Power performance in industrial control
28 February 2011
Colm Slattery looks at improving system performance while reducing power for process control output modules.
This article outlines some typical output architectures for industrial control.
While discrete architectures have typically been used in the past, there is a clear advantage from using system level integrated IC devices.
These devices offer better overall system accuracy, better drift performance and crucially have integrated some novel power control techniques that allow for increased channel density, while simultaneously decreasing the overall system power budget.
Care also needs to be taken to ensure that the chosen semi-conductor components can be designed to operate within their maximum ratings of junction temperature, while still offering the highest levels of performance available.
System specifications
System specifications are typically denoted in percentage accuracy with respect to the full scale range of that system (%FSR). Some lowend systems may have accuracies of only 1% FSR where high-end systems may have accuracies up to 0.1% FSR.
It is important to know if temperature drift is included in the system specifications, as this can typically be the dominant source of error.
In general, system specifications can be discussed with respect to three error sources:
•Errors which can be calibrated at 25°C nominal, such as gain and offset errors.
•Errors which cannot be calibrated at 25°C, such as INL – basically accuracy in the linear operational region of the DAC, device noise, etc.
•Drift over temperature – Can be calibrated out but expensive as needs factory calibration over temperature (test time cost), temperature monitoring (analogue cost) and some linearisation algorithms (digital cost) during normal operation.
Consider the circuit in Figure 1, which uses a 16-bit 5 V DAC with external signal conditioning to generate bipolar outputs in the range of ±10 V.
S1 and S2 are external switches which set bias points in the circuit to control the output ranges.
In this way the 12-bit resolution can be maintained across different output ranges. The circuit works as follows:
With both S1 and S2 closed, the bias points are GND and 5 V respectively. In this case the equation which dictates the output transfer function for this circuit is:
Vout = 4 x VDAC –10 V
/* Output maps to ±10 V */
Opening switch S2 but leaving switch S1 closed changes the transfer function of the circuit to:
Vout = 2 x VDAC
/* Output maps to 0-10 V */
In this way we can see how using switches in the circuit enables different output ranges to be software configured.
Some conclusions can be made from analysing these results. Firstly, the noise contribution for each device is typically quite low compared to the least significant bit (LSB) resolution, and therefore can be ignored.
Errors at nominal temperature can be removed via a factory system calibration. The remaining error is due to drift over temperature, in this case 0.1128%.
Using a lower drift reference, such as the ADR435 (3 ppm/°C max) and a more accurate DAC such as the AD5060 (1 ppm/°C), would improve the post-calibration system specification to <0.05%.
Power to the output stage
As high voltage output ranges are required, high voltage supplies will be needed. For typical PLC systems, power is made available to the module board via the system backplane. Typically, 5 V, 12 V, and 24 V (or some combination) are available for this purpose.
The chosen supply is DC-DC converted to provide bipolar supplies. The analogue I/Os are typically isolated, but not always. There are many different power architecture topologies available, including feed-forward, flyback, etc.
There are some important considerations to make when choosing a part to generate the output supplies, including efficiency, robustness and cost.
The ADP1864 provides a high efficiency, low cost solution and offers a number of features making it suitable for industrial applications.
Take the example where the ADP1864 controller is adapted to provide ±15 V supplies from a 5 V input. The output voltage is easily adjustable using two external resistors.
The part provides system flexibility by allowing accurate setting of the current limit with an external resistor - a useful feature to limit power dissipation under fault conditions.
The ADP1864 also includes an internal soft start to allow quick power-up while preventing high inrush current. Critical for industrial applications, the controller provides additional safety features including output over-voltage protection, and input under-voltage protection.
Finally, another feature is that this architecture provides for an output disconnect.
This means that under conditions where the part is not enabled, but does have voltage applied to it, there is no current path from the input to the output and so the output loads are protected.
Two important power design specifications are efficiency and ripple. The efficiency of the ADP1864 can reach up to 94%. Low output ripple is important to minimise system noise.
Figure 3 shows a plot of output ripple under load conditions of 30 mA to the ±15 V outputs when driven from a 5 V supply.
In this case the ripple is ~30 mV pk-pk at the switching frequency of the DC-DC controller.
Given the fact that the switching frequency is high (in the order of 580 kHz) the circuit enables small filter components to be employed. As both current and voltage outputs are always presented with capacitive loads, up to some nF in real systems, calculated ripple will typically be attenuated in practice.
On-chip/Off-chip power concerns
Most Industrial outputs need to provide current output as well as voltage outputs.
Careful consideration is needed to ensure power and thermal considerations are met when designing such circuits. There are two areas for consideration, IC die temperature (to ensure the IC itself is protected), and module power dissipation (how efficient is the module).
If we consider the AD5422 single channel device discussed earlier, we can do some simple calculations to specify the supply voltage required. Assume in this case we are driving a 20 mA output into a 1 kOhm load.
In this case the output compliance voltage needed to be generated would be in the order of 20 V. With the necessary headroom for the IC, the minimum supply voltage needed would be in the order of 24 V.
For industrial applications, current outputs driving into a zero Ohm load (short circuit) is a valid operating condition. In this case, the full power dissipated will be handled by the AD5422 IC, therefore we need to ensure the part is not over heated.
Doing the maths, the total power dissipated under this condition is ~0.6 W (24 mA x 24 V).
Referring to the AD5422 datasheet, the thermal impedance of the part is 28°C/W. Such low thermal impedance is achievable as the AD5422 is housed in an LFCSP (QFN) package which has both low thermal impedance characteristics and an under paddle that is soldered to the PCB to increase the heat sinking effect.
The die temperature increase due to the 24 mA driving a zero Ohm load can now be calculated as:
Die Temperature Increase = 0.6 W x 28°C/W = 16.8°C
Assuming the junction temperature can reach a maximum of 125°C, then the maximum ambient temperature of the system cannot exceed 108.2°C (this is under a short circuit condition).
The AD5755 is one part that solves all the above challenges, while delivering the highest performing DC accuracy in the market.
The AD5755 integrates four output channels, each of which deliver up to 16 programmable ranges (current and voltage) on a per-channel basis.
Each channel can deliver up to 24 mA into a 1k load and can also handle all outputs driving into short circuit loads.
It achieves this by the addition of DC-DC converters on each channel, these monitor each output load and dynamically adjust the compliance voltage to provide only the minimum voltage needed. Hence minimal power is dissipated in the IC - and therefore in the module – allowing the device to both self protect and ensure maximum channel density is achieved for the same power budget.
Conclusion
Integration of System Level features and functions brings many benefits to new designs.
While increasing system accuracy, devices like the AD5755 also reduce board area and power consumption on a perchannel basis, thus enabling increased channel density and lower cost in the final system.
COLM SLATTERY is Applications Engineer Industrial and Instrumentation Segment Group, Analog Devices
Contact Details and Archive...
Related Articles...
Most Viewed Articles...