Chocolate machinery

The machines between a sack of beans and a wrapped bar, in the order the cocoa meets them. For each: what it is for, how its inventors described it working, and what has actually been measured about what it does to chocolate.

10 machines and processesStudies and patentsNo makers, models or prices

How to read this page

  • A study reports what a machine did to a product under stated conditions, usually in a laboratory or pilot plant. It is linked to its summary and limits.
  • A patent is used only for how an inventor described a machine and when. It is not evidence that the machine was built or used.
  • This is not a buying guide. Nothing here recommends a machine, and choosing one depends on scale, product and budget that no study addresses.

For what each step is and how to do it, see making chocolate.

Roasters

The first machine that changes flavour, and the one that makes raw beans safe. Beans or nibs are heated in a rotating drum, in a stream of hot air, or by newer means. What is controlled is temperature, time and how the heat is delivered.

  • The type of roaster changes the bean physically. Roasting in a fluidised bed, where hot air lifts the nibs, was twelve times faster than an oven and tripled the porosity of the nib where the oven doubled it. Laboratory scale. (Peña-Correa et al. 2023)
  • Microwave roasting gave a browner bean with a more pronounced cocoa aroma in less time than convection, and a trained panel could tell the two chocolates apart although colour, hardness and flow did not differ. (Lemarcq et al. 2022; Schouteten et al. 2023)
  • Roasting creates no new aroma compounds in the bean, in one study of Criollo cocoa: the same 42 were present before and after, in different amounts. (Frauendorfer and Schieberle 2008)
  • Flavour and polyphenols pull in different directions. The roast of cocoa powder that scored best with tasters had the least polyphenol; in another study polyphenols rose before they fell; and at equal final moisture a hotter, shorter roast lost less than a cooler, longer one. (Asefi et al. 2023; La Mantia et al. 2023; Ioannone et al. 2015)
  • Roasting in superheated steam left fewer total phenolics than hot air and more of some individual compounds. (Navare et al. 2026)
  • A 1927 patent application describes roasting as a continuous stream: beans roasted, cracked and shelled, the nibs roasted again and ground, without batches. (source 1)

Not shown: A best roast. The studies disagree on what roasting does to polyphenols and each used its own beans and equipment.

Not shown: How drum roasters of different sizes compare. No such study was found.

Not shown: Whether any of these roasts is sufficient for safety. That is a separate measurement, covered with Salmonella.

Crackers and winnowers

Roasted beans are broken and the light shell is blown or sieved away from the heavier nib. How clean the separation is matters for flavour, for contaminants and in law.

  • A 1903 patent application claims one machine combining cracking, grading and winnowing, and says separate machines for each had been customary. (source 2)
  • How much shell is left can be estimated without chemistry: near-infrared light predicted shell content in nibs to within about two percentage points. (Drees et al. 2023)

Not shown: How well any winnower separates. No study measuring separation efficiency was found.

Grinders and refiners: ball mills, roll refiners and melangers

Nibs are ground to liquor, and liquor with sugar is ground again until the particles are too small for the tongue to find. Two industrial routes exist: steel rollers that crush a thin film of paste, and ball mills that stir it with steel balls. Small makers use stone melangers.

  • The two routes do not make the same chocolate. From one milk chocolate recipe, roller refining gave a broad, two-peaked spread of particle sizes and ball milling a narrower one short of fine particles, and the flow differed accordingly. (Bolenz and Manske 2013)
  • Ball-milled chocolate had to be made by blending separately ground coarse and fine fractions to flow like the roller-refined kind. (Bolenz et al. 2014)
  • At the same particle size, roller-refined compounds flowed more easily than dry-ground ones, and small-scale ball-milled chocolate had a higher yield value than roll-refined and conched chocolate at similar viscosity. (Gallery and Agoda-Tandjawa 2026; Saputro et al. 2019)
  • Settings matter as much as the machine. Re-tuning the speed and running time of one factory ball mill raised its output by 43%. (Alamprese et al. 2007)
  • A 1951 patent application describes the standard process of its day in passing: the mixture ground on roll refiners, cocoa butter added, then worked in a conche or melangeur with air so that unwanted aromas escape. (source 3)
  • Stone melangers, the machine most small makers actually own: the one comparison found had an abstract too damaged to use.

Not shown: Which route tastes better. No study compared the flavour of the two.

