IRI Filler Pro - Drive Performance Requirement (drive-neutral)
Confidential - under NDA - YAVU - 2026-08-10
This is the governing requirement for the drive. VOICE_COIL_DRIVE.md is subordinate to it.
You choose the drive. We specify the result.
The existing manual IRI device is the reference standard, and YAVU supplies a physical reference unit. The electric device must deliver the same injection pressure and the same jet performance as that unit. Any drive principle that demonstrably achieves this is acceptable.
Acceptance is decided by a side-by-side test against the physical reference unit, not by matching numbers on paper.
| Requirement | |
|---|---|
| Injection pressure / jet | Equivalent to the manual reference unit. Penetration depth in a standardised gel or skin surrogate within ±10 % of the reference, measured side by side, same day, same chamber, same product. |
| Delivered volume | Within ±10 % of the reference at 1, 5 and 10 units (shot weight vs commanded volume). |
| Dosing | 1-10 units at 6 µl. Mechanical hard stop at 60 µl per shot. Chamber 0.3 ml = 5 full shots. |
| Reference figures | ~2.25 J / ~354 bar / ~265 m/s — orientation only, not acceptance criteria. Taken together they imply ~94 % end-to-end efficiency, which is not achievable, so at least one of the three is not exact. The physical unit is the standard; a measurement that clarifies which figure is reliable would be welcome. |
| Ready / re-cock | Fast ≤ 1.5 s, quiet ~2.5 s from shot to ready. |
| Noise | ≤ 45 dB(A) at 30 cm, quiet mode. |
| Envelope | Handpiece Ø 35-40 mm preferred; a justified deviation is acceptable. |
| Safety | No shot while docked; stored energy safely held with power removed and released under control; no dry fire without a chamber. See HAL_SPEC.md §4b. |
Drive options
Update 10 Aug 2026 — our preference has changed. The measured data of the reference device is now published in
MEASURED_DATA.md, and with it the voice-coil path is far more feasible than earlier revisions of our documents suggested: 360 N over a 5.90 mm stroke, not 563 N over 3.8 mm. YAVU would prefer the voice-coil direct drive if you can build it at a sane size and cost — it removes the cocking noise entirely and allows a programmable shot. The motor-cocked spring remains the lower-risk fallback. Please quote both.
A - Motor-cocked spring (v1.3). Lower risk, our fallback. Keeps the proven firing system and motorises only the cocking action. Injection pressure is equivalent by construction — the same spring doing the same work. Cumulative plunger advance across the five shots of one chamber is already solved mechanically. Lowest technical risk and lowest NRE.
B - Voice-coil direct drive. PREFERRED, if feasible. No spring, no cocking step, no
cocking noise, programmable bi-phasic injection. Force and stroke are now known from the
measured geometry — see VOICE_COIL_DRIVE.md rev. D. What remains is the actuator class
and an indexing mechanism for the cumulative plunger advance. The control firmware for this
path is already written and tested.
C - Any other principle. Pneumatic, cam, gas spring, hybrid. If equivalence per the table above and the safety requirements are met, we are open to it — tell us early so the HAL interface can be aligned.
The firmware does not constrain this choice
/core is drive-agnostic and selects the path automatically at runtime: if
halVoiceCoil() returns a driver, the direct-drive path runs; if it returns nullptr, the
spring path runs. Switching between A and B needs no change to core. Both paths are
covered by the host test suite (tools/host_test.sh), including a build with no cocking
motor present at all.
If you choose principle C, the work is a new HAL interface plus a firing controller
alongside core/vc_shot — tell us and we will write it.
Open physical questions (measurements, not blockers)
None of these prevent quoting path A today.
- ~~Ampoule bore has never been measured.~~ Resolved — see
MEASURED_DATA.md. Plunger Ø 3.60 mm, effective area 10.18 mm², nozzle 0.18 mm, all measured. Force 75–360 N over a 5.90 mm stroke. The dosing chamber is the nozzle, it is proven, and it does not change. - Multi-shot advance. 300 µl chamber, ≤ 60 µl per shot: the plunger must advance cumulatively over five shots. A voice coil returns to centre and needs an indexing mechanism that path A already has.
- ~~Actuator size makes path B impractical.~~ Withdrawn. That estimate used an assumed geometry and continuous force ratings. A 5.90 mm stroke at 1.25 % duty is the favourable regime for a voice coil — short coil, high force density.
- Nozzle discharge coefficient sets delivery timing. Exposed as a calibration value written to the device at bench time; no firmware rebuild.
- The firing pressure is the last open number — between 74 bar (measured spring, static) and 354 bar (simulated). One measurement on the reference unit closes it and will probably let us lower the force target. This is the single most useful result your bench test can deliver.
Source: firmware repo docs/DRIVE_REQUIREMENT.md · confidential · NDA