mdr@reed.UUCP (02/07/87)
What is the "phase jitter of a free-running oscillator" and what does it depend on? Does anyone have any revealing references? Mike -- Reed College -- Portland, Oregon -- 503/774-9192
gibson@unc.UUCP (02/14/87)
In article <5246@reed.UUCP> mdr@reed.UUCP (Mike Rutenberg) writes: >What is the "phase jitter of a free-running oscillator" and what does it >depend on? Does anyone have any revealing references? I'm not sure what your context is, but the only context in which I've seen that phrase is in digital oscillators. If you use a digitally- stored waveform and sequence through its points to generate a waveform (sending each point to a DAC), I know of 2 definitions : 1) sample period jitter - unevenness in time between output samples. The reference implies noise from this jitter is 83 dB below the RMS level of a generated sine wave iff the sample period jitter is < (0.0000118 / F) (sec), where F is the output frequency. "Design of a digital oscillator...", John Snell, Computer Music Journal 1(2):4-25, 1977 2) phase jitter - errors due to finite length of the stored waveform. If you use a 256-point table and a 16-bit phase register, you will get noise because the 8-bit address and the 16-bit phase often refer to different points on the waveform (since the phase must be truncated or rounded). "Table Lookup Noise...", Richard Moore, Computer Music Journal 1(2):26-29, 1977 Bill Gibson gibson@unc ...[akgua,decvax,philabs]!mcnc!unc!gibson
keithl@vice.UUCP (02/17/87)
In article <5246@reed.UUCP> mdr@reed.UUCP (Mike Rutenberg) writes: >What is the "phase jitter of a free-running oscillator" and what does it >depend on? Does anyone have any revealing references? Phase jitter in an analog oscillator can be caused by environmental variations such as supply ripple, temperature fluctuations, or mechanical vibration, or simple thermal noise in the oscillator or resonant device. Here at Tek, we have folk designing spectrum analyzers and various sampling instruments who have sleepless nights about phase jitter, and who attack the problem with surface acoustic wave oscillators and other such esoterica. They assure me that sub-picosecond movement of edges on 100 MHz clocks are quite measurable and upset some of our customers. -- Keith Lofstrom MS 59-316, Tektronix, PO 500, Beaverton OR 97077 (503)-627-4052
mdr@reed.UUCP (02/18/87)
In article <1444@vice.TEK.COM> keithl@vice.TEK.COM (Keith Lofstrom) writes: >Phase jitter in an analog oscillator can be caused by environmental variations >such as supply ripple, temperature fluctuations, or mechanical vibration, or >simple thermal noise in the oscillator or resonant device. How can I obtain large fluctuations in a *low* speed oscillator? I want an oscillator at about 3 kHz with large unpredictable variation in the lengths of individual cycles. Mike -- Reed College -- Portland, Oregon -- 503/774-9192
markf@amc.UUCP (02/18/87)
> How can I obtain large fluctuations in a *low* speed oscillator? I > want an oscillator at about 3 kHz with large unpredictable variation > in the lengths of individual cycles. Well... If rectangular waves are OK, you could use a pair of oscillators, set to relatively prime frequencies. They could be combined in different ways for different effects, such as XOR'ing, or using the clock & "D" inputs of a flip-flop, or "phase comparator II" of the 4046 PLL. Or, use a VCO, and feed the voltage-control input with a noise source, say the RC-filtered output of a pseudo-random sequence generator. Or, a Wein-bridge oscillator, with the resistive-tuning element being a photocell, pointed at a lava lamp. Or, use a high-gain amplifier I built some time ago. Or... -- Mark S. Freeman Applied Microsystems, Inc. markf@amc
jewett@hpl-opus.UUCP (02/18/87)
> What is the "phase jitter of a free-running oscillator" and what does it > depend on? Does anyone have any revealing references? Mike To determine phase jitter of a waveform, compare the time of its zero crossings with the zero crossings of an ideal sinewave of the same frequency. The difference in time can be plotted as a function of time: P | ** ** h | ** * ** * a | * * * * ** s |* * ** * * e +---------------*---------***--------*---*------> Time | * * * * * E | * * * * * *** r | * *** r | Electronic circuits called phase detectors can compare the phases of two oscillators, and give a voltage proportional to the difference. Often a very quiet crystal-controlled oscillator is used as a reference standard. You can do varous kinds of statistics on the resulting waveform, such as spectrum analysis to find probable noise sources (60Hz/120Hz due to the power supply is common). See McGraw-Hill's "Electronic Instrument Handbook" for more details and references. Bob Jewett hplabs!jewett