Not shown: Anything measured about melangers.

Conches

A heated vessel that stirs and shears the refined mass for hours, driving off moisture and acids and coating the particles in fat so that the mass flows.

  • It does both jobs at once and they conflict. Conching lowered viscosity and lowered every odorant measured, wanted ones included; acetic acid fell most, by about 60%. Raising the temperature improved flow and stripped up to 78% of some odorants from the fat. (Guckenbiehl et al. 2022; Guckenbiehl et al. 2024)
  • Longer is not always more. In one factory's refiner-conche nothing measured, by instrument or by trained panel, changed between 12 and 20 hours. (Valverde-Ayllon et al. 2022)
  • Patents trace attempts to shorten it. One filed in 1920 says “fondant” chocolate was until then treated for many hours in conches and claims a short, violent agitation in its place; one of 1922 improves the conche with a reciprocating roller; one granted in 1993 claims conching continuously, with material fed in and discharged without stopping. (source 4; source 5; source 6)
  • The invention of the conche is usually dated to 1879 and credited to Rodolphe Lindt. No patent was found. The company's own history page gives the date and calls the familiar story of the machine left running “allegedly” the result of chance. (source 7)

Not shown: How long to conche. One factory, one machine and one recipe found 12 hours enough.

Not shown: That any of the patented alternatives replaced the conche in practice. A patent is a claim.

Cocoa presses and alkalising

Pressing liquor separates cocoa butter from a cake that is milled into cocoa powder. Alkalising, or Dutching, treats nibs, liquor or powder with alkali to darken the colour and soften the taste.

  • The principle of an alternative to the hydraulic press is described in a 1910 patent application: roasted beans put through a continuous screw expeller until the residue can be ground directly to powder. (source 8)
  • Alkalising was already described as well known in a patent of about 1917, which claims doing it with ammonia gas in place of dissolved alkali. (source 9)
  • It costs flavanols, in proportion. Flavanol content fell in a straight line as the powder's pH rose, and about 60% of the flavonoids measured were lost, more of the epicatechin than of the catechin. (Miller et al. 2008; Andres-Lacueva et al. 2008)
  • It is fast: most of the change happened in the first minute. Alkali concentration mattered more than temperature or time for colour and pH, and theobromine was not significantly affected. (Sioriki et al. 2021; Demirci et al. 2024)
  • The pressing of cocoa butter is usually credited to a Dutch patent of 1828. No patent text or archive record was found.

Not shown: Anything measured about presses: pressure, yield or the fat left in the cake.

Not shown: That alkalised cocoa is worse. It has fewer flavanols and a different taste; which is better is a matter of use.

Tempering machines

Tempering cools and warms chocolate under shear so that its fat sets in the stable crystal form. The existing walkthroughs give the temperatures; this is what has been measured and claimed about the machine.

  • The principle has not changed in a century. A patent applied for in 1913 carries chocolate as a thin film on a drum against air held at a controlled temperature by a thermostat; one of 1931 pumps it through a water-jacketed device of large surface area on its way to the coating machine; one of 1988 claims continuous tempering through several cooling stages followed by heating stages. (source 10; source 11; source 12)
  • Seeding with ready-made stable crystals gave a denser structure and better resistance to bloom than conventional tempering, up to a point. (Svanberg et al. 2013)
  • Tempering may not always need a machine: adding 0.1% of a saturated phospholipid and cooling quickly without stirring gave chocolate in the stable form, in the laboratory. (Chen et al. 2021)

Not shown: How temper meters perform, or how tempering machines of different designs compare. No study was found.

Not shown: That the additive route works outside a laboratory.

Depositors, moulds and cooling tunnels

Tempered chocolate is deposited into moulds, shaken to release air, cooled in a tunnel and knocked out.

  • Cooling can be too fast as well as too slow. In a pilot tunnel, moulded dark chocolate released from the mould soonest at an intermediate rate of heat removal. (Grob et al. 2021)
  • Sticking is a matter of conditions: temperature, contact time and the humidity of the surrounding air all changed how firmly chocolate adhered to a polycarbonate mould. (Keijbets et al. 2010)
  • Hollow shells can be pressed instead of poured out: a patent granted in 1997 claims filling the mould and plunging in a member cooled below 0°C, which presses the chocolate into a shell. (source 13)
  • Vibrating tables, and depositing a shell and its filling in one shot: no study or early patent was found.

