmware: read the preamp as ads.readADC_Differential_1_3() (A1 − A3 = PRE_OUT − VREF), then I = −V / 100e6. The ADS1115 only offers the pairs 0-1, 0-3, 1-3 and 2-3, which is why VREF is on A3. SIGNAL IN (J1) → A0 still works as before for a voltage input.

J4 changed: pin 1 is now PRE_OUT and pin 3 is VREF (monitor only: connect a meter or scope, never a source). Pin 2 (A2) carries the QPS read-back while JP1 is bridged (rev E); cut JP1 to use it as a free input.

Still 0-5 V on QPS OUT. The Extrel 150-QC wants a 0-10 V mass command; that needs a gain-2 amplifier on a supply above 10 V (e.g. a 12 V adapter on the UNO's DC jack). Not on this revision.

Handling: the input works at picoamps. After soldering, wash off all flux around U3, R6, C2, R4 and J5 (isopropyl alcohol, then dry), handle the board by its edges, and measure the zero offset with the cable connected and the beam off before every scan.

Names on the drawing and the board (2026-09-24, copper unchanged)

  • The UNO is ARD1 and the clamp diodes are CR1, CR2 (were A1, D1, D2), so the names no longer clash with the UNO's pins A1 and D2.
  • The UNO symbol shows only the pin names printed on the UNO (D2, D8, SDA ...), not KiCad pad numbers.
  • SDA/A4 and SCL/A5 are the same signals on the UNO R3. Since rev E the shield connects A4/A5 as well (for clones without the SDA/SCL pins), so never use A4 and A5 as analog inputs.

What changed in rev C

Three parts on the signal input, in the free band under the ADS1115 socket:

  • C1 (100 nF) to ground. With R2 it is a single-pole low-pass at 339 Hz, and it supplies the charge the ADC's switched-capacitor input takes at each sample. It takes the edge off fast noise and limits the bandwidth, so leave it out if you need to measure something faster than ~300 Hz. It does not remove mains hum: at 50 Hz it attenuates by only 1.1 %. See the firmware notes for what actually works. C1 must be X7R or C0G, not Y5V (a Y5V part loses most of its value when warm, which moves the corner frequency anywhere up to ~2 kHz).
  • CR1 and CR2 (1N4148) clamps. They divert most of an over-voltage to the rails. Be clear about what does the work: R2 is what keeps the chip legal (1.3–2.4 mA at ±12 V against its 10 mA limit), because the 1N4148 only starts conducting around 0.5–0.7 V, by which point the input is already past the ADC's absolute maximum of VDD + 0.3 V and the chip's own internal protection diode is conducting alongside CR1. The diodes are there for the energy of a static zap or a badly wrong connection, not to hold the pin inside spec. TI also warns that overdriving one input can upset conversions on the others: while A0 is overdriven, readings on A1–A3 are not valid either.
  • What the diodes cost: up to 25 nA of leakage each, which through R2 is up to 117 µV of offset (15 counts at GAIN_SIXTEEN, and several times worse when hot). Fine at low gain; calibrate the offset if you use the high gain ranges.
  • One side effect: with the Uno unpowered, a signal above 5 V feeds up to 2.4 mA (at +12 V) back into its 5 V rail through CR1. Don't leave a live preamp connected to an unpowered board.
  • No amplifier. The ADS1115 has a programmable gain amplifier built in, so a small signal is handled in software with ads.setGain(...) instead of another chip. Note that at the highest gains the chip's differential input resistance falls to about 710 k, where R2 costs ~0.66 % of gain (under 0.1 % at the ±6.144 V range, 0.2–0.4 % at ±2.048 and ±1.024 V): calibrate if you use GAIN_SIXTEEN. Note TI suggests keeping a filter resistor under 1 k; 4.7 k is deliberate here, chosen for protection, and C1 at the pin supplies the sampling charge that the series resistance otherwise would.

Watch the stripe when soldering the diodes. The silkscreen prints K at the cathode end of each one, and that is the end with the black band on the part. CR1's K points right, towards +5 V. CR2's K points left, towards the signal. Backwards, they short the signal to a rail.

