AN-1186: Radiated Immunity Performance of the AD7192 in Weigh Scale Applications

Introduction

The AD7192 is an ultralow noise, low drift, 24-bit sigma-delta converter which includes a PGA. The AD7192 is used in high-end weigh scale systems. The radiated immunity of the weigh scale system is tested as part of the qualification for release.

This application note describes how to achieve the best radiated immunity performance from the AD7192, taking into account the effects of board layout and component placement when designing a printed circuit board (PCB). The radiated immunity testing is performed as per standard IEC 61000-4-3 and the complete system (ADC, PCB, and load cell) is tested.

Radiated Immunity

The radiated immunity test is performed as described in the standard IEC 61000-4-3. The field strength is 10 V/m and the RF frequency is swept from 80 MHz to 1 GHz. According to the specification, a device is classified as follows:

• Class A: Normal performance within limits specified by the manufacturer, requestor, or purchaser.
• Class B: Temporary loss of function or degradation of performance, which ceases after the disturbance ceases, and from which the equipment under test recovers its normal performance, without operator intervention.
• Class C: Temporary loss of function or degradation of performance, the correction of which requires operator intervention.
• Class D: Loss of function or degradation of performance, which is not recoverable, owing to damage to hardware or software, or to loss of data.

The ADC converts continuously during the frequency sweep. The error as referred to throughout this application note is the maximum deviation between the ADC conversions when an RF frequency is present versus when there is no RF frequency present.

For a weigh scale system to be Class A, the allowable error e in the presence of the RF interference is:

where n is the number of counts for the weigh scale system.

Radiated Immunity Test Analysis

Setup

Figure 1 is a block diagram of the circuit used for the radiated immunity testing. The AD7192 is configured as follows:

Sinc⁴ filer
Chop off
Output data rate = 10Hz
Gain = 128

The AD7192 operates from a 3.3 V power supply. This supply is also used to excite the load cell. The load cell is 6-wire with a sensitivity of 2 mV/V. For more details on weigh scale design using the AD7192, refer to Circuits from the Lab® reference circuit (CN-0119).

Error

As discussed in the Radiated Immunity section, the allowable error e for a Class A system is:

where n is the number of counts. The error is equivalent to ±0.5 counts.

In this application note, the goal is to design a weigh scale system that has 3000 display counts and is classified as Class A when the load cell is excited with 3.3 V. With a sensitivity of 2 mV/V and an excitation voltage of 3.3 V, the maximum signal from the load cell is 6.6 mV. Often, to use the most linear portion of the load cell’s span, only two-thirds of this range is used. This reduces the full-scale output voltage from the load cell to 4.4 mV.

For an accuracy of 3000 counts, one count is:

1 count = 4.4 mV/3000 = 1.46 µV
±0.5 counts = ±1.46µV /2 = ±0.73 µV

The error must be less than +0.73 µV while the RF frequency is present. The load cell used in the application accepts a full weight of 2 kg so the error needs to be less than +2 kg/(2 × 3000) = +0.33 grams—this ensures that the digital display is not affected by the RF interference.

Printed Circuit Board

The standard AD7192 evaluation board is designed to give optimum analog-to-digital conversion performance. However, it is not optimized for EMC. For example, the standard AD7192 evaluation board includes links (vertical pins) to allow different power supply options and links are present for the noise test connection; these links act as antenna. In addition, filtering on the analog and digital inputs is not optimized in terms of location and component size (0603 components are used). However, using this board as a starting point, an investigation was performed to highlight any adverse effects due to EMC. See the Results section for details. The grounding, component location, and addition of extra filtering were all reviewed. The ADC performance was maintained at all stages.

In summary, the key findings were:

• The link options (vertical pins) should not be included on the board. These act as antenna. Therefore, replace link options with a solder link option.
• The printed circuit board should be 4-layer, with the analog inputs and reference inputs buried in the inner layers. A single ground plane should be used. Flood the top and bottom sides of the board with ground. Also, flood the inner layers with ground. Multiple vias should be included to minimize any potential differences across the board. There is no hard rule on the density of vias required. On the AD7192 board, a ring of vias was included around the ADC and the filtering on the analog and reference inputs. In general, any islands on the board should have vias also, the number of vias being in excess of one. Any tracks on the top and bottom sides should be as short as possible since tracks will also act as antenna.
• Filtering is recommended on the analog and reference inputs. Figure 1 shows the R and C values that are normally recommended on the analog and reference inputs. This filtering provides attenuation at the AD7192’s sampling frequency (307.2 kHz) and multiples of the sampling frequency. The AD7192 itself does not provide any attenuation at these frequencies. The capacitors need to be as close as possible to the AD7192’s analog inputs and reference inputs so that the track length from the component to the ADC is minimized. Using components that are physically smaller allows the user to place the components closer to the pins. The layout should ensure that track lengths from the pins to the components are well matched.
• In addition to these filters, adding additional filtering in the R and L locations shown in Figure 1 improved the immunity further. This filtering is located at the connector to the load cell. Various combinations for the L (L2, L3, L4, and L5) and C (C1, C9, C12, and C13) values were evaluated to achieve the best results. The Bill of Materialssection lists the final components selected.
• The power supplies are decoupled with a 10 µF capacitor in parallel with a 0.1 µF capacitor. Again, the components should be as close as possible to the power pins of the AD7192. The analog power supply is used as the excitation voltage to the load cell which, in turn, is used as the reference to the ADC. Therefore, the power supply tracks are also buried in an inner layer.

Results

Following the investigation, a printed circuit board optimized for radiated immunity was developed (see Figure 3). The artwork and schematics for the board are included in the Evaluation Board Schematics and Artwork section of this application note. Using this board and the components listed in the Bill of Materials, the maximum error measured during radiated immunity testing was less than e. Figure 4 shows the conversions read from the AD7192 while the RF frequency is swept from 80 MHz to 1 GHz. A constant weight is placed on the load cell during the testing.

The error measured is 0.45 µV, which is higher than e. This is equivalent to 0.2 grams.

For comparative reasons, Figure 5 shows the conversions read from the standard AD7192 evaluation board when tested for radiated immunity. The board has an error of 356 µV when the RF interferer is present which is equivalent to 161 grams.

This comparison highlights the importance of layout, component selection, and component placement to achieve optimum performance in terms of radiated immunity.

To further improve the device’s immunity to radiation, a copper shield can be placed over the AD7192 and the auxiliary components.

Conclusion

Key factors in optimizing the performance of a weigh scale system for radiated immunity are the board layout and the component placement and selection. When the layout practices discussed in this application note are used, the weigh scale system is Class A as per IEC 61000-4-3. Therefore, a weigh scale with an accuracy of 3000 counts continues to function correctly in the presence of radiated immunity, that is, the weigh scale will not react to the interferer.

Evaluation Board Schematics and Artwork

Build of Materials

作者

Li Ke

Li Ke于2007年加入ADI公司，担任精密转换器产品线应用工程师，任职地点在中国上海。他曾在Agilent Technologies公司的化学分析部门担任过四年的研发工程师。Li于1999年获得西安交通大学电子工程学士学位，并于2003年获得西安交通大学生物医学工程硕士学位。他在2005年成为中国电子学会专业会员。

Mary McCarthy

Mary McCarthy是ADI公司应用工程师。她于1991年加入ADI公司，就职于爱尔兰科克市的线性与精密技术应用部，专注于精密Σ-Δ转换器产品。她于1991年从科克大学毕业，获得电子与电气工程学士学位。