--- title: "Part Counting in CNC Shops: Track Part Count Automatically" description: "Part counting for CNC shops: compare controller signals, sensors and cameras, see a Fanuc example, and run a pilot to trust your counts." image: https://www.jitbase.com/hubfs/part-counting-cnc-shops.webp --- [+1 438 942-8501](tel:+1%20438%20942-8501) [Log In](https://app.jitbase.com/) - [English](https://www.jitbase.com/blog/part-counting-cnc-shops) - [Français](https://www.jitbase.com/fr/blog/comptage-pieces-atelier-cnc) [![JITbase black logo](https://www.jitbase.com/hs-fs/hubfs/Logos/jitbase-logo-black-horiz.webp?width=175&height=88&name=jitbase-logo-black-horiz.webp)](https://www.jitbase.com) - Solutions Back ##### [Machine Monitoring](https://www.jitbase.com/machine-monitoring) - ##### [Machine Monitoring](https://www.jitbase.com/machine-monitoring) Back - [Live Dashboard](https://www.jitbase.com/machine-monitoring) - [Track OEE](https://www.jitbase.com/machine-monitoring#Track-OEE) - [Track Downtime Reasons](https://jitbase.com/machine-monitoring#Track-Downtime-Reasons) - [Reporting](https://jitbase.com/machine-monitoring#Reporting) - [Timelines](https://jitbase.com/machine-monitoring#Timelines) ##### [Production Monitoring](https://www.jitbase.com/production-monitoring) - ##### [Production Monitoring](https://www.jitbase.com/production-monitoring) Back - [Accurate Standard Times](https://jitbase.com/program-monitoring#Accurate-Standard-Times) - [Tool path & M00 tracking](https://jitbase.com/program-monitoring#Tool-path-M00-tracking) - [Performance tracking](https://jitbase.com/program-monitoring#Performance) - [CNC Program Improvement](https://jitbase.com/program-monitoring#CNC-Program-Improvement) - [Overtime Reasons](https://jitbase.com/program-monitoring#Overtime-Reasons) - [Lights-out Operations](https://jitbase.com/program-monitoring#Light-Out-Ope) ##### [Scheduling & Workforce Management](https://www.jitbase.com/workforce-management) - ##### [Scheduling & Workforce Management](https://www.jitbase.com/workforce-management) Back - [Operator Workload](https://jitbase.com/workforce-management#Operator-Workload) - [Workforce Scheduling](https://jitbase.com/workforce-management/#Operator-Daily-Allocation) - [Live Job Tracking](https://jitbase.com/workforce-management#Live-Job-Tracking) - [Production Planning and Scheduling](https://jitbase.com/workforce-management#Job-Operation-Planning) - [Live guidance for Operators](https://jitbase.com/workforce-management#Live-guidance-for-Operators) ##### [Edge Platform](https://www.jitbase.com/edge-platform) - ##### [Edge Platform](https://www.jitbase.com/edge-platform) Back - [Manage your own infrastructure](https://www.jitbase.com/edge-platform#Manage-your-own-infrastructure) - [Send Machine data to local server](https://www.jitbase.com/edge-platform#Machine-data-to-local-server) - [Drivers for standard protocols](https://www.jitbase.com/edge-platform#Drivers-for-standard-protocols) - [Integrations](https://www.jitbase.com/edge-platform#Integrations) - Resources Back ##### [Our Downloadable Resources](https://www.jitbase.com/download) - ##### [Our Downloadable Resources](https://www.jitbase.com/download) Back - [Case Studies](https://jitbase.com/download#case-studies) - [White Papers](https://jitbase.com/download#white-papers) - [Brochures](https://jitbase.com/download#brochures) - [Product & Integration](https://jitbase.com/download#Product-Integration) - [In the Media](https://jitbase.com/download#in-the-media) ##### [Education](https://www.jitbase.com/education) - ##### [Education](https://www.jitbase.com/education) Back - [Workforce Management vs. Machine Monitoring](https://jitbase.com/education#Workforce-Management-vs-Machine-Monitoring) - [How to start?](https://jitbase.com/education#How-to-start) - [Infrastructure & Cyber-Security](https://jitbase.com/education#Infrastructure-Cyber-Security) - [Why the Cloud?](https://jitbase.com/education#Why-the-Cloud) - [FAQ](https://jitbase.com/education#Education-FAQ) ##### [Forms](https://www.jitbase.com/demo) - ##### [Forms](https://www.jitbase.com/demo) Back - [Return on Investissement](https://www.jitbase.com/return-on-investment) - [Machines we connect](https://www.jitbase.com/machines-we-connect) - [Demo Request](https://www.jitbase.com/demo) ##### [Blog](https://www.jitbase.com/blog) - ##### [Blog](https://www.jitbase.com/blog) Back - [Machine Monitoring](https://www.jitbase.com/blog/tag/machine-monitoring) - [Labor Management](https://www.jitbase.com/blog/tag/labor-management) - [Production Planning and Scheduling](https://www.jitbase.com/blog/tag/production-planning-and-scheduling) - [Quality and Inspection](https://www.jitbase.com/blog/tag/quality-and-inspection) - [Plans](https://www.jitbase.com/plans) - Why JITbase Back - ##### [Why JITbase](https://www.jitbase.com/why-jitbase) - ##### [About