spectroscience.com / blog / instrument path / part 2
Instrument path · Part 2 of 4Choose the technology

FT-NIR vs dispersive vs filter NIR: which instrument fits your lab?

Vitaly Kirkpatrick · first published August 21, 2025 · re-templated sample · 10 min read

Short answer: all three measure the same near-infrared light; they differ in how they pick wavelengths apart. FT-NIR uses a moving-mirror interferometer and a laser reference, so its wavelength axis is the most exact and repeatable: the strongest choice for method development and for sharing calibrations across instruments. Dispersive instruments spread light with a grating; diode-array versions have no moving parts and suit at-line and online work. Filter instruments read a handful of fixed wavelengths: fast, rugged and simple, but only for the analytes they were built for.

Same sample, three ways of reading itInteractive · simulated spectrum
The grey curve is a made-up absorbance spectrum with four broad bands, like most food and feed spectra, plus one narrow band near 2310 nm. Each button shows what that instrument type records: a few fixed points, a few hundred pixels blurred by the slit bandwidth, or a fine, laser-referenced grid. Slide the wavelength axis to see why axis stability matters more than raw resolution: a small shift moves every point, and a filter instrument cannot see it happen. Simulated in your browser to show the idea, not data from a real instrument.

What is FT-NIR, and how is it different from "regular" NIR?

FT-NIR is not a different kind of light. It is near-infrared spectroscopy (roughly 780 to 2500 nm) read through a Fourier-transform instrument. Light passes through an interferometer, one mirror moves, and the detector records an interferogram: every wavelength at once, encoded as a pattern. Software turns that pattern into a spectrum with a Fourier transform.

Two things follow. A helium-neon or diode laser tracks the moving mirror, so the wavelength axis is set by physics rather than by mechanical alignment, and it stays put from scan to scan and from unit to unit. And because the whole range lands on the detector together, FT-NIR collects a full, fine-grained spectrum quickly. When I train QC teams, this is the one point I ask them to remember: the advantage of FT-NIR is less "more resolution" and more "the same wavelength every time."

A compact dark benchtop near-infrared analyzer with a cup of ground grain on its sample window
A benchtop FT-NIR is typically a closed box with a sample window or cup on top, run from software on a PC. The interferometer is sealed inside. Illustration, not a specific product.
Where the wavelengths come fromHow each design works
Simplified optics. Top: the interferometer splits the beam, one mirror moves, and the recombined light hits a single detector. Middle: a fixed grating fans the light across a row of detector pixels (diode array); a scanning version rotates the grating instead. Bottom: a wheel of interference filters passes one narrow band at a time.

How does a dispersive NIR instrument work?

A dispersive instrument uses a diffraction grating to spread light by wavelength, the way a prism makes a rainbow. There are two common builds:

  • 1Scanning monochromatorThe grating rotates and steps each wavelength across a single detector. Full-range coverage, but there is a moving part and a scan takes longer.benchtop, lab
  • 2Diode arrayThe grating is fixed and a row of detector pixels reads every wavelength at once. No moving parts, very fast, tolerant of a conveyor or a process line.at-line, online
An at-line diode-array NIR analyzer mounted above a conveyor belt of feed pellets
A diode-array analyzer at-line: a sealed stainless head looking down at product on a conveyor, reading every few seconds with no moving parts. Illustration, not a specific product.

The wavelength axis of a dispersive instrument depends on how precisely the grating and detector are aligned, and that can move a little with temperature or after service. It is not a weakness in routine QC, as long as you check it. Run a certified wavelength standard on the schedule your vendor sets and after every repair.

What does a filter-based NIR instrument measure?

A filter instrument does not record a spectrum at all. It reads a short, fixed list of wavelengths through interference filters, often on a rotating wheel. The wavelengths are chosen for the analytes the instrument was built for, such as protein, moisture and oil in grain.

A rugged dark grain analyzer with a hopper of wheat at a receiving bay, a scoop lying beside it
A filter grain analyzer at intake: grain goes into the hopper, the result appears in under a minute, and the operator records it. Illustration, not a specific product.

That design is the reason filter units are quick, simple to run and hard to break. It is also their limit. You cannot add a new analyte that needs wavelengths the filters do not cover. And with only a few numbers per sample, the instrument has little information to tell you when a sample does not look like the ones the calibration was built on. Outlier detection is the part people underestimate.

