[time-nuts] OCXO Voltage Input?
dan at irtelemetrics.com
dan at irtelemetrics.com
Fri Sep 5 22:54:54 EDT 2014
You have some good observations. Spread spectrum clocking is one I
hadn't considered when looking at this problem. In that case the
crystal is pulled a bunch. (It's also cheating in my opinion!)
Correcting for mechanical vibration in aircraft would also tend to
indicate it's possible
In the schematic for the 10544 that Tom posted the link to, it appears
that when both of the EFC lines are available, only two 20K resistors
are in series with the varactor. One would guess that changes on the
EFC line could quite easily modulate that varactor. Assuming a similar
scheme in other oscillators, one would think that modulation could
quite easily happen there also.
I supposed that the manual telling the operator to avoid noise on the
EFC line, due to FM modulation happening, supports this theory as well!
(Big Exclamation point there!)
Once the question is asked, you have to ask how does one measure such a
modulation? At least with simple equipment easily available. Some
searching didn't result in anything promising, at least for what I have
access to. Initially my thoughts are because the varactor is acting on
the crystal to change frequency the assumption is the modulation is FM
(Again, the HP manual backs this up). The specified peak to peak
deviation is only +/- 20Hz at most. No matter what the modulating
frequency is the FM modulation index is virtually zero, so there are no
side bands to look for! If they are there, they are very close
together. With a lack of FM side bands, one would postulate the low
deviation modulation is going to look like just like phase noise.
Obviously very hard to measure without a lot of good equipment.
Is there a way to tease the data out, to at least get a frequency
response plot by disturbing the EFC line with a signal generator? Maybe
for low frequencies, a TIC and known reference could do it. It's
something I'd like to test, but fear it requires more equipment than is
The other thought which you brought up is spread spectrum. If spread
spectrum is taken as an example, the amplitude of the 10Mhz may change
with modulation on a spectrum analyzer. It's an easy enough test to
The bottom line is, at this point, the examples on line and provided
here point towards the fact modulation can happen. Given this
background information it only makes sense to keep the EFC line as
clean as possible.
More reading is necessary to understand what the implications are. Any
other input regarding real numbers, or actual testing is very welcome!
> Some VCXOs actually specify their bandwidth. High audio is sometimes
> useful. I haven't seen anything beyond that, but I'm just listening
> to discussions like this one. There could well be applications that
> use a higher frequency.
> One application is correcting for mechanical vibrations. This is
> interesting in radar used on helicopters. (They do Doppler filtering
> to remove clutter. The lower speed of objects that can get through
> the filter depends on the clock stability.)
> PCs often FM modulate their clocks. It's a hack to get past the
> FCC EMI requirements. It spreads a spike in the frequency domain
> into a blob with a lower peak. I think 30 KHz is typical. The PCI
> specs were tweaked to allow this so they probably say something about
> the legal frequency limit.
> PCs probably don't use expensive OCXOs, but that technology might get
> used in other applications.
> How do FM modulators work?
More information about the time-nuts