Beyond timing, accuracy, and repeatability
Bonnie Baker, Texas Instruments -- EDN, July 29, 2010
| View as PDF |
When you decide which converter will work best
for your application, you may first think about the
speed, accuracy, and repeatability of your perspective
system. OK, but let’s try to get beyond the obvious.
Your e-mails over the last few months related to this
topic are insightful. Thank you very much for taking
the time to send your ideas my way.Harvey Wiggins sent an e-mail discussing the difficulty behind trying to make a group of delta-sigma converters operate simultaneously. “Our application calls for more than 200 simultaneous channels at low power,” he writes. “We felt as though 24 bits would have enough dynamic range to use a fixed gain stage. Upon further investigation, it proved difficult to design [these devices] into our system because [delta-sigma] ADCs have to run continuously, with a continuous clock to operate the [internal]-filter stages. It would have been a tough job to start up 200-plus ADCs simultaneously across four or more boards. The synchronization of the startup with a system sample clock is tricky because the clock for the delta-sigma converters is not necessarily a nice multiple of the desired sample period. There usually is a defined clock-synchronization protocol, but all the edges have to line up exactly across multiple converters and boards.”
Wiggins took the easier and lower-power route of going to 18-bit SAR converters, even though they are more expensive.
“For time-domain signal acquisition, which requires edge preservation, I would choose the PGA-SAR [programmable-gain-amplifier successiveapproximation-register ADC],” writes Larry Bodnar. “For frequency-domain signals, such as music audio, sigma delta would be the obvious choice.”
These thoughts are astute. The SARADC system takes a “snapshot” of the input entity you are monitoring. The delta-sigma converter repeatedly samples the input entity, combining these samples through a digital filter for a final conglomerate-output digital value. This digital value represents an input signal over time.
Al Welch believes that delta-sigma converters are better. “I have used them for years,” he writes. “I have not used external gain in a long time. I worry about resistor drift when you have external gain resistors. I would guess that even those with programmable gain are going to have some limitations due to resistor drifts. I know we can find delta-sigma converters that are pretty good in the low-ppm [parts-per-million] range. Conversion speeds have risen with delta-sigma converters, as well. Of course, if you need high sample rate you may be looking at a SAR, but not with the accuracy of a delta sigma.”
Welch is enthusiastic about delta-sigma converters; however, the delta-sigma units have problems meeting all of the PGA-SAR performance levels. Welch questions the stability of external gain resistors—and that concern may be well-founded—but an integrated solution has matched on-chip gain resistors.
What about other issues, such as cost, reliability, and power for battery-powered applications? Although PGA-SAR systems have multiple chips, they generally cost less than delta-sigma devices. On the downside, a multiple-chip PGA-SAR converter may exhibit some reliability issues. Battery-powered applications are challenges for both systems: Delta-sigma devices require continuous clocks, and PGA-SAR units may have components lacking shutdown pins.
Summarizing the last four articles from this series—“Which system is best for handheld meters, data loggers, automotive systems, and monitoring systems: a PGA-SAR ADC or a delta-sigma ADC?”—I can only say that the delta-sigma ADC has arrived. It is a viable alternative to the traditional SARADC system.
| References |
|
Talkback
