With respect to data acquisition systems, measurement devices can be either scanning or parallel architectures. Parallel architectures utilize an independent analog to digital converter on a per channel basis while scanning architectures typically use a multiplexer to scan each channel to one analog to digital converters.
Parallel architectures are generally preferred as they provide superior accuracy and better synchronization when compared to a scanning system. However, they are also more expensive, and are only justified in those applications that require a higher level of accuracy
The phenomenon that introduces inaccuracies in a multiplexed data acquisition system is commonly referred to as “channel bleeding.” Specifically, when a measurement on one channel affects the accuracy of the measurement on the following channel in the system. For example, consider the case where a very large voltage measurement is immediately followed by a low voltage measurement on the same ADC. In this case, if the scan rate is too fast, the ADC may not have a chance to dissipate the large voltage before taking the smaller voltage reading. The result is inaccuracies in the smaller voltage measurement. Another scenario to consider is when a valid temperature channel is followed by an open transducer (damaged thermocouple) condition. The front end of the instrument, being high impedance, will only slowly drift. Again, if the scan rate is too fast, this measurement will appear to be very close to the proceeding channel, and will appear valid even though in reality, the transducer is faulty.
There are strategies to mitigate the effects of this phenomenon. The EX10XXA family of products utilize a scanning architecture to take both temperature and voltage measurements. Despite the scanning nature of this system, this device is known for it’s high level of accuracy in delivering temperature and voltage measurements.
In addition to analog and programmable, digital filtering, this product is designed with independent channel to channel signal conditioning which goes a long way to prevent against channel bleeding effects. Specifically, independent gain stages on each channel allow some channels to measure large voltages with respect to adjacent channels with no negative effects on the measured data.
Furthermore, each channel can independently source a very small amount of current to the transducer. This source will have negligible effect on the accuracy of the measurement, but in the case of an open transducer, the current will quickly drive the measurement amplifier into saturation, creating a reliable and indicative overload condition.
Users receive a low cost solution that delivers many of the same advantages in terms of accuracy as a parallel implementation.