Keyboard typing height: standards, the case stack and keycap profiles

How high a keyboard’s keytops sit above the desk, what the ergonomics standards and studies say about that height, and how the case, the switch and the keycap profile add up to it. The numbers are worked through on auspex, an open 5x12 ortholinear Hall-effect keyboard (the series hub covers the whole board), but the limits and the stack arithmetic apply to any board with MX-style switches.

Three kinds of number appear below, and each is labelled where it appears. Standard or study: from a published standard or paper, footnoted. Measured: by the author on 2026-11-05, either with calipers (three MT3 keycap rows) or from the desk to the keytop of the author’s own keyboard. Modelled: stack arithmetic, the wrist-geometry model, and keycap heights taken from manufacturer drawings or community CAD. Absolute modelled heights carry about +-1 mm from inputs nobody has measured yet: the switch’s keycap seat, the thickness of the pads as fitted (they are foam and compress under load), and the MT3 R4 cap. Differences between case variants and wedge angles for one keycap set are exact, because those terms cancel. Differences between profiles also carry the keycap-height uncertainty: manufacturer drawings for DSA and the SA home row, official CAD for KAT, calipers for three MT3 rows, and community clone models for the rest (the MT3 clone was up to 0.8 mm off the caliper values).

TL;DR:

• Standard: all four documents read here put the preferred home-row height at 30 mm or less above the desk. ISO 9241-410 adds “as low as possible” and a 35 mm limit the keyboard “shall not exceed”; ANSI/HFES 100-2007 and the HFES 100-2026 draft say it “should not exceed” 35 mm; DGUV 215-410 gives 30 mm and no second figure. Slope: ISO recommends 0-12 deg and requires 0-15 deg positive unadjusted, HFES asks for 0-15 deg, DGUV 0-12 deg (15 deg with the feet out).
• Modelled: in the wrist-geometry model each millimetre of keytop height costs about 0.34 deg of wrist extension, so 30.5 -> 25 mm saves about 2 deg; finger curl alone moves the answer by +-6 deg. Study: in Rempel’s typing data carpal tunnel pressure rises at every step of extension, and 30 deg is the first step whose pressure the post-hoc test separates from neutral. Modelled: a palm rest level with the keytops cuts extension to about 11 deg, more than any change in keyboard height. The model ignores contact pressure on the palm: Rempel’s subjects were not allowed to rest their hands, because palm contact can raise carpal tunnel pressure.
• On an MX-style board the switch fixes two terms above the PCB: plate top 5.0 mm above the PCB top (datasheet) and keycap rim 6.2 mm above the plate (modelled, +-0.5 mm). Only the pads, the floor, the standoff and the keycap profile are free.
• With a bottom-mounted USB-C receptacle the lowest PCB position is set by the cable plug overmold (7.0 mm at the USB-IF maximum, up to 7.4 mm on real cables, centred on the receptacle), once the case’s plug window is cut through the floor: the PCB bottom has to sit about 5.8 mm above the desk (modelled).
• auspex, flat (modelled): home row 26.01 mm with Cherry caps, 26.85 mm with DSA, 30.21 mm with MT3; the previous case put the MT3 home row at 34.05 mm. The author’s current keyboard measures 30.5 mm (measured) at the second row from the bottom, an MT3 R4 cap; auspex’s MT3 cap in that row sits at 31.06 mm (modelled).
• No study found keycap profile affecting typing speed, keycap wobble or rapid-trigger behaviour. Choose a profile on height and on how its rows map to the layout: Cherry is the lowest of the six profiles compared here (DCS and the low-profile families are lower but were not modelled), DSA is uniform so any cap fits any key, and MT3’s home row is 0.21 mm over the 30 mm preferred height.

---

Which profile, in one table

All heights modelled, desk to the keytop at the key centre, auspex case flat on 1.6 mm pads. The tilt column is the largest wedge angle (0.5 deg steps) that keeps the home row under the 35 mm cap on the 3D-printed case.

Profile	Home row	Row 2 from bottom	Largest tilt under 35 mm	Pick it when
Cherry	26.01 mm	26.81 mm	7 deg	typing is the main use: lowest of these six, room for a tilt
DSA	26.85 mm	26.85 mm	6 deg	layers and game remaps move caps around: uniform, any cap fits any key
XDA	28.56 mm	28.56 mm	4 deg	uniform rows with a larger top
KAT	30.13 mm	28.52 mm	2 deg	a spherical dish like MT3 with lower bottom rows
MT3	30.21 mm	31.06 mm	2 deg	you already own a set; home row 0.21 mm over the 30 mm preferred height
SA	31.19 mm	32.38 mm	1 deg	rarely on a low board: tallest of the six

The table comes from the stack and keycap sections further down; the decision guide at the end puts it as a flow.

