Recycling Machinery Hub
Sorting & Separation

How to Tune Optical Sorters for High Purity Recovery

Published 11 min read

An optical sorter unit on a conveyor line in a recycling facility.
Quick answer

To tune an optical sorter for high purity recovery, you must verify the feed stream, adjust camera and lamp settings, calibrate material profiles, and test purge cycles. This process requires consistent feed, proper sensor placement, and regular maintenance to ensure accurate sorting and reliable material recovery.

Key takeaways
  • A tuned optical sorter depends on a clean, consistent feed stream and well-maintained sensors.
  • Calibration requires matching the software profile to the actual material present in the bin.
  • Purge cycles and reject gates must be adjusted to balance purity with throughput.
  • Verification involves sampling the clean and reject streams to confirm the sorting performance.

Prerequisites for Effective Tuning

Before touching the software or adjusting the air knives, you need a stable and representative feed stream. An optical sorter cannot compensate for a chaotic input. If the material is wet, matted, or mixed with non-target contaminants, the sensor will miss targets or misidentify them.

Check the feed consistency first. Ensure the material size falls within the operating range of the sensor. Large, dense objects that shadow smaller targets will reduce accuracy. Wet plastic that clumps together presents a single, irregular shape to the camera. The camera captures the top surface, so if the material is folded, the sensor sees a crumpled sheet rather than a flat, identifiable target. This creates false negatives where the machine thinks it is sorting a clean target but actually rejects a valid piece, or worse, accepts a contaminated piece because the contamination is hidden beneath the fold.

You also need access to the sorter’s control interface. You must be able to view the camera feed, adjust the lamp intensity, change the sorting profile, and monitor the reject stream. Without these controls, tuning is guesswork. Verify that the operator has permission to edit the parameters. On many systems, the engineer sets the limits, and the operator only has read-only access. If you cannot write to the parameters, you cannot tune the machine.

Step 1: Clean the Sensor and Optics

Dust, plastic film, and moisture on the camera lens or lamp housing degrade image quality. Wipe the lens with a clean, lint-free cloth. Check the lamp housing for condensation.

If the lamp is on a short maintenance cycle, replace it. A dimming lamp changes the spectral output, which confuses the sensor’s identification algorithms. Clean optics are the baseline for any calibration work.

Start with the lens. Use a microfiber cloth dampened with a lens cleaning solution. Dry it immediately. Do not use household glass cleaners, as they may leave a residue that scatters light. Check the lamp housing. If there is condensation inside, power down the unit and allow it to dry. Condensation creates a fog that reduces contrast in the image.

Inspect the air knife nozzles. They often accumulate dust and fine plastic particles. Blocked nozzles change the airflow pattern. Instead of a focused jet, you get a diffuse spray. This weakens the sorting force and causes material to drift into the wrong bins. Use compressed air or a brush to clear the nozzles. Be careful not to damage the gaskets.

Step 2: Verify the Feed Stream Composition

Know exactly what is entering the sorter. Review the upstream processing steps. If the feed is a mixed PET stream, confirm it is primarily PET bottles and not mixed with other clear plastics.

If the feed contains mixed materials, the optical sorter must be configured for a multi-material profile. If the feed is a single material, use a single-material profile. Mismatched profiles cause low purity.

Take a representative sample of the feed. Note the size, color, and condition of the material. This sample will be your reference for the verification step later.

Walk the conveyor. Look for materials that are not part of the expected stream. A stray piece of wood, a metal cap, or a chunk of styrofoam can confuse the sensor. If the upstream shredder is producing uneven flakes, the optical sorter will struggle. The sensor needs a predictable shape to identify the material. If the feed is a mix of bottles, film, and containers, the machine must be set to handle that complexity. A profile set for flat film will not work well on cylindrical bottles.

Check for moisture. Wet material absorbs light differently than dry material. A dry PET bottle reflects light in a specific way. A wet PET bottle absorbs more light and appears darker. The algorithm may misidentify the wet piece as a different material or reject it as an impurity. If the feed is wet, the profile must be adjusted, or the feed must be dried upstream.

Step 3: Adjust Camera and Lamp Settings

Start with the manufacturer’s default settings for the material type. If you have a specific material profile, load it. If not, use the generic profile for the material class.

Adjust the lamp intensity if the material is highly reflective or transparent. High-gloss PET can create glare that washes out the image. Lowering the lamp intensity or adjusting the camera angle can help.

Check the camera focus. Use the live feed to ensure the material edges are sharp. Blurred images reduce identification accuracy.

