In gaseous carburizing, measurement of the carbon content is an important process. Accurate control of carbon potential and the amount of carbon that is diffused into workpieces are part of that process. It relies on measurements obtained from furnace control instrumentation and carbon analysis to verify the measurement’s validity and reliability. Carbon-analysis methods like shim stock essentially verify that readings obtained by the oxygen probe or other control devices are accurate.

Understanding Carbon Control

Typically, a carburizing furnace is equipped with:
  • An oxygen probe emitting a voltage signal proportional to the partial-pressure ratio pH2O/pH2
  • A controller to pick up the oxygen-probe signal – effective carbon potential is calculated, assuming there is equilibrium of the water-shift reaction in the furnace atmosphere
  • Other control units relying on NDIR analyzers measuring the percentage of gases such as CO2, CO and occasionally CH4
Since the carbon potential in this setup is only a calculated value, it may possess an intrinsic error. By correlating instrumen-tation readings with direct measurements obtained from the shim stock, it is possible to improve the accuracy of the carburiz-ing process.

Fig. 1. C-Detect with sensor head for rolled shims. Head for flat shims also available.

Analyzing Shim Stocks

To ensure precise heat-treating results, the calculated value has to be checked periodically and the controller calibrated accordingly. Furnace-atmosphere analysis by iron shim stock is still the best test method to measure carbon percentage. With shim-stock analysis, either the weight-gain or combustion method has heretofore been used.

In the weight-gain method, the initial weight of the shim is compared to the total weight after exposure to the furnace atmos-phere. As shims are thin and light, the difference in weight gain between untreated/treated shims is no more than a few milli-grams. Special care should therefore be taken when handling shims to prevent transmitting impurities that could affect read-ings. The combustion method, a second way to determine the carbon content, uses combustion of the shims where the carbon content is measured by IR-spectrometry. Shims tested by combustion require the same careful handling as is used for the weight-gain method.

A third process is now available that offers the same precise analytic functions as the other two methods but requires less caution in handling. C-Detect measures the magnetic properties of iron in shim stock even with the presence of impurities such as soot, grease, dust or oxides.

Fig. 2. Iron structure and magnetic response

Measuring Principle

The crystallographic structure of the shim-stock material is dependent on its carbon content and the cooling/quenching conditions. If shims are not quenched in oil or water and the cooling rate is low enough to prevent the formation of mart-ensite, the structure will be mostly ferritic with low carbon contents. With an increasing amount of carbon, the structure will turn more pearlitic. If increased even further, a precipitation of secondary cementite will result. As the increasing carbon content changes the crystallographic structure, it also changes the electromagnetic properties of the shims to be analyzed (Fig. 2).

Fig. 3. Schematic illustrates the principle of measurement employed by C-Detect to calculate the carbon content in shim stock.

How C-Detect Works

C-Detect analyzes variations in the electromagnetic properties of the shim – a result of the modified carbon content – by applying eddy current testing in combination with analysis of higher harmonics. The carburized shim is placed in the measuring head close to a circular emitting coil that generates an AC magnetic primary field of discrete frequencies and amplitudes (Fig. 3). Depending on the electromagnetic properties, eddy currents are induced in the shim material, overlaid by nonlinear magnetization processes. This nonlinear signal transfer leads to an AC magnetic secondary field, inducing a measuring voltage inside the receiving coil of the measuring head. The magnitudes and phases of the induced voltage result in a multiparameter signal describing the electromag-netic properties corresponding to the shim structure.

Fig. 4. C-Detect data Processing

C-Detect Hardware Configuration

C-Detect is composed of a measuring unit with integrated power amplifier, a measuring head and the C-Detect software application. The software controls all hardware components and monitors the sequence of events (e.g. parameter setting, data acquisition and processing, system calibration and carbon-content determination) (Fig. 4).

The C-Detect’s measuring unit is connected to a PC via USB connection. It generates, converts and amplifies the signal sent to the emitting coil of the measuring head. If a shim stock is positioned inside the measuring head, the existing magnetic primary field magnetizes and induces eddy currents in the shim. The superposition of time-dependent changes in the resulting secondary field induces a measuring voltage in the receiving coil, which is amplified, filtered and converted. The digitalized signal is transferred to a standard personal computer via USB connection, where the C-Detect software demodulates, pre-processes and analyzes the data.

