Bastak
Sampling and measurement, Dağyaka, Kahramankazan, Ankara, Turkey.
History
Bastak Teknoloji Sistemleri is based in Dağyaka, an industrial zone of Kahramankazan, north of Ankara. The firm publishes no founding date. Its own pages give three different durations, twenty-four years, twenty-five years and twenty-six years, which places the start between 2000 and 2002. Its trade was never to roast the bean, but to measure it. It makes the control instruments for flour mills and animal feed plants: falling number, gluten, moisture, ash, sedimentation. The falling number measures the time a rod takes to pass through a hot flour paste. This time says whether the grain sprouted on the stalk, because a sprouted grain liquefies the paste and makes the flour unfit for breadmaking. It also produces mycotoxin assay kits, for those toxic molds that develop in poorly dried grain. Its catalog counts seventy-two control instruments and thirty-five additives for flour, bread and pasta. Making food ingredients requires a food safety certification, and the firm displays FSSC 22000. Finally, it covers sampling, upstream of analysis, for a simple reason: an accurate measurement made on a bad sample is worth nothing. It claims 265 employees, 195 of them engineers, on 10,000 square meters of which 7,000 are covered.
What it makes
Bastak builds no roaster. None of its machines heats the bean to roast it, either by conduction or by convection. What it makes lies entirely upstream. First the sampler in the strict sense, that is, the hollow tool that goes to fetch material from the mass. Then the divider, which reduces this sample to an analyzable quantity without changing its composition. Its 10000 series probe does not work by hand but by suction. A tube goes down into the load and a pump draws the grain up along an internal helix. The manufacturer presents this vacuum and blowing system as patented. The interest comes down to one word, depth. The probe samples at any level of the load, and not only at the surface. The arm that carries the tube is hydraulic, driven by rams rated at two tonnes, and is controlled remotely. The 13000 divider consumes no energy: it works by gravity alone. The sample falls into thirty-eight channels that send it alternately to the right and to the left. The split is thus made without anyone having to choose a handful.
The models
The question of capacity per batch does not arise here: no model roasts anything. The sampling range has six probes. The 10000 equips a truck or silo reception. The 10100 offers an extended reach. The 10200 is the fully automatic model, announced for trucks and ships. The 10350 samples fertilizer, the 10400 coal, according to the Turkish standard TS 2390. The 10500 is the portable probe, for sites where no fixed installation is possible. It also works by suction, through aluminum tubes, and the manufacturer intends it for hazelnut as much as for wheat. For the 10000, the manufacturer announces an adjustable rotation of 180 to 360 degrees and a reach of about 9.7 meters. It announces sampling in 2 to 6 seconds, six points in 40 seconds, and remote control up to 40 meters. These same figures appear word for word on the pages of the 10350 and the 10400. The table divider 13000 separates the sample into two halves through its thirty-eight copper and brass channels. The announced accuracy is plus or minus 1 percent over 1,000 grams. The firm announces more than nine hundred probes in service worldwide.
What you gain
The gain is measurable, and it is played out entirely before the analysis. The manufacturer points out that a hand-held probe goes down only 30 or 50 centimeters. Yet broken grains, dust and foreign matter, finer or denser, sink toward the bottom during transport. Road vibrations are enough to perform this sorting. A surface sample therefore underestimates the defect systematically, and not at random. The error always goes in the same direction, in favor of the seller, since what remains at the bottom is precisely what downgrades the lot. Drawing over the full height, at six points and in forty seconds, removes this bias. The gravity divider follows the same logic downstream: thirty-eight alternating channels split the lot without the operator having to choose a handful.
What you lose
None of this transposes to a cupping table. The 10000 probe is a fixed reception installation, with a mast, two-tonne rams and a remote control. It samples a truck, not a sixty-kilo bag. The gravity divider, for its part, depends on a freely flowing product. A damp, oily or sticky material bridges in the channels, and the split then ceases to be equal. Copper and brass resist corrosion but remain soft metals. A channel edge deformed by a knock durably skews the distribution, and nothing signals it to the operator. Finally, the announced accuracy holds for 1,000 grams and does not carry over as is to a sample ten times smaller. The fewer beans there are left to split, the more the chance of the split weighs on the result.