Writing User Manuals

How to Read Electrical Wiring Diagram of a Control Panel ? When you’re looking for a motor (e.g., MTR-101) inside a large electrical diagram, it can feel like searching for a needle in a haystack. But diagrams are designed with a navigation system to make it easier: ✅ 1. Use the Motor Tag Every motor has a unique ID or tag (like M1, MTR-101, or P-201). This tag will appear in the power circuit (showing the motor connection) and in the control circuit (showing how the motor is started/stopped). ✅ 2. Follow Cross-References Next to the motor symbol or its contactor coil, you’ll usually see a reference like “See Pg. 4, B5”. Pg. 4 → the page number. B5 → the location on that page (column B, row 5). This tells you exactly where to continue reading. ✅ 3. Columns & Rows on a Page Each page of a diagram is divided into a grid (like an Excel sheet): Rows → numbers (1, 2, 3, …) Columns → letters (A, B, C, …) So if a symbol says “C3”, you go to column C, row 3 of that page to find the next related symbol. ✅ 4. Example Motor M1 shown on Page 2, cell A4. Its contactor coil shown on Page 5, cell D7. The reference note on Page 2 will say → “→ Pg. 5, D7”. This way you can quickly jump across pages and follow the full motor circuit without confusion. 📌 In short: Motor tag = identity card of the motor. Cross-reference = address of where to find its other parts. Page grids = map to navigate the diagram. 🔖 Hashtags #ElectricalEngineering #MotorControl #IndustrialAutomation #ControlPanels #ElectricalDesign #EngineeringTips #PanelWiring #Troubleshooting #PLC #Maintenance

▎Understanding Wiring Diagrams 🛠️✨ Hello, fellow engineers! 👋 Welcome to another exciting journey into the world of electrical wiring diagrams! Today, we’re diving into #EWD241225-E, which continues from our previous post #EWD241225-D. Let’s get started! 🚀 ▎DAY FIVE 🌟 1. Component Identification 🔍 The diagram uses standard European electrical symbols (IEC style). Here are some key components to look out for: • K1 K5: These are your contactors, crucial for controlling power. • K25: This is the phase sequence relay that helps ensure everything runs smoothly. • The square boxes labeled "A1/A2" at the bottom represent the coils that activate these switches. • Y1: This symbol (a box with a diagonal line and an 'X' pattern) represents a Solenoid Valve, which regulates the flow of refrigerant. ❄️ • EWD1: This refers to an Electronic Water Drain, featuring connections for Power (L, N) and a potential-free contact for status or alarm signaling. • S2: This represents a Pressure Switch or sensor, indicated by the 'P' inside the symbol. • Terminal Strip (1X5): The numbers at the top (1 through 12) represent a terminal block 2x30, often referred to as a relay block because it has a common (C) and Normally Open (NO) contact where external wiring connects to the internal control panel. 2. Reading the Logic 📊 The diagram is organized in "rungs" (vertical lines), similar to ladder logic but drawn vertically. • Power Supply: The top horizontal line is your "hot" or control voltage line, while the bottom horizontal line serves as the neutral or common line. • Cross-Referencing: At the very bottom, under the relay coils, you’ll see grid references like 03.2 or 05.8. These indicate where else in the full manual (which page and column) those specific relay contacts are being used. ▎Troubleshooting Tips 🛠️💡 If you’re using this or any other diagram to troubleshoot a machine, here are some handy tips: • Check the Coils: If a component isn’t turning on, make sure that the coil (A1/A2) of its corresponding relay (like K1) is receiving voltage. • Trace the Wire Numbers: Look for the small numbers next to the lines (like 20, 21, 16). These are wire markers that should match the physical tags on the wires inside the electrical cabinet. Happy wiring, everyone! 🔌✨ #Troubleshooting #AtlasCopco #ElectricalWiringDiagram #ElectromechanicalEngineer #Refrigeration #PanelBuilder #HVACSystem #RAISL