What the standards say

All four documents put the preferred home-row height at 30 mm or less; they differ on the upper limit and on slope. ISO 9241-410 Annex B.2.3.1 b: “The preferred home row height is as low as possible and not greater than 30 mm. However, the home row height of the keyboard shall not exceed 35 mm.”¹ B.2.3.2, for compact keyboards (which a 60-key 5x12 is closer to), repeats the 30 and 35 mm figures. ANSI/HFES 100-2007 section 6.2.2.3 gives the same pair and defines the measuring point: “Keyboard height is the height measured from the table surface to the middle of the home row of the keyboard.”² The HFES 100-2026 public review draft keeps both figures and the measuring point, and makes the unadjusted 0-15 deg slope range a “shall”.³ The German workplace guidance DGUV Information 215-410 asks for at most 30 mm in the middle key row and 0-12 deg of slope with the feet folded in, 15 deg with them out.⁴

Item	ISO 9241-410	ANSI/HFES 100-2007	HFES 100-2026 draft	DGUV 215-410
Home-row height	preferred as low as possible, at most 30 mm; shall not exceed 35 mm	preferred 30 mm or less, should not exceed 35 mm	same as 2007	at most 30 mm (should)
Slope, unadjusted	recommended 0-12 deg; shall be 0-15 deg positive	at least one setting in 0-15 deg	shall be 0-15 deg	0-12 deg; at most 15 deg with feet out
Negative slope	not provided for (the slope range is positive only)	allowed for alternative designs	allowed for alternative designs	not mentioned
Key pitch	19 +- 1 mm	not checked	not checked	-
Palm rest	if present, 50-100 mm deep in front of the bottom row	supports should be provided for prolonged use; under 3.8 cm deep may raise contact pressure	optional; match or exceed the keyboard’s front-edge height	100-150 mm of free desk in front; an extra rest can hinder
Wrist posture limit	-	extension and flexion each under 30 deg	extension and flexion each under 30 deg (posture list, Fig. 5-5)	-

HFES measures to “the middle of the home row”, read here as the centre of the home-row keytops; it does not say whether the feet are extended. DGUV states its 30 mm for the keyboard with the feet folded in. The ISO text read here does not define the point. Positive slope means the bottom row is lower than the top row. A 19.05 mm ortholinear grid meets the ISO 19 +- 1 mm pitch.

One commercial reference point: Wooting’s design guideline for its 80HE is a front height with keycaps of at most 30 mm. The board measures 19.4 mm at the front without caps and 28.4 mm with its OEM caps at 2.8 deg. Wooting says it built a 0 deg prototype under 30 mm, but that at 0 deg “you need to go lower, which was physically not possible”; the post lists the daughterboard and the gasket mount among the parts it had to make space for.⁵ Wooting measures the front edge, the standards the home row.

What the studies show

Carpal tunnel pressure against wrist extension

Rempel, Keir and Bach measured carpal tunnel pressure (CTP) in 20 subjects while they typed at set wrist postures.⁶ Their results table:

Target posture	Measured posture	CTP while typing
15 deg flexion	11.2 deg flexion	1.9 kPa (14 mmHg)
0 deg	2.6 deg extension	1.9 kPa (14 mmHg)
15 deg extension	16.1 deg extension	2.3 kPa (17 mmHg)
30 deg extension	27.9 deg extension	3.1 kPa (23 mmHg)
45 deg extension	38.1 deg extension	4.0 kPa (30 mmHg)
split keyboard set to a conventional layout (halves together, 7 deg slope; not in the analysis)	24.8 deg extension	2.8 kPa (21 mmHg)

Typing raised CTP over the same static posture (p = 0.001). Keir, Bach, Hudes and Rempel (n = 37) worked out the 25th-percentile angles, the postures that keep 75% of people under a pressure threshold: 26.6 deg of extension for 25 mmHg and 32.7 deg (95% CI 27.2-38.1) for 30 mmHg. Ulnar deviation has the tightest limits, 12.1 and 14.5 deg.⁷ The lowest CTP in controls sat at 2 +- 9 deg of extension, not exactly 0,⁸ and people asked to find a rest posture blindfolded chose 7-9 deg of extension.⁹ In Rempel’s data pressure rises with every step of extension, but the first step the post-hoc test separates from neutral and from flexion is 30 deg (45 deg is above all the others), and the authors advise avoiding extension over 30 deg. Keir’s thresholds put a quarter of people at 25 or 30 mmHg by about 27-33 deg. By this data a few degrees of extension are not harmful.

Pressure on the palm matters for rest design: in five cadaver hands, 1 kg on the flexor retinaculum raised CTP by 103 mmHg, on the thenar area by 75 mmHg and on the hypothenar area by 37 mmHg.¹⁰ Loading the proximal palm is therefore a pressure risk, and no study read here identifies a safe contact site. ANSI/HFES 100-2007 warns that supports under 3.8 cm deep may raise contact pressure.²

Slope

The slope studies cited here find less wrist extension at 0 deg or negative slope than at positive slope. On a split keyboard, Rempel and colleagues (n = 24) found slope effects mixed, with the least extension at 0 to -4 deg.¹¹

• Simoneau and Marklin (n = 30) varied slope from +15 to -15 deg. Mean extension fell from 22 to 9 deg, about 0.43 deg of extension per degree of slope (derived, 13 / 30). Height relative to the elbow had a similar effect: 21.8 deg with the keyboard below elbow height against 7.3 deg above it.¹² That height is the desk and chair setup, not the keyboard’s own thickness.
• Simoneau, Marklin and Berman (n = 16): about 12 deg of extension at +7.5 deg slope became 3 deg of flexion at -15 deg; forearm muscle activity changed by only 1-3% MVC.¹³
• 90 professional touch typists on a conventional keyboard with a 5 deg positive slope typed at 21.2 +- 8.8 deg (left) and 17.0 +- 7.4 deg (right) of extension.¹⁴
• Hedge and Powers: a self-chosen negative-slope tray averaged 12 deg below horizontal and took extension from 13 deg to -1 deg.¹⁵ A 3-week field trial of a tilt-down tray raised the share of neutral-zone wrist movements from about 50% to over 80%.¹⁶ Both tested negative-slope keyboard supports that hold the keyboard off the desk (a tray in the 1999 trial). A tray can lower the whole keyboard; a desktop case tilted negative has to stand on its back edge with the front lifted, which raises every row in front of that edge.
• Eghujovbo and Kim (n = 30): users prefer positive angles, but negative tilt reduced muscle activity and awkward wrist posture “while maintaining performance”.¹⁷

A flat case is inside the standards’ slope ranges (ISO 0-12 deg recommended, HFES 0-15 deg) and at the low end the posture studies favour, though Eghujovbo and Kim’s subjects preferred positive angles.