Begin with the default settings. Do not change everything at once. Change one parameter, observe the effect, and then move to the next. This method isolates the variable that causes the change in performance.

If the image is too bright, lower the lamp intensity. If the image is too dark, increase it. The goal is to balance the exposure so that the material is clearly visible against the conveyor background. If the background is too bright, the contrast is low. The sensor struggles to distinguish the target from the belt. Adjust the exposure to maximize the difference between the target and the background.

Check the color balance. If the material is a specific color, such as green HDPE, the color balance must be set correctly. A misaligned color balance can cause the sensor to misidentify the material as a different plastic. Use the live feed to verify the colors match the actual material.

Adjust the camera focus. Use a high-contrast object to test the focus. Place a piece of material with a sharp edge on the conveyor. Zoom in on the live feed. If the edge is blurry, adjust the focus until it is sharp. A sharp image allows the software to identify the material’s shape and texture accurately.

Step 4: Calibrate the Material Profile

Load the correct material profile for your stream. If you are sorting PET, select the PET profile. If you are sorting HDPE, select the HDPE profile.

If the sorter allows custom profiles, create a new one. Use the live feed to capture the specific material you are sorting. Train the software to recognize the shape, color, and spectral signature of your material.

Save the profile. Name it clearly, such as “PET 50-200mm” or “HDPE Mixed.” This makes it easier to switch profiles if the feed changes.

Select the appropriate material class. For example, if you are sorting PET bottles, select the “PET Bottle” profile. If you are sorting PET film, select the “PET Film” profile. The spectral signatures of these two materials are different. The algorithm must be trained on the correct class.

If the sorter has a “learn” function, use it. Place a clean, representative piece of material on the conveyor. Activate the learn function. The sensor will capture the spectral data of that piece. Repeat this process with several pieces of the same material. The more samples the software sees, the better the profile becomes.

Avoid using dirty or damaged material for training. A dirty piece of plastic has a different spectral signature than a clean piece. If you train the software on a dirty piece, it will reject clean pieces in the future.

Save the profile. Use a clear name. “PET 50-200mm” is better than “Profile 1”. A clear name helps operators understand what the setting is for. If the feed changes, you can quickly select the correct profile.

Step 5: Set the Reject Threshold

The reject threshold determines how sensitive the sorter is to impurities. Set it to the highest purity you require. If you need 99% purity, set the threshold high. If you need 95% purity, set it lower.

Start with a conservative setting. Run the sorter for a short period. Observe the reject stream. If the reject stream is too large, you are rejecting too much clean material. If it is too small, you are letting impurities through.

Adjust the threshold in small increments. Record the settings you use. This creates a baseline for future reference.

The reject threshold is the sensitivity setting. A high threshold means the sorter rejects more material. It is conservative. It will reject material that is only slightly impure. A low threshold means the sorter accepts more material. It is lenient. It will only reject material that is clearly impure.

Start with a high threshold. Run the sorter for ten minutes. Check the reject bin. If the bin fills quickly with clean material, lower the threshold. If the bin is empty and the clean bin contains impurities, raise the threshold.

Adjust in small steps. Change the threshold by one unit at a time. Run the sorter for five minutes after each change. Observe the effect. This prevents you from overshooting the setting.

Record the threshold you use. Write down the setting, the feed composition, and the resulting purity. This data is useful for troubleshooting. If the purity drops later, you can compare the current settings to the known good settings.

Step 6: Adjust the Air Knife and Purge Cycle

The air knife blows the material out of the stream. If the air is too weak, the material will not move. If it is too strong, it will push clean material into the reject bin.

Adjust the air pressure to match the material weight and size. Lighter materials need less air. Heavier materials need more.

Set the purge cycle. The purge cycle is the time between sorts. A short purge cycle allows for higher throughput but may reduce purity. A long purge cycle improves purity but reduces throughput.

Test different purge cycle lengths. Find the balance that meets your purity target without sacrificing too much volume.

The air knife is the actuator. It pushes the material into the reject bin. The pressure must be high enough to move the material but low enough not to disturb the clean stream.

Adjust the air pressure. Start with the manufacturer’s recommendation. If the material is not being rejected, increase the pressure. If the material is being blown into the clean bin, decrease the pressure.

Test with different material sizes. A small, light piece of plastic needs less air than a large, heavy piece. If the feed is a mix of sizes, the pressure must be set to handle the heaviest piece. This may cause the lighter pieces to be over-blown.

