# Basic crystal oscillator specifications question

1. Jul 21, 2012

### bmxicle

I need to create a 10 MHz square wave clock signal which gets terminated into a 50 ohm resistor for use by a waveform generator. The specifications for the waveform generator suggest a 12 to 14 dBm power level for the input clock. ie. a power level of ~16-25mW.

So I've been looking to buy an oscillator circuit component online of which there are plenty but I can't find anything at all which meets those power specifications. The way I'm calculating the power is by:

$$p_{rms} = \dfrac{v_{rms}^2}{R}$$
$$v_{rms} = V_H \sqrt{D}$$

Which means i need a high level of $$V_H = 1.12V$$

Is the power output calculated in this way for RF specifications of this sort or am I doing something wrong? Or are Oscillator ICs just generally not available with these power levels? If so what are my options? I'm a physics student currently doing a bunch of electronics stuff so I'm still not incredibly familiar with a lot of specifications given on data sheets but I'm definitely willing to do some reading.

2. Jul 21, 2012

### jim hardy

theer are 50 ohm line drivers like SN74138

might be overkill but if you want something to experiment with,,,,,

3. Jul 22, 2012

### bmxicle

Ahh yes that looks like it could be useful. Since I don't really have many requirements other than generating a clean clock signal overkill is probably fine. I guess will probably just buy a few components and try to see what works best.

4. Jul 22, 2012

### vk6kro

Getting a square wave at 10 MHz is more than power levels. It needs a bandwidth close to 100 MHz to achieve this.

Even if you got it, you would not be able to see it as a square wave on an oscilloscope unless it had a bandwidth like this.

So. I would be more worried about bandwidth than exact power levels.

One way of doing this is to use very fast frequency dividers (like the 74 F family) and divide from 100 MHz to 10 MHz or even from 20 MHz to 10 MHz.

Crystals you see on Ebay vary a lot in accuracy and stability. If you need an accurate reference frequency, be prepared to pay \$20 or more for a suitable oscillator. The "Bliley" brand is pretty good and they make oven controlled oscillators which give a crystal a stable temperature. Of course the oven uses some power (about 500 mA at 5 volts).
Computer grade oscillators usually have very poor accuracy, (+/- 100 Hz) but they are cheap and may be good enough for a signal generator.

Last edited: Jul 22, 2012
5. Jul 23, 2012

### jim hardy

i once made myself an oscillator for calibrating oscilloscope, used Epson SPG8640 which is a crystal oscillator and frequency divider in one IC. It lets you set the divisor so is quite versatile.

But i couldn't locate an IC at Digikey anymore
surely somebody still makes the little fellow.

I took it into work and compared it against our frequency standard and it was good to five digits.

What a tinkerer's delight.
Maybe one of the electronics professionals here knows of something similar.

old jim

edit aha found