Height in the field

A prospective study of 632 new computer users found that “use of a keyboard with the ‘J’ key > 3.5 cm above the table surface, key activation force > 48 g, and radial wrist deviation of > 5 degrees while using a mouse was associated with a greater risk” of hand and arm symptoms or disorders, in one multivariable model, and that a J key more than 12 cm from the desk edge was associated with lower risk (read here as room for the forearms on the desk).¹⁸ The study’s NIOSH final report defines J-key height as table surface to J keytop.¹⁹ The 3.5 cm threshold coincides with the 35 mm cap, but it is a risk association: the keyboards of that period differed in more than height, and no mechanism was isolated.

The cross-sectional study before it (n = 379) found wrist extension “weakly correlated with keyboard height (r = -0.24) and virtually not at all with keyboard thickness (r = 0.07)”, and wrist rest use associated with less flexion (21.9 against 25.1 deg).²⁰ Serina, Tal and Rempel (n = 25) found that wrist angles “were not predictable based on anthropometry or typing speed and varied widely between subjects”.²¹ A best height for one person does not follow from these data.

Floating against supported hands

• Forearm support gave less ulnar deviation and discomfort than floating; wrist support lowered shoulder muscle activity (n = 13).²²
• A 1-year randomised trial with 182 call-centre operators: a wide forearm support halved neck and shoulder disorders (hazard ratio 0.49, 95% CI 0.24-0.97).²³
• Hedge and Powers found that full-motion forearm supports did not change wrist posture.¹⁵

DGUV takes the other side: a low, flat keyboard makes a palm rest unnecessary, and a rest can get in the way.⁴ ANSI/HFES 100-2007 says supports should be provided for prolonged use.²

Gaming

No study found varies keyboard height or slope for gaming. One small motion-capture study (n = 5, first-person shooter, 5 minutes per device) found the keyboard among the devices with the highest left-wrist ulnar deviation.²⁴ The model below adds an inference: a hand curled over WASD flexes its fingers more than a touch-typing hand, and at the same keytop height that needs about 5 deg more extension. In the model a low board helps the gaming hand slightly more: 0.39 deg per mm with that curl against 0.34 deg per mm with the typing curl.

Speed and keycap profile

No source found a typing-speed penalty from flat or negative slope. HFES 100-2007 6.2.2.4: “Research examining stepped and sloped keyboard profiles found no notable differences in user performance (Kinkead & Gonzales, 1969).”² No study was found that measures keycap wobble per profile, rapid-trigger false actuations against cap height or mass, or typing speed across modern keycap profiles.

Reasoning only, unmeasured: 2013 forum weigh-ins of single ABS and PBT keys put common caps at 0.80-1.35 g,²⁵ and a 2 g cap on a spring with 30 gf of initial force is under 7% of the preload, so cap weight alone cannot move a switch at rest. Whether a taller, heavier cap makes rapid-trigger false releases more or less likely depends on stem play and the firmware’s sensitivity, and is a bench question.

The wrist-geometry model

To put the standards’ millimetres into degrees, a sagittal-plane model: a rigid hand pivots about the wrist joint centre, and the fingertip has to land on the home-row keytop. The model is a sensitivity analysis, and every input is an assumption:

Assumption	Value
Forearm	horizontal (the ISO/HFES reference posture); extension = hand axis angle above the forearm line
Palm length (wrist centre to knuckle)	116 mm male, 108 mm female: ANSUR II medians26
Finger length	hand length minus palm length: 193 - 116 = 77 mm male, 180 - 108 = 72 mm female
Finger curl	MCP 20, PIP 45, DIP 20 deg flexion; phalanges split 0.45 / 0.30 / 0.25 of finger length (assumed)
Wrist joint centre	20 mm above whatever the wrist or palm heel rests on (about half a wrist depth)
Result of the curl	fingertip 160 mm ahead of and 52 mm below the wrist centre (male median; female 149 / 49 mm)

Wrist resting on the desk, male median (female within 0.6 deg):

Home-row keytop above desk	15 mm	20 mm	25 mm	30.5 mm	35 mm	40 mm
Wrist extension	16.3 deg	18.0 deg	19.7 deg	21.6 deg	23.1 deg	24.8 deg

• The slope near 30 mm is 0.341 deg per mm. 30.5 -> 25 mm saves 1.9 deg; 30.5 -> 20 mm saves 3.6 deg.
• At a 30.5 mm keytop, finger curl changes the answer more than height does: 10/30/10 deg curl gives 14.5 deg of extension, 30/60/30 deg gives 26.8 deg. Moving the wrist centre to 15 or 25 mm gives 23.3 or 19.9 deg. That fits the weak correlations in Gerr 2000 and the spread in Serina 1999.
• The model’s 21.6 deg at 30.5 mm falls inside the 17-25 deg measured on conventional keyboards by Simoneau 1999, Serina 1999 and Rempel 2008. The agreement is a sanity check only: the three studies used different keyboards and setups.
• A tighter gaming curl (30/60/30) gives 22.7 / 24.7 / 26.8 deg at 20 / 25 / 30.5 mm, about 5 deg over the typing curl.

