Recommendations and the Bench — What to Buy, What to Decide, and Nine Measurements

6.1 The Plan in One Page

Nothing here changes a decision already made. The first gun runs from the GST60A24-class brick (D14, D16, D17); the connector at the gun is one GX (D31); the first build is printed (D35). PD enters as a second supply option, tried on the bench first.

1. Contract. 20 V fixed, 3 A, from a charger of 60 W or more, for a gun that should run from an ordinary laptop charger; EPR AVS at 24.0 V, for a gun that should behave exactly as it does on the brick (Volume 2). 28 V fixed is rejected on the D14 rail because it sits on the D33 clamp.
2. Coil. If 20 V is to be supported, the coil chosen on the Solenoid Recoil bench must keep a forcing ratio of 1.5 at 20 V: a hot I·R of no more than about 12.7 V typical and 11.2 V worst case at 2 A, that is about 5.6 to 6.4 Ω hot. The Ledex B14’s 6.9 Ω winding does not (Volume 2, Section 2.2). This is a criterion to add to the D15 ranking, not a new purchase.
3. Where. In a box at the far end of the gun’s cable, with the gun and its GX unchanged (Volume 5, architecture A).
4. Chip. ST STUSB4500 for the 20 V box, configured to connect its output only on a 20 V contract; Diodes AP33772S for AVS at 24 V, with a small host. TI TPS25751D if PD later moves into the gun on the custom board (Volume 3).
5. Front end. In every case: at most 100 µF on VBUS, a soft-start switch that blocks reverse current, then the SMBJ28A and the 1000 µF reservoir (Volume 3, Section 3.5).
6. Never trust the label. The rail must not be connected on anything below the accepted contract, whether by the chip’s own fallback rule or by an under-voltage lockout (Volume 3, Section 3.4).

6.2 Proposed Decision Rows

For Light-Guns/DECISIONS.md. All are Proposed; none is Jeff’s call yet.

Table 1 — 6.2 Proposed Decision Rows

#	Choice	Candidates	Proposed	Why, and why not the others	Vol
UP-1	PD contract for the rail	15 V; 20 V fixed; PPS 21 V; 20 V + boost; EPR AVS 24 V; 28 V; 36/48 V	20 V fixed (ordinary charger) and EPR AVS 24.0 V (brick-equivalent)	20 V: universal above 45 W, one stage fits under 3 A; AVS 24 V: the brick’s forcing and clamp margin. PPS optional in chargers; boost exceeds 3 A at 2.5 A; 28 V up to 29.9 V at the sink, over the SMBJ28A stand-off; 15 V below the window	2
UP-2	Coil criterion for PD	ignore PD; add a PD criterion	Add to the D15 ranking: k ≥ 1.5 at 20 V typical (hot I·R ≤ about 12.7 V at 2 A)	The B14 6.9 Ω winding gives 1.37 at 20 V; a coil chosen for the brick alone may not suit PD	2
UP-3	Where PD lives	A box at the cable’s far end; USB-C at the base plate; captive cable	A: far-end sink box, D31 unchanged	No change at the gun; brick and PD interchangeable; B and C need a host that sources 20 V or a PD source box	5
UP-4	Sink controller, 20 V box	STUSB4500; CYPD3177; HUSB238A; IP2721; CH224K	STUSB4500	Exact-match request and output only on the configured contract; no D+/D−; English data sheet; SparkFun board. CYPD3177 the alternative; HUSB238A and IP2721 fall back to lower voltages; CH224K unavailable	3
UP-5	Sink controller, AVS 24 V	AP33772S; HUSB238A; CH224Q/A; TPS26750	AP33772S	Certified EPR AVS to 28 V, protected N-FET driver, English data sheet; HUSB238A the alternative; CH224 no switch, Chinese-only manual; TPS26750 overkill	3
UP-6	Sink controller, in the gun (Stage 2)	TPS25751D; STUSB4500 + discrete switch	TPS25751D	Integrated 20 V 5 A switch with reverse-current protection, OVP, UVP and slew control; TPS25750 is NRND	3
UP-7	Bench trigger boards	SparkFun DEV-15801; Adafruit 5807; marketplace decoys	Buy one DEV-15801 ($31.92) and one 5807 ($5.95)	DEV-15801 is the box prototype; 5807 reads charger capabilities over I2C; decoys only after the five checks	4
UP-8	PD analyser	buy TI-PD-ANALYZER; buy a ChargerLAB KM003C; ADP3450 + sigrok	Start with the ADP3450 and sigrok’s usb_pd decoder	No purchase; the lab already has the instrument; TI’s analyser is an EVM accessory and its price was not shown; KM003C’s price was not confirmed on a seller’s page	6