€𝟒𝟎𝐊 𝐬𝐲𝐬𝐭𝐞𝐦 𝐟𝐚𝐢𝐥𝐞𝐝 𝐨𝐧 𝐝𝐚𝐲 𝐨𝐧𝐞. 𝐄𝐯𝐞𝐫𝐲 𝐝𝐢𝐚𝐠𝐫𝐚𝐦 𝐫𝐞𝐭𝐫𝐢𝐞𝐯𝐞𝐝. 𝐍𝐨𝐧𝐞 𝐨𝐟 𝐭𝐡𝐞𝐦 𝐫𝐞𝐥𝐞𝐯𝐚𝐧𝐭. 𝑽𝒊𝒔𝒊𝒐𝒏 𝑳𝑳𝑴𝒔 𝒅𝒆𝒔𝒄𝒓𝒊𝒃𝒆 𝒚𝒐𝒖𝒓 𝒎𝒂𝒏𝒖𝒇𝒂𝒄𝒕𝒖𝒓𝒊𝒏𝒈 𝒅𝒊𝒂𝒈𝒓𝒂𝒎𝒔 𝒂𝒔 "𝒂 𝒕𝒆𝒄𝒉𝒏𝒊𝒄𝒂𝒍 𝒅𝒊𝒂𝒈𝒓𝒂𝒎 𝒔𝒉𝒐𝒘𝒊𝒏𝒈 𝒎𝒆𝒄𝒉𝒂𝒏𝒊𝒄𝒂𝒍 𝒄𝒐𝒎𝒑𝒐𝒏𝒆𝒏𝒕𝒔." That description is useless for retrieval. I learned this building RAG for 1000+ page service manuals. The diagrams looked perfect in the index, were extracted cleanly, were described by GPT-4V, and were embedded properly. But searches like "where is the oil drain plug on the 1255" returned no relevant results. 𝐓𝐡𝐞 𝐩𝐫𝐨𝐛𝐥𝐞𝐦: 𝐕𝐢𝐬𝐢𝐨𝐧 𝐋𝐋𝐌𝐬 𝐚𝐧𝐚𝐥𝐲𝐳𝐞 𝐢𝐦𝐚𝐠𝐞𝐬 𝐢𝐧 𝐢𝐬𝐨𝐥𝐚𝐭𝐢𝐨𝐧. They don't know page 47 is in "Section 3.2: Hydraulic System." They don't see the caption "Figure 3-2: Pump Assembly." They can't tell that this diagram only applies to models 943 and 1255. 𝐓𝐡𝐞 𝐟𝐢𝐱 𝐭𝐚𝐤𝐞𝐬 𝟑𝟎 𝐦𝐢𝐧𝐮𝐭𝐞𝐬 𝐭𝐨 𝐢𝐦𝐩𝐥𝐞𝐦𝐞𝐧𝐭 𝐚𝐧𝐝 𝐜𝐡𝐚𝐧𝐠𝐞𝐬 𝐞𝐯𝐞𝐫𝐲𝐭𝐡𝐢𝐧𝐠: Extract page context BEFORE calling the Vision LLM. Feed the model three things: → The image → The section heading, caption, and surrounding paragraphs → The document metadata (manual title, model applicability) Now, instead of "technical diagram showing mechanical components," you get: "Exploded view of hydraulic pump assembly from Section 3.2, showing 12 components including housing (P/N 45-2234), impeller (P/N 45-2235), and seal kit (P/N 45-2240). Applicable to models 943, 1055, and 1255." Now, the description is useful. Here's the complete pipeline I use: 🩸Quality gate ▶️ filter out logos and decorative graphics 🩸Page context extraction ▶️ section, caption, surrounding text 🩸High-res PNG extraction + blob storage 🩸Vision LLM analysis WITH page context + separate OCR pass 🩸Callout mapping ▶️ link numbered callouts to parts list rows 🩸Rich metadata assembly ▶️ type, models, components, part numbers 🩸Cross-reference linking ▶️ connect figures ↔ tables ↔ procedures 🩸Constructed embedding text ▶️ not raw OCR, structured context 🩸Multi-vector indexing ▶️ dense + sparse + ColBERT 🩸Query routing ▶️ diagram queries vs. text queries Step 1 is where most pipelines fail. They extract the image, analyze it in isolation, and wonder why "show me the fuel filter location" returns wiring diagrams. 𝐂𝐨𝐧𝐭𝐞𝐱𝐭 𝐢𝐬𝐧'𝐭 𝐨𝐩𝐭𝐢𝐨𝐧𝐚𝐥 𝐦𝐞𝐭𝐚𝐝𝐚𝐭𝐚. 𝐂𝐨𝐧𝐭𝐞𝐱𝐭 𝐢𝐬 𝐰𝐡𝐚𝐭 𝐦𝐚𝐤𝐞𝐬 𝐝𝐢𝐚𝐠𝐫𝐚𝐦𝐬 𝐫𝐞𝐭𝐫𝐢𝐞𝐯𝐚𝐛𝐥𝐞