With a palm rest, the wrist centre sits 20 mm above the rest top:

Keytop, rest top	0 mm	10 mm	18 mm	25 mm	30.5 mm	35 mm
20 mm keytop	18.0 deg	14.6 deg	11.8 deg	9.4 deg	7.5 deg	6.0 deg
25 mm keytop	19.7 deg	16.3 deg	13.6 deg	11.2 deg	9.3 deg	7.7 deg
30.5 mm keytop	21.6 deg	18.2 deg	15.4 deg	13.0 deg	11.2 deg	9.6 deg

A rest whose top is level with the keytop gives about 11 deg at either height, close to the 7-9 deg people choose blindfolded and far under the 26.6 deg CTP threshold. Zero extension with this curl needs the wrist centre about 52 mm above the keytop, 82.5 mm above the desk at a 30.5 mm keytop: that is the floating posture, or a keyboard below elbow level. The model ignores contact pressure: Rempel’s subjects were not allowed to rest their hands because palm contact can raise carpal tunnel pressure, and the authors expect resting at a workplace to raise it above their measured values.⁶

Tilt costs height. Against the flat case, a wedge raises the Cherry home row by 5.4 / 7.2 / 8.9 / 11.4 / 13.0 mm at 3 / 5 / 7 / 10 / 12 deg (modelled: 1.5 mm wedge front, the 1.6 mm pads moved from 0.4 mm proud in the case wells to fully under the wedge, home row 52.1 mm behind the case front; the stack formula is in the next section). By the height model, 5 deg adds about 2.4 deg of extension; by the measured slope coefficients (0.43-0.67 deg per degree of slope, Simoneau 2001 and 2003) it adds 2.2-3.4 deg. The two overlap in mechanism, because the studies’ slope changes also moved the rows up and down, so do not add them.

The stack, desk to keytop

For a flat tray-mounted case:

keytop = pad proud of case bottom + floor + standoff + PCB + 5.0 + 6.2 + keycap height

The two fixed terms come from the switch. The Cherry MX datasheet puts the plate seat 5.0 mm above the housing bottom and the stem top 3.6 mm above an 11.6 mm housing, so 10.2 mm above the plate top.²⁷ Gateron’s KS-20T magnetic switch drawing gives the same 5.00 +- 0.05 mm plate seat, so the plate top sits 5.0 mm above the PCB top whatever the plate thickness.²⁸ The keycap rim sits 6.2 mm above the plate top at rest, the MX value in the Cosmos keyboard generator;²⁹ that leaves 4.0 mm of the 10.2 mm stem inside the keycap socket. The Gateron drawing does not dimension the stem, so for a magnetic switch the 6.2 mm is Cherry-derived and carries +-0.5 mm.

The worked stack, mm, for the previous auspex case and the current one (modelled):

Term	Previous case	Current case	Why it changed
Pad proud of the case bottom	1.3 (2.5 mm pad assumed in a 1.2 mm well)	0.4 (1.6 mm 3M SJ5202-class pad in the same well)30	thinner pad; SJ5202 is soft polyurethane foam, so it sits lower under load31
Floor	2.5	2.0	0.51x the bending stiffness, still 0.8 mm under the foot wells
Standoff, floor top to PCB bottom	6.75 (SWD header 4.25 + 2.5 clearance)	4.26 (USB-C receptacle 3.26 + 1.0 clearance)	SWD header moved over its own floor opening; clearance cut
PCB	1.6	1.6	-
PCB top -> plate top	5.0	5.0	fixed by the switch
Plate top -> keycap rim	6.2	6.2	fixed by the switch
Desk -> keycap rim	23.35	19.46	3.89 mm lower at every key
MT3 home row (cap R3)	34.05	30.21	R3 10.7 mm (clone) before, 10.75 mm (caliper) now

Only the first three rows are free. Plate thickness does not move the keycaps.

With a wedge under the case, the case rotates about its front-bottom edge and the pads move to the wedge’s underside:

keytop = pad + wedge front + y sin(a) + z cos(a)

where y is the row centre’s distance from the case front and z the keytop’s height above the case bottom. On auspex the home row is 52.13 mm behind the front face of the 3D-printed case (3.0 mm wall, 0.5 mm gap, 1.0 mm PCB margin, 2.5 key pitches).

Keycap heights per row

Keycap height here is rim to the highest point of the top surface, mm. On a 5x12 the rows from the back are number, top alpha, home, bottom alpha and modifiers; sculpted profiles call them R1 to R4 with the modifier row taking R4, except KAT, which numbers from the bottom (R1 bottom row, R5 number row).³²³³

Profile	Number row	Top alpha	Home	Bottom alpha	Modifiers	Source and kind
MT3	13.89 (R1)	11.08 (R2)	10.75 (R3)	11.6 (R4)	11.6 (R4)	R1-R3 measured: calipers on the author’s MT3 Susuwatari set; R4 modelled, from the KeyV2 clone34
Cherry	9.8	7.45	6.55	7.35	7.35	modelled: KeyV2 clone; no primary Cherry drawing found
DSA	7.39	7.39	7.39	7.39	7.39	manufacturer drawing, 0.291 in35
KAT	13.54 (R5)	11.65 (R4)	10.67 (R3)	9.06 (R2)	10.68 (R1)	bounding boxes of Keyreative’s official STEP model33
SA	14.89	12.925	11.73	12.925	12.925	home row from the manufacturer drawing (0.462 in);36 other rows modelled, KeyV2
XDA	9.1	9.1	9.1	9.1	9.1	modelled: riskable keycap_playground clone37

