A batch is ready to ship. The labels are printed, the records look tidy, and then a spot check shows a short count in one bottle out of a hundred. In a regulated line that single miss becomes a deviation, a root cause meeting, and a long afternoon you did not plan.
The weak point was not the team. It was the method. When you infer counts from weight or rely on manual tallies, tiny shifts in unit mass or a moment of fatigue can break compliance. In pharma, medtech, nutraceuticals, and aerospace supply, that risk touches GMP, customer clauses, and every audit you will face this year.
Validation removes guesswork. A modern optical counter moves small parts in a single file stream, detects each unit cleanly, and closes a preset with a time stamped record. When you prove that process with clear requirements, repeatable tests, and clean data, piece counts stop being a promise and become evidence.
This article gives you a practical playbook for validation. You will see how to write user requirements that match your products, how to run IQ OQ PQ with simple acceptance tests, and how to capture logs that satisfy data integrity rules. We will use concrete examples from capsule lines, medtech kitting, and aerospace fasteners so every step maps to real work on your floor.
If your goal is fewer deviations, faster audits, and a counting method you can defend, you are in the right place.
Why validation matters and how to frame user requirements
Validation is the contract between your process and your proof. In regulated lines the promise on the label is a piece count, while much of daily control still relies on mass. That gap creates deviations. A bottle can meet a gram target and still miss the true count when unit mass drifts with moisture, coating, or supplier mix. Validation closes the gap by defining what the process must prove, then testing that proof in a way an auditor can follow without a second meeting.
What your process must prove
At minimum your counting method must show that it delivers the promised quantity under real conditions, not just on a quiet bench. That means three things. First, accuracy at the preset with a visible target for doubles and misses. Second, stability across shifts with dust, static, and routine cleaning. Third, a data trail that ties every closed container to time, lot, and operator so QA can answer who did what and when without detective work.
Translate product reality into limits
A good user requirement starts with the parts you actually run. If you pack one hundred capsules that average zero point five gram with a five percent spread, a gram target will swing by about five pieces at the extremes. The requirement should say that the method controls pieces directly and holds a double rate below one percent at the intended speed. If you kit micro fasteners that shed fines, add a requirement for stable counts after a defined dust exposure and a scheduled window clean. If your facility sees low humidity, require static controls and a quick acceptance test that catches drift before it reaches customers.
Draft the first pass URS the way an auditor reads
Keep the document short and specific.
- Size window the counter must handle, from your smallest unit to your largest
- Target counts per container and the minimum speed you will accept in units per hour
- Accuracy target at the preset, with clear limits for double rate and jam rate
- Environmental limits for humidity, dust load, and cleaning frequency
- Data fields to be captured at each preset completion, including time, lot, operator, work order, and exception notes
- Changeover rules that include saved recipes and a two minute acceptance test
- Accessories that affect capacity, such as dual outlet for alternating fills or a carousel for vials and small jars
Write each point so a technician can test it in a single session and a reviewer can trace the evidence without extra files.
Choose the reference device with eyes on evidence
A modern optical counter is the practical base for this URS because it controls the variable you promise. The bowl singulates, the optical gate treats one pass as one event, and the preset closes at the exact number. That is the method you will prove in Operational and Performance Qualification. Where you have very small parts or mixed shapes, an optical counting machine Elmor C1 is a sensible reference example because it pairs calm singulation with simple presets and clean logs. For larger or more awkward items the C1 MAXI keeps the same daily workflow with more headroom. The point is not the logo. It is the ability to map every URS point to a setting, a test, and a line in the record. Use an optical counting machine Elmor C1 to control pieces directly while keeping a clean log for audits. For alternative formats and accessories, review Elmor counting machines.
Build proof into daily work from the start
Validation fails when it lives in a binder. Put the acceptance test on the bench. Two minutes at intended settings with three numbers recorded each time. Items per minute, visible doubles in the last few centimeters before the sensor, and any stops. Save those three numbers with the lot and the operator. That single habit will do more to protect compliance than a perfect document that no one opens during a shift.