Us](https://www.jitbase.com/about-us) - ##### [Partners](https://www.jitbase.com/partners) [Connect Machines for Free](https://sso.jitbase.com/realms/webapp/protocol/openid-connect/registrations?client_id=website-registration&response_type=code&scope=openid&redirect_uri=https://admin.jitbase.com) [+1 438 942-8501](tel:+1%20438%20942-8501) [Log In](https://app.jitbase.com/) - [English](https://www.jitbase.com/blog/part-counting-cnc-shops) - [Français](https://www.jitbase.com/fr/blog/comptage-pieces-atelier-cnc)   [Book a demo](https://www.jitbase.com/demo) [Connect Machines for Free](https://sso.jitbase.com/realms/webapp/protocol/openid-connect/registrations?client_id=website-registration&response_type=code&scope=openid&redirect_uri=https://admin.jitbase.com) ![Machinist inspecting a finished part beside a CNC lathe with a bin of turned steel and brass parts: part counting with JITbase](https://www.jitbase.com/hubfs/part-counting-cnc-shops.webp) Blog [Machine monitoring](https://www.jitbase.com/blog/tag/machine-monitoring) # Part Counting in CNC Shops: Track Part Count Automatically [*written by* Judicael Deguenon on October 14, 2026](https://www.jitbase.com/blog/author/judicael-deguenon) The lathe ran unattended last night. Did it make 17 parts or 23? Automatic part counting gives you a number you can trust, without operators stopping to write it down. This guide covers the main part counting methods, how to read counts from CNC controllers and sensors, and how to check them before you rely on them. **TL;DR:** - The best automatic part counting method depends on what you are counting. Controller events work well for stable programs, while sensors or cameras are often better when physical part movement matters. - Start with one pilot machine, compare automatic counts with a short manual tally, and only send it to your ERP or MES software after the count source proves reliable. - Good part counts do more than update inventory; they also support tracking of parts in progress (WIP), scheduling, and operator workload decisions. ## Why automatic part counting matters for CNC shops ### Operational benefits: output, inventory accuracy, traceability Automatic counts remove one of the most common weak spots in small and mid-sized machine shops: production reporting that depends on memory, paper notes, or delayed updates. If a lathe ran unattended for part of the shift, the planner still needs the real number. That number changes the next setup, the job queue, and in some cases the shipment. Reliable counts help with: - Output tracking (parts per hour) by machine, cell, and shift - Better visibility of work in progress (WIP) between machining, inspection, and packing - Lot traceability when counts are tied to timestamps or job records - Faster ERP or MES updates with fewer manual entries - Less interruption for operators who would otherwise stop to record production For planners, the value is simple: the schedule reflects reality more quickly. For finance and ERP teams, production transactions become easier to reconcile. For supervisors, the count becomes a live signal instead of a once-per-shift estimate. ### Common pain points manual counting creates on small shop floors Manual counting creates work that often goes unseen. An operator finishes a cycle, removes parts, updates the job traveler (the paper that follows the job), answers a question, clears chips, then gets pulled to another machine. That’s how counts get skipped or entered late. Operators who already juggle [hidden-time tasks](https://www.jitbase.com/blog/hidden-time-operator-tasks-cnc-shops) such as inspection, deburring, and tool presetting have little spare attention left for production reporting. The damage from a bad count is rarely isolated. A count error can make a job look ahead of schedule when it is actually short. It can also trigger a premature material release, a wrong reorder, or a packing mismatch. And if the shop is using machine events for analysis, bad counts can distort output-per-hour calculations and make a machine look less stable than it really is. In practice, automatic counting matters most where one small reporting step affects several downstream decisions. That’s common in high-mix CNC shops, bar-fed turning work, and unattended machining where nobody wants to audit every parts bin by hand. **See production live, machine by machine.