What changed in rev B

  • R2 (4k7) in series with SIGNAL IN. The ADS1115 input may not go above VDD + 0.3 V (TI datasheet, absolute maximum ratings). If the preamp swings to its ±12 V rails, R2 keeps the current into the chip's input clamp to 1.3–2.4 mA (its limit is 10 mA). Gain error: under 0.1 % on the ±6.144 V range, 0.66 % on the ±0.256 V range.
  • R3 (100k) pull-down on D8. A TRIG IN with nothing connected now reads LOW instead of picking up noise and giving false triggers.
  • Pin-1 triangles and "which way round" text on both sockets. Both breakouts fit in their sockets rotated 180°. For the MCP4725 that puts 5 V on its GND pin. Now the MCP4725 outline is the real 15.2 × 15.2 mm board.
  • Firmware and BOM fixes. dac.begin(0x60) (the library default is 0x62), and the BOM no longer orders the new two-row ADS1115 board that doesn't fit.

Before you order: three things to check by hand

  1. Which ADS1115 board do you have? Use the one from the class kit; do not order a new one. This shield fits the original Adafruit board with one row of 10 pins (VDD GND SCL SDA ADDR ALRT A0 A1 A2 A3). Adafruit now sells a newer STEMMA QT version (two rows of 6, "VIN" instead of "VDD") under the same product number 1085, and it does not fit. Your team's self-test sketch says "VDD", so you probably have the original, but count the pins.
  2. Print fab/print_1to1_top.pdf at 100 % and lay the UNO and both breakouts on it. The header holes must line up with the UNO's pins and with each breakout's pins.
  3. Height over the UNO's chip area. CR1, CR2, C1, the ADS1115 socket and J4 all sit above it. On the UNO R3 SMD the ATmega328P is a flat TQFP (about 1.2 mm), so the chip itself is no problem; check that nothing else on your board (crystal, USB chip, capacitors) is taller than the gap, and trim the leads flush.

Parts (full list in BOM.csv)

RefPartWhere
U1Adafruit ADS1115 breakout (original, 1×10), from the class kitplugs into a 1×10 female header (Sullins PPTC101LFBN-RC, Digi-Key S7008-ND). Don't order 1528-1085-ND: that is the new two-row board
U2SparkFun MCP4725 breakout BOB-12918plugs into a 1×6 female header (PPTC061LFBN-RC, S7004-ND). Retired at SparkFun, still Active at Digi-Key 1568-12918-ND (656 in stock on 2026-09-25). If it is gone, use the class-kit board: Adafruit's MCP4725 breakout is not pin-compatible (VIN/GND/SCL/SDA/A0/VOUT, no second GND)
J1, J2, J3Phoenix 1984617 screw terminals, 3.5 mmDigi-Key 277-1721-ND
J41×4 female header for the spare inputsS7002-ND
R11k ¼ W resistor on the trigger1.0KQBK-ND
R24k7 ¼ W resistor in series with SIGNAL IN13-CFR-25JB-52-4K7-ND
R3100k ¼ W pull-down on D8, mounted standing up13-CFR-25JB-52-100K-ND
C1100 nF 50 V X7R disc, 2.5 mm lead pitch (not Y5V: BC1160-ND is the wrong one)BC1084CT-ND
CR1, CR21N4148 signal diodes (DO-35)1N4148FS-ND
—UNO stacking headers (1×10, 2× 1×8, 1×6, 10.5 mm legs)Adafruit 85, Digi-Key 1528-1074-ND (legs just clear the USB jack)
U3, U4, CR3, R5–R11, C2–C7, J5, R4the preampall Digi-Key, see BOM.csv / fab/digikey_bom.csv

No other parts. Both breakouts already carry the I²C pull-ups: the ADS1115 board has 10k on SDA and on SCL (hard-wired), the MCP4725 board has 4.7k on each through jumper SJ1 (closed by a trace from the factory). In parallel that is 3.2k per line, about 2.9k once the Arduino's own internal pull-ups are counted, which draws 1.6 mA at the low level against the 3 mA an I²C device must sink. The minimum sensible value here is about 1.5k, so this is fine and the shield adds no pull-ups of its own. If you ever chain more I²C boards onto this one, each brings its own pull-ups and the total drops: cut SJ1 on the MCP4725 board to take its 4.7k pair off the bus. The ADS1115 breakout also pulls ALRT (UNO pin D2) up with 10k.