KeyV2 says of its own MT3 file: “This is an imperfect attempt to clone the MT3 profile”. The caliper values bear that out: KeyV2 gives 13.1 / 10.7 / 10.7 mm for R1-R3, against the measured 13.89 / 11.08 / 10.75. Where KeyV2 and the Cosmos table disagree on Cherry (Cosmos gives 9.3 / 7.0 / 6.0 / 7.0), the taller KeyV2 set is used.³⁸

matt3o, MT3’s designer, gives the home-row (R3) top angle as -7 deg, against 0 deg for SA.³⁹ He writes that the R3 top is meant to sit parallel to the desk on keyboards at 5-7 deg,⁴⁰ and that the goal was a set that “worked perfectly on a 5 to 9-ish degree angle”.⁴¹ On a flat case MT3’s home row faces the typist slightly.

Uniform profiles (DSA, XDA) have no row assignment: any cap goes on any key, which suits an ortholinear board whose layers move keys around. A sculpted set needs an ortho kit or enough spare R4 caps; whether a set has one varies by set, and the MT3 Susuwatari set has one.³²

Desk to keytop on auspex

Row 2 from the bottom / home row, mm, modelled. “Previous” is the case before the redesign, with 2.5 mm pads assumed; its CNC variant had a 10 mm floor carrying a brass weight. The current CNC variant moved the brass into the walls and uses the same 2.0 mm floor as the printed case, so both sit at the same height when flat.

	MT3	Cherry	DSA	KAT	SA	XDA
previous, printed	34.95 / 34.05	30.70 / 29.90	30.74 / 30.74	32.41 / 34.02	36.27 / 35.08	32.45 / 32.45
previous, CNC (10 mm floor)	42.45 / 41.55	38.20 / 37.40	38.24 / 38.24	39.91 / 41.52	43.77 / 42.58	39.95 / 39.95
current, printed and CNC, flat	31.06 / 30.21	26.81 / 26.01	26.85 / 26.85	28.52 / 30.13	32.38 / 31.19	28.56 / 28.56
current, printed, 3 deg wedge	35.45 / 35.60	31.20 / 31.40	31.24 / 32.24	32.91 / 35.52	36.77 / 36.58	32.95 / 33.95
current, printed, 5 deg wedge	36.53 / 37.34	32.29 / 33.16	32.33 / 33.99	34.00 / 37.26	37.85 / 38.32	34.04 / 35.70
largest wedge, home at most 35 mm, printed	2 deg	7 deg	6 deg	2 deg	1 deg	4 deg
largest wedge, home at most 35 mm, CNC	1.5 deg	5 deg	4.5 deg	1.5 deg	0.5 deg	3 deg

The CNC limits are lower because its 18 mm front wall puts every row 15 mm further from the wedge’s pivot. The previous CNC case had every profile over the 35 mm cap before any tilt.

The design targets were a home row of about 25-27 mm with Cherry or DSA caps, an MT3 home row no higher than 30.5 mm, flat by default, and the 35 mm cap for every profile and variant. The 30.5 mm is the author’s measurement of the current keyboard at the second row from the bottom (an MT3 R4 cap), which the case’s height check uses as a ceiling for the MT3 home row: 30.21 mm, 0.29 mm under it. Row for row, auspex’s MT3 cap in that second row sits at 31.06 mm, 0.56 mm above the measured board. Both margins are inside the +-1 mm modelling error, so the unmeasured R4 cap and keycap seat decide them.

What stops a board going lower

The receptacle of auspex’s USB-C port is on the bottom of the PCB. A cable plug is centred on the receptacle, 1.63 mm below the PCB bottom (half of the 3.26 mm receptacle), and its overmold is 7.0 mm tall at the USB-IF maximum, up to 7.4 mm on real cables. So the plug reaches 5.13-5.33 mm below the PCB bottom. While the case’s plug window starts at the floor top, the standoff has to be at least 6.13 mm (5.13 mm for a 7.0 mm overmold plus 1.0 mm of margin). Cut the window down through the floor and the desk becomes the limit: the PCB bottom has to sit about 5.8 mm above the desk (5.33 + 0.5 mm of air).

The current case is about 0.9 mm above that limit (PCB bottom 0.4 + 2.0 + 4.26 = 6.66 mm above the desk, computed); a 7.4 mm overmold clears the desk by 1.33 mm. Lowering the PCB the remaining 0.86 mm, with a floor pocket under the receptacle, would put auspex’s MT3 second row from the bottom at about 30.2 mm, level with the author’s current keyboard (modelled). Below that, only a PCB change (moving the USB-C receptacle) or a right-angle cable helps.