** Follow what your CNC machines produce in real time. [Discover machine monitoring](https://www.jitbase.com/machine-monitoring) ## Overview of methods for automatic part counting ### Machine-origin counts: controller and cycle counters, and program increments A machine-origin count comes from the CNC control or the part program. The controller may expose a part counter, a cycle-complete event, or a macro increment tied to the end of a successful program. This is often the lowest-hardware path because it uses signals the machine already produces. But a controller count measures a machine event, not always a physical finished part. If the machine increments at cycle end before a part is ejected, or if one program makes multiple parts, the logic needs review. This is where [cycle time monitoring implementation](https://www.jitbase.com/blog/implement-cycle-time-monitoring-cnc-shop-minimal-hardware) often overlaps with counting, because both depend on identifying true cycle boundaries. ### External sensors: photoelectric, proximity, and part-presence switches External sensors count physical movement. A photoelectric beam can count parts dropping down a chute, and many shops feed that signal into a standalone parts counter or an I/O module. A proximity sensor can detect a metal part passing a fixed point. A limit switch or part-present switch can confirm that a finished part reached a tray or transfer point. These methods are often better than controller-only logic when the part must physically leave the machine before it should count. Still, installation matters a lot. Chips, coolant spray, bouncing parts, and reflective surfaces can cause false triggers. Shops already working on [automated downtime detection](https://www.jitbase.com/blog/automated-downtime-detection-cnc-setup) will recognize the same issue: event detection is only as good as the signal quality. ### Vision systems: single-camera and area-coverage counting Vision-based counting uses a camera to detect completed parts in a field of view. Sometimes it counts one part at a time as items pass a zone. In other cases, it counts parts in a tray, conveyor area, or bin opening. This approach can work well for small parts, mixed orientations, or situations where a simple beam sensor would miss or double-count. The trade-off is setup effort. Lighting, lens choice, mounting stability, coolant mist, and background contrast all affect reliability. Vision usually needs more tuning than a simple sensor, but it can solve problems that discrete sensors can’t. ### Weight and scale-based counting Scale-based counting uses weight change to estimate part count. If each part has a known mass and low variation, the system can infer count from the total weight in a bin or at a transfer point. This can be useful for heavy castings, forgings, or uniform blanks. It’s less effective for mixed batches, oily chips in the bin, or parts with high weight variation. And it does not directly tell the shop *when* each part was produced, only how much weight accumulated between measurements. ## How to extract an accurate part count from CNC programs and controllers ### Which controller signals to look for (part complete, cycle end) The first step is to identify what event actually represents a good part. Shops often start with a controller part counter, cycle complete bit, or an end-of-program event tied to M30 or another program state. Some controls also support macro variables (values the program can read and update) or parameter-based counters that increment inside the program. However, “cycle end” and “part complete” are not always the same thing. On a bar-fed lathe, one cycle may produce one part, several cutoff pieces, or one part plus a remnant. On a twin-spindle or mill/turn machine, the count logic may need to account for transfer operations or multiple finished components per cycle. So before using any signal, review: - The machine’s I/O list or signal map - Program end behavior on normal and aborted cycles - Whether one cycle equals one finished part - How rework, dry runs, and prove-out runs should be handled ### Example: part counting on a Fanuc controller Fanuc controls are a useful reference because they expose several counting mechanisms. Exact parameter numbers and options vary by model and machine builder, so confirm each item in your machine’s documentation before changing anything. - **Default behavior:** on many Fanuc controls, the built-in parts counter goes up by one each time the program reaches M02 or M30. Parameter settings can change this, so ask the machine builder how the counter is set on your machine. - **Looped programs:** a program that repeats with M99 (return to the start of the program), as on many bar-fed lathes, never reaches M30, so the default counter does not move. A custom M code, or a macro line placed before the M99, solves it. - **Multi-part fixtures:** when one cycle produces two or three parts, the macro variable #3901, which holds the part count, can add the right number each cycle, for example `#3901=#3901+2`. This relies on the Macro B option. If M30 also counts, a cycle can be counted twice, so have the setup confirmed on the machine first. Changing controller parameters or program logic should be done or approved by whoever is responsible for the machine, since some builders restrict these settings. Then compare the