Build1 distinct publisher3 min readPublished
Stack the coin cell above the board and the budget runs 0.65 mm over before tolerances get a line. Move it into a window and the 1.60 mm plane left behind is what picks the radio package and the sensor's thermal path.
The Engineer · Build desk
Compiled by The EngineerSomething wrong?How this is made
Once the cell sits in a hole in the board, thickness stops being a sum of stacked parts and becomes cell height plus two walls [6]. That is why the CR2032 never gets a second look. Put the taller cell in the same window and the stack is 3.2 for the cell, 0.10 of film and two 0.25 shells, so 3.80 mm, or 1.40 mm over a 2.4 mm target [1]. You can buy 2.5 times the energy for twice the height [2], which is a fine deal in a catalogue and a non-starter in this envelope [4][5].
The reallocated stack closes at 2.20 mm nominal and leaves 0.20 mm, about 8% of the envelope [7][4]. That 0.20 mm is not margin. It has to cover the cell contacts, any bonding layer, PCB-thickness and moulding tolerances, and the bow every thin shell carries [8]. Note the caveat the write-up puts on its own table: every line except the cell height is an allocation, illustrative rather than measured [3]. So read the shape and not the digits. The shape is a 27% overrun on target for the stacked version [5], and no component substitution recovers it.
Six features compete for the single plane [13]. Two of them are card-scale and get fixed first. The CR2016 is 20.0 mm across [4], and Nordic's published nRF52 guidance describes a reference monopole about 23 mm long that needs a minimum of 5 mm clearance to the ground plane [15], which makes the antenna 3 mm longer than the cell is wide [7]. Nothing conductive may enter that clearance, and a coin cell is a metal disc that detunes it [16]. Opposite short edges is the only placement that does not eat the middle of the board [14].
Package selection is where this gets an electrical label on a mechanical decision [19]. A 0.40 mm board on the 1.60 mm plane leaves 1.20 mm, split between the two faces by wherever the board sits inside the cell's height [18]. Nordic lists the nRF54L15 in a CSP47 at 0.42 mm and QFN options for that family at 0.85 mm [17], which is 71% of that interior against 35% [3]. Both fit geometrically. The QFN pays for its fit by spending most of one face's share and then competing with the cell contacts and the shell's inner bow [18].
The sensor figures deserve the same treatment as any benchmark table, because they describe a die under someone else's test conditions [23]. The TMP117 datasheet quotes +/-0.1 C maximum over a 70 C band and +/-0.2 C maximum over a 140 C one [20][6]; Sensirion's STS40 and STS41 quote a typical 0.2 C over a 165 C span [21][6]. For either number to transfer, the thermal path has to cooperate: current low enough that self-heating stays small, the package thermal pad left unsoldered, and slots routed so the board stops carrying SoC heat into the die [23][24]. Decide the plane first, because the datasheets all argue downstream of it [1].
Ranked by verification strength, evidence, and original report placement.
A sealed logging card is a thickness problem before it is an electronics problem: every subsystem sits downstream of the decision about how the millimetres are handed out.
Taking 2.4 mm as the target and allocating it with the cell stacked above the board makes the thickness budget go negative before the last line.
The stacked-cell allocation overruns by 0.65 mm on a 2.4 mm target, and every line in that table except the cell height is an allocation described as illustrative rather than measured.
The CR2016 cell datasheet gives 20.0 mm diameter, 1.6 mm maximum height and 90 mAh.
A 2.4 mm card cannot stack a coin cell on top of its PCB because the board and two shell walls have already spent most of the envelope; the cell drops into a window cut through the board instead, and thickness becomes cell height plus two walls.
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · September 1, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
build
Three manual interventions in a month, and every guard was working as designed1 distinct publisher
build
Six MariaDB versions, one real difference: the only reason to leave 10.6 is the July 2026 clock1 distinct publisher
build
Force the tool call, then hand Lightsail a long-lived key1 distinct publisher
build
Separating moderation rejections moved one API gateway's success rate from 95.5% to 98.9%1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Borrowed numbers solid, own numbers illustrative
Every figure that could be falsified comes from someone else's published document — CR2016 and CR2032 dimensions and capacity, Nordic's 23 mm monopole with 5 mm clearance, the 0.42 mm CSP47 against the 0.85 mm QFN, TMP117's ±0.1 °C over a 70 °C band, Sensirion's typical 0.2 °C over 165 °C, IEC 60529's 0.15–1 m immersion. What holds the story together is not borrowed: the four-layer 2.20 mm stack is the author's own allocation, described as illustrative, and the 0.20 mm it leaves is asked to absorb contacts, bond line, PCB and moulding tolerance and shell bow all at once. Checkable inputs, unverified assembly.
No card, no field, no data
This card does not exist anywhere in our coverage. There is no prototype, no measured stack-up, no shipped logger, no deployment in a real cold chain and no lifetime measurement; the energy discussion ends mid-sentence on a bound it never finishes. Nothing here can be scored as uptake without inventing it.
Hedged harder than it needed to be
The headline promises a closed thickness budget and the body immediately undercuts the promise: the allocations are called illustrative, the leftover 0.20 mm is denied the word margin, Nordic's dimensions are demoted to order of magnitude, and the reader is told a card inherits no sensor accuracy automatically. Even the light-sensor trade-off is left open pending polling energy and switch leakage. That is a piece under-selling its own arithmetic, not over-selling it.
Nothing for sale, affiliation unstated
No product is being pitched: three vendors' parts are quoted and none is anointed, and the piece keeps arguing against its own convenient conclusions. What is missing is who is writing — a self-published developer post with no stated affiliation, recommending specific silicon by name, is exactly where an undisclosed preference would sit if there were one. Low pressure, unverifiable source of it.
One voice, verifiable pieces
Our confidence divides the way the piece does. The quoted datasheet and standards figures can be looked up by anyone and we would expect them to hold. The synthesis — that the cell window is what picks the radio package and dictates the sensor's thermal path — comes from a single unaffiliated author with no second publisher, no reviewer and no hardware to point at. Believable engineering, uncorroborated.