Engineering Implementation

This page is for implementation and integration engineers, covering the complete engineering process of the automated water meter reading solution — from selection and deployment to business integration.

1. Solution Scenarios

Field sites typically include meter rooms, sewers, pump rooms and stairwell meter boxes — damp, dark, off-grid with degraded signal; selection and deployment are organized around these conditions.

Typical Environments

Environment	Characteristics	NE101 response
Meter room / pump room	Damp, normally dark, some with mains power	Fill light for darkness; protective housing against condensation; higher capture rate where mains power exists
Stairwell meter box	Relatively benign, usually powered	Simplest installation; watch for space and tamper protection
Sewer / meter pit	Damp, completely dark, no mains, heavy signal attenuation	IP67 protection; battery life first; Cat.1 / HaLow backhaul
Outdoor meter position	Open air, wide temperature swings, possible flooding	IP67 protection; fill light for night; sun and freeze protection

Communication Options

NE101 supports three communication modules; choose by site signal and power conditions (battery life figures are official theoretical values, default 5 captures/day):

Option	Battery life (default / optimized)	Best for
Wi-Fi	2.39 yr / 6.20 yr	Meter within router coverage; short-to-medium range (recommended starting point)
Wi-Fi HaLow	1.46 yr / 4.30 yr	Long-range backhaul for remote or obstructed sites
Cat.1	0.83 yr / 2.08 yr	Direct cellular where no LAN exists; external power advised for high-frequency capture

Capture Frequency

• Frequency is driven by the billing / business cycle: 1–4 captures/day for daily settlement, lower for monthly
• Battery life scales linearly with frequency (1 capture/day ≈ 5× the life of 5 captures/day)
• Images upload with each capture — no extra configuration

High-Frequency Capture (Custom)

For minute-level or continuous capture, battery power no longer applies — contact CamThink for a custom Type-C powered NE101 (battery removed, powered directly over Type-C) for sustained high-frequency operation. For custom requirements, reach the technical support team via Contact Us; see also Technical Support.

2. Meter Types & Selection

Lens Selection

• NE101 is a fixed-focus module with two FOV lenses: 60° (nominal working distance 15cm) and 120° (nominal working distance 8cm)
• The lens must face the digit wheel squarely; keep the distance within the bracket's nominal range
• For older meters with small digit wheels, watch the framing ratio (digit area ≥ 1/3 of the frame)

Mounting Distance (60° / 120° FOV)

Both lenses are factory fixed-focus; the nominal working distance is the focus reference. The actual mounting distance need not match it exactly — small deviations are fine as long as the digits stay sharp in a test snapshot. Aim the lens at the digit wheel via the bracket so the digit area occupies ≥ 1/3 of the frame (1/2 recommended):

Lens	Nominal working distance	Frame coverage width	Best for
60° FOV	15cm	≈17cm	Standard residential / industrial meters, single-meter close-up (recommended)
120° FOV	8cm	≈28cm	Large dials, multi-meter overview

Principle: coverage width ≈ 2 × nominal working distance × tan(FOV/2). At 15cm the 60° lens covers ≈17cm; at 8cm the 120° lens covers ≈28cm. The higher the digit share of the frame, the denser the OCR pixels.

Mounting verification: take a manual snapshot before tightening the bracket — confirm sharp digits, no glare, proper framing — then lock the screws; if the distance deviates from nominal, fine-tune based on the snapshot.

Bracket Selection

• Use the official water meter bracket: it fixes the lens-to-dial distance, angle and view in one shot — the prerequisite for stable recognition
• No drilling; about 10 minutes per meter; re-installation after a battery swap never drifts

3. Bill of Materials (BOM)

Thumbnail	#	Item	Model / spec	Qty	Purpose
	1	NE101 AI Camera	Battery powered (4× AA) · scheduled capture · Wi-Fi/Cat.1/HaLow selectable	1 per meter	Dial capture
	2	Water meter bracket	Official NE101 accessory	1 set per meter	Fixes the lens-to-dial relative position
	3	NG4500 AI Box (platform host)	Linux host / NG4500	1	Runs NeoMind: ingest, OCR, rules, data egress
—	4	Batteries	4× AA	1 set per camera	Power supply
—	5	Wi-Fi HaLow gateway	As needed: HaLow backhaul scenario	1	Bridges long-range backhaul into the LAN
—	6	SIM (IoT) card	As needed: Cat.1 backhaul scenario	1 per camera	Cellular backhaul

Starter setup (≤10 meters): one NE101 (with bracket) per meter + one Linux host running NeoMind; as the meter count grows, add cameras only — the platform scales horizontally by capacity.