controller’s count with a physical tally, as described below, because even a correct macro counts a cycle, not necessarily a good part. ### Standards and adapters: MTConnect and OPC UA as data pathways Once the count source is identified, the data has to leave the machine. Two common pathways are *MTConnect* and *OPC UA*. MTConnect is an open standard widely used to collect machine data. OPC UA, maintained by the [OPC Foundation](https://opcfoundation.org/about/opc-technologies/opc-ua/), is common for industrial communication across controls, PLCs (programmable logic controllers), and gateways. Shops may also use controller-specific connectors, middleware, or machine monitoring platforms. The important point is not the brand. It’s whether the path preserves timestamps, machine identity, and event order without duplication. ### When machine-origin counts need external validation Controller-based counts are efficient, but they still need checking. A short manual tally during pilot runs is usually enough to catch bad assumptions. For example, a shop might compare the automatic count with a manual bin count across two or three shifts as an illustrative sample, then review every mismatch by timestamp. External validation is especially useful when: - A cycle can end without ejecting a finished part - One cycle creates more than one part - Operators restart from mid-program - Setup mode and production mode use the same program structure **Not sure your machines can be connected?** Check which CNC machines and controllers JITbase connects to. [See the machines we connect](https://www.jitbase.com/machines-we-connect) ## Quick checklist: what to evaluate before automating part counting Before buying a sensor or connecting software, a planner or shop manager should review the process itself. A simple checklist prevents expensive rework later. - Confirm whether one program cycle equals one finished part, more than one part, or sometimes zero parts - Review part size, shape, reflectivity, and material, since tiny polished parts behave very differently from large steel castings - Check the machine environment for coolant mist, chips, vibration, heat, and limited mounting space - Decide where a count should happen: at cycle end, at cutoff, at ejection, or when the part reaches a bin - Define the required update frequency, such as live dashboards, end-of-shift totals, or lot-close reconciliation - Identify which system needs the data next, such as ERP, MES, PLC, or a standalone machine monitoring layer - Clarify how to handle scrap, rework, setup pieces, and dry runs - Estimate who will maintain the hardware, clean lenses, inspect sensor mounts, and investigate mismatches - Train operators on what the system counts and what they should do if a jam or missed ejection occurs - Pilot on one machine first so the shop can test logic, maintenance, and reporting before scaling That pilot-first approach saves time. It also makes it easier to separate signal problems from process problems. ## Step-by-step implementation guide for automatic part counting ### Plan a pilot: select machines and use cases Start with one machine where the count matters and the process is reasonably stable. Good pilot candidates include a bar-fed lathe, a vertical machining center with a consistent unload step, or a machine feeding parts into a controlled chute. Avoid the messiest, most variable job first. Set a narrow pilot goal: count finished parts at the chute, then compare with manual totals for several shifts. Keep the pass mark simple. For example, the gap should be small enough that a supervisor can check each exception one by one. ### Hardware selection and mounting best practices For sensors, review response time, sensing range, ingress protection, cable routing, bracket rigidity, and how easy the device is to clean or replace. A photoelectric sensor near coolant splash may need shielding. A proximity sensor may need careful distance control to avoid picking up nearby fixtures. A switch must be mounted where it won’t be damaged by chip buildup or part bounce. For vision, lens selection and lighting matter as much as the camera. A stable background, fixed focal distance, and enclosed lighting often improve results more than software changes do. If parts are shiny, test glare before final mounting. ### Capturing and aggregating counts: local edge vs gateway The count signal can be read directly from the controller, from a sensor wired to a PLC or I/O module, or from an edge gateway that timestamps each pulse. Local edge collection is useful when the shop wants fast capture near the machine and then sends normalized events upstream. A central gateway can work well too, as long as it doesn’t lose short pulses or merge events. The main design question is simple: where should raw events become production counts? Some shops count at the