Each lever, and what it saved at every key:

Lever	Saving	Condition
SWD header (4.25 mm, the tallest bottom part) over its own floor opening, exempt from the standoff	0.99 mm	the debug plug enters from below anyway; the header must stay inside the opening
Component clearance under the PCB 2.5 -> 1.0 mm	1.5 mm	the 2.5 mm was the SWD plug’s room; needs the plug window through the floor
Plug window cut through the floor	enables the two above	cable plug clears the desk
Floor 2.5 -> 2.0 mm	0.5 mm	0.51x bending stiffness
Pads 2.5 mm (assumed) -> 1.6 mm in the same 1.2 mm wells	0.9 mm	0.4 mm proud, still clear of the desk
Case rim plate top + 2.0 -> + 1.0 mm	1.0 mm at the case edge, 0 at the keys	at 3.2 mm of travel the keycap skirt is still 2.0 mm above the rim
CNC floor 10.0 -> 2.0 mm, brass moved into the walls	8.0 mm (CNC)	mass from the walls instead of the floor

The case build that carries these changes is the subject of the build123d case guide; the PCB they sit under is the one ordered in the JLC 4-layer order guide.

The reference board

The author types on a Boardsource Equals 60 (a 5x12) with MT3 Susuwatari caps. Measured from the desk (bottom of the rubber bumpers) to the top of the second row from the bottom, it stands about 30.5 mm, and feels a bit high: a reference, not a target. On a 5x12 that row is the bottom alpha row, an MT3 R4 cap, not the home row the standards measure.

Its case is 14 mm tall.⁴² A sleeve maker describes the board as “completely flat with no incline”,⁴³ and the working below assumes a flat case. Working back from the 30.5 mm (inferred, no teardown): 30.5 - 11.6 (R4 cap) - 6.2 (keycap seat) = 12.7 mm from desk to plate top, which leaves about 4-4.5 mm of floor and standoff under the PCB, against 9.25 mm in auspex’s previous case. The 14 mm is the case outline; the product page does not say whether it includes the bumpers or where the plate sits in it, so this inference is the weakest number on the page. Its home row would sit at about 29.6 mm (30.5 minus the MT3 R4-R3 difference). The current auspex case with the same MT3 caps sits 0.56 mm above it at that row (31.06 mm, modelled); with Cherry or DSA caps it sits about 3.7 mm below. Each comparison subtracts one measurement from a modelled height, so it carries the model’s +-1 mm.

The author’s keymap for the same board is the subject of the QMK and VIA keymap guide.

Choosing a profile

The same decision as a list (auspex case, heights modelled):

1. If layers or game remaps move caps between rows, use DSA: uniform, home row 26.85 mm, tilt up to 6 deg under the 35 mm cap.
2. Otherwise, if you want MT3’s spherical dish, use KAT (home 30.13 mm, its bottom alpha row 2.5 mm lower than MT3’s) or MT3 (30.21 mm). Both home rows sit 0.1-0.2 mm over the 30 mm preferred height and allow at most 2 deg of tilt.
3. Otherwise use Cherry: the lowest of the six profiles compared, home row 26.01 mm, tilt up to 7 deg.
4. For any of them, the model gives a palm rest more effect than any height change: top about level with the keytops (about 25 mm here), at least 38 mm deep (HFES) or 50-100 mm (ISO). The model ignores palm contact pressure, and the cadaver data above show pressure rising wherever the palm near the carpal tunnel is loaded. Leave forearm room on the desk: a J key more than 12 cm from the desk edge was associated with lower risk, which on auspex puts the case front about 7 cm back.

Keep the default flat. A tilt adds height at every row behind the pivot and adds extension; the standards allow it, and the slope studies show the least extension at 0 deg or below.

Evidence

Claim	Kind	How it was checked
Home row 30 mm or less preferred (all four); 35 mm limit (ISO “shall”, HFES “should”, none in DGUV); slope 0-12 / 0-15 deg	standard	ISO 9241-410 Annex B.2.3, ANSI/HFES 100-2007 6.2.2.3, HFES 100-2026 draft 3.3.4, DGUV 215-410 p. 46, primary texts read
CTP rises from 1.9 to 4.0 kPa between 0 and 45 deg target extension while typing	study	Rempel 2008 results table, full text read
26.6 / 32.7 deg extension thresholds for 25 / 30 mmHg	study	Keir 2007, abstract read
J key over 3.5 cm associated with more hand/arm disorders	study	AJIM 2002, abstract and NIOSH final report read
Extension against keyboard thickness r = 0.07	study	Gerr 2000, abstract read
0.341 deg of extension per mm near 30 mm	modelled	wrist-geometry script output; all inputs assumed, ANSUR II medians for lengths
MT3 R1 / R2 / R3 13.89 / 11.08 / 10.75 mm	measured	calipers on the author’s MT3 Susuwatari set, 2026-11-05
MT3 R4 11.6 mm, Cherry rows	modelled	KeyV2 community clone at a pinned commit; not a manufacturer number
DSA 7.39 mm, SA home 11.73 mm	manufacturer drawing	Signature Plastics family drawings
KAT row heights	official CAD	bounding boxes of Keyreative’s STEP file
Plate top 5.0 mm above PCB top	datasheet	Cherry MX datasheet; Gateron KS-20T drawing (5.00 +- 0.05)
Keycap rim 6.2 mm above plate top	modelled, +-0.5 mm	Cosmos generator’s MX value; Gateron does not dimension the stem; not yet measured on the auspex switch
Pad 1.6 mm, 0.4 mm proud	datasheet, +-0.5 mm	3M Bumpon selection guide (SJ5202 11.00 x 1.60 mm); soft foam, compression under load and the pads fitted not measured
Reference board 30.5 mm at row 2 from the bottom	measured	the author, desk to keytop, 2026-11-05
Reference board floor + standoff about 4-4.5 mm	inferred	from the 30.5 mm and keycap and switch heights; the 14 mm case outline does not say where the plate sits; no teardown or drawing found
auspex heights per profile and tilt limits	modelled	the case’s height model, run against the board file
No speed, wobble or rapid-trigger effect of profile	absence	searched; nothing measured found; HFES 100-2007 cites one 1969 study on stepped vs sloped

Related docs

• auspex: an open 5x12 Hall-effect keyboard - the series hub.
• PCB as code: verification gates - how the board file the height model reads is generated and checked.
• Ordering a 4-layer PCBA from JLC - the PCB whose USB-C receptacle sets the floor.
• A keyboard case in build123d - the case these heights come from, and its printability checks.
• QMK and VIA keymap guide - the keymap on the reference 5x12.