6.3 What to Buy for the Bench

Table 2 — 6.3 What to Buy for the Bench

Item	Why	Price (date, seller)
SparkFun Power Delivery Board, DEV-15801	The STUSB4500, configured for 20 V only; the first cable-end box	$31.92, 2026-10-19, SparkFun1
Adafruit 5807 USB-C PD Dummy Breakout	Capability reader over I2C from a Pico	$5.95, 2026-10-19, Adafruit2
One USB-C cable, CC2-3 class, about 1 m	Charger to box; a known cable for test T7	price at order time
One EPR (CC2-5E) cable	Only if test T1 finds a charger that grants AVS	price at order time
A 140 W EPR charger	Only if UP-1’s AVS route is pursued and no charger in the house has EPR	Anker 717 listed “Sold Out” in the EU on 2026-10-19; price at order time

The household’s existing laptop and phone chargers are the test population for T1; none needs buying.

6.4 Seeing the CC Wire: Analyser or ADP3450

Most of what goes wrong with PD happens on the CC wire, so the bench needs some way to see it.

TI-PD-ANALYZER. TI sells a PD analyser board that logs CC1 and CC2 traffic and reports VBUS and CC voltages and currents, decoding to PD revision 3.2 v1.1, with a Windows GUI (version 0.9.0.0, May 2025). TI’s own page calls it “intended to be used with a paired EVM and not licensed for use as a stand-alone product”, and showed no price on 19 October 2026.³

ChargerLAB POWER-Z KM003C. A widely used USB-C meter with PD protocol capture; retail listings describe it as 50 V and 6 A capable, covering EPR. Its price was not confirmed on ChargerLAB’s or a retailer’s page for this dive (the pages refused automated reading); one undated listing showed $220.88.

The ADP3450 and sigrok. BMC on CC is a 300 kbit/s, roughly 1.1 V signal, well within what the Analog Discovery Pro captures on an analog channel or, thresholded, on a digital one. sigrok’s protocol-decoder library includes a supported USB Power Delivery decoder (usb_pd), which takes a logic input.⁴ A capture exported from WaveForms and decoded with sigrok-cli gives the same message list an analyser shows, after the fact rather than live. Tapping CC means a breakout between charger and sink: a USB-C pass-through breakout board with CC brought out to pads, or the SparkFun board’s own receptacle pins.

Recommendation (UP-8): start with the ADP3450 and sigrok. It costs nothing, it puts the CC messages and the rail on the same time base (which a standalone analyser does not), and it is enough for every test below. Buy a dedicated analyser only if live decoding turns out to matter, for instance while bringing up the AP33772S.

6.5 Bench Setup and Safety

• Probing. The rail and VBUS run to 21.5 V on a 20 V contract and to 29.9 V on a 28 V one; the coast lifts the reservoir by another 2 to 3 V. Use ×10 probes on every power node above 20 V (the LiPo Batteries dive notes the ADP3450’s direct range does not cover a full 6S pack either). CC is a low-voltage signal and can be probed directly, but it must never be shorted to VBUS: a CC-to-VBUS short at 20 V is exactly the fault every chip in Volume 3 protects its own pins against.
• Grounding. Chargers are floating; the ADP3450 is earthed through the PC. The same rule as Power Supplies Vol 5 applies: one ground reference, at the gun’s board, and no second path through a probe ground clip.
• The coil. Run the regulated stage from the bench carrier, current limit set by its VREF divider (Reference Electronics Vol 5), never a bare coil on a charger: a charger’s over-current response is unknown until test T3 has measured it.
• Hot-plugging. The GX rule still holds (D31): mate the gun’s connector, then plug the charger into the box.