source. Others store raw pulses and apply business rules later. ### Validate, calibrate, and roll out Validation should be boring. That’s a good sign. Run the pilot, compare automatic results to manual counts, inspect mismatches, and document what caused them. Common causes include double-triggering, missed ejection, bin bounce, or program logic that counts setups as production. Only after the pilot proves stable should the shop scale to more machines. And even then, repeat the validation on each machine family. A sensor setup that works on one lathe may fail on another because the part path is different. ## Comparison table: sensors and systems for part counting ### How to read the comparison table The table below uses relative labels instead of invented dollar figures because actual cost depends on machine access, installation labor, and existing controls. “Reliability” assumes normal setup and maintenance, not best-case lab conditions. | Method | Relative cost | Reliability in common conditions | Installation complexity | Best use case | | --- | --- | --- | --- | --- | | Controller-based counters | Low to medium | High when one cycle truly equals one part | Low to medium | Stable programs, unattended cycles, shops already collecting machine data | | Photoelectric sensors | Low to medium | Medium to high, affected by coolant and reflectivity | Medium | Parts crossing a known path, chutes, conveyors, eject stations | | Proximity sensors or switches | Low to medium | Medium to high, affected by placement and bounce | Medium | Metal parts at fixed transfer points, trays, stops, escapements | | Vision or camera systems | Medium to high | High when lighting and scene control are good | High | Small parts, mixed orientation, tray counting, complex physical flows | | Scale or weight-based counters | Low to medium | Medium, depends on part weight consistency | Low to medium | Uniform heavy parts, bin accumulation, slower count intervals | ### Choose a solution by shop use case Turning centers often fit controller counts or chute-mounted sensors well because parts usually follow a defined exit path. Mill/turn machines and transfer-heavy workflows may need more logic or a camera. There, “cycle complete” doesn’t always mean “good part delivered.” Loading cells and parts bins can sometimes use scale-based checks as a second validation layer rather than the primary source. False positives deserve special attention. Chips can break a beam. A part can bounce and count twice. A camera can drift if the mount loosens. And controller counters can overstate output if the event fires on an incomplete cycle. ## Integrating part counts with ERP/MES and using counts to improve operations The payoff on the shop floor is simple: the planner sees what is finished without calling the shop or walking the floor. That is what connecting your counts to the ERP (order and inventory software) and the MES (shop-floor tracking software) delivers. ### Data flows and common integration patterns Once counts are reliable, the next decision is where they go. Common patterns include middleware pushing updates to ERP or MES, periodic batch syncs at set intervals, or direct API calls from a data collection layer. The best pattern depends on how fast the business needs the count and how tightly it must match lot, job, or operation records. A few rules help avoid common problems: - Keep timestamps in one agreed timezone - Prevent duplicate events during reconnects - Decide whether the machine count or the lot-close count is the source of truth - Separate good-part counts from scrap or rework counts where possible For a deeper look at the ERP and MES side, see our guides on [automating manual interventions and streaming events to ERP/MES](https://www.jitbase.com/blog/automate-manual-interventions-cnc-erp-mes-integration) and on the [shop-floor data integration checklist](https://www.jitbase.com/blog/integrate-shop-floor-data-erp-mes-checklist-template). If the shop wants near-live production visibility, validated counts can also feed [real-time OEE dashboards](https://www.jitbase.com/blog/implement-real-time-oee-dashboards-small-cnc-shops) or a [real-time shop-floor KPI dashboard](https://www.jitbase.com/blog/real-time-shop-floor-kpi-dashboard-no-code). ### Using counts to feed OEE, WIP reconciliation, and operator workload balancing Part counts become much more useful when combined with time data. A *CNC part counter* by itself answers “how many.” Add timestamps and machine state, and you can see the real output rate, compare expected and actual cycle counts, and spot stalls sooner. ![Wall-mounted screen showing live part count and utilization for each CNC machine on a shop floor: automatic part counting with JITbase](https://www.jitbase.com/hubfs/JITbase%20dashboard%20on%20TV%20screen.png) That supports: - WIP reconciliation without constant floor walks - Better promise dates because planners see actual completions - Faster response to machines that are running but not producing good output - Staffing decisions based on who is covering too many manual check points Count data also helps [balance operator workload](https://www.jitbase.com/blog/7-shift-planning-techniques-balance-operator-workload), especially in cells where one person tends several machines. If one operator’s machines are producing frequent unload events while another has long unattended cycles, the staffing pattern may need to change. **Put a number on reliable production data.