Set the purge cycle. The purge cycle is the time between sorts. A short purge cycle means the sorter is ready to reject the next piece quickly. This allows for higher throughput. However, it may allow impurities to pass through if the sensor is not ready.

A long purge cycle gives the sensor more time to identify the material. This improves purity. However, it reduces throughput because the sorter is not rejecting material as quickly.

Test different purge cycle lengths. Start with a long purge cycle to establish a purity baseline. Then, gradually shorten the cycle until the purity drops. The shortest cycle that meets your purity target is the optimal setting.

Step 7: Test and Verify Purity

Run the sorter for a set period, such as one hour. Collect the clean stream and the reject stream.

Sample each stream. Weigh the samples. Count the number of impurity particles per kilogram. This gives you a quantitative measure of purity.

If the purity is below target, return to Step 4 or Step 5. Adjust the profile or the reject threshold. Repeat the test.

Document the results. Record the feed composition, settings, and purity results. This data is valuable for troubleshooting and future tuning.

Run the sorter for one hour. Collect the clean stream in one bin and the reject stream in another. Do not mix them.

Take a sample from the clean stream. Weigh the sample. Count the number of impurity particles per kilogram. This gives you a quantitative measure of purity.

If the purity is below target, return to Step 4 or Step 5. Adjust the profile or the reject threshold. Repeat the test.

Document the results. Record the feed composition, settings, and purity results. This data is valuable for troubleshooting and future tuning.

Parameter Low Setting High Setting Effect
Lamp Intensity Dim Bright Dim reduces glare; Bright improves resolution
Reject Threshold Lenient Strict Lenient increases throughput; Strict increases purity
Air Pressure Low High Low moves light material; High moves heavy material
Purge Cycle Short Long Short increases throughput; Long increases purity

Common Mistakes in Optical Sorter Tuning

The most common mistake is assuming the sensor is the only variable. If the feed is inconsistent, no amount of software adjustment will fix the problem.

Another mistake is over-tuning. Changing too many variables at once makes it impossible to know what caused a change in performance. Change one variable, test it, and record the result.

Operators often ignore the reject stream. If the reject bin is overflowing, the sorter is working too hard. If it is empty, the sorter is not working at all. Monitor the reject stream continuously.

Finally, do not skip the verification step. An optical sorter can appear to be working while failing to meet purity targets. Sampling and counting impurities is the only way to confirm performance.

The most common mistake is assuming the sensor is the only variable. If the feed is inconsistent, no amount of software adjustment will fix the problem.

Another mistake is over-tuning. Changing too many variables at once makes it impossible to know what caused a change in performance. Change one variable, test it, and record the result.

Operators often ignore the reject stream. If the reject bin is overflowing, the sorter is working too hard. If it is empty, the sorter is not working at all. Monitor the reject stream continuously.

Finally, do not skip the verification step. An optical sorter can appear to be working while failing to meet purity targets. Sampling and counting impurities is the only way to confirm performance.

Final Verification and Documentation

Once you have reached your target purity, lock in the settings. Save the profile. Document the feed composition, the settings used, and the purity results.

Schedule regular maintenance. Clean the lens and lamp housing weekly. Check the air knife for wear. Review the reject stream daily.

If the feed changes, re-tune the sorter. A profile that works for clean PET bottles will not work for wet, matted PET film. Keep the tuning process active and responsive to changes in the feed stream.

Once you have reached your target purity, lock in the settings. Save the profile. Document the feed composition, the settings used, and the purity results.

Schedule regular maintenance. Clean the lens and lamp housing weekly. Check the air knife for wear. Review the reject stream daily.

If the feed changes, re-tune the sorter. A profile that works for clean PET bottles will not work for wet, matted PET film. Keep the tuning process active and responsive to changes in the feed stream.

Frequently asked questions

How often should I clean the optical sorter lens?

Clean the lens weekly or whenever you notice a drop in sorting accuracy. Dust and moisture accumulate quickly in a MRF environment.

Can I tune an optical sorter for multiple materials at once?

Yes, but it reduces purity. A multi-material profile is less specific than a single-material profile. Use single-material profiles when possible.

What should I do if the reject stream is too large?

Lower the reject threshold or increase the purge cycle length. This will reduce the number of clean materials being rejected.

How do I know if the lamp is failing?

Look for inconsistent image brightness or color shifts in the live feed. If the sensor misses materials that were previously sorted, check the lamp.

Can I use an optical sorter for wet materials?

Yes, but performance will drop. Wet materials are harder to identify and may clump together. Dry the material before sorting if possible. ===END===