[ARC5] Drift in LC oscillators (and particularly the BC-453 series)

[Leslie Smith]

I make a few brief comments about oscillator stability.

(1)  My comments are based on direct experience building (and monitoring) home-brew analog VFOs.  

(2)  Without doubt, temperature is the main cause of oscillator drift.  Some time ago another list reader stated that frequency drift came from the aggregate effect of the momentary variation in every component in the oscillator circuit.   I can say whoever wrote that had direct practical experience with oscillator stability (or instability).    At a frequency of 5000kHz femto-farads  count, and even "stable" capacitors change value by a few femto-farads with varying temperature.   Check this yourself - do the calculation.  Ditto for coil dimensions, inter-electrode capacity etc.  This is why "those that knew 'summat' wound the coil (or a coil) using warmer wire.  On cooling the wire became taut on the former.  This contributed to a frequency-stable signal.  

(3)  I doubt that "Q" plays much part in the longer term frequency stability.    I suspect that "Q" has a role in short term "jitter".   (to Ian W).

(4)  Looking at the frequency of an oscillator over an interval of 10 seconds or so (this is the shortest period available to "count" frequency to a resolution of 0.1Hz at a few MHz) I see "jitter".  Either my oscillator is "noisy" or the counter is noisy (or both are noisy).  For a long time I have suspected that there is a relationship between the Gaussian distribution of observed "jitter" and the longer-term stability.  If anyone has a handle on how the two are related, I would like to know.  I suspect if this relationship is known, then the longer term stability can be estimated in a shorter time.  Any-one here know the maths?

(5)  Any oscillator "in the open" will drift.  By this I mean "not in a box" This is because each component contributes to the overall frequency of the oscillator and random air movement "tweaks" the value of each component moment by moment.  Even dropping a plastic sheet over an oscillator "in the open" will markedly decrease the measured "jitter" and improve longer term stability.

(6)  I built a number of oscillators and subjected them to a "standard 100mm (4 inch) drop test.  I have the results somewhere, but my recollection is that the smallest "jump" I measured (after dropping) was in the order of 25Hz.  The greatest "jump"I measured  was many hundred Hz.  A typical "jump" (on applying a "standard 100mm stability drop test") was in the order of 60-100Hz.   This was my best effort, but construction did not involve any form of shock absorber.   Just a straight drop onto a folded blanket.  (A standard military blanket, of course.  Dropping onto a rug or tea-towel will invalidate the result of this test.)  

(7)  I measured the temperature stability of a BC-211 oscillator (using a Fluke counter).  I believe the BC-211 had a temperature coefficient of drift of about 160Hz per degree C.   When I use the word "temperature stability" I mean stability of the oscillator frequency wrt temperature.  I think every-one know the meaning of "temperature stability" except my English teacher.  I had to explain this phrase for her benefit or she may give me a failing grade.  (I hope no-one else here has a stroppy  English teacher.) 

(8)  The National receiver (seen in the "you-tube" demonstration) was (in my view) faulty in some way (wrt stability) and should not be used for a comparison or reference.  A receiver  that exhibits more instability from the BFO signal than from the H.F.O. signal is faulty.  The set should be serviced before being used in a demonstration.    HA!

73 de Les Smith
   vk2bcu at operamail.com