** Estimate the return on investment of JITbase for your shop. [See the return on investment](https://www.jitbase.com/return-on-investment) ## The bottom line Automatic **part counting** works best when the count source matches the physical reality of the process. Start with one pilot, verify the signal against manual tallies, and only then connect it to ERP, MES, or dashboards. For most shops, the next steps are straightforward: review controller signals, inspect where a physical count could happen, choose one pilot machine, and define a short validation plan. After that, the data becomes much more useful in scheduling and reporting. ## Frequently asked questions ### How accurate are different automatic part counting methods? It depends on what is being measured. Controller-based methods can be very accurate when one cycle always equals one finished part and aborted cycles are excluded. Sensor-based methods can also be accurate, but they are more exposed to chips, coolant, part bounce, and mounting errors. Vision systems often perform well in complex part flows, though they need stable lighting and regular checks, and scale-based methods work best when part weight barely varies. No method is accurate in general, only on a given machine, part family, and environment. The safest approach is to validate any method against a manual tally during a pilot, spread across several shifts. A shop can then set its own acceptance rule, for example investigating every mismatch by timestamp. From there it can decide whether the count is reliable enough to feed planning, WIP reconciliation, or ERP records. ### Can a shop use the CNC program alone for part counting? Sometimes, yes. If the program has a dependable part-complete event, program-based counting may be enough. The control must expose it through a counter, a macro variable, MTConnect, OPC UA, or another data path. This works well on stable programs where each cycle produces exactly one finished part and an aborted cycle never increments the count. The limit is that a program counts cycles, not physical parts. If a cycle can stop early, restart from mid-program, make several parts, or count before the part leaves the machine, the count drifts from reality. In those cases the shop should either validate the program logic against a physical tally or add a sensor at the point where the finished part actually exits. ### How should mixed batches or multiple parts per cycle be handled? The count logic should follow the actual production rule for each job, not a generic machine event. If one cycle produces two parts, the system has to increment by two, or count the physical exits separately. On a Fanuc control, for example, a macro line can add the right number each cycle, as described earlier in this guide. For mixed batches, the count also has to be tied to the right job. Many shops tie counts to the active job ID, a pallet ID, or an operator confirmation at each lot change. That way, parts from one job are never credited to another. Mixed-flow production exposes weak assumptions quickly, which is why a pilot on a single machine, with a manual tally alongside, is worth the effort. ### What maintenance do vision and sensor systems require? Most setups need basic upkeep. Clean the lenses, wipe the sensor faces, check the brackets for movement, and inspect the cables for damage. Also confirm that the count still matches the process after tooling or fixture changes. Shops with coolant mist or fine chips should expect to clean more often, and a loose bracket is a common cause of a count that slowly drifts. The effort is light, but it needs an owner. Name a person who checks the hardware on a fixed routine and who investigates when automatic and manual totals disagree. A counting system with no inspection routine will drift over time, and the first sign is often a planner who stops trusting the numbers. ### How can a shop verify counts after installation? Run a short comparison between automatic counts and manual totals, ideally across different shifts and normal production conditions. Review every mismatch by time and machine event. The team can then tell whether the issue came from program logic, physical sensing, or operator handling, and fix the cause rather