From risk assessment to Installation Qualification
Map the risks before you touch the machine
Start with a simple grid that lists what can go wrong, how likely it is in your facility, and how hard it is to detect quickly. Doubles near the sensor, jams in the last hand width, static cling on dry days, dust loads from coating lines, and operator setup errors make a useful first pass. Rank each risk as low, medium, or high. Add one real example from your line for each item so the team sees the problem, not a theory.
Define preventive controls that live on the bench
For each high or medium risk, write a control that an operator can apply in under two minutes. Examples include a grounded mat and glass containers for static, a scheduled window wipe every two hours during dusty runs, a gentle throat just one item wide for final singulation, and a barcode driven recipe that loads bowl amplitude, rail position, throat width, and preset. Tie every control to a quick check so it does not become a poster on the wall.
Select challenge parts that expose the limits
Pick the smallest item in each SKU family, plus a difficult shape or a coated version if you use one. These are your challenge parts for qualification. Store a small sealed bag of each with a label that records date, lot, and who pulled the sample. You will reuse these during OQ and later during periodic reviews so results are comparable across months.
Installation Qualification checklist
Treat IQ as proof that the device sits in the right place, with the right utilities, talking to the right data flow. Keep the checklist short and specific.
- Site and power
Stable table, level to manufacturer tolerance, no shared outlet with a motor that could add vibration, power at the rated voltage with surge protection. Record outlet ID and a photo of the setup. - Grounding and static control
Bench and bowl grounded, grounded mat under the station, use of glass or metal containers confirmed. Record room humidity and your acceptable band for operations. - Environment and lighting
No direct sunlight on the sensing zone, no fan or vent that drives dust across the window, housekeeping plan for nearby coating or sanding areas. Record a short note on airflow and nearby sources of fines. - Barcode link and presets
Scanner tested with work order, lot, and SKU codes and barcode driven recipes on Elmor C1 confirmed in the log. Confirm that scans load the correct preset and recipe and that the same fields appear in the log when a preset completes. Capture two test lines from the log with timestamps. - Mechanical setup
Bowl firmly mounted, rails tightened, throat insert installed to the intended width, chute or funnel aligned with container positions. Record photos and the final throat width in millimeters for each recipe tested. - Optical health
Window inspected and cleaned, sensor test run performed with a handful of parts to confirm a clear signal and no bounce in the lane. Record a thirty second video of the last few centimeters before the sensor for the file. - Data capture and storage
CSV export enabled to the correct network location or device storage, file naming convention agreed, daily folder structure created. Confirm that a second system can read the file without manual edits. - Training and access
Operators and QA sign the IQ page to confirm access levels, recipe edit rights, and the location of the acceptance test sheet. Store one printed quick start at the bench.
Put IQ in context with a small example
A capsule line installs an optical counting machine Elmor C1 at a kitting cell. During IQ the team documents power and grounding, sets the humidity band at forty to fifty five percent, and positions the station away from a door that caused dust drift. Barcodes now load the preset for sixty count bottles and the recipe for the small capsule family. A test preset writes a log line with time, lot, SKU, and operator. Photos show the throat width and the chute alignment. The file path for CSV export is recorded and support confirms read access. IQ closes in one session because every point is observable at the bench.
Why this approach survives audits
Auditors read evidence chains. The risk map shows you thought about your parts and your room. Each IQ line ties a risk to a control or a verification. Photos and short videos reduce debate. Barcode links and logs prove that identification and traceability are not promises but settings that work. Most important, every step is framed so a technician can repeat it without a meeting.
Operational Qualification you can run in one shift
Operational Qualification proves that the counting method works under real conditions with real people. The goal is not a perfect day. The goal is a documented day that shows the process stays in control when small stresses appear.
Define acceptance limits before you start
Write limits that a technician can measure at the bench.
- Preset accuracy
For a preset of N pieces, the container must contain N pieces with zero tolerance. Any miss is an exception that triggers investigation and a repeat of that cycle after the fix. - Double rate
Visible doubles in the last few centimeters before the sensor divided by total items observed in that clip. Start with a limit below one percent at intended speed. - Jam rate
Operator touches required to restore flow divided by total minutes of run time. Start with a limit of zero in a ten minute run. - Throughput
Items per minute at least equal to the minimum promised in the URS for each SKU family. - Data integrity
Each preset completion must create a log line with time, SKU, lot, operator, and target count in the correct fields.
