The 'old' metal detector learns new tricks: application advancements solve three nagging inspection challenges

Food & Drug Packaging, April, 2004 by Martin Lymn

In recent years, improvements in metal detection technology have been more evolutionary than revolutionary. Much of the progress has been in areas of enhanced user-friendliness and improved durability in hostile washdown environments. True performance improvements have been achieved more through intuitive set-up and more reliable operations than through new system capability.

But pocket-protector equipped whiz kids in the metal detector manufacturers' design labs have not been idle. New technologies, techniques and designs are continually under review.

Three innovations--some new, some not widely known--are shared here that can make a difference in online inspection.

Multiple units across a wide belt

Metal detection of wide-belt systems has always presented a challenge to the user. For example, in a row of 20 granola bars, maybe one is contaminated. All 20 bars in that row (and maybe the 20 in front and the 20 behind) have to be rejected as suspect as it's not known which particular bar was contaminated, in which lane. This is simply because the traditional food industry metal detector uses a single pair of detector "receiver" coils that encircle the entire belt and product, thus unable to distinguish between one section of belting and another.

Using a variant on an old technology originally used extensively in the textile, carpet and recycled board industries, the "single face" detector, two products have been developed that have a unique twist; one is currently in use for detection of metal contaminants in the flow of crushed glass and plastic regrinds in recycling facilities while the other product is already available to the food industry.

This approach, rather than using a single coil wrapping around the entire width of the belt, uses a series of flat miniature coils positioned beneath the belt. With the first technology, each of these coils scans a mere 1.25 inch of the belt width, reporting back to the control system on metal contamination in that section of the product alone.

In this way, a 30-inch wide belt can be scanned with a single detector monitoring 24 individual "zones," potentially one zone per lane of product. The additional information from this approach allows for an equal number of small reject devices to be installed such that only product from the contaminated "zone" is eliminated from the product stream. Additionally, this concept can develop a profile over time allowing identification of any condition that might be causing contaminant introduction always in the same lane of product.

Sensitivity in the order of 1.0mm of magnetic and non-magnetic contaminants can be achieved in most such applications with the first technology.

The other product goes a step further to eliminate screen issues by encircling the entire belt, in the traditional fashion with the metal detector case.

More expensive? Yes, but, especially for higher value products produced in this fashion, a good use of older technology to provide added protection, less waste and more information on the problem.

Automatic frequency optimization

As long as metal detectors have been in use in the food industry, we've wrestled with the challenge of setting operating frequency. Assigning the all-important operating frequency of the magnetic field is central to the detector's ability to "see" nonmagnetic metals, such as stainless steel.

The inspection system's ability to detect non-magnetic metals depends on another property-conductivity, the ability of the piece of metal to conduct electricity. The higher the operating frequency (measured in thousands of cycles per second; KHz) the greater the signal from non-magnetic metals can be expected to be.

Here's the problem ... moisture, salt and other elements within the product will make the product conductive, and high product temperatures can exacerbate this.

Today's metal detectors have automatic calibration or "learn" features to teach the unit what product signal looks like so it can be ignored. But these techniques are not foolproof, especially when product cross section, temperature or moisture levels will vary, as is likely in most situations.

For this reason, you need to optimize the detector operating frequency to keep the "product effect" at manageable levels, while getting as much signal from nonmagnetic metals as possible. If the frequency is too high, instability and false rejects due to product effect could result; too low, and stainless steel detection capability is diminished.

In 1997, "dual frequency" metal detectors were pioneered that were capable of operating at either a high (say, 300 KHz) or low (100 KHz) frequency depending on product.

Many other detector manufacturers have subsequently developed similar capabilities and the concept has been expanded to include three frequencies. However, these relative extremes to select from, while meeting a specific need, do not help in those occasional applications where a difference between 100 KHz and 150 KHz makes the difference between success and failure.