0
\$\begingroup\$

Many lab measurement equipment for analog, radiou and RF purposes, such as frequency counters, arbitrary signal generators, oscilloscopes, etc. have the option to input an external reference clock (almost always 10 MHz, though some micro/mm-wave equipment is switching to a higher frequency reference).

From what I know, using this has two advantages:

It allows me to lock different instruments to each other - if my synthesizer and spectrum analyzer are using the same 10 MHz reference clock, they will relative to each other have no error except for their own internal jitter. This can be very useful since it removes (well, significantly reduces) uncertainty in my measurements.

A second advantage is that this allows me to use a very high-performance clock and share it over multiple instruments, instead of needing a very high performance clock in every instrument.

This brings me to my question:

If I have a lab with multiple devices, (why) would I still want a separate precision timebase (such as a TCXO) in every instrument, instead of just getting a stand-alone reference and provide all of my instruments with this reference?

(Ofcourse, we assume that all our gear can take an external reference clock. I think that this is possible with almost all modern gear that needs a timebase and thus a valid assumption)

\$\endgroup\$
  • \$\begingroup\$ Sometimes one internal precision oscillator provides an output that can be used to drive the other devices. \$\endgroup\$ – Brian Drummond Oct 3 '17 at 10:39
  • \$\begingroup\$ @Joren Vaes: if it is good enough that each instrument has its own clock and it is already there then the advantage is that you don't have to wire a master clock signal to each instrument. \$\endgroup\$ – Curd Oct 3 '17 at 10:48
  • \$\begingroup\$ @Curd ofcourse, but often getting a TCXO timebase in an instrument adds to the cost of the instrument (very often it is an upgrade). I was wondering why I would want that, instead of going for a single master, which might be a much more economical cost when building up a lab with a hand full of instruments that can use a external reference. \$\endgroup\$ – Joren Vaes Oct 3 '17 at 10:50
  • 2
    \$\begingroup\$ @Joren Vaes: but lab instruments are designed and bought to be used as flexible as possible, i.e. without assuming any overall configuration together with other instruments, e.g. you don't want a synthesizer or SA that is not able to work on its own. Maybe next month the instruments are used in a completely different configuration or stand-alone. \$\endgroup\$ – Curd Oct 3 '17 at 10:57
  • \$\begingroup\$ The reason the TCXO is a cost-added option is because some customers want or need it and others don't. If your application always allows use of an external reference, then you're one of the ones who doesn't need it. If your application requires the instrument to work stand-alone or as the reference for other instruments, and you need the accuracy, then you're one of the ones who does. \$\endgroup\$ – The Photon Oct 3 '17 at 15:54
1
\$\begingroup\$

If I have a lab with multiple devices, (why) would I still want a separate precision timebase (such as a TCXO) in every instrument, instead of just getting a stand-alone reference and provide all of my instruments with this reference?

My answer would be: flexibility

At the moment I work at a small company and we only have a few measurement instruments which would need an accurate clock. We also have no need for a more precise clock so we don't want to spend money to buy one.

Also we do not need very precise frequency synchronization, the few ppms frequency difference we get from each instrument using its own clock is perfectly acceptable to us.

So we do not have and do not need a common reference clock.

Couple of years ago I worked at a larger company where we had lots and lots of measurement equipment and there it was standard practice to use an external clock. Obviously the money was already spent on this 10 MHz source so we better use it as much as possible. This also helped avoiding small frequency differences between setups.

Nice thing is, you can do both with the same set of instruments. And since for most equipment an internal oscillator does not add that much to the price I think it makes sense to include one by default.

\$\endgroup\$
  • \$\begingroup\$ I guess the flexibility is a very good argument. Coming from a single, central lab where nothing ever moves, I forgot how often you might want to move equipment around or even take it with you to do on-site measurements. My background was for a home lab, where I wanted to upgrade a counter with a TCXO, and then started wondering why I wouldn't just put that TCXO in an external system and include a clock distribution network and allow all my other gear to also use this reference. Thanks. \$\endgroup\$ – Joren Vaes Oct 3 '17 at 11:40
1
\$\begingroup\$

I have a spectrum analyser, sometimes it is in the air conditioned lab, and hooked up to a GPS disciplined reference, sometimes it is thrown in the back of the car and goes out to a transmitter site, it is nice if it is still reasonably accurate.

Also, phase noise is often lower with an OCXO then it is with a cheap jellybean XO, and locking to an external source only helps with this within the loop bandwidth of the PLL, sometimes this really matters.

Regards, Dan.

\$\endgroup\$
  • \$\begingroup\$ If the loop bandwidth of the PLL is an issue, would the OCXO not face the same issue? Most applications I have gone over seemed to just use a 10 MHz TCXO or OCXO. How would this be different for the PLL than an external 10 MHz signal? \$\endgroup\$ – Joren Vaes Oct 3 '17 at 12:34
  • \$\begingroup\$ Within the look bandwidth the phase noise is controlled largely by the reference oscillator, outside the loop bandwidth it is controlled by the VCO. If you have some box that multiplies the reference up by a few hundred times and need fast locking (say for frequency agility), then much of the close in phase noise from the reference will appear close to the output carrier due to the wide loop bandwidth required. OCXOs generally have better close in phase noise, so this translates directly to a quieter output close in. \$\endgroup\$ – Dan Mills Oct 3 '17 at 13:01
  • \$\begingroup\$ Within the LOOP bandwidth.... \$\endgroup\$ – Dan Mills Oct 3 '17 at 14:42
1
\$\begingroup\$

Bimpelrekkie already pointed a valid benefit and I would like to mention another:
known parameters.

  1. You might want and expect your instrument to be able to operate without external equipment unless it is some kind of a module of a larger system.
  2. Properties of internal oscillator are known (stability, accuracy, aging, you name it) and instrument can be calibrated and specified as a whole even if there is a possibility to use external time-base.
\$\endgroup\$
  • \$\begingroup\$ Very good point. I would imagine that the characterization of a instrument as a black box, independent of externals can be a very valuable asset. It's also easier to send one device out for call than it would be to send your main reference out for cal, and have all your measurements take a big hit in performance in it's period of abscence. \$\endgroup\$ – Joren Vaes Oct 3 '17 at 11:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.