FT-NIR vs dispersive vs filter: side by side

CriterionFT-NIRDispersiveFilter
How wavelengths are separatedinterferometer + Fourier transformdiffraction grating (scanning or diode array)fixed interference filters
What you get per samplefull spectrum, fine wavelength gridfull spectrum, a few hundred pointsa few fixed wavelengths
Wavelength-axis stabilityhighest (laser referenced)good; check with a standardfixed by the filters
New analytes lateryesyesonly within the existing filters
Outlier detectionstrong (full spectrum)strong (full spectrum)limited
Moving partsmoving mirrornone (diode array) or rotating gratingfilter wheel in many designs
Typical homecentral lab, method development, process analyzers with probesroutine QC lab, at-line, online on a conveyorreceiving stations, single-product lines
Key point

Fit beats specs. Wavelength-axis stability, sample presentation and calibration quality decide your results more than the technology label on the box.

Notice what is missing from the table: a "best" column. In my work with food manufacturers the expensive mistakes rarely come from a bad instrument. They come from a good instrument chosen for the wrong job.

Is FT-NIR worth the extra cost?

FT-NIR is worth it when one of three things is true. You develop methods or run many different products, so you need full spectra and the flexibility to add analytes. You run more than one instrument or more than one site, and you want calibrations to move between them with as little rework as possible; a stable wavelength axis makes calibration transfer much easier. Or you need a process analyzer with fibre-optic probes and a long life.

It is usually not worth it for a single product and two or three analytes measured at a receiving station. There, a well-maintained dispersive or filter instrument will give the same answer at a lower total cost. Remember that most NIR bands in food are broad. Past a certain point, extra resolution adds data points, not information. Before you compare prices, compare total cost of ownership: calibration development or transfer, reference analysis, service contracts, training and integration with your LIMS. Our guide to calculating NIR ROI walks through the full list.

Which technology fits your job?Interactive · rule of thumb
A transparent rule of thumb, not a vendor ranking: each answer adds points to the technologies it favours, and the reasons are listed under the result. Use it to frame the questions for your vendors, not to replace a trial on your own samples.

NIR vs FTIR: are they the same thing?

No, and the confusion is common because "FTIR" and "FT-NIR" share the same Fourier-transform technique. In everyday use, FTIR means mid-infrared spectroscopy (about 2.5 to 25 µm, or 4000 to 400 cm⁻¹). That region holds the strong fundamental vibrations of molecules, so it needs thin samples or an ATR crystal. NIR covers about 780 to 2500 nm, where the bands are weaker overtones and combinations. Weak bands are what let NIR shine light through or into a cup of whole grain or ground feed with little or no sample preparation.

For food and feed QC that means: FTIR is the stronger tool for identifying compounds and studying structure, and NIR is the faster tool for routine composition (moisture, protein, fat, fibre) on thousands of samples. We compare NIR with mid-infrared and Raman in more depth in NIR vs MIR and Raman in oilseed labs.

Can a calibration move between technology types?

Between two instruments of the same design, usually yes, with a standardization step. Our guide to transferring NIR calibrations covers how. Across technology types it is much harder. A filter instrument sees only a few wavelengths, so a full-spectrum model cannot be loaded onto it. Going from dispersive to FT-NIR, or the other way, changes the wavelength grid, the bandwidth and the detector response together. Plan for a rebuild with your own reference samples rather than a transfer, and put the reference analysis for it in the budget before you sign.

Watch out

Budget the rebuild before you sign. Switching technology types usually means building new calibrations from your own reference samples, not transferring the old ones.

Try it free

Not sure what an interferometer, a grating or SEP means? The free NIR glossary explains 29 NIR terms in plain English. Comparing instruments on your own validation set? Enter the reference standard deviation and SEP in the free RPD and SEP calculator to see whether each one is good enough for QC.

Worked example: one feed mill, two jobs

ILLUSTRATIVE A made-up mill to show the reasoning; not a customer case.

A feed mill has two needs. At the intake, trucks of soybean meal and corn need moisture and protein in under a minute, run by whoever is on shift. In the lab, a nutritionist wants amino acids and fibre fractions across a dozen ingredients and finished feeds, and the company plans a second mill next year.

