Developing a Seasoning Calibration for Snack Foods with NIR

How potato chip, pretzel, popcorn, cheese puff, and other snacks makers build their own  NIR seasoning calibrations

Moisture and oil have been measured in snack plants for decades. Seasoning, in many plants, still is not. Manufacturers frequently instead rely on the applicator running as it was set — even though applicators clog and drift, and a calculation based on production volume does not reflect what actually stuck to the product.

Seasoning is often the most expensive component in the process. A blend running one to two percent over target does not immediately look wrong on the line but can add thousands of dollars a week to the cost of goods. Running under target does not look wrong either, until consumers decide the product tastes flat.

The good new - NIR analyzers, like the Phoenix 5000 NIR snack food analyzer, are now precise enough to measure applied seasoning directly, in about 30 seconds, from the same sample cup already used for moisture and oil.

How NIR Measures Moisture and Oil in Snack Foods

Near-infrared spectroscopy reads molecular bonds. O–H bonds in water and C–H bonds in fat / oil each absorb light at characteristic wavelengths between 1100 and 2500 nm. This absorbance is proportional to concentration.

Water and oil are ideal targets — both are present at high concentration, and both are strongly absorbing. That is why NIR took hold decades ago across potato chips, tortilla chips, pretzels, popcorn, extruded cheese puffs, and other snacks alike. A 30-second scan replaced oven drying and solvent extraction that took hours. NIR analysis can check both incoming raw material against spec, as well as measure moisture and oil in finished product.

Why Seasoning Is Harder to Measure Than Moisture or Oil

Salt, a primary component of seasoning, does not absorb in the NIR spectrum. However, all seasonings contain individual components (Sugars, dairy solids, spices, carriers) that do absorb in the near infrared. The signal is small, spread across the spectrum, and tangled with the much larger moisture and oil signals sitting beside it.

Pulling a reliable number out of that variation takes a high quality instrument with low noise

performance. Most lab or at-line NIR units were not acceptable, so plants relied on workarounds: the applicator setpoint, which assumes the equipment is doing what it was told; color, which fails on seasonings with little visible color; or salt titration, which takes 5–15 minutes, requires silver nitrate to buy and dispose of, and varies from operator to operator.

The Phoenix 5000  changes that arithmetic. This high performance instrument is a scanning monochromator-based NIR instrument which covers the full 1100–2500 nm range. This provides accurate and repeatable results. Seasoning comes off the same 30-second scan that reports moisture and fat.

What a Seasoning Calibration Is, and Why Yours Must Be Custom

A seasoning calibration is a mathematical model that relates the NIR spectrum of a ground snack sample to the percentage of seasoning applied to it. It is built by scanning samples whose seasoning level is already known, and it is specific to one seasoning blend on one base product.

That specificity is important. For moisture and oil you can often start from a broadly applicable calibration — water is water in anyone's potato chip. Seasoning is different, for the same reason the product sells: your blend's ratio of dairy solids to onion powder to salt is proprietary, your target level is your decision, and your base chip carries its own spectrum underneath it.

So, no vendor can hand you a finished seasoning calibration. You develop one per seasoning — five flavors would mean five calibrations.

Developing one is easier than it sounds. Just collect samples spanning the range you care about, record the NIR spectrum, add a trusted reference value for each, and let the chemometric software find the relationship between them. From then on, the instrument applies that relationship to every scan.

Sample off the line. Pull finished product, grind it, scan it, and measure its seasoning level by whatever reference method your lab already trusts — either by titration or by making samples at known seasoning levels. Pair each spectrum with its reference value and repeat across production days, seasoning lots, and applicator settings. The reference must come from the same ground material that was scanned; including process variation teaches the model sampling error as though it were real. This anchors the calibration in genuine production variation, which is what makes it hold up.

Two cautions: your reference method's own error propagates into the model, so it is best to run duplicates and average them; and normal production will cluster tightly around the target, so deliberately run the applicator high and low for short intervals to spread the range out.

How to Make a Calibration Hold Up in Production

• Bracket the range generously. Cover target plus or minus 1-2%.

• Grind identically every time. Particle size changes how light scatters before it reaches the sample. Use a timer rather than operator judgment.

• Hold samples back for validation. Predicting samples the model never saw is the only honest test of how it will perform in production.

• Maintain it. Check against your reference method routinely.

  
  

Benefits of Measuring Seasoning with NIR

• Seasoning cost under control. The most expensive ingredient becomes a measured input instead of an assumption.

• Consistent flavor. Application rate is what makes a bag taste the same in September as it did in March.

• Faults caught in minutes. A drifting applicator costs one adjustment instead of a hold on finished goods.

• No hazardous chemistry. No silver nitrate to buy, store, or dispose of, and no operator-to-operator variation.

• Three results from one scan. Moisture, fat, and seasoning from a single sample cup.

• A record, not a spot check. Export to Excel, LIMS, or Modbus turns individual readings into a trend line.

Getting Started

The hardest part of developing a seasoning calibration is believing it is hard. In practice it is a scale, a grinder, a few afternoons of sample collection, and an instrument precise enough to see the difference.

Start with your highest-volume SKU, where a percentage point of overage is worth the most. Build the set, validate it, and run it alongside your current method until the plant trusts the number. Then do the next flavor.

Blue Sun Scientific builds every Phoenix 5000 in Maryland, and our support team averages more than ten years of NIR experience each. We work through calibration development with customers regularly — sample set design, modeling, and validation. If you are weighing whether seasoning measurement is worth the effort in your plant, we are glad to help you build the applications.

Frequently Asked Questions

How long does it take to develop an NIR seasoning calibration? 

Most manufacturers collect a usable sample set over a few days of normal production.

How many samples do you need? 

Plan on 30 to 50 samples spanning target plus or minus a full percent, with a portion held back for validation. More samples help, but range coverage and lot-to-lot seasoning variation matter more than raw count.

How accurate is NIR for seasoning? 

A well-built calibration should predict routine production to roughly 0.3% - 0.5% or better — comfortably inside the tolerance most plants need to hold.

Can you buy a ready-made seasoning calibration? 

No. Seasoning blends are proprietary and application targets differ by manufacturer, so each company must develop its own. Blue Sun can help design the method development for ease of implementation.

Do you need a separate calibration for every flavor?

Yes — one per seasoning blend. Similar seasonings (for example, similar types of BBQ flavors) may be able to be combined to save some time and effort. You do not need one per SKU, per line, or per shift.

Does NIR measure salt directly? 

No. Salt is ionic and has no NIR absorbance. The instrument tracks the organic and hydrated material that travels with it in the blend, which is why the calibration is tied to a specific formulation.

What sample preparation does it require? 

Grind approximately 100 - 200 g of product to a consistent particle size, fill the sample cup, and scan. Using a timer on the grinder matters more than any other handling step. Care should be taken to not grind so much that the oil starts to separate from the chips.