HP 1500W Server PSU Breakout Overcurrent OVP Trip

Confirm your PSU revision off the side label - `HP DPS-1500EB A`, `DPS-1500FB A`, `HSTNS-PL11`, `HSTNS-PR16`, `HSTNS-PD19`, or similar. Pull the canonical pinout PDF from a reputable surplus-PSU vendor (Parallel Miner / Deep In The Mines publish per-PN documents). The 1500 W family pinout is similar to the 1200 W family but I_share pin numbering and the I2C / PMBus pins are different - don't assume cross-compatibility. Spend 5 minutes on this step before touching any wire and you save a Bitaxe-shaped repair bill later.

Inventory your breakout's PCIe DC output ports. Most generic HP 1500 W breakouts ship with 4-8 PCIe `6+2`-pin connectors. Count them. Each connector is rated `~25-30 A` continuous (PCIe spec - `8 A` per pin, three `12 V` pins, with thermal derating). The PSU itself outputs `~125 A` total. The math: `125 A ÷ 5-8 ports = 16-25 A per port`. Concentrating load on fewer ports than the PSU's headroom allows is asking for OCP.

Redistribute load across multiple PCIe outputs. S9 hashboard with `~70 A` peak: split across 3 PCIe outputs minimum, `~23 A` per output, comfortably inside spec. Multi-Bitaxe rig: one Bitaxe per PCIe output, period. Bitaxe Hex on XT30: still one output per Hex even though Hex's nominal draw is only `~9 A` - the inrush spikes are what matter. Use the breakout's full output count. This is the single highest-impact 'no soldering, no parts' fix on the page.

Verify the sense lead is mechanically secure. PSU off. With breakout open, give the `+12V_SENSE` and `12V_SENSE_RTN` leads a gentle tug at both ends - at the back-edge connector pad and at the load termination. Anything that wiggles or pulls free needs strain-relief - hot-glue the leads to the breakout PCB and to the load entry point. Sense leads break = OVP trigger; mechanical security is non-negotiable.

DMM-test enable signals before re-enabling. PSU plugged in, AC switch on, no load. DMM common probe on `COM`, test probe on `PS_ON`: must read `0.00 V`. Same for `PS_KILL` / `PRESENT`. Anything above `0.1 V` = breakout's enable jumper is via a resistor instead of a hard short. Replace with `0 Ω` jumper before continuing.

Strain-relieve the sense leads at both ends with a soldering iron. Open the breakout. Solder `28 AWG` silicone-jacket wire-wrap leads onto the `+12V_SENSE` and `12V_SENSE_RTN` pads. Add a hot-glue blob over each solder joint to lock the lead mechanically. Run the leads alongside (but not bundled with) the heavy `12 V` cables to the load - bundled, the magnetic field from the heavy cable can induce noise on the sense lead. Terminate at the load's input terminals (Bitaxe barrel-jack solder pads, S9 hashboard 6-pin), again with hot-glue strain relief at the load end.

Confirm sense routing with a scope. Scope on the `+12V` bus terminal of the breakout, `1 V/div`, `100 ms/div`. Enable the PSU under partial load (`30-50 A`). Healthy = clean line at `12.1-12.2 V`, no excursions above `12.3 V`. If you see `13 V`+ peaks, sense-pin is still oscillating - re-trace the sense lead, check for partly-seated connectors on either end, re-strain-relieve. If you don't own a scope, borrow one - a `~$300` Rigol DS1054Z pays for itself the first time it diagnoses an OVP event correctly.

Add bulk capacitance at each load. A `2200 µF` low-ESR electrolytic across each load's input terminals (Bitaxe barrel jack, S9 hashboard input) absorbs transient current spikes that the cable + breakout combo can't respond to in microseconds. The 1500 W HP PSU's loop bandwidth is finite; bulk caps fill the gap. Cost: `$3` per cap. Eliminates 60-80% of 'PSU drops out during high-difficulty work' complaints in our repair queue.

Upgrade individual cable runs to appropriate gauge per-port. Each PCIe DC output should feed a load through `12 AWG` minimum for runs under `30 cm`, `10 AWG` for longer. Silicone-jacket wire only - PVC stiffens and cracks at `60+ °C` connector temps. Crimp ring lugs with a real ratcheting crimper, not pliers. Pull-test each crimp at `25 lb` minimum.