References

1. ISO, “ISO 9241-410:2008 Ergonomics of human-system interaction - Part 410: Design criteria for physical input devices,” Annex B.2.3; clause text read from the DIN EN ISO 9241-410:2012-12 English edition, which includes Amendment 1:2012. https://www.iso.org/standard/38899.html ; amendment: https://www.iso.org/standard/54107.html ↩

2. HFES, “ANSI/HFES 100-2007 Human Factors Engineering of Computer Workstations,” sections 5.2.1.1, 5.2.3.5, 6.2.2.3 and 6.2.2.4. HFES technical standards page: https://www.hfes.org/Publications/Technical-Standards ↩ ↩² ↩³ ↩⁴

3. HFES, “BSR/HFES 100-2026,” public review draft (may change before publication), sections 3.3.4, 3.4 and 5. https://www.hfes.org/Portals/0/Uploads/BSR%20HFES%20100-2026%20Draft%20%28003%29.pdf ↩

4. DGUV, “DGUV Information 215-410 Bildschirm- und Bueroarbeitsplaetze,” 2019 edition, p. 46. BGHM. https://www.bghm.de/fileadmin/user_upload/Arbeitsschuetzer/Gesetze_Vorschriften/Informationen/215-410.pdf ↩ ↩²

5. Wooting, 80HE ergonomics and feet, Wooting blog. https://wooting.io/post/wooting-80he—ergonomics-and-feet ↩

6. Rempel, Keir, Bach, J Orthop Res 26(9):1269-73, 2008, typing and static carpal tunnel pressure by wrist posture. PubMed Central. https://pmc.ncbi.nlm.nih.gov/articles/PMC2649727/ ↩ ↩²

7. Keir, Bach, Hudes, Rempel, Hum Factors 49(1):88-99, 2007, wrist posture thresholds for carpal tunnel pressure. PubMed. https://pubmed.ncbi.nlm.nih.gov/17315846/ ↩

8. Weiss, Gordon, Bloom, So, Rempel, J Bone Joint Surg Am 77(11):1695-9, 1995, wrist position of lowest carpal tunnel pressure. PubMed. https://pubmed.ncbi.nlm.nih.gov/7593079/ ↩

9. Fagarasanu, Kumar, Narayan, Clin Biomech 19(7):671-7, 2004, self-selected wrist rest posture. PubMed. https://pubmed.ncbi.nlm.nih.gov/15288452/ ↩

10. Cobb, An, Cooney, J Hand Surg Am 20(2):181-5, 1995, external palm pressure and carpal tunnel pressure. PubMed. https://pubmed.ncbi.nlm.nih.gov/7775748/ ↩

11. Rempel, Nathan-Roberts, Chen, Odell, “The effects of split keyboard geometry on upper body postures,” Ergonomics 52(1):104-11, 2009. PubMed. https://pubmed.ncbi.nlm.nih.gov/19308823/ ↩

12. Simoneau, Marklin, “Effect of computer keyboard slope and height on wrist extension angle,” Hum Factors 43(2):287-98, 2001. PubMed. https://pubmed.ncbi.nlm.nih.gov/11592669/ ↩

13. Simoneau, Marklin, Berman, Phys Ther 83(9):816-30, 2003, downward keyboard slope, wrist posture and muscle activity. PubMed. https://pubmed.ncbi.nlm.nih.gov/12940768/ ↩

14. Simoneau, Marklin, Monroe, “Wrist and forearm postures of users of conventional computer keyboards,” Hum Factors 41(3):413-24, 1999. PubMed: https://pubmed.ncbi.nlm.nih.gov/10665209/ ; full text (keyboard slope in the methods), CDC Stacks: https://stacks.cdc.gov/view/cdc/191620/cdc_191620_DS1.pdf ↩

15. Hedge, Powers, Ergonomics 38(3):508-17, 1995, negative-slope keyboard system and forearm supports. PubMed. https://pubmed.ncbi.nlm.nih.gov/7729392/ ↩ ↩²

16. Hedge, Morimoto, McCrobie, Ergonomics 42(10):1333-49, 1999, field trial of a downward-tilting keyboard tray. PubMed. https://pubmed.ncbi.nlm.nih.gov/10582503/ ↩

17. Eghujovbo, Kim, Ergonomics 67(8):1134-46, 2024, keyboard tilt in sitting and standing. PubMed. https://pubmed.ncbi.nlm.nih.gov/37966342/ ↩

18. Marcus et al., Am J Ind Med 41(4):236-49, 2002, prospective study of 632 new computer users, part II: postural risk factors. PubMed. https://pubmed.ncbi.nlm.nih.gov/11920967/ ↩