State these limits on a single page next to the counter. Operators should not need a binder to know what good looks like.
Build an OQ matrix that reflects your floor
Test the corners of your reality, not only the easy center.
- Parts
Smallest item in each SKU family, one awkward shape, and one coated or dusty variant if relevant. - People
Two operators with different experience levels. - Time
Two short windows that reflect your climate and shifts, for example early morning and late afternoon when humidity changes. - Accessories
One run with the gentle throat installed, and if you plan to use them, one run that exercises dual outlet or carousel filling.
Layout a small table before you begin that maps parts, people, time, and accessories to planned cycles.
Run OQ with challenge parts and short, repeatable cycles
For each line in the matrix:
- Acceptance test
Mix a handful from the lot. Run two minutes at intended settings. Record items per minute, visible doubles, and any stops. If either ratio is out of limit, clean the window, calm static, reduce feed one step, and repeat the test. Only proceed when the acceptance test passes. - Preset cycles
Run ten consecutive containers at the production preset. Alternate between the two outlets if fitted. Record cycle time per container from start to close and capture the log lines. If you plan alternating fills, confirm the dual outlet setup shown under counting machines before timing the set. - Visual clip
Film the last few centimeters before the sensor for thirty seconds during cycle three or four. This provides evidence that the lane and throat shape create true single file flow at speed. - Count check
Manually check one container out of ten for exact count using a fast recount method. Record the check without slowing the line. A miss requires immediate investigation and a repeat of the ten cycle set after the fix. - Data check
Open the log file and confirm that time, SKU, lot, operator, and preset count are present and readable by your downstream system.
Handle exceptions with a short decision tree
When OQ reveals drift, act in minutes and record both cause and fix.
- Doubles above limit and no jams
Lower amplitude one step and retest the two minute clip. If doubles persist, narrow the throat slightly, retest, then polish a short patch where stacks form. - Jams present
Inspect for burrs or debris, clean the window and track, verify chute alignment, then retest. If jams continue, reduce bowl angle or add a shallow S curve before the sensing point. - Static effects
Ground the bench, replace plastic cups with glass or metal, increase room humidity if within your range, and place a small ionizer near discharge. Retest. - Dust effects
Brush and vacuum the lot, wipe the window, implement a scheduled wipe every two hours during dusty runs, then retest.
Close the loop with one sentence in the record. Problem seen, action taken, result in numbers.
Lock the method with recipes and operator notes
Save a simple recipe for each SKU family that lists bowl amplitude, rail position, throat width, preset, and the acceptance test targets. Print one page and place it at the bench. Add two operator notes that capture small tricks, for example tilt the vial slightly for coated parts, or swap to the wider funnel for the washer kit. Good OQ turns skill into something that can be trained in an hour.
A short example that shows the shape of a pass
A medtech team validates a sixty count bottle on an optical counting machine such as Elmor C1. Morning and afternoon windows pass acceptance tests at nine hundred items per minute with zero jams and a double rate below one percent. Ten cycle sets close at sixty every time. One manual recount confirms the preset. Logs show clean lines with time, SKU, lot, and operator. An afternoon spike in doubles appears on the coated variant. The team adds a narrow throat insert and repeats the acceptance test. Doubles drop below the limit and the set passes. The exception note ties the fix to the recipe so the next shift inherits the improvement.
Performance Qualification on live products
Performance Qualification proves that the method survives real volume, real shifts, and small surprises. It extends the discipline of OQ into the pace and variety of daily work so your numbers speak for themselves when an auditor asks.
Define the run in terms that matter to your line
Pick a window that reflects normal production. Many teams choose two to five consecutive days or a fixed count of containers per SKU family. Include at least one change of shift and a changeover. Record room humidity and temperature at the start of each block so you can read results in context. If you plan to use dual outlet or a carousel, run them exactly as you will in routine work.
