Product · Matvision Grade
Separate a mixed metal stream into saleable fractions, in one pass.
Multi-sensor identification and five to twelve robots to split a mixed stream into saleable grades, family by family and, with LIBS, alloy by alloy.
> 95 %
Final quality rate per output
1 – 3 t/h
Throughput
10 – 150 mm
Particle size
5 – 12
Robots, one output per category
Throughput and purity depend on the stream and on the configuration. They are assessed during the performance study. A line is configured for the value it recovers, not for the number of sensors it carries.
Applications
Built for your stream.
Non-ferrous first, where one stream holds the most prices; then everything worth splitting rather than cleaning.
Zorba and Zurik
Every family in one pass: nine outputs where an optical sorter gives two.
WEEE and batteries
The rest split by material family once the batteries are out.
E40 steel scrap
An E40 load whose contaminants are worth their own outputs rather than one reject.
How it works
Twelve robots,
one belt.
- 01
Hopper, incline, vibrating feeder
The feeder spreads the layer and sets the working speed. It is what makes the presentation usable by the sensors, so it conditions every measurement downstream.
- 02
Read in one pass
3D, colour and X-ray transmission on every piece. With LIBS, the elemental composition as well.
- 03
Every piece gets a destination
At the exit of the sensing block each piece carries its class: copper, zinc, lead, aluminium, stainless, nickel-plated, and with LIBS the grade inside a family.
- 04
The robots pick in cascade
They work from upstream to downstream, each on the pieces assigned to it. The line runs continuously and in parallel.
- 05
One output per category, in one pass
What an optical sorter would need one pass per category to achieve.
- 06
One report per output
Each output leaves with its own composition: what it holds, in pieces and in weight, and the software version, model and thresholds behind the figures. The line does not stop to produce it.
The problem it solves
An optical sorter splits in two.
A stream is rarely worth two prices.
After the eddy current separator, and often after an optical sorter has removed the bulk of the aluminium, what remains is the difficult part of the Zorba: copper, zinc, lead, brass, stainless and nickel-plated pieces, all of similar size, all mixed. Sorting it one category per pass means one pass per category. Sorting it on surface colour alone means missing what the surface does not show.
Matvision Grade identifies each piece on volume, colour, density and, with LIBS, composition, then places it in the output where it is worth the most. One pass.
Application cases
One pass, every fraction.
Zorba by family, in one pass
Zorba contains the non-ferrous metals recovered after shredding and eddy-current separation. Once the easier fractions have been removed upstream, what is left is difficult to separate by physical properties alone. Matvision Grade separates copper, zinc, lead, aluminium, stainless and nickel-plated pieces, and detects unwanted items such as printed circuit boards and lead.
Base configuration: colour camera, 3D profilometer, X-ray transmission. One to three tonnes per hour on 10 to 100 mm material, depending on stream composition and average piece size
Inside a family: alloys by grade with LIBS
LIBS reads elemental composition, which colour and density cannot give. Stainless steel by grade, typically 304 against 316, a recurring request. Aluminium cast against wrought, then wrought by series: 1000, 2000, 3000, 4000, 5000, 6000, 7000. Two visually identical pieces, two different outputs.
Sensors: LIBS, on top of the base configuration
Zurik
A more heterogeneous mixed metal stream: steel, aluminium and copper with zinc, lead, nickel and iron, usually pre-treated by magnetic separation and screening. Matvision Grade resolves the residual stream into metal families that a buyer will price separately.
WEEE by material family
Shredded electrical and electronic equipment sorted by material family. Batteries sorted by chemistry once they have been set aside upstream.
Production scrap, back to its own families
A production stream is a known set of alloys in unknown proportions. Split back into its families, it returns to the melt at the value of each family rather than at the price of a mixed lot.
IBA non-ferrous, split in one pass
The non-ferrous fraction recovered from bottom ash holds several families at once. One pass separates the fractions a refiner buys separately, instead of selling them together at the price of the least valuable.
Configuration
Five to twelve robots,
one output per category.
| Sensing core | Matvision Sense: 3D, colour, XRT. LIBS for alloy grades. |
|---|---|
| Robots | Five to twelve. |
| Ejection | Robotic picking, one output per category, in combination with pneumatic ejection where a class only needs removing. |
| Conveyor | Included. |
| Particle size | 10 to 150 mm. |
| Quality | Final quality rate above 95 percent per output, measured on the material concerned. |
| Throughput | One to three t/h, depending on composition and piece size. |
| Outputs | As many as the stream holds categories of value. Sorting containers adapted to the stream. |
| Streams | Zorba, Zurik, production scrap, e-waste, IBA. |
| Return on investment | One to three years, depending on the composition of the stream. |
| Naming | Matvision Grade, then robots and sensors: Grade 6 CDX, Grade 12 CDXL. |
Matvision Grade is built from the same modules as Matvision Refine. A line can start with fewer robots and grow with the stream; sensors can be added later if market conditions change what is worth separating.
Configured, not chosen
One line, many ways to run it.
Two plants with the same stream rarely want the same machine. What varies is not the sensors alone: it is the strategy, the number of arms, how the work is shared between them, and what actually takes a piece off the belt.
Modes and recipes
A line holds several recipes and moves between them. A stream that changes with the delivery does not need a new machine each time; it needs the right recipe selected, and the operator to be able to select it.
Sorting strategies
Positive or negative: pick what you want, or take out what you do not and let the rest run. A class can be given to one robot, or spread across several when its share of the stream justifies the redundancy.
A hazardous class can be routed straight out on its own, on a different rule from the rest: lithium cells before a shredder, for instance.
Five to twelve robots
The number follows how many categories the stream is worth splitting into, and how dense each of them is. A line can start with fewer arms and grow with the stream.
Robots, air, or both
Picking is not the only way off a belt. A pneumatic ejector handles what is small, fast or too abundant to be worth an arm, and the two work on the same line: robots on the pieces that need placing, air on the ones that only need removing.
Grippers, and who takes what
The gripper is chosen for the material. The work is then shared between the arms: some take everything that passes, others only the large pieces, others the small ones. How that sharing is set is often what decides the rate a line actually holds.
The sensing core
Matvision Sense sits at the head of the line, in the configuration the stream justifies: from a single sensor to the full stack. It is the same core, and the same models, as a Matvision Sense unit on its own.
What we measure, what we promise
Purity per output,
recovery per class.
Purity per output and recovery per class are measured during the performance study on your material, then written into the offer as the KPIs we commit to. Precision is a setting between purity and recovery, chosen with you according to what you sell and at what price.
Not sure how far to sort? Start with the measurement. Matvision Sense ↗