6.6 The Tests

Table 3 — 6.6 The Tests

#	Test	Method	Pass / what it decides
T1	Capability survey of every charger in the house	Adafruit 5807 in I2C mode from a Pico; print every object, including PPS and AVS; repeat with an EPR cable if one is available	Which chargers offer 20 V at 3 A, PPS to 21 V, EPR and AVS. Decides whether UP-1’s AVS route is reachable without buying a charger
T2	Voltage at the sink under a shot	SparkFun box at 20 V, bench carrier, chosen coil; ADP3450 on VBUS at the receptacle and on VM, ×10	Minimum VM during the rise; compares with the 17.75 V worst case. Also measures how far real chargers sit above nominal
T3	Charger response to the shot	As T2, set point stepped 1.0, 2.0, 2.5, 2.8 A, single shots then OpenFIRE’s default autofire (45 ms on, 80 ms off)	Does any charger trip, fold back or reset below 3 A? Above it? This is the PD equivalent of Power Supplies’ trip-band test
T4	Forcing ratio in practice	Same shot from the brick and from the PD box; IPROPI and OUT2 as in Reference Electronics Vol 5	Arrival time at 20 V against 24 V; checks Volume 2’s 1.37 / 1.59 / 1.91 for the coil actually bought
T5	Fallback	SparkFun box (20 V only) and one marketplace decoy on a 30 W and a 45 W charger	SparkFun: no output. Decoy: what voltage, and does its output switch on? Confirms Volume 3, Section 3.4
T6	Hard reset	Force a reset (pull and reinsert the charger, or send one from an analyser if bought); capture VBUS, the switched rail and the reservoir	Rail must go to zero and return only at 20 V; the stage must not fire from a partial rail
T7	Inrush and the 100 µF rule	Soft-start switch in front of the 1000 µF reservoir; current probe or shunt in VBUS during contract completion	Peak inrush under 3 A; the charger does not reset. Without the soft start, expect it to
T8	Coast energy	Capture the reservoir and VBUS at the end of a shot	Reservoir rises about 2.2 to 2.5 V; VBUS must not, which proves the switch blocks reverse current
T9	AVS at 24 V (if T1 finds an EPR charger)	AP33772S evaluation board or a breakout with a Pico host; request 24.0 V; repeat T2 to T4	Whether real EPR chargers grant AVS, and whether the gun on AVS is indistinguishable from the brick

6.7 The Numbers This Dive Is Waiting For

• Real chargers’ 20 V under load (T2): how far above 17.75 V the worst real case sits. If every charger tested holds 19.5 V or more under a shot, Volume 2’s coil criterion relaxes.
• Charger over-current behaviour (T3): the specification says nothing about what a source does above its operating current except that peak-current windows end at 10 ms; each charger’s answer is its own.
• AVS in the field (T1, T9): whether EPR chargers bought in 2022 to 2024 actually grant an AVS request. The 2023 capability lists show none.
• The coil (D15 bench): its hot resistance at its set current decides whether 20 V PD is a supported input for this gun at all.

Until then the dive’s position is the one in Section 6.1: the brick for the first gun, a PD box on the bench beside it, and the decision on PD left to the measurements.

Sources

Footnotes

1. SparkFun Electronics, DEV-15801, https://www.sparkfun.com/products/15801 (read 2026-10-19). ↩

2. Adafruit, product 5807, https://www.adafruit.com/product/5807 (read 2026-10-19). ↩

3. Texas Instruments, TI-PD-ANALYZER, https://www.ti.com/tool/TI-PD-ANALYZER (read 2026-10-19). ↩

4. sigrok, “Protocol decoders”, https://sigrok.org/wiki/Protocol_decoders (read 2026-10-19): “USB Power Delivery”, id usb_pd, input logic, status supported. ↩