What datasheets leave out
The method behind every page here: why a number without a measurement boundary is not a specification — starting with one of ours we withdrew.
Compare
Comparisons, with
our own row held to
the same standard.
Comparison pages are usually written to win. These are written to be checkable: every figure attributed to another manufacturer names its source and retrieval date, and our own figures name theirs.
The comparisons
Start with the method.
The first page is the one worth reading before the others — it sets out what a comparable specification actually requires, and applies it to us first.
Moog
The same architecture at a completely different scale. What an incumbent's full-cabin motion base and a single-seat instrument do and do not have in common.
D-BOX
Different categories, not different grades. A haptic actuator array under a seat and a parallel platform moving the whole rig are answering different questions.
Head to head
Four platforms, four different questions.
Qubic QS-S25
The closest like-for-like in this category — both real 6-DOF parallel kinematics — and the places the two genuinely diverge.
MOZA HMA150
Shared intent and one shared spec, different kinematics. Three axes on four posts against six axes on six legs.
DOFReality H6 / P6
Both ship as kits you build yourself. The difference is what is inside the actuator.
Our own row
The row we ask you to check.
Every page in this section puts the PDK in the right-hand column. Here it is on its own, with the basis of each figure written out — so the row can be audited without reading six comparisons first.
| ARK PDK | Basis | |
|---|---|---|
| Architecture | 6-DOF Stewart–Gough, six direct-drive legs | Patent CA 3217174 C |
| Actuation | Direct-drive BLDC outrunners, field-oriented control | Design of record |
| Static load | Carried pneumatically, 3 × 50 mm × 140 mm cylinders at 6.0 bar max | Published specification |
| Envelope | Surge 158.23 mm · sway 143.85 mm · heave 117.35 mm · pitch 16.07° · roll 17.36° · yaw 19.60°, single-axis | T1 firmware boot banner, envelope of record |
| Payload | Up to 250 kg | Published specification |
| Feedback | 6 × 21-bit magnetic encoders, 0.000171661° per count | verified by arithmetic |
| Control rate | 8 kHz drive loop per motor · 3.3 kHz inverse kinematics · 1 kHz position commands — all internal | verified |
| Transport delay | Under 1 ms — command issued to actuator moving; end-to-end path published on /technology/latency/ | boundary stated |
| Power | 3 kW universal PSU, 100–290 V single-phase domestic mains · 30 kW burst from the capacitor bank | verified |
| Footprint | 1.00 × 1.00 m platform · 1.20 × 1.00 m PDK Racer · 0.50 m retracted | verified |
| Price | $4,899.99 platform · $6,299.99 PDK Racer · $8,499.99 Ecosystem | published, not quoted |
| Network | Direct, plus a community. No dealer network. | Stated |
What the labels mean
- verified — checkable against a primary source in this repository: firmware, arithmetic, or a measured drawing
- Competitor figures are read from published manufacturer material, named to their source. We do not re-measure another company’s product, and we do not characterise anyone’s machine as worse house rule
The rules we wrote them under
Sourced, dated,
and reversible.
Every figure attributed to another manufacturer is read from their published material and named to its source. Manufacturers revise specifications and a figure here can go stale — if you find one that has, tell us and it gets corrected. We do not characterise anyone's product as worse; we state what each machine is and let the architecture speak. Where a rival publishes something we do not, that is on our row too.
Keep reading
The engineering
Our own architecture, ten pages, one subject each.
Latency
The figure we withdrew, and why.
Contact
Found an error in a competitor figure? Tell us.