Ordering from Digi-Key

Everything is bought from Digi-Key (university procurement). Checked on digikey.com on 2026-09-25: every part number below exists, matches the MPN and package in BOM.csv, is Active and is in stock.

Parts: fab/digikey_bom.csv

./build.sh writes fab/digikey_bom.csv in Digi-Key's BOM Manager / "Upload a list" format (Digi-Key Part Number, Manufacturer Part Number, Manufacturer, Quantity, Customer Reference), 24 lines, quantities for ONE board. Not in it (do not order): the UNO (ARD1) and the ADS1115 breakout (U1), both from the class kit. Upload it, then set the quantities you want (below).

RefDigi-Key P/NMPNStock 2026-09-25Unit price (USD, qty for 1 board)
C1BC1084CT-NDVishay K104K15X7RF5TL2209,2080.29
C21276-1027-1-NDSamsung CL10C100JB8NNNC289,3980.10
C3, C7732-8013-1-NDWürth 885012206095508,7770.10
C4587-4334-1-NDTaiyo Yuden TMK212BBJ106KGHT13,2420.29
C5, C61276-1102-1-NDSamsung CL10A105KA8NNNC7340.10
CR1, CR21N4148FS-NDonsemi 1N4148663,4430.10
CR3BAV199LT1GOSCT-NDonsemi BAV199LT1G290,3530.16
J1–J3277-1721-NDPhoenix 198461744,5690.59
J4S7002-NDSullins PPTC041LFBN-RC46,1500.42
J5ACX1230-NDAmphenol RF 13213423,3809.98
R11.0KQBK-NDYAGEO CFR-25JB-52-1K200,9420.10
R213-CFR-25JB-52-4K7-NDYAGEO CFR-25JB-52-4K7109,0750.10
R313-CFR-25JB-52-100K-NDYAGEO CFR-25JB-52-100K92,5970.11
R413-CFR-25JB-52-10K-NDYAGEO CFR-25JB-52-10K148,6160.10
R5, R7, R11311-1.00KHRCT-NDYAGEO RC0603FR-071KL3,768,0970.10
R6HVCB1206FKC100MCT-NDStackpole HVCB1206FKC100M19,6744.01
R8, R9311-100KHRCT-NDYAGEO RC0603FR-07100KL916,0170.10
R10311-100HRCT-NDYAGEO RC0603FR-07100RL260,2610.10
U21568-12918-NDSparkFun 12918 (BOB-12918)6566.95
U3LMP7721MA/NOPB-NDTI LMP7721MA/NOPB2,1917.34
U4MCP6002-I/SN-NDMicrochip MCP6002-I/SN11,6650.42
X1S7008-NDSullins PPTC101LFBN-RC24,6050.64
X2S7004-NDSullins PPTC061LFBN-RC30,0300.47
stacking headers1528-1074-NDAdafruit 858141.95

Changed on 2026-09-25 for Digi-Key: C2 (Murata GRM1885C1H100JA01D is Not For New Designs) → Samsung CL10C100JB8NNNC, same 10 pF C0G 50 V 0603. C3/C7 (Samsung CL10B104KB8NNNC, 0 in stock, 61-week lead time) → Würth 885012206095, same 100 nF X7R 50 V 0603. C4 (Samsung CL21A106KAYNNNE, 0 in stock) → Taiyo Yuden TMK212BBJ106KGHT, same 10 µF X5R 25 V 0805 (1.35 mm max height). R4: the old number 10KQBK-ND no longer shows up at Digi-Key; same Yageo part under 13-CFR-25JB-52-10K-ND. The LCSC column in BOM.csv is only for the optional JLC route and may name an equivalent part (R6: Uniroyal), not the Digi-Key MPN.

Watch before ordering: C5/C6 had only 734 in stock (2 needed); if it has run out, any 1 µF ≥ 16 V X5R/X7R 0603 will do. The LMP7721 has no pin-compatible substitute: if LMP7721MA/NOPB-ND is gone, LMP7721MAX/NOPB is the same part on a reel (drop-in; check whether Digi-Key sells it as cut tape); anything else (e.g. ADA4530-1, LMC6001) has a different pinout/package and means a PCB respin.