*Deciphering Electrical Wiring Diagrams: A Step-by-Step Guide* Have you ever found yourself lost in a complex electrical wiring diagram, searching for a specific motor or component? You're not alone! Understanding electrical diagrams is a crucial skill for electrical engineers, technicians, and professionals in related fields. In this post, we'll break down the navigation system used in electrical diagrams, making it easier for you to find your way around. *4 Essential Steps to Read Electrical Wiring Diagrams:* *Use the Motor Tag*: Every motor has a unique ID or tag. Look for this tag in the power circuit and control circuit to identify the motor connection. *Follow Cross-References*: Next to the motor symbol or contactor coil, you'll find a reference like "See Pg. 4, B5". This tells you exactly where to continue reading. *Understand Columns & Rows*: Each page is divided into a grid. Use the column letters (A, B, C, ...) and row numbers (1, 2, 3, ...) to navigate. *Practice with an Example*: Let's say Motor M1 is shown on Page 2, cell A4, and its contactor coil is on Page 5, cell D7. The reference note on Page 2 will guide you to the next related symbol. By mastering these steps, you'll be able to: Efficiently navigate complex electrical diagrams Quickly identify motor connections and control circuits Streamline your troubleshooting and maintenance processes What are some common challenges you've faced when reading electrical wiring diagrams? Share your experiences, tips, and best practices in the comments below! #ElectricalWiringDiagrams #ElectricalEngineering #Navigation #Troubleshooting #Maintenance #ElectricalSafety

DOCAI : Solved a major problem in document AI: cross-referenced pages 👉 “Refer Page 5” While you’re reading Page 1. And just like that… the model loses context. I solved this in my project. 🚧 The Problem Traditional PDF AI systems: • Read pages linearly • Ignore cross-page references • Never fetch the actual linked context Result → incomplete or wrong answers. And this becomes a huge issue in: Legal documents Research papers Technical docs 🚀 What I Built (DocAI) I built a system that actually understands cross-referenced documents. Instead of treating PDFs like plain text: 👉 break the Pdf into tree/subnodes using pageindexed tree nodes 👉 tree nodes stored in db 👉 Store raw pages separately for deep retrieval (s3 bucket) 👉 queue for upload route ⚙️ How It Works ----check flow in comment section When a user asks a question: LLM analyzes the query Then decides: 🟢 Normal query → use normal (hi,hello) (fast) 🟡 Enough context → no extra fetch(no raw pages) 🔴need content → fetch linked raw pages for better context • Identify referenced pages (e.g., Page 5) • Fetch raw page data • Resolve cross-links • Send ONLY required pages to LLM 👉 Not the whole PDF. Just the relevant context. 🧠 The Real Insight It was this: Let the system decide when to fetch more context. Benefits • Faster • Cheaper • More accurate • Works for real-world documents Piyush Garg you said to create this and its done 🙂 lots of love to PageIndex Try it — link in comments 👇🙂 #AI #LLM #SystemDesign #BuildInPublic #DocAI