Issue link: http://www.e-digitaleditions.com/i/316458
Tablets & Capsules May 2014 25 5 Initial or actual weight 3 Target tablet Target weight cam depth weight cam depth Initial tablet weight Over-blending in the paddle feeder can also affect tablet compactability, especially in formulations with a high amount of lubricant. Over-blending can also affect disintegration times. If either becomes a problem, see how slowly you can operate the paddle feeder while still producing tablets of correct weight consistently. Cross-contamination Cross-contamination is the intermingling of the first and second-layer powders or vice versa. Dust collection goes a long way toward preventing cross-contamination but it needs to be balanced against excessive material loss. There are several steps you can take to prevent loss: • Ensure turret run-out is less than 0.003 inch; • Set the feeder table at a height that prevents exces- sive powder from reaching the turret; • Examine the scraper bars for wear and replace as necessary; • Ensure that the dies fit tightly and don't allow pow- der to lodge in crevices • Verify that the vacuum is set as the manufacturer recommends. If the vacuum is too strong, it will pull powder from the feeder. If set too low, it will allow the powder to contaminate the other layer; and • Secure all vacuum hoses and ports tightly. The importance of instrumentation Making good bi-layer tablets consistently requires using force instrumentation during development and manufacture. Without high-quality data acquisition, you cannot know how the press is running. In fact, you'll waste time, effort, and powder as you test tablet samples just to discover that the press needs adjustments. After that, it may take you hours to optimize the operation. Opt for a high-quality data-acquisition system instead. While a complete system has many aspects, let's focus on the importance of dedicating a compression transducer to monitor the creation of the tablet's first layer. Most bi-layer tablet presses exert as much as 80 to 100 kilonewtons (kN) of compression force on both layers, and a force transducer rated for 80 kN typically has a full- scale error margin of 0.25 percent, a range of ±200 Newtons (N). Furthermore, the target tamping force for the first layer is typically in the range of 1 to 2 kN. Thus, an error of 200 N represents a measurement error of 20 percent at 1 kN of force. To reduce the magnitude of this error: • Use a force transducer designed for a full scale of 25 kN instead of 80 kN. You can do this by reducing the web thickness in the strain pockets, which results in a similar stress at a lower force; • Change the transducer material from steel to a softer material, such as aluminum bronze, which will gen- erate a greater electrical output due to a reduction in the modulus of elasticity; • Use platinum-tungsten or semi-conductor strain gages instead of constantan (copper-nickel alloy) to generate a greater electrical output; • Tighten the restraints on the software filter to im - proved the signal-to-noise ratio; and • Install a mechanism that shuts down the tablet press when force exceeds 25 kN of compression to protect the transducer. These improvements will provide the measurement accuracy you need to monitor and maintain the compres- sion force of layer one. It's also good practice to display force in real time because you can set up the press faster and quickly identify weight variations. Figure 3 shows the real-time force readings of a simu- lated compression of layer one. Note that it shows 16 compression events and the average peak force is 994.7 N. By displaying each peak force, you can gauge the vari- ability between stations. The relative standard deviation (RSD) percentage of the compression force appears at the bottom of the chart. During development, the press commonly does not use an automatic weight control system, and thus any variabil- ity in tablet weight will directly affect the compression force. In this example, the compression forces are uniform, which means the tablet weights are consistent. Figure 3 Real-time force readings of layer-one compression (simulated) g-Sedlockart_22-27_Masters 5/14/14 10:16 AM Page 25