Cost (Digi-Key qty-1 prices, USD, parts only, excl. PCB, shipping and VAT): 1 board $36.40, 2 boards $72.80 (Teams 4+5 may want a spare). The two biggest items are J5 SMA ($9.98) and U3 ($7.34).

Recommended spares (cheap, and small parts get lost or damaged in hand-soldering):

  • 0603/0805 passives and CR3: order 10 of each line instead of 1–3 (price per piece drops to $0.03, so 10 cost about the same as 3–4), i.e. +20–50 % is the minimum.
  • 1 extra LMP7721 ($7.34): it is the critical, ESD-sensitive part; 1 extra R6 ($4.01): it is easy to contaminate. 1 extra MCP6002, +2 of each THT resistor and 1N4148, +1 C1.
  • For 2 boards plus these spares: about $87.

The bare PCB: DigiKey PCB Builder

Digi-Key does not make or assemble boards itself; its PCB Builder (digikey.com/en/pcb-builder, needs a My DigiKey login to check out) uploads your files and shows live offers from partner board houses, which ship directly to you: DKRed (Digi-Key's own budget programme, built in Hollister CA), Summit Interconnect, Advanced Circuits and Precision PCBs.

  • Files: upload fab/mie1001_daq_simple_gerbers.zip (RS-274X Gerbers + Excellon drill, < 10 MB). Digi-Key requires at least 2 copper layers, a drill file, 2 solder-mask layers and an outline; the zip has all of them (F/B_Cu, F/B_Mask, F/B_Silkscreen, Edge_Cuts, PTH + NPTH drill). Digi-Key's FAQ lists extensions such as .gts/.gto/.drl/.gko and does not name KiCad's .gtl/.gbl/.gm1: check the viewer and assign top copper = .gtl, bottom copper = .gbl, outline = .gm1 by hand if it guesses wrong. F_Paste and the .gbrjob are not needed (no stencil: hand-soldered).

  • Capabilities vs this board (checked from the .kicad_pcb, not just the rules):

    this boardDKRedAdvanced Circuits / Summit / Precision
    layers, thickness2, 1.6 mm FR42 or 4, 0.062" (1.6 mm) FR4 Tg1702/4/6, 0.031/0.062/0.093"
    min track / space0.5 mm / ≥ 0.3 mm (DRC min), zone min width 0.2 mm5 mil (0.13 mm)5, 6 or 8 mil
    min drill0.4 mm (vias)8 mil (0.20 mm)8, 11 or 12 mil
    annular ring≥ 0.2 mm (vias 0.8/0.4), ≥ 0.27 mm pads≥ 3 mil–
    board size68.58 × 53.34 mm (2.70 × 2.10 in)0.5–10 inAC 0.75–10 in, Summit 2–16 × 2–22 in, Precision 1–18 in
    finish / maskguard copper deliberately without maskENIG only, red mask, white silkHASL (leaded or lead-free) or ENIG; green/black/red/blue/purple/clear mask

    Nothing fails: every rule has ≥ 2× margin. The unmasked guard ring is just openings in the F/B_Mask Gerbers, which every house reproduces; the mask web between the bare guard and any other copper is ≥ 0.4 mm. ENIG (DKRed's only option) is the better finish here anyway: flat pads for the SOIC/0603 parts and gold on the exposed guard copper. If you pick another house, choose ENIG or lead-free HASL.

  • Price: DKRed is $1.50 per square inch per board, minimum 4 boards: 5.67 in² → $8.50/board, $34 for 4 (you cannot order just 3; 4 is the cheapest option). Build time 10 business days (5 quick-turn). The other houses quote live after the upload (not published; typically more). DKRed ships free only within the US: shipping to the Netherlands, customs and VAT are extra: check the total at checkout before committing.

Optional: JLCPCB instead

Only if procurement allows it. Upload the same gerber zip (2 layers, 1.6 mm, HASL or ENIG). To have JLC assemble the SMD preamp parts, turn on PCB Assembly (top side) and upload fab/jlcpcb_bom.csv and fab/jlcpcb_cpl.csv (16 parts, 11 lines, from the LCSC column); check U3/U4/CR3 rotation in JLC's preview and that the LCSC parts are in stock. Then solder only the THT steps below.

