Premium Content

New Signal Chain Resources from Texas Instruments:

Measuring Stability In A Logic-Gate-Based Oscillator

Date Posted: March 11, 2010 12:00 AM

Designers seeking a quick and easy frequency source often use a relaxation oscillator (Fig. 1) based on digital gates such as the 74HC14 inverter with hysteresis. The oscillator is also useful in sensor signal digitization when using resistive or capacitive sensors instead of fixed components in the circuit. The oscillator frequency will track changes in the sensor’s resistance or capacitance, providing a basis for measuring the corresponding physical parameters.

This utility of the gate-based oscillator raises the question of how reliably such an oscillator maintains frequency stability over time and temperature variations. I measured the frequency stability of such an oscillator built using metal film resistors and an NPO capacitor to minimize the possibility of component drift. Capacitor C1 had a value of 0.01 µF and R1 was a parallel combination of 1100 Ω and 1 MΩ, giving a frequency between 11 kHz and 12 kHz.

I tracked temperature using an Intersema barometric pressure and temperature recorder. Atmospheric pressure remained stable at 999.95 mbar and relative humidity at 75% during the test. The results show that the oscillator’s stability over temperature (Fig. 2) was no better than 1%/°C and that frequency increased with temperature.

No time-dependent drift is evident. The changes seen over time appear to be the result of small fluctuations in temperature. The most likely cause of the oscillator’s limited stability is a change in the internal threshold voltage of the hysteresis circuit.

Thus, this type of oscillator doesn’t provide a frequency source with stability better than 1%/°C. Time-dependent drift wasn’t seen and fell within the fluctuations caused by temperature. I associate this drift with the internal threshold-voltage change in the hysteresis circuit.

Part Inventory
Go
powered by:
 

 
You must log on before posting a comment.

Are you a new visitor? Register Here
  • kb6nu
    2 years ago
    Mar 31, 2010

    Do you have any thoughts about how stable a crystal oscillator built with a 74HC14 would be? Did you perhaps make any measurements on this type of circuit?

  • kb6nu
    2 years ago
    Mar 31, 2010

    Do you have any thoughts about how stable a crystal oscillator built with a 74HC14 would be? Did you perhaps make any measurements on this type of circuit?

  • kb6nu
    2 years ago
    Mar 31, 2010

    Do you have any thoughts about how stable a crystal oscillator built with a 74HC14 would be? Did you perhaps make any measurements on this type of circuit?

  • bcarso
    2 years ago
    Mar 31, 2010

    The possibly more significant issue with these parts as oscillators: the thresholds are all over the map. So even if they didn't drift that much (and they do drift as you have determined) they are not going to serve for any but the most undemanding applications.

    In the heyday of CMOS, National Semi managed to do a fairly good job of making parts with roughly 1/3-2/3 thresholds as a fraction of the supply voltage. But the guaranteed specs were already awfully loose. Some other manufacturers didn't even bother.

  • bcarso
    2 years ago
    Mar 31, 2010

    The possibly more significant issue with these parts as oscillators: the thresholds are all over the map. So even if they didn't drift that much (and they do drift as you have determined) they are not going to serve for any but the most undemanding applications.

    In the heyday of CMOS, National Semi managed to do a fairly good job of making parts with roughly 1/3-2/3 thresholds as a fraction of the supply voltage. But the guaranteed specs were already awfully loose. Some other manufacturers didn't even bother.

  • bcarso
    2 years ago
    Mar 31, 2010

    The possibly more significant issue with these parts as oscillators: the thresholds are all over the map. So even if they didn't drift that much (and they do drift as you have determined) they are not going to serve for any but the most undemanding applications.

    In the heyday of CMOS, National Semi managed to do a fairly good job of making parts with roughly 1/3-2/3 thresholds as a fraction of the supply voltage. But the guaranteed specs were already awfully loose. Some other manufacturers didn't even bother.