19. NIOSH, final report of the prospective study of computer users, CDC Stacks. https://stacks.cdc.gov/view/cdc/218084/cdc_218084_DS1.pdf ↩

20. Gerr et al., AIHAJ 61(2):223-30, 2000, cross-sectional study of 379 computer operators. PubMed. https://pubmed.ncbi.nlm.nih.gov/10782194/ ↩

21. Serina, Tal, Rempel, Ergonomics 42(7), 1999, wrist and forearm postures while typing. Taylor and Francis. https://www.tandfonline.com/doi/abs/10.1080/001401399185225 ↩

22. Cook, Burgess-Limerick, Papalia, Appl Ergon 35(3):285-92, 2004, forearm and wrist support against floating. PubMed. https://pubmed.ncbi.nlm.nih.gov/15145291/ ↩

23. Rempel et al., Occup Environ Med 63(5):300-6, 2006, randomised trial of forearm support for call-centre operators. PubMed. https://pubmed.ncbi.nlm.nih.gov/16621849/ ↩

24. Mariano, Priadythama, Loh, Muraki, ERGONOMICS 2022 proceedings, 2023, gaming controllers and wrist angles. Kyushu University. https://kyushu-u.elsevierpure.com/en/publications/effect-of-gaming-controllers-on-wrist-angles/ ↩

25. geekhack, “Keycap Stats,” forum thread, 2013. https://geekhack.org/index.php?topic=47882.0 ↩

26. “2012 Anthropometric Survey of US Army Personnel” (ANSUR II), public since 2017; project page at the OPEN Design Lab, Penn State. Medians computed here from a public combined copy of the male and female data. https://www.openlab.psu.edu/ansur2/ ↩

27. Cherry, “MX Series” datasheet, Digi-Key. https://media.digikey.com/pdf/Data%20Sheets/Cherry%20PDFs/MX%20Series.pdf ↩

28. Gateron, “KS-20T series PRO” (Magnetic Jade Pro) drawing DS-02-001-A0, p. 7. https://www.gateron.com/u_file/2505/21/file/GateronMagneticJadeProSwitchKS-20TF10B045NW-Y106PRO1.pdf ↩

29. Cosmos Keyboards, “switches.ts,” GitHub, commit 21f2333. https://github.com/rianadon/Cosmos-Keyboards/blob/21f233377fa5451fd398c2a5a473329dc6f49e08/src/lib/geometry/switches.ts ↩

30. 3M, “Bumpon Protective Products” selection guide, hexagon SJ5202: 11.00 mm wide, 1.60 mm high, +-0.5 mm. Digi-Key. https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/7161/3m-bumpon-protective-products.pdf ↩

31. 3M, “Bumpon Protective Products SJ5202,” product page (soft foam). https://www.3m.com/3M/en_US/p/d/b5005035181/ ↩

32. Drop, MT3 Susuwatari ortholinear kit, product image. https://massdrop-s3.imgix.net/product-images/drop-matt3o-mt3-susuwatari-custom-keycap-set/FP/mb7Jjm3NRfRvMpRV1A9C_SUSUWATARI-ortholinear-copy.jpg ↩ ↩²

33. Keyreative, “KAT Profile Design and Information,” Keyreative blog, 2024-06-26 (links the official STEP model). https://keyreative.store/blogs/news/keyreative-new-keycaps-height-profile-kam-preview ↩ ↩²

34. rsheldiii, KeyV2 key profiles, GitHub, commit 19f0d2f. https://github.com/rsheldiii/KeyV2/tree/19f0d2faadd4949634c93f38d1a66869d29e8f43/src/key_profiles ↩

35. Signature Plastics, DSA family drawing. https://cdn.shopify.com/s/files/1/0793/6996/1750/files/DSAFamily.pdf ↩

36. Signature Plastics, SA family drawing. https://cdn.shopify.com/s/files/1/0793/6996/1750/files/SAFamily.pdf ↩

37. riskable, keycap_playground profiles.scad, GitHub, commit 8c16a84. https://github.com/riskable/keycap_playground/blob/8c16a84839c6974a4471db5eb137436de2cab4cc/profiles.scad ↩

38. Cosmos Keyboards, “keycaps.ts,” GitHub, commit 21f2333. https://github.com/rianadon/Cosmos-Keyboards/blob/21f233377fa5451fd398c2a5a473329dc6f49e08/src/lib/geometry/keycaps.ts ↩

39. matt3o, “A new keycap profile is in the work,” matt3o.com. https://matt3o.com/a-new-keycap-profile-is-in-the-work/ ↩

40. matt3o, “About MT3 profile and /dev/tty set,” matt3o.com, 2017. https://matt3o.com/about-mt3-profile-and-devtty-set/ ↩

41. matt3o, “MT3 keycap profile: a brief history,” matt3o.com, 2022. https://matt3o.com/mt3-keycap-profile-a-brief-history/ ↩

42. Boardsource, Equals 60 product page, archived. https://web.archive.org/web/2025id_/https://boardsource.xyz/products/equals-60 ↩

43. Modern Coup Make (a sleeve maker), 40% case for the Boardsource Equals 60, blog post. https://www.moderncoupmake.com/2024/05/22/40-case-for-the-boardsource-equals-60-ortholinear-mechanical-keyboard/ ↩

Linked from

• A keyboard case in build123d, checked before the vendor checks it
• auspex: an open 5x12 Hall-effect keyboard

Link graph: 2 docs link here, this page links to 5