4. Network Topology (by Communication Option)

Network requirements differ by communication option — confirm them before deployment:

Option	Networking
Wi-Fi	NE101 and the NG4500 (running NeoMind) join the same LAN; the host needs no public internet — ideal for campus / community intranet deployments
Wi-Fi HaLow	Requires a HaLow gateway: NE101 → HaLow gateway → the network where NeoMind lives; for long range and obstructed sites
Cat.1	NE101 connects directly to the cloud over cellular; we recommend deploying NeoMind in the cloud (public reachable) — for scattered meters without a LAN

General requirement: NE101 must be able to route to NeoMind's MQTT port (built-in broker, TCP 1883 by default, MQTTS supported — the host must open this port to the device subnet); recognition and ingestion complete locally on the host, and business systems integrate via OpenAPI / Data Push / Webhook.

5. Solution Setup

Build in this order: install NeoMind (NG4500 / PC) first, then add the NE101 device on the platform, then configure the device to report data, and finally commission and integrate.

5.1 Install NeoMind (NG4500 / PC)

The NG4500 AI Box is the recommended host; for evaluation, NeoMind can also run on a PC or any Linux machine.

• Platform install: one-line script / manual deployment / HTTPS setup — see Install & Upgrade
• Install the OCR extension: one-click install paddle-ocr-v6 from the extension marketplace — the local OCR engine that turns NE101 dial captures into readings
• Verify the camera component: confirm the built-in ne101_camera component is available — it receives NE101 uploads and feeds captures into the recognition extension

For detailed steps see Install Extensions & Components.

5.2 Add the NE101 Device in NeoMind

On the platform side, add the NE101 in the ne101_camera camera component and bind it to a project; note the MQTT details at the top of the component — Server Address and Data Reporting Topic — you will need them for the device-side configuration. Full steps: OCR Use Case — Capture Images from NE101.

5.3 Report Data from the NE101

1. Long-press the NE101 shutter button for 2s to enable the device Wi-Fi AP; connect from a laptop or phone
2. In the NE101 Web UI → System Settings → Communications, select the site router's Wi-Fi and make sure the device can reach the NeoMind host
3. In Application Management, fill in the Data Reporting Topic and Server Address (the MQTT details noted in 5.2), then click connect
4. From now on, every press of the shutter button uploads the image to the bound project

For device activation and Web UI configuration, see the NE101 Quick Start.

5.4 Commissioning

Verify link by link; if one fails, troubleshoot that link first:

1. Capture: press the NE101 shutter manually and confirm the image reaches NeoMind (visible in the component's image list)
2. Recognition: confirm the OCR pipeline produces the reading field (visible in the dashboard component)
3. Rules: confirm the reading is parsed into a number by Transform and stored as the meter_reading metric (see Data Transforms)
4. Egress: verify Data Push / Webhook delivers readings to the business endpoint

5.5 Data Storage & Display

• Capture photos: stored on the NeoMind host; retention is controlled in the data retention settings and auto-purged on expiry
• Reading metrics: ingested as virtual metrics (e.g. meter_reading) with history queries
• Dashboard: three recommended cards — reading card (latest reading + time), consumption trend (day/week/month), device health (battery / signal / online); see Using Dashboards

5.6 Data Forwarding

Two outbound options for readings and alerts — choose by latency needs:

Option	Description	Best for
Data Push	The platform pushes new data points of selected metrics to a business HTTP endpoint in real time (with retries)	Real-time linkage, big screens
OpenAPI pull	The business system pulls readings via REST API per settlement cycle	Daily / monthly settlement systems

• Configuration: Data Forwarding / Platform API
• Alerts such as abnormal readings, low battery and offline devices go through the Notifications Webhook / IM / email channels

Field mapping example: meter no. ↔ device ID; reading ↔ meter_reading; reading time ↔ data point timestamp; evidence ↔ capture photo URL.

6. Technical Support

• Community: Discord / GitHub Discussions
• Solution customization & volume deployment: Contact Us — handled by the CamThink technical support team
• High-frequency capture / Type-C powered version: same as above; specify the capture frequency and deployment scale in your request
• Bracket customization: for non-standard meters or constrained mounting positions (size limits / multi-meter sharing), same as above; specify the meter dimensions and site photos