NerdMiner – Submits Invalid Shares Below 1e-9

Confirm the symptom is `set_difficulty` related and not a wallet / WiFi / authorize problem. Open a serial monitor at `115200` baud. Look for the sequence: `mining.subscribe` -> `mining.authorize` -> `mining.set_difficulty <value>` -> `mining.notify`. If you see any `error`, `unauthorized`, `bad worker name`, or `invalid address` line, you are NOT on this page - jump to the related-errors block (`mining-authorize-invalid-credentials`, `nerdminer-taproot-bc1p-address-rejected`, `bitaxe-stratum-authorize-failed`).

Read the actual `mining.set_difficulty` value the pool sent. The serial line looks like: `[Stratum] mining.set_difficulty 0.0000000005`. Convert to scientific notation: `5e-10`. Compare against the NerdMiner firmware floor of `1e-9` (i.e. `0.000000001`). If the pool value is BELOW `1e-9`, you are in the right spot - the firmware is silently discarding every nonce because none can clear the local difficulty filter. If the pool value is AT OR ABOVE `1e-9`, the issue is elsewhere - check `nerdminer-no-block-templates-received` or `nerdminer-public-pool-io-shares-stalled`.

Switch to a NerdMiner-friendly solo pool. The fastest fix is to leave whatever vardiff Stratum farm pool you were on and point the NerdMiner at a pool whose difficulty floor matches NerdMiner-class hashrate. Boot into the NerdMiner setup AP (`NerdMinerAP_XXXXXX` or `MineYourCoins`), open `http://192.168.4.1`, change pool to `public-pool.io:21496` (community solo) or `solo.ckpool.org:3333` (Con Kolivas solo), set Bitcoin payout to your `bc1q...` SegWit address (NOT a `bc1p...` Taproot - both pools reject Taproot). Save, reboot. Public-Pool.io accepts shares down to `1e-9` and has been tuned for NerdMiner / Bitaxe-class workers since 2023. solo.ckpool.org has accepted ultra-low-difficulty workers since the original Bitaxe launched in 2022.

If you must stay on your current pool, set a static minimum difficulty using the worker-name suffix trick. Most Stratum-V1 pools accept a difficulty hint appended to the worker name in the form `<address>.<workername>+<diff>` or `<address>.<workername>;d=<diff>`. Common forms: `bc1q...address.nerdminer1+1e-9`, `bc1q...address.nerd_lobby;d=1e-9`, `bc1q...address.worker/d=0.000000001`. Pool-specific - check the pool's docs page. NiceHash, OcEAN, F2Pool, ViaBTC, and Ant Pool all support some form of static-diff suffix; the syntax varies. Set the minimum at exactly `1e-9` or slightly above (`1e-8`, `1e-7`) to give yourself headroom.

If you self-host the pool, configure the floor server-side. On Public-Pool.io self-hosted (Node.js fork from `benjamin-wilson/public-pool`), set `min_difficulty` in `.env` or `config.json` to `0.000000001` and restart the daemon. On a private CKPool instance, edit `ckpool.conf` -> `"mindiff":1` is the field; on solo CKPool the default is already low enough. On stratum-mining.py forks, edit `MERGE_CONF['vardiff']['minimum_difficulty']`. The setting names differ by pool flavor, but the principle is identical: the firmware's `1e-9` floor must be at or above the pool's `min_difficulty`.

Verify share submission resumes. After any pool / config change, watch for 30 minutes. Expected behaviour on a healthy 50-90 kH/s NerdMiner against a `1e-9` pool target: roughly **1 share every 5 to 15 minutes**, sometimes longer due to Poisson variance, but the pool dashboard's `last share` timestamp must update within an hour. If after 30 minutes you still see zero shares submitted, return to Step 2 and re-read the actual `set_difficulty` value - some pools advertise a static diff in their docs but apply vardiff anyway.

Cross-check with a known-good reference workload. Disconnect the NerdMiner. Point a second device (a Bitaxe at ~1 TH/s, a phone CPU miner, anything that submits to the same stratum endpoint) at the same pool / address / worker name. If the Bitaxe submits shares cleanly, the pool is healthy and your problem is the difficulty floor mismatch - finish Step 3 or Step 4. If the Bitaxe also submits zero shares, the pool itself is broken or your worker name / address is rejected silently - escalate to `mining-authorize-invalid-credentials` or `bitaxe-stratum-authorize-failed`.

Check NerdMiner firmware version and whether your build has the share-discard fix. The 1e-9 floor itself is by design and exists in every BitMaker-hub/NerdMiner_v2 build to date. What changed across versions is the *visibility* of the discard event in the serial log. v1.7.x and later print `Share below pool target` lines to serial when this happens; v1.6.x and earlier silently dropped the share. If you cannot see why your shares are missing, flash the latest stable release from `bitmaker-hub.github.io/diyflasher` (matching your variant - see `nerdminer-boot-loop-after-web-flash` for variant-matching warnings) and re-read the serial log.

