Hi

I am currently looking for an external quarzoscillator that can be used
to clock a chip. The chip has a SMA socket, so I would need in some way
an oscialltor that could be connected to this SMA socket. The frequency
I am looking for is 5-10 MHz. Would be thankful if someone could point
me out where to get such a device

Thanks

On Jul 3, 9:56 am, Rob Berger <RobBer...@myweb.de> wrote:

You can buy packaged oscillators with SMA output connectors, and then
you can use an SMA patch cable to connect the two.  See, for example,
http://www.wenzel.com/pdffiles1/Oscillators/STR_4_to_30.pdf.  Be aware
that fully packaged oscillators are generally quite a bit more
expensive than simple oscillator "cans".  But first, I think you need
to establish better just what output you want.  An oscillator may have
sinewave output, or logic level output in various formats (ECL, PECL,
LVPECL, CMOS and LVCMOS all come immediately to my mind).  If the
output is something like PECL, it may well need a particular external
termination (generally 50 ohms to Vcc-2V).  Then there are the
questions of stability and spectral purity.  Is it a simple clock, and
you don't much care about the frequency, or do you want it stable to
one part in 10^9?  Does phase noise matter to you?  If your
application is non-critical, I would suggest getting a common crystal
oscillator "can" with appropriate output level and patching it onto
the board you have, and forget about using the SMA on the board,
though of course it's up to you to decide what's appropriate.

Cheers,
Tom

Hi Tom,

Thanks for your comment!

Looks pretty good I have to admit on first sight!

An oscillator may have

  If the

The thing is it should be quite stable to get proper measurement
results. I have intended to use it as a clock frequency for a
circuit that is implemented on an FPGA. As I am unfortunately no
electronics guru I havent heard the expressions ECL, PECL etc
except CMOS. The most important thing is that the frequency is stable,
that i have proper edges and that it can be connected via SMA ;)

Thanks,
Rob

On Jul 3, 10:47 am, Rob Berger <RobBer...@myweb.de> wrote:

OK, so if it is driving an FPGA input, I suppose that LVCMOS should
work.  The clock requirements should be stated in the board and FPGA
documentation, though I know from personal experience that it is not
always trivial to extract them.

When it comes to crystal oscillators, "quite stable" has very little
meaning.  Can you quantify that?  Does it need short-term stability,
over a period of a second or a minute?  Does it need long-term
stability, over a period of a year or a decade?  Just HOW stable?  The
very cheapest of crystal oscillators should manage 100 parts per
million initial accuracy (with respect to the marked frequency) and
stability over their operating temperature range at least that good
(and commonly quite a bit better).  It is not terribly difficult to
get down to a few parts per million stability at a little higher
cost.  With compensation techniques, you can get down in the one part
per million range or a bit better.  With oven temperature control and
(expensive) attention to detail, you can get down to a few parts per
billion, where long-term drift will dominate in a relatively short
period.  So, what do you need?  The difference in cost from one end of
the range to the other is a few orders of magnitude, easily 1000:1.

A Google search on "SMA crystal oscillator" gives several useful
hits.  But again, be aware that two things are in play here:  the
package with SMA connector is inherently considerably more expensive
than a simpler leaded package (and you will have to supply power to
each anyway), and the sort of oscillator that you will find packaged
with an SMA connector will practically always be at the more expensive
end of the price range (have tighter specs on stability, initial
frequency accuracy, phase noise, etc.).

Cheers,
Tom