Fig. 5. Sensor heads for rolled and flat shims

Available Measuring Heads

Two different types of measuring heads are available to fit standard rolled or flat shim stocks. The sending and receiving coils of the measuring heads for rolled shims are positioned centrically inside the casing. For flat shims, the sending and receiving coils are aligned axially. While carrying out measurements, rolled shims are positioned on a special designed measuring pin (Fig. 5), and flat shims are placed inside the measuring slot.

Fig. 6. Calibration model

System Calibration

Carbon content is determined by comparative method using a set of calibration shims that can be adjusted to the customer’s requirements such as type and thickness of foils as well as cooling conditions. To increase the accuracy of the calibration, the following multiparameter calibration algorithm – based on a multidimensional regression analysis – is implemented in the C-Detect software guiding the user through the following calibration steps (Fig. 6):

Step 1: Data Acquisition – The induced voltages for discrete measuring frequencies and amplitudes are recorded. To increase the signal-to-noise ratio, the time signals are averaged.

Step 2: Data Preprocessing – In the field of data preprocessing, the recorded time signals are demodulated and multiple complex properties of higher harmonics are analyzed. Applying an algorithm based on the theory of complex data rotation, the amount of properties can be halved while the information content inside the remaining properties increases. To avoid possible system instabilities due to incorrect use of higher dimensional calibrations, as well as for further property reduction and separation of best-optimized values, tests are carried out evaluating the statistical criteria:
  • One-dimensional correlation
  • Measuring value spreading
  • Measuring value deviation
Step 3: Jack-Knife Test – Increasing the accuracy of calibration models by use of multidimensional property combination is the main aim of the final calibration step. The advanced optimization algorithm detects the best available multidimensional regression equation inside the pool of all calculated combinations. The calibration quality is determined by evaluating the mul-tidimensional correlation coefficient in standard and weighted form.

To fulfill customer’s requirements for more precise calibrations, the generation and handling of multiple calibration sets is supported as well.

Fig. 7. Emitted and measured signals at various carbon contents

Measuring Results

Measurements on shims with varying carbon contents using low test frequencies lead to time signals of sinusoidal primary field and induced voltages as shown in Fig. 7.

Due to the nonlinear signal transfer, the induced voltages are distorted comparing to the sinusoidal primary field emitted by the sending coil of the measuring head. These visible distortions are caused by the varying electromagnetic properties of the shim materials and can be analyzed in the frequency domain as changes in amplitudes and phases of the higher har-monics signal of parts.

Fig. 8. Amplitudes of 1st and 3rd harmonic versus carbon content

A strong dependence is seen (Fig. 8) between the analyzed amplitudes of the 1st and 3rd harmonics and carbon contents of corresponding shims. While the amplitudes of 1st harmonics show direct nonlinear dependence to the carbon content caused by increasing eddy current and magnetization losses, the amplitudes of 3rd harmonics decrease with indirect but strong linearity due to decreasing distortions of the induced voltages. To further optimize the accuracy of the correlation, especially while cali-brating with higher amounts of shims, the information contents inside the different harmonics properties can be combined using multiregression analysis algorithms. The hard copy of a calibration and measuring-result diagram (Fig. 9) shows an example of a two-dimensional C-Detect calibration. A set of six shims has been used to calibrate within a carbon content range from 0.48% up to 1.06% and a high correlation coefficient nearly 100% could be reached.

Experiments made on various tested shims showed that the accuracy of the results depend mostly on proper calibrations, which typically range within +/-0.02%C absolute.

Fig. 9. C-Detect measurement screen showing the calibration and the measured shim in a graph.

Conclusion

For proper carburizing treatments, the carbon potential of the furnace atmosphere should be accurately measured and veri-fied. However, a single measurement is not sufficient to get a truly reliable reading. Beyond atmosphere control devices such as probes and analyzers, carbon-analysis methods like shim stock are necessary to validate measurement data.

The C-Detect is a new, nondestructive testing tool from Process-Electronic. It quickly and accurately verifies the carbon content of carburizing atmospheres. C-Detect's uniqueness lies in its ability to provide accurate readings regardless of shim handling and its surface condition. Gloves are not mandatory since fingerprints, dust, soot and oxidation will not affect results. IH

For more information: Mr. Paul Oleszkiewicz, vice president, Nitrex Metal Inc., 3474 Poirier Blvd., Montreal, QC H4R 2J5, tel: 514-335-7191; fax: 514-335-4160; e-mail: paul.oleszkiewicz@nitrex.com

Additional related information may be found by searching for these (and other) key words/terms via BNP Media SEARCH at www.industrialheating.com: shim stock, carburize, oxygen probe, IR spectrometry, eddy current, harmonics, electromagnetic properties