Audit your pool's vardiff retarget interval. Some pools (NiceHash, large industrial farms) retarget downward aggressively for a worker submitting fewer than 1 share per 30 seconds; their floor is calibrated for TH/s-class workers and the floor often sits at `1e-12` to `1e-15`. NerdMiner at 50-90 kH/s is **billions of times slower** than the lowest worker class those pools were designed for. The right move is to leave that pool entirely - no amount of static-diff hacking makes a 50 kH/s lottery box well-matched to a difficulty floor designed for a 100 TH/s S19. Use a NerdMiner / Bitaxe-aware solo pool instead.

If you are running the NerdMiner against a self-hosted Umbrel `public-pool` or BTC-RPC-Explorer instance, verify the daemon's `bitcoind` is fully synced and serving block templates. Run `bitcoin-cli getblockchaininfo` and confirm `verificationprogress` is `1.0` (or > `0.99999`) and `initialblockdownload` is `false`. Run `bitcoin-cli getmininginfo` and confirm `currentblockweight` is a non-zero value. If the daemon is mid-IBD or pruning, pool will accept a stratum connection but never serve a real `mining.notify` job, and there is nothing for the NerdMiner to hash on - looks identical to the difficulty-floor symptom from outside.

Sanity-check the NerdMiner Bitcoin payout address. Even if everything else is right, an invalid `bc1q...` payload, a malformed checksum, or a `bc1p...` Taproot address will cause some pools to silently accept the worker connection but discard every share. NerdMiners.com docs and `public-pool.io` both reject Taproot addresses currently. Verify your address opens cleanly in a SegWit-aware explorer (`mempool.space/address/<your-bc1q>`); if mempool.space shows it as a malformed query, regenerate from a fresh wallet and reconfigure the NerdMiner.

If shares now submit but block-found events don't credit, follow `solo-block-found-but-not-credited` for the verification checklist (block-height, payout address registered to coinbase, pool's notification settings). The 1e-9 floor only governs share submission; lottery wins are independent and credited from the coinbase output of the block your NerdMiner happens to win.

Open a fresh issue or comment on `BitMaker-hub/NerdMiner_v2` issue #97 if you have evidence of a pool that NerdMiner cannot work with even after Step 4. Include: pool name, observed `mining.set_difficulty` values, your firmware version, and serial log excerpts. The maintainer may add a per-pool override or expose the difficulty floor as a runtime config knob in a future release. The community-maintained fork space (matlen67, NerdMiner-S3) sometimes ships with a configurable floor; switching forks is a valid workaround if upstream cannot be made to play nicely with your pool.

If you want to lottery-mine with absolute determinism about share submission, run a private CKPool instance on your own Umbrel / Start9 / Raspberry Bitcoin node. You set `mindiff = 1` (CKPool's lowest), point the NerdMiner at it, and you control end-to-end. Solo block reward goes 100% to you (minus a small ckpool fee on the public solo.ckpool.org; zero on a private instance). D-Central's solo-mining education content covers this end-to-end - it is the Mining Hacker's path: own your pool, own your stratum, own your shares. No vardiff surprises.

Stop DIY and ship to D-Central if: shares are still missing after 60 minutes on `public-pool.io` or `solo.ckpool.org` with a verified `bc1q...` address, the NerdMiner serial output is gibberish or unreadable, the firmware will not flash with `esptool.py` even via UART, or you suspect the NerdMiner is hashing into a void due to a corrupted internal state we cannot debug from your couch. D-Central runs the unit on a bench fixture with Wireshark capturing the live stratum frames against a known-good reference pool, side by side with a healthy reference NerdMiner. We can prove in 20 minutes whether the issue is firmware, hardware, or pool-side configuration.

What D-Central does at the bench: re-flashes the latest stable NerdMiner firmware for your exact variant on a known-good 3.3V harness, points the unit at solo.ckpool.org with a test `bc1q` address, captures stratum traffic with Wireshark for 60 minutes, confirms `mining.submit` traffic against pool acks. If shares submit, the unit is healthy and the original problem was pool-side - we hand it back with a config sheet for your home pool. If shares still don't submit, we triage NVS corruption, swap the ESP32-S3 module if needed, replace the LCD if the pin map is wrong on a board variant we know, and re-burn-in for 24 hours before return. D-Central stocks every NerdMiner variant - we own the bench infrastructure for this exact diagnostic.

Ship safely to D-Central. ESD bag inside a padded bubble mailer or small rigid box. NerdMiners are physically tiny but the USB connectors are delicate. Include the USB cable and power brick you were using (rules out a bad cable), plus a written note: pool name and stratum URL, observed `mining.set_difficulty` values, firmware version, exact variant flashed, your `bc1q` payout address (so we test against a real config, not a placeholder), and the time window during which the pool reports zero shares. Canada Post small-packet or UPS Ground; D-Central HQ receives and opens within 48 hours on business days. Canadian / US / international all welcomed.