INA monitors/The two boards/03. The two boards
The two boards · 03 of 11

The two boards

One block with one channel and one wide board with three. They share pull-ups, address pads and wiring habits, fit different shunts — 100 milliohms and 50 — and their four-pin headers disagree about which of SDA and SCL comes first.

Same instrument, two resistors

Different resistor, different header
TK119 · 22.4 × 30.4 mm
Shunt
100 mΩ
Full scale
3.20 A
One count
0.1 mA
The header reads GND · 3V3 · SCL · SDASCL before SDA. The INA3221 next to it reads GND · VCC · SDA · SCL. Two boards in one box, and their last two wires are the other way round, so a set of jumpers moved straight from one to the other swaps SDA and SCL. Nothing is damaged by that — the bus simply goes quiet, and an I²C scan finds nothing at all.

Much is identical on the two boards: 10 kΩ pull-ups on both bus lines, solder pads to move the address, a red power LED off the supply rail through 5.1 kΩ, and a ground shared between the rail being measured and the microcontroller.

What differs is the chip, and the resistor that goes with it. The TK119 carries one INA219, which has a gain stage in front of its converter and so four current ranges you pick in software, on a 100 milliohm shunt. The wide board carries one INA3221, which has three channels and no gain stage, so one range for all three — and because that one range is narrow, each channel gets a 50 milliohm shunt to reach 3.28 A. The back of the board prints exactly that.

The header, which is the trap

Pin 1Pin 2Pin 3Pin 4
TK119GND3V3SCLSDA
INA3221GNDVCCSDASCL

Four wires either way, and the last two swap. If you wire one board, get it working, then move the same four jumpers onto the other, the bus goes silent and a scan reports nothing. No damage — I²C lines are open drain and the pull-ups are the same on both — but nothing at all to tell you why.

The habit worth forming with these two boards is reading the silkscreen every time. It is printed on the front of both.

What the INA3221 adds

Three things beyond the extra channels.

Four alert pins, brought out on the 14-pin header and marked CRI, WAR, TC and PV. Critical and warning compare the current against limits in hardware, so the board notices a problem whether your sketch is looking or not — and each of those two already has a red LED wired to it, so it also notices visibly.

A cuttable pull-up pad, marked I2C PULL-UP. Cut it and the board's two 10 kΩ resistors come off the bus, which is what you want when several devices share the wires and their pull-ups add up.

A three-way ground terminal, so three channels' negatives can each have a screw of their own rather than sharing one.

The back of the TK119 block: a black PCB with a gold-plated edge, printed POWER + on the left and + LOAD on the right with an arrow between them and a bar marked GND below, two small solder-bridge pads labelled A1 and A0 in the middle, a boxed label reading TK119 INA219, and the line MEASURE CURRENT AND VOLTAGE along the bottom.
The TK119's two address pads are on the back, and so is the reminder of which terminal is which way round.

What the TK119 adds

Two things, and they make it the better first board. The range is yours to choose, and the steps are fine. The INA219's gain setting spans ±40, ±80, ±160 or ±320 millivolts, which across 100 milliohms is ±0.4 to ±3.2 amps — so the same block can watch an LED strip and a motor starting — and every count is 0.1 mA, against 0.8 mA on the INA3221.

It also has a power register, so watts arrive without arithmetic. The INA3221 has none, and you multiply.

When it does not work

An I²C scan finds nothing on either board

Check the last two pins of the header against the silkscreen, not against your wire colours. The TK119 reads GND, 3V3, SCL, SDA and the INA3221 reads GND, VCC, SDA, SCL, so the two are the other way round. Swapped SDA and SCL do no damage at all — the bus simply never completes a transaction, which is indistinguishable from a dead board until you look.

Both boards answer at the same address

They both leave the factory at 0x40, because every address pin has a 10 kΩ pull-down fitted. Two devices at one address on one bus is two devices neither of which works. Bridge the INA3221's I2C ADDR pad to move it to 0x41, or one of the TK119's two pads on the back. See the next page.

The INA3221's 14-pin header duplicates the screw terminals

It does, deliberately: pins 4 to 9 are the same six nets as the three channel terminals, so a channel can be wired with a jumper instead of a screwdriver. Do not use both for one channel — two paths in parallel across one shunt means only some of the current goes through the resistor, and the reading is low by whatever fraction took the other route.

Which board is bigger, and will it fit?

The TK119 is 22.4 × 30.4 mm and the INA3221 is 48.0 × 30.6 mm, both with two mounting holes. The screw terminals overhang the edge of the board slightly and stand about 10 mm tall, so allow for them when measuring a case. The figure above draws both to scale.

Where this goes next

Power and address first

Edit this page — content/books/inaset/the-two-boards.mdx

Community

Questions about this product

See what other owners have asked, and read their solutions.

This page covers several products. Choose yours to see the right discussions.

Discuss this article

Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.

Browse Modules and blocks on the forum