than adjusting the number. After go-live, keep periodic spot checks in place, especially after program edits, new part introductions, or hardware adjustments. Verification should be part of change control, not a one-time launch task. A simple habit is to compare the count with a bin tally at lot close and to record the result, so any slow drift becomes visible early. #### SHARE [mailto:email@example.com?subject=Mail%20from%20our%20Website](mailto:email@example.com?subject=Mail%20from%20our%20Website) ## Related Articles [![Operators using a tablet on the CNC shop floor to monitor production, illustrating shop-floor management software and real-time manufacturing visibility.](https://www.jitbase.com/hs-fs/hubfs/blog-images/best-shop-floor-management-software-top-10-hero.jpg?width=352&name=best-shop-floor-management-software-top-10-hero.jpg)](https://www.jitbase.com/blog/best-shop-floor-management-software-top-10) [Machine monitoring](https://www.jitbase.com/blog/tag/machine-monitoring) ### [Best Shop-Floor Management Software: Top 10 Picks for 2026](https://www.jitbase.com/blog/best-shop-floor-management-software-top-10) Read more [![Technician connecting an Ethernet cable to a CNC machine controller panel in a machine shop — shop floor ERP integration with JITbase](https://www.jitbase.com/hs-fs/hubfs/blog-images/integrate-machines-erp-mes-checklist-hero.jpg?width=352&name=integrate-machines-erp-mes-checklist-hero.jpg)](https://www.jitbase.com/blog/shop-floor-integration-validation-checklist-cnc) [Machine monitoring](https://www.jitbase.com/blog/tag/machine-monitoring) ### [Shop Floor Data Validation Checklist: OEE, Cycle Time and WIP](https://www.jitbase.com/blog/shop-floor-integration-validation-checklist-cnc) Read more [![Engineer reviewing a shop-floor checklist while CNC machines exchange data, illustrating MES and ERP integration and structured production workflows with JITbase.](https://www.jitbase.com/hs-fs/hubfs/blog-images/integrate-shop-floor-data-erp-mes-checklist-template-hero.jpg?width=352&name=integrate-shop-floor-data-erp-mes-checklist-template-hero.jpg)](https://www.jitbase.com/blog/integrate-shop-floor-data-erp-mes-checklist-template) [Machine monitoring](https://www.jitbase.com/blog/tag/machine-monitoring) ### [Checklist + Template: Integrate Shop-Floor Cycle Data with ERP/MES to Boost Throughput Without Adding Staff](https://www.jitbase.com/blog/integrate-shop-floor-data-erp-mes-checklist-template) Read more [![CNC machines connected through automated data flow, illustrating reduced manual interventions, MES and ERP integration, and real-time shop-floor automation with JITbase.](https://www.jitbase.com/hs-fs/hubfs/blog-images/automate-manual-interventions-cnc-erp-mes-integration-hero.jpg?width=352&name=automate-manual-interventions-cnc-erp-mes-integration-hero.jpg)](https://www.jitbase.com/blog/automate-manual-interventions-cnc-erp-mes-integration) [Machine monitoring](https://www.jitbase.com/blog/tag/machine-monitoring) ### [Implementation Guide: Automate Manual Interventions on CNC Machines and Stream Events to ERP/MES](https://www.jitbase.com/blog/automate-manual-interventions-cnc-erp-mes-integration) Read more [![JITbase block logo](https://www.jitbase.com/hs-fs/hubfs/eca01011d7169e8c6395a8d7eebe802d.webp?width=317&height=158&name=eca01011d7169e8c6395a8d7eebe802d.webp)](https://www.jitbase.com) ![image-2](https://www.jitbase.com/hs-fs/hubfs/image-2.webp?width=156&height=78&name=image-2.webp) Solutions - [Machine Monitoring](https://www.jitbase.com/machine-monitoring) - [Production Monitoring](https://www.jitbase.com/production-monitoring) - [Scheduling & Workforce Management](https://www.jitbase.com/workforce-management) - [Edge Platform](https://www.jitbase.com/edge-platform) - [Plans](https://www.jitbase.com/plans) Resources - [Download](https://www.jitbase.com/download) - [Education](https://www.jitbase.com/education) - [Return on Investment](https://www.jitbase.com/return-on-investment) - [CNC Machines we connect](https://www.jitbase.com/machines-we-connect) - [Demo Request](https://www.jitbase.com/demo) Why JITbase - [Why JITbase](https://www.jitbase.com/why-jitbase) - [About Us](https://www.jitbase.com/about-us) - [Partners](https://www.jitbase.com/partners) - [Contact Us](https://www.jitbase.com/contact) [Book a demo](https://www.jitbase.com/demo) [Connect Machines for Free](https://sso.jitbase.com/realms/webapp-ngen/protocol/openid-connect/registrations?client_id=website-frontend&response_type=code&scope=openid%20email&redirect_uri=https://admin.jitbase.com/next-gen) © 2026 All rights reserved. - [Privacy Policy](https://www.jitbase.com/privacy-policy) ```json { "@context" : "https://schema.org", "@type" : "BlogPosting", "author" : { "@type" : "Person", "name" : "Judicael Deguenon", "url" : "https://www.jitbase.com/blog/author/judicael-deguenon" }, "dateModified" : "2026-10-14T20:22:03.461Z", "datePublished" : "2026-10-14T20:22:03.000Z", "headline" : "Part