Hand-soldering order

Nothing is assembled by the board house. Work on an ESD mat with a grounded wrist strap, fine tip (≤ 1 mm chisel), 0.5 mm solder, flux pen, fine tweezers and a magnifier. Solder all SMD parts first, while the board still lies flat, then the THT parts from lowest to tallest.

  1. U4 MCP6002 (SOIC-8): tack one corner pin, check pin 1 (dot) and alignment, solder the opposite corner, then the rest. Remove any bridge with wick.
  2. U3 LMP7721 (SOIC-8): ESD-sensitive (wrist strap; leave it in its bag until needed), pin 1 at the dot. Pins 1 and 8 sit in the bare guard pocket: use little solder, no bridge to the guard copper.
  3. CR3 BAV199 (SOT-23): small; tack the single pin first, then the two others.
  4. 0603 parts: R5, R7, R11 (1k), R8, R9 (100k), R10 (100R), C2 (10 pF, unmarked: keep it apart from the other caps), C3, C7 (100 nF), C5, C6 (1 µF). Ceramic caps are unmarked: open one tape at a time.
  5. 0805 C4 (10 µF) and 1206 R6 (100 MΩ; don't touch the resistor body with bare fingers).
  6. Clean the preamp area now: scrub U3, R6, C2, R4/CR3 and the guard ring with IPA (≥ 99 %) and a clean brush, rinse with fresh IPA, let it dry fully (heat gun on low / 30 min). Critical for the pA input: any flux residue across the guard is a leakage path.
  7. THT, lowest first: CR1, CR2 (lying flat, stripe towards the K), R1, R2 (flat), C1, R3 and R4 (standing, body on the circle).
  8. J5 SMA: the four ground legs are on the GND pour and soak heat: use a larger tip / higher temperature (~370 °C) and let the solder flow through; the centre pin sits in the bare VREF guard ring: no solder bridge from a ground leg to the ring (check with a meter).
  9. Female headers X1 (1×10), X2 (1×6), J4 (plug the breakout in while soldering so they stay straight), terminal blocks J1–J3, then the stacking headers (last: tallest).
  10. Trim every lead under the board flush. This is required, not optional. The ADS1115 socket and J4 sit right above the UNO's ATmega chip, and CR1, CR2 and C1 are in the same band. An untrimmed socket lead sticks out about 1.6 mm and a terminal pin about 2.9 mm. Check the clearance over the rest of that area with the real boards before you solder everything down.
  11. Clean the whole board (IPA) once more around J5 and the guard ring; handle the board by its edges.
  12. Plug the breakouts in component side up, matching the pin names on the breakout to the names on the shield: VDD on the left for the ADS1115, and OUT at the top (at the ▼) for the MCP4725. Rotated 180° the MCP4725 gets 5 V on its GND pin.

Hard to hand-solder: the 0603 parts (small, unmarked caps: C2 mixed up with another cap ruins the preamp's stability), CR3 (SOT-23), and J5 (ground legs on a big pour, next to the bare guard). U3 is easy to solder but easy to kill with ESD and easy to spoil with flux: this is why it gets a spare. None of the parts needs hot air or reflow.