Not shown: A cooling tunnel setting. The optimum found belongs to one pilot tunnel and one chocolate.

Enrobers

A centre passes on a wire belt through a curtain of tempered chocolate and over a bath that coats its base.

  • A patent applied for in 1910 claims exactly that arrangement: a container holding chocolate at temperature and a pump forcing it through slotted conduits above and below a conveyor carrying the articles. (source 14)
  • The enrober is often said to be a French invention of 1903. No such patent was found.

Not shown: Anything measured: coating weight, tails, or the effect of belt speed and air. No usable study was found.

Aerating and 3D printing

  • Aerated chocolate is made by dispersing gas in molten chocolate and letting it expand. A British patent applied for on 11 July 1935, assigned to Rowntree, claims a honeycomb structure made by distributing bubbles in the fluid chocolate and reducing the pressure during moulding. The text names no product. (source 15)
  • A patent of 1981 claims dissolving a gas that is largely soluble in the chocolate under pressure and releasing it. (source 16)
  • The gas matters. Carbon dioxide and nitrous oxide gave more and larger bubbles than nitrogen and argon, and the chocolates tasted and felt different. (Haedelt et al. 2007)
  • The bubbles can be measured by X-ray scanning. (Sarfarazi et al. 2024)
  • Printing: blowing cooler air across the part let an extruded strand of chocolate span a wider gap, and printing speed made no difference to it. (Lanaro et al. 2017)

Not shown: That the 1935 patent is the origin of any particular product. It names none.

Making chocolate that survives heat

Not a machine but a process problem that six companies have patented answers to over seventy years. The approaches fall into three families.

  • Add a little water so the sugar and milk solids form a skeleton: claimed by General Foods in an application of 1951, with water limited to about 3%, and by Mars in 1956 with control of air humidity during refining. (source 3; source 17)
  • Disperse an emulsion or gel through the chocolate: a water-in-fat emulsion at 2 to 10% (Battelle, filed 1981), hydrated lecithin (Mars, filed 1991), a polyol gel in particles (Nestec, 1996). (source 18; source 19; source 20)
  • Make the fat discontinuous: fat held as separate particles in a sugar glass (Cadbury, 1978). (source 21)
  • Outside the patents, two laboratory routes have been measured: ethylcellulose, which gave milk chocolate that kept its shape at 40°C, and curing white chocolate in warm humid air. (Stortz and Marangoni 2013; Laughter et al. 2024)
  • One of the patents states as background that the fat of ordinary chocolate begins to soften at 28°C. (source 18)

Not shown: That any of these is in a product on sale. A patent records a claim, and the laboratory studies report no tasting.

Not shown: What heat-resistant chocolate tastes like.

Sources

Studies are cited by author and year and summarised, with their limits, on the research pages. The official documents are listed here. Each was downloaded and read, and the passage relied on was checked to be in it.