Counting in CNC Shops: Track Part Count Automatically", "image" : [ "https://www.jitbase.com/hubfs/part-counting-cnc-shops.webp" ], "mainEntityOfPage" : { "@id" : "https://www.jitbase.com/blog/part-counting-cnc-shops", "@type" : "WebPage" }, "publisher" : { "@type" : "Organization", "logo" : { "@type" : "ImageObject", "url" : "https://www.jitbase.com/hubfs/logo%20jitbase%201.svg" }, "name" : "JITbase" } } ``` ```json { "@context" : "https://schema.org", "@type" : "FAQPage", "mainEntity" : [ { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "It depends on what is being measured. Controller-based methods can be very accurate when one cycle always equals one finished part and aborted cycles are excluded. Sensor-based methods can also be accurate, but they are more exposed to chips, coolant, part bounce, and mounting errors. Vision systems often perform well in complex part flows, though they need stable lighting and regular checks, and scale-based methods work best when part weight barely varies. No method is accurate in general, only on a given machine, part family, and environment. The safest approach is to validate any method against a manual tally during a pilot, spread across several shifts. A shop can then set its own acceptance rule, for example investigating every mismatch by timestamp. From there it can decide whether the count is reliable enough to feed planning, WIP reconciliation, or ERP records." }, "name" : "How accurate are different automatic part counting methods?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "Sometimes, yes. If the program has a dependable part-complete event, program-based counting may be enough. The control must expose it through a counter, a macro variable, MTConnect, OPC UA, or another data path. This works well on stable programs where each cycle produces exactly one finished part and an aborted cycle never increments the count. The limit is that a program counts cycles, not physical parts. If a cycle can stop early, restart from mid-program, make several parts, or count before the part leaves the machine, the count drifts from reality. In those cases the shop should either validate the program logic against a physical tally or add a sensor at the point where the finished part actually exits." }, "name" : "Can a shop use the CNC program alone for part counting?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "The count logic should follow the actual production rule for each job, not a generic machine event. If one cycle produces two parts, the system has to increment by two, or count the physical exits separately. On a Fanuc control, for example, a macro line can add the right number each cycle, as described earlier in this guide. For mixed batches, the count also has to be tied to the right job. Many shops tie counts to the active job ID, a pallet ID, or an operator confirmation at each lot change. That way, parts from one job are never credited to another. Mixed-flow production exposes weak assumptions quickly, which is why a pilot on a single machine, with a manual tally alongside, is worth the effort." }, "name" : "How should mixed batches or multiple parts per cycle be handled?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "Most setups need basic upkeep. Clean the lenses, wipe the sensor faces, check the brackets for movement, and inspect the cables for damage. Also confirm that the count still matches the process after tooling or fixture changes. Shops with coolant mist or fine chips should expect to clean more often, and a loose bracket is a common cause of a count that slowly drifts. The effort is light, but it needs an owner. Name a person who checks the hardware on a fixed routine and who investigates when automatic and manual totals disagree. A counting system with no inspection routine will drift over time, and the first sign is often a planner who stops trusting the numbers." }, "name" : "What maintenance do vision and sensor systems require?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "Run a short comparison between automatic counts and manual totals, ideally across different shifts and normal production conditions. Review every mismatch by time and machine event. The team can then tell whether the issue came from program logic, physical sensing, or operator handling, and fix the cause rather than adjusting the number. After go-live, keep periodic spot checks in place, especially after program edits, new part introductions, or hardware adjustments. Verification should be part of change control, not a one-time launch task. A simple habit is to compare the count with a bin tally at lot close and to record the result, so any slow drift becomes visible early." }, "name" : "How can a shop verify counts after installation?" } ] } ```