Fix the acceptance limits before you begin
Carry forward your OQ limits and make them explicit for PQ.
- Preset accuracy is absolute. Each container must contain the target count. Any miss triggers an exception record and an immediate investigation.
- Double rate remains below the threshold you set during OQ at the intended speed.
- Jam rate stays at zero or within the small allowance you defined for your products.
- Throughput meets or exceeds the URS minimum without operator workarounds.
- Data completeness reaches one hundred percent. Every closed preset generates a readable log line with time, lot, SKU, operator, and target.
Post these limits next to the station so operators and QA see the same rules.
Use a sampling plan that proves control without slowing the line
Adopt a simple interval count check for each SKU family. For example, recount one container out of every twenty, then roll the starting point by five containers each hour so the sample is spread across the run. Record the check result and the time in the same file that stores preset completions. If a miss appears, pause that SKU, fix the cause, and repeat the last block.
Track metrics that reveal drift early
Go beyond pass or fail and collect a small set of indicators.
- Items per minute per outlet, reported by hour
- Visible doubles in a short clip near the sensor at the top of each hour
- Operator touches per hour for any reason
- Exceptions by type, with a one sentence cause and fix
- Log integrity checks that confirm no missing or corrupt lines
Plot the first three as simple control charts. If a point lands outside your band, capture an action and a retest in the same hour. This habit turns PQ into a foundation for routine process control.
Prove the edge cases that matter to your auditors
Run at least one block with the smallest item in the family and one block with a coated or dusty variant. If your facility experiences low humidity afternoons, schedule one block then. If a customer requires a specific kit sequence, prove that sequence with scans and a short video of the last few centimeters before the sensor. Evidence is strongest when it looks like your real day.
What a passing PQ looks like
Over the chosen window every preset closes at the target count. Recount samples confirm the count with no misses. Double and jam rates remain within limits while items per minute stay at or above the URS minimum. The log contains a complete, continuous set of lines that link time, lot, SKU, and operator. Exceptions are few, explained, and tied to a change that prevented recurrence. Operators can run the method without a supervisor at their shoulder.
A brief example from practice
A nutraceutical line validates a sixty count bottle across three days. Two operators alternate shifts. Morning blocks run at nine hundred items per minute with a double rate below one percent. Afternoon blocks see lower humidity, so the team grounds the bench and wipes the window on a two hour schedule. Recount sampling of one in twenty shows no misses across nine hundred containers. A brief spike in doubles on a coated capsule is fixed by narrowing the throat insert and logged with numbers before and after. PQ closes with complete logs, stable throughput, and a short list of small actions that became part of the recipe.
Conclusion and next steps
Validation is how you turn a counting promise into proof. When you control pieces instead of guessing from grams, deviations fall, audits move faster, and your team works with calm confidence. The method is not exotic. It is a short acceptance test at the bench, recipes that make changeovers repeatable, and a log that ties every closed preset to time, lot, and operator.
Make it real on your floor this month. Choose two live SKUs, write a one page URS with size window, preset targets, and limits for doubles and jams, then run IQ in a single session with photos and a short video of the sensing zone. Complete OQ in one shift using challenge parts and the two minute acceptance test before every set. Close PQ over two to five days with recount sampling of one in twenty and control charts for items per minute, doubles, and touches.
If you want a ready path, pilot on an optical counting machine Elmor C1 for tiny and mixed parts, or move to C1 MAXI for larger components while keeping the same daily workflow. Use barcode driven recipes so operators load the right preset without memory, and store the CSV log from Elmor C1 in your validated repository so QA can answer who did what and when in seconds. When the pilot shows zero misses at the preset, a stable double rate, and complete logs, publish the SOP and make the acceptance test part of every shift.
The payoff is simple to measure and hard to argue with. Fewer deviations, fewer complaints, and fewer meetings spent reconstructing what happened. Start with one cell, prove it in your own numbers, then scale the method wherever the label promises a count. Start a two day pilot with Elmor and document IQ, OQ, and PQ on two live SKUs.