  1. United States Patent Office (scan via Google patent images). US1862166: Continuous chocolate making process (1932-06-07). patentimages.storage.googleapis.com/pdfs/US1862166.pdf
    Patent: evidence of a claim and a date, not of use · US
  2. United States Patent Office (scan via Google patent images). US762705: Combined machine for nibbing, grading, and winnowing cocoa-beans or the like (1904-06-14). patentimages.storage.googleapis.com/pdfs/US762705.pdf
    Patent: evidence of a claim and a date, not of use · US
  3. United States Patent Office (text via Google Patents). US2760867A: Finished chocolate product (1956-08-28). patents.google.com/patent/US2760867/en
    Patent: evidence of a claim and a date, not of use · US
  4. United States Patent Office (scan via Google patent images). US1387377: Manufacture of high-class 'fondant' chocolate and similar chocolates (1921-08-09). patentimages.storage.googleapis.com/pdfs/US1387377.pdf
    Patent: evidence of a claim and a date, not of use · US
  5. United States Patent Office (scan via Google patent images). US1563898: Conch for the manufacture of chocolate (1925-12-01). patentimages.storage.googleapis.com/pdfs/US1563898.pdf
    Patent: evidence of a claim and a date, not of use · US
  6. United States Patent and Trademark Office (scan via Google patent images). US5215771: Process for continuously conching chocolate mass (1993-06-01). patentimages.storage.googleapis.com/pdfs/US5215771.pdf
    Patent: evidence of a claim and a date, not of use · US
  7. Chocoladefabriken Lindt & Sprüngli AG. History (company history page). www.lindt-spruengli.com/about-us/history
    The company's own account · Switzerland
  8. United States Patent Office (scan via Google patent images). US1006913: Process of producing cocoa-powder and cocoa-butter (1911-10-24). patentimages.storage.googleapis.com/pdfs/US1006913.pdf
    Patent: evidence of a claim and a date, not of use · US
  9. United States Patent Office (scan via Google patent images). US1243244: Method of making so-called alkalized cocoa (1917). patentimages.storage.googleapis.com/pdfs/US1243244.pdf
    Patent: evidence of a claim and a date, not of use · US
  10. United States Patent Office (scan via Google patent images). US1141683: Chocolate tempering machine (1915-06-01). patentimages.storage.googleapis.com/pdfs/US1141683.pdf
    Patent: evidence of a claim and a date, not of use · US
  11. United States Patent Office (scan via Google patent images). US1898771: Apparatus for tempering chocolate (1933-02-21). patentimages.storage.googleapis.com/pdfs/US1898771.pdf
    Patent: evidence of a claim and a date, not of use · US
  12. United States Patent and Trademark Office (scan via Google patent images). US4859483: Method for continuous processing of substances containing cocoa butter or similar fats (1989-08-22). patentimages.storage.googleapis.com/pdfs/US4859483.pdf
    Patent: evidence of a claim and a date, not of use · US
  13. United States Patent and Trademark Office (scan via Google patent images). US5635230: Method and apparatus for producing shells of fat-containing, chocolate-like masses and food articles produced thereby (1997-06-03). patentimages.storage.googleapis.com/pdfs/US5635230.pdf
    Patent: evidence of a claim and a date, not of use · US
  14. United States Patent Office (scan via Google patent images). US1132253: Chocolate-enrobing machine (1915-03-16). patentimages.storage.googleapis.com/pdfs/US1132253.pdf
    Patent: evidence of a claim and a date, not of use · US
  15. UK Patent Office (text via Google Patents). GB459583A: Improved process for manufacturing articles of food or confectionery (1937-01-11). patents.google.com/patent/GB459583A/en
    Patent: evidence of a claim and a date, not of use · UK
  16. United States Patent and Trademark Office (scan via Google patent images). US4272558: Method and apparatus for producing a cellular product (1981-06-09). patentimages.storage.googleapis.com/pdfs/US4272558.pdf
    Patent: evidence of a claim and a date, not of use · US
  17. United States Patent Office (text via Google Patents). US2904438A: Chocolate product and process (1959-09-15). patents.google.com/patent/US2904438/en
    Patent: evidence of a claim and a date, not of use · US
  18. United States Patent and Trademark Office (text via Google Patents). US4446166A: Chocolate composition for the preparation of heat-resistant chocolate articles (1984-05-01). patents.google.com/patent/US4446166/en
    Patent: evidence of a claim and a date, not of use · US
  19. United States Patent and Trademark Office (text via Google Patents). US5149560A: Heat-resistant chocolate and method of making same (1992-09-22). patents.google.com/patent/US5149560/en
    Patent: evidence of a claim and a date, not of use · US
  20. United States Patent and Trademark Office (scan via Google patent images). US5523110: Chocolate heat-resistance by particulate polyol gel addition (1996-06-04). patentimages.storage.googleapis.com/pdfs/US5523110.pdf
    Patent: evidence of a claim and a date, not of use · US
  21. United States Patent and Trademark Office (scan via Google patent images). US4081559: Edible composition and method of manufacturing same (1978-03-28). patentimages.storage.googleapis.com/pdfs/US4081559.pdf
    Patent: evidence of a claim and a date, not of use · US

Reviewed 9 October 2026.

Machines with no entry

These are in everyday use. They have no entry because no study measuring them and no patent describing them was read, and an entry from general knowledge would look the same as one that was checked.

  • Bean cleaners and destoners
  • Cocoa butter presses, as measured
  • Panning and coating drums
  • Guitar cutters and other hand tools
  • Vibrating tables
  • Wrapping and packaging machines
  • Temper meters and other instruments used on the line

Related: chocolate technology in patents and records, the processing studies, the science of making chocolate, the troubleshooter and how cocoa and chocolate are tested. All data.