Oscillators

[Froskoy]

Hi,

I want a really stable oscillator at around 50Hz. Is it possible to use a crystal oscillating at around 3MHz to control another circuit that oscillates at around 50Hz.

Please could you give me a starting point so I can Google through the problem?

Thanks very much,

Matt.

 

[MrAl]

Hello,


Yes, you can use a digital divider or a phase locked loop.

 

[BrownOut]

Check this out https://www.electro-tech-online.com/threads/zezj-zero-error-zero-jitter-period-algorithm.100299/

 

[RCinFLA]

Define your stability need first. There is absolute frequency over some period of time and short term jitter (from cycle to cycle, or several cycles)

An AT cut crystal oscillator will give you the best performance over a reasonable 0 to 40 deg temp range. The best you can get over the temp range for a temp uncompensated AT cut oscillator is about 10 ppm. Uncompensated modules like sold for computers are typically spec'd at 60 ppm accuracy that takes into account make tolerance and temp variation. There are also 'oven' oscillators that keep the crystal oscillator at about 45 deg C to improve performance over ambient temp.

You can buy a temp compensated oscillator module that will give you about 0.5 ppm accuracy. This is what is used in a GPS receiver. GPS oscillator has a tight jitter requirement also.

Simplest method is to use ripple counter to divide down to 50 Hz. Ripple counter is just a series of cascaded divide by two flip-flops. You can buy these in a single I.C. Typically used in time clocks. It has lowest current drain because successive stages are operating at lower frequency. Jitter performance of ripple counter however is statistically cumulative caused by cumulative edge triggering time variation on successive stages. If you want best jitter performance then a syncronous counter may be used.

If you want better yet, then a 'dissiplined' oscillator can be made that is a voltage controlled crystal oscillator that is adjusted slightly and slowly by some other reference. For example, you can take a very accurate 1 pulse per second signal from a GPS receiver and use it to adjust the average frequency of a voltage controlled crystal oscillator over 10 to 30 minutes of averaging time. This will yield an oscillator frequency accuracy of better then 0.1 ppm. The average frequency accuracy of the GPS one pulse per second is better then a part per billion but the receiver output logic will give it some jitter that must be averaged out over time.

If your 50 Hz requirement relates to European power grid frequency, the accuracy of the grid is 'dissiplined' over time. As power loads go up and down they can can slow or speed up the generators slightly. More steam has to be made to take up new heavier loads that takes some time to build up so the generator may slow up a bit until this happens. The power companies keeps tract of the highs and lows and will try to compensate the variations over time by over speeding or underspeeding the generators slightly to get back to a pretty accurate 50 Hz long term average. At any given time they are typically much better then 1 Hz accuracy from 50 Hz nominal.