Put through the selector above, the intake scores filter or diode-array dispersive: two analytes, two products, operators, a dusty bay. The lab scores FT-NIR: many analytes, a specialist, and calibrations that must travel to the second site. One expensive instrument would be wrong for at least one of the two jobs. Two instruments chosen for their jobs, with the lab instrument as the place where calibrations are developed and checked, cover both.

Try this first

Run a trial on your own samples. Ask every vendor to scan the same 20–30 of your samples, including your hardest ones, and compare against your reference lab values.

Course · Lesson 19

Different Types of NIR Instruments: FT-NIR, dispersive and filter-based systems, with their uses and limits in grain, feed and food labs.

Before you call a vendor

Take the questions to ask a NIR vendor before you sign to your first demo.

Quick check

1. Why do FT-NIR instruments make calibration sharing easier?
2. A receiving station measures moisture and protein in one grain. Which is usually enough?
3. You want to add a new analyte next year. Which design is most likely to block you?
Answer the three questions above to see your score.
Where you are on the Instrument path

Step 2 of 4: Choose the technology

2 of 4 steps
  1. 1Inside the instrumentLight source, sample interface, detector.Read part 1 →
  2. 3Benchtop, at-line or process?Where the instrument should live.Read part 3 →
  3. 4Move a calibrationTransfer and standardization.Read part 4 →

Frequently asked questions

What does FT-NIR stand for?

Fourier-transform near-infrared spectroscopy. The instrument records an interferogram with a moving-mirror interferometer and converts it into a near-infrared spectrum with a Fourier transform.

Is FT-NIR more accurate than dispersive NIR?

Not automatically. Accuracy for protein or moisture depends mostly on the calibration, the reference method and sample presentation. FT-NIR's real advantage is a very stable, laser-referenced wavelength axis, which helps most when calibrations are shared between instruments or developed for many products.

What is the difference between NIR and FTIR?

FTIR usually means mid-infrared spectroscopy (about 4000 to 400 cm⁻¹), which reads strong fundamental vibrations and needs thin samples or ATR. NIR (about 780 to 2500 nm) reads weaker overtone and combination bands, which lets it measure thick, unprepared samples. FT-NIR applies the Fourier-transform technique to the NIR region.

Can FT-NIR be used online in a process?

Yes. Process FT-NIR analyzers with fibre-optic probes or flow cells are used online. They need stable mounting and a protected enclosure, so plan the installation with the vendor. Diode-array dispersive instruments are the other common choice for online work because they have no moving parts.

Which NIR instrument is best for feed analysis?

It depends on the job, not the brand. Intake checks of a few analytes favour a filter or diode-array instrument run by operators. A central lab developing calibrations for many ingredients and finished feeds favours FT-NIR or a full-range dispersive benchtop. Trial any candidate on your own samples before you buy.

Does more resolution give better calibrations?

Only up to a point. Most NIR bands in food and feed are broad, so beyond a few nanometres extra resolution adds data points rather than information. Wavelength-axis stability and a good sample presentation usually matter more.

Next questions people ask

NIR Fundamentals course · 32 lessons

Choose, run and trust the right NIR instrument

Lesson 19 compares FT-NIR, dispersive and filter systems, and the other 31 lessons cover everything from Beer-Lambert to PLS calibration and transfer. One payment, lifetime access.

Sources

  1. Pasquini, C. "Near infrared spectroscopy: a mature analytical technique with new perspectives — a review." Analytica Chimica Acta 1026 (2018): 8–36. doi:10.1016/j.aca.2018.04.004
  2. Ozaki, Y., Huck, C., Tsuchikawa, S., Engelsen, S. B. (eds.). Near-Infrared Spectroscopy: Theory, Spectral Analysis, Instrumentation, and Applications. Springer, 2021.
  3. Workman, J. "A review of calibration transfer practices and instrument differences in spectroscopy." Applied Spectroscopy 72 (2018): 340–365. doi:10.1177/0003702817736064
  4. Burns, D. A., Ciurczak, E. W. (eds.). Handbook of Near-Infrared Analysis, 3rd ed. CRC Press, 2007.
  5. United States Pharmacopeia. General Chapter <1119> Near-Infrared Spectroscopy (instrument qualification and wavelength checks).