Firmware notes

  • ADS1115 at 0x48, MCP4725 at 0x60. Libraries: Adafruit ADS1X15 and Adafruit MCP4725.
  • dac.begin(0x60): the Adafruit_MCP4725 library defaults to 0x62, so a plain dac.begin() finds nothing. ads.begin() (0x48) is correct as is.
  • Read the signal on A0. ads.setGain(GAIN_TWOTHIRDS) gives ±6.144 V full scale. For a small signal turn the gain up instead of adding an amplifier: GAIN_SIXTEEN gives ±0.256 V full scale and ~7.8 µV steps.
  • The input is filtered at 339 Hz (R2 with C1), so the board cannot follow anything faster. Leave C1 out if you need the bandwidth.
  • For 50 Hz mains hum, average in the sketch over whole 20 ms windows, not over "a slow data rate". At 8 SPS one conversion takes 125 ms = 6.25 mains cycles, and the ADS1115's filter nulls sit at multiples of its data rate (8, 16, 24 Hz ...), not at 50 Hz, so 8 SPS does not by itself cancel hum. Instead take readings at a fixed rate that fits a whole number of times into 20 ms and average an integer number of 20 ms windows: e.g. 250 SPS (RATE_ADS1115_250SPS, 4 ms) and average 5, 10, 50 ... consecutive readings (20, 40, 200 ms). Time the samples with the ALRT/RDY pin (D2), not with delay(). C1 barely touches 50 Hz. Measure the result with the input shorted.
  • J4 is a monitor header, not three spare inputs any more. J4-1 is PRE_OUT (the preamp output via R11) and J4-3 is VREF (the MCP6002 output via R10): never connect either to GND or to a source, that shorts an op-amp output. Connect only a meter or scope. J4-2 is ADS1115 A2, which carries the QPS read-back while JP1 is bridged; only if you cut JP1 is it a free input, and then it has no resistor, clamp or filter: keep it within 0 V … VDD. Leaving A2 open is fine (the firmware just doesn't read it); there is nothing to tie down.
  • The useful input range is 0 V to VDD (the USB 5 V, often only 4.8 V). This board expects a voltage from Team 3's preamp, not the raw detector current. R2 protects the chip from a preamp that overshoots, but readings above VDD are not valid.
  • The MCP4725 powers up at mid-scale (2.5 V on QPS OUT). Write 0 to its EEPROM once.
  • Trigger in: pinMode(8, INPUT), active HIGH. R3 holds D8 low when nothing is connected. Use 5 V logic if you can: the ATmega needs 3.0 V for a HIGH, so 3.3 V logic (3.27 V at D8 after the R1/R3 divider) has under 0.3 V to spare. R1 limits the current if the trigger goes a little above 5 V. Never feed it 12 V: that pushes 6.5 mA into the ATmega's input clamp, past its absolute maximum.
  • ALRT goes to UNO pin 2: pinMode(2, INPUT) is enough, the breakout has the pull-up. (The diodes are named CR1 and CR2 and the UNO is ARD1, so "D2" and "A1" on this board always mean the UNO's pins.)

Compared with the full board (../mie1001_daq_board, rev E)

Since rev D this board also has an electrometer preamp (U3 LMP7721, 100 MΩ, guarded input on J5), so it can measure a Faraday-cup current directly, with a single fixed range (~25 nA full scale). SIGNAL IN (J1) still digitises a voltage that someone else has made, with basic protection (R2, CR1/CR2) and a 339 Hz filter (R2, C1). The full board (../mie1001_daq_board) adds a jumper-selectable second feedback range (2.2 MΩ / 100 MΩ); this one stays easy to build, with the breakouts from class plugged in.

Files

generate_schematic.py and generate_pcb.py generate everything (the PCB script reuses the UNO geometry and the router of the full board); ./build.sh rebuilds and re-runs all checks. The possum and the prayer to the machine gods are in generate_pcb.py, possum().

Parts

40
QtyValueFootprintReferences
61kResistor_SMD:R_0603_1608MetricR3, R4, R9, R13, R16, R17
510kResistor_SMD:R_0603_1608MetricR7, R8, R10, R11, R14
4100nFCapacitor_SMD:C_0603_1608MetricC4, C5, C6, C8
31uFCapacitor_SMD:C_0603_1608MetricC2, C9, C10
2100RResistor_SMD:R_0603_1608MetricR12, R15
210uFCapacitor_SMD:C_0805_2012MetricC3, C7
24k7Resistor_SMD:R_0603_1608MetricR5, R6
2MountingHole_3.2mm_M3MountingHole_3.2mm_M3H1, H2
1100MResistor_SMD:R_1206_3216MetricR2
110pFCapacitor_SMD:C_0603_1608MetricC1
12M2Resistor_SMD:R_0603_1608MetricR1
1ADS1115IDGSRPackage_SO:MSOP-10_3x3mm_P0.5mmU2
1Arduino UNO R3Module:Arduino_UNO_R3A1
1BAV199LT1GPackage_TO_SOT_SMD:SOT-23D1
1Detector inConnector_Coaxial:BNC_Amphenol_031-6575_HorizontalJ1
1LMP7721MA/NOPBPackage_SO:SOIC-8_3.9x4.9mm_P1.27mmU4
1MCP4725A0T-E/CHPackage_TO_SOT_SMD:SOT-23-6U1
1MCP6002-I/SNPackage_SO:SOIC-8_3.9x4.9mm_P1.27mmU3
1QPS cmd outTerminalBlock_Phoenix:TerminalBlock_Phoenix_PT-1,5-2-3.5-H_1x02_P3.50mm_HorizontalJ2
1Range selectConnector_PinHeader_2.54mm:PinHeader_1x03_P2.54mm_VerticalJP1
1Trigger inTerminalBlock_Phoenix:TerminalBlock_Phoenix_PT-1,5-2-3.5-H_1x02_P3.50mm_HorizontalJ4
1VREF/AIN3Connector_PinHeader_2.54mm:PinHeader_1x03_P2.54mm_VerticalJ3

