Optically isloated and powered 1-kHz ADC

March 23, 1998
Some data-conversion applications require complete galvanic isolation of the signal source from system ground. Examples include industrial process monitoring using ion-selective and pH electrodes, and biological and medical diagnostics such as...

Some data-conversion applications require complete galvanic isolation of the signal source from system ground. Examples include industrial process monitoring using ion-selective and pH electrodes, and biological and medical diagnostics such as electrocardiography and electroencephalography (EKG, ECG) in which sensor electrode isolation is needed for both noise reduction and safety reasons. Optical coupling, of course, is the gold standard of signal isolation techniques, but it doesn’t solve the sticky problem of providing a power source for converter circuitry on the transducer side of the isolation barrier.

The ADC introduced here is unique because it uses standard optoisolator devices to accomplish both tasks. Achieving this dual objective is an interesting exercise in sub-micropower signal-handling tricks.

The converter is based on pulse-width-modulation (PWM) techniques. A1 compares the filtered input signal (±1-V full scale) to the voltage on C1. A1’s output is smoothed by the R4C4 time constant and compared by A2 to multivibrator A3’s approximately 1-kHz “triangle” waveform. The resulting variable-duty-factor squarewave is scaled and averaged by R1, R2, and C1, and fed back to A1. This feedback loop continuously adjusts A2’s duty factor to maintain V(C2) = V(C1). In doing so, A2’s output squarewave is forced to track the unique T+/(T+ + T) duty factor that maintains balance at A1’s inputs.

The A2 squarewave is differentiated by C5 to provide bipolar drive pulse to the anti-parallel LEDs of high-speed, low-current optoisolator OI2. In turn, OI2 produces groundreferred pulses that the HC02 NORgate flip-flop converts back to a squarewave with the same duty factor as A2’s output.

Extracting a signed numeric conversion result from this signal can be done in a number of ways, the easiest (zero glue logic) of which is to use the “edge-capture” feature implemented in 68HC05, HC11, and HC12 microcontrollers. If, for example, one of the eight counter-timer channels of an HC12 is programmed for 8-MHz-resolution, dual-polarity edge capture, the timer-channel hardware will explicitly capture values for the T+ and T intervals of each A2 cycle. Note that up to eight such converters could therefore be connected in parallel to one HC12. Direct conversion of these times to Vin voltage is then as easy as:

Vin = (T+ − T)/(T+ + T)

Isolated power for the converter comes from OI1—an International Rectifier PVI5100 photovoltaic optoisolator. This device is intended by the vendor to serve in isolated MOSFET gate drive applications. It can reliably source about 20 µA of current (80 µW), which is shunt-regulated by A4 to provide a stable 4.0 V ratioed against the MAX924 internal ±1%, 1.2-V reference.

Conversion resolution for this ADC is pretty good: 12 bits + sign in the HC12 example cited (8 MHz/1 kHz = 8000 counts/conversion cycle). Peak-to-peak conversion noise, on the other hand, isn’t so wonderful: approximately 10 mV due to A1 and A2’s rather large input-referred noise as is common in sub-micropower devices like the MAX924. But converter linearity and stability is better than 0.05%, so high-resolution applications can generally be accommodated by software averaging of multiple conversions. Thus, overall signal-tonoise ratio to the sub-millivolt level is achievable. Converter dc input impedance is about 1 TW with less than 1 pA bias. This permits good accuracy to be expected even when working with very high impedance signal sources. Converter span and input offset errors are trimmable to zero.

See associated figure

About the Author

W. Stephen Woodward

Steve Woodward has authored over 50 analog-centric circuit designs. A self-proclaimed "certified, card-carrying analog dinosaur," he is a freelance consultant on instrumentation, sensors, and metrology freelance to organizations such as Agilent Technologies, the Jet Propulsion Laboratory, the Woods Hole Oceanographic Institute, Catalyst Semiconductor, Oak Crest Science Institute, and several international universities. With seven patents to his credit, he has written more than 200 professional articles, and has also served as a member of technical staff at the University of North Carolina. He holds a BS (with honors) in engineering from Caltech, Pasadena, Calif., and an MS in computer science from the University of North Carolina, Chapel Hill.

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!