5 MHz and 10 MHz

ULN oscillators at 5 MHz and 10 MHz are the easiest to manufacture.  Below is a plot of a CFS10B frequency standard supplied to a satellite company.  The frequency was 10 MHz.

Here we can see phase noise of -112 dBc/Hz at a 1 Hz offset with a -168 dBc/Hz floor noise.

It is possible to go as low as -130 dBc/Hz at a 1 Hz offset and a -175 dBc/Hz floor noise, but sadly not in the same oscillator (unless a multiple oscillator design is used).  We achieve lower phase noise at close in offsets (1 and 10 Hz) but sacrifice phase noise at wider offsets (10 kHz).

However, the plot below is a compromise of good phase noise at close in and far out offsets.

Other Frequencies

It is much harder to achieve ultra low phase noise at other frequencies such as 80 MHz or 100 MHz.  So what we do in these cases is lock a fundamental 100 MHz oscillator with average phase noise, to a ULN 10 MHz oscillator.  The 10 MHz reference provides the 100 MHz accuracy and low ageing, while at the same time improving the phase noise of the 100 MHz oscillator, especially at close in offsets.

Below, shows another CFS10B frequency standard we supplied, this time with a 75 MHz output.  Here we locked a 75 MHz oscillator to a 10 MHz ULN reference.  The customers specification is shown by the yellow line.  We easily beat this specifications as shown by the white trace.  And, in fact, the plot below is of two of these oscillators.  So the individual phase noise is up to 3 dB better than this.

This isn't the best that we can do, simply we met the customers specification goals.  It's very important not to ask for more than you need.  Low phase noise is a very expensive hobby!

100 MHz Oscillator

Our latest design is for a 100 MHz oscillator.  The customer has a limited budget and has asked for the following phase noise. The table also shows what we currently have offered and what may be possible.

 

100 MHz Oscillator Design
Phase Noise dBc/Hz @ Offset	1 Hz	10 Hz	100 Hz	1 kHz	10 kHz	Remarks
Customer's Specifications	-98	-117	-131	-155	-155
Our Offer so far	-90	-125	-140	-155	-170
Theoretical Best Achievable	-100	-130	-155	-165	-170

In the "Our offer so far" line, we lock a 100 MHz oscillator to a 10 MHz ULN oscillator in a normal analog phase lock loop.

The "theoretical best achievable" is a complicated combination of digital phase lock loops and multiple oscillators.

Currently, we meet the customers specification except for the 1 Hz phase noise.  However, we have already exceeded his budget with this first offer.

The theoretical figures meet the customers entire specifications, but at five to ten times the customers budget!  This design is only for a quantity of one, and as it is a special design, costs to add up.

So something has the give, either the phase noise requirement or the budget.

The above is a typical scenario and often a design is a compromise between what the customer wants and what can be achieved, within the budget.

Multiple Outputs Often Required

Usually the customer wants multiple outputs of the ULN signal.  This involves a distribution amplifier to isolate the required outputs and protect the loading on the main oscillator's output.

Obviously the distribution amplifier must have ultra low phase noise itself, otherwise the signal will be degraded.

Luckily we have developed our DA1-100-10 range. This is a distribution amplifier offering multiple outputs (up to 20 in a 1U rack size) each with excellent isolation from the other outputs (-90 dB typical) and a large reverse isolation figure (typically 130 dB) to protect the main oscillators output.

The plot below shows our DA1-100-10's phase noise.  It can be seen the close in noise at an 1 Hz offset is 10 to 20 dB better than most ULN oscillators.  The far out noise is about -170 dBc, again more than enough to meet the requirements of most projects.

The phase noise below is our standard product.  It is possible to drop the 1 Hz noise to -140 dBc/Hz and the far out noise to -175 dBc/Hz if special attention is paid to gains, output levels etc.