More from Chaospossum/mie1001-daq

1
  • 2 layers68.6 × 53.3 mmMie1001 daq simple, the printed circuit board in 3D

    Mie1001 daq simple

    Chaospossum/mie1001-daq

    MIE1001 Team 5: DAQ firmware and Arduino shield designs (preamp, ADS1115, MCP4725)

    35 parts

Similar boards

  • 6 layers160 × 100 mmCourant.before colored actual models 20260918T190457, the printed circuit board in 3D

    Courant.before colored actual models 20260918T190457

    KrasForge/Courant

    A 2D finite-difference physical-modeling synthesis engine in structural VHDL: a vibrating membrane/plate/gong solved on FPGA fabric, with a nonlinear chaos term, playable polyphonically over MIDI/CV.

    199 partsMIT
  • 4 layers112 × 90 mmNova pcb v6 power v2, the printed circuit board in 3D

    Nova pcb v6 power v2

    Ace2932/Fennec

    This project is a heavily modified fork of the open-source NovaSM3 quadruped, redesigned to serve as a research platform for embodied AI.

    61 parts1MIT
  • 4 layers119.4 × 119.4 mmUDuino, the printed circuit board in 3D

    UDuino

    duncmccl/ARMStrong

    Microcontroller Learner board replacement for the BASYS MX3

    128 partsGPL-3.0
  • 4 layers95 × 55 mmGalvanik, the printed circuit board in 3D

    Galvanik

    viridivn/galvanik

    OSHW galvanic hair electrolysis device

    103 parts2
  • 4 layers111.8 × 59.1 mmControleurPulve, the printed circuit board in 3D

    ControleurPulve

    Mobilab-AgroTIC/debitdouille

    Système pour controler le fonctionnement d'un pulvérisateur

    107 parts
  • 4 layers62 × 50 mmBattery monitor rev a, the printed circuit board in 3D

    Battery monitor rev a

    bogdan-kovalchuk/esp32-battery-charging-monitoring

    An ESP32-based 12 V battery voltage monitor. The firmware measures battery voltage, while a Flask backend delivers notifications to a Signal group through signal-cli.

    66 partsMIT
  • 2 layers30.5 × 68.6 mm16bit-5v, the printed circuit board in 3D

    16bit-5v

    bread-modular/bread-modular

    Modular Synths for Everyone

    62 parts67MIT
  • 4 layers64 × 115 mmAmp board, the printed circuit board in 3D

    Amp board

    splogdes/Audio-Engine-A5-Input-Stage

    198 parts

Questions

How big is the Mie1001 daq board PCB?

It measures 68.6 × 53.3 mm and has 2 copper layers, 1.6 mm thick.

What parts does the Mie1001 daq board use?

40 parts, built around the ADS1115IDGSR, MCP6002-I/SN, RC0603FR-071KL and RC0603FR-0710KL. The full list of its parts is on this page, with links to each part's symbol and footprint where PCBKit's library has them.

Where can I download the Mie1001 daq board KiCad files?

From the Chaospossum/mie1001-daq repository on GitHub, at the commit this page was made from: the board is hardware/mie1001_daq_board/mie1001_daq_board.kicad_pcb and its schematic hardware/mie1001_daq_board/mie1001_daq_board.kicad_sch. PCBKit opens the board in the browser, with nothing to install.

Can I use the Mie1001 daq board design in my own project?

Its repository states no licence, so its author keeps all rights: ask Chaospossum before reusing the design.