For S9-class loads, derate the breakout output with a soft-start. Even with load split across multiple PCIe ports, an S9 hashboard pulls violent inrush at startup that can momentarily exceed each port's continuous rating. Add a small `12 V` MOSFET-based soft-start circuit (`IRFB7430` + RC delay = `~$8` of parts) between the breakout and the hashboard input. Limits inrush to `~50 A` ramp over `100 ms` instead of `200+ A` for `<5 ms`. Dramatically reduces nuisance OCP trips on S9 cold-starts.

For multi-PSU rigs, bond `I_share` between all PSUs. Two HP 1500 W units paralleled for a `200+ A` build (multi-S9, large Space Heater) must have their `I_share` pins tied together with a short, low-impedance lead - `16 AWG` is plenty since I_share carries an analog reference voltage, not bulk current. Generic breakouts leave this floating. Without the bond, the PSUs fight each other on the bus and one trips OVP / the other trips OCP within seconds. Consult the HP service note for the exact PN - DPS-1500EB A and HSTNS-PL11 differ on I_share pin numbering.

For paralleled rigs, also bond the sense leads to a single common load point. Both PSUs' `+12V_SENSE` leads tie to the same physical point on the load - typically the bus terminal where the heavy cables converge into the load. If each PSU senses at a different point, they regulate to different references and fight each other into OVP-OCP cycles. One sense point. Both PSUs reference it.

For Bitcoin Space Heater enclosures, design airflow before sealing. The HP 1500 W brick needs `~80 CFM` of intake at full load. In an enclosed build, the PSU intake must pull from a different airstream than the miner's exhaust - otherwise the PSU is sucking `60+ °C` air, derating internally, and tripping thermal-OCP. Use a separate intake duct for the PSU or mount it on the cool-air side of the airflow path. IR-thermometer the PSU shell at `5 minutes` of full-load operation: must stay under `55 °C`. If you can't hit that, redesign before sealing.

For DCENT_OS users on Antminer S9 / S9-class builds powered by HP 1500 W breakouts: DCENT_OS is D-Central's open-source Antminer firmware exposing per-chip HW% and rail-voltage telemetry on the Antminer side. It surfaces the rail voltage seen by the hashboard in real time, which is exactly the diagnostic you want when chasing intermittent breakout-side trips. Stock Bitmain firmware shows nothing useful at this layer. DCENT_OS is Antminer-only - does not run on Bitaxe. For Bitaxe rail diagnostics use AxeOS's `Vin` readout. (DCENT_OS for Whatsminer / Avalon is on D-Central's roadmap but not yet shipping.)

Add a `30 A` automotive blade fuse on each PCIe DC output between the breakout and the load. `$2` of insurance against a Bitaxe-side short blowing back into the breakout. Fuses blow faster than the PSU's OCP, isolate the failure to the affected load, and let the surviving Bitaxes / hashboards keep running. Mining Hacker discipline says: protect every branch.

Document the build before re-energizing. Photograph the wiring, write down which PCIe output feeds which load, record the sense-lead routing path. Two months later when troubleshooting an unrelated issue, this saves you hours. The documentation is the cheapest diagnostic tool you'll ever own.

Stop and consult D-Central if a sense-pin OVP event already destroyed a Bitaxe or fried a buck regulator on a hashboard, you're planning a paralleled-PSU build for a Space Heater or multi-S9 rig and want the wiring vetted before powering up, or the breakout PCB shows visible browning / lifted pads / arc-tracks. The safe path is replace, not rework. D-Central's ASIC Repair Service covers Bitaxe and hashboard damage from OVP events; the shop carries server-PSU breakouts vetted for sense-pin routing and per-port current handling. Book at https://d-central.tech/services/asic-repair/ - free quote, parts at cost, return shipping included for North American customers.

Decide repair vs rebuild for the breakout. If the breakout has any visible thermal damage, replace it. The '$25 cheap breakout that took out a $190 Bitaxe' math is unforgiving. Quality breakouts (Parallel Miner, Deep In The Mines, Mining-Heaven, D-Central-stocked) are `$45-65 CAD` for the 1500 W form factor and get the per-port current handling, sense routing, and PCB copper right. Document the failure mode for warranty claims, then upgrade.

Consider the turnkey alternative for new high-current builds. Mean Well RSP-1500-12 is the alternative for the 1500 W regulated 12 V slot - proper screw terminals, certified continuous-duty, no signal-pin gymnastics, `~$185 CAD` new. The HP-1500 W-on-breakout approach is for the budget-conscious DIY operator who enjoys repurposed-enterprise-hardware builds and accepts the 1-2 hours of breakout rework that comes with it. Both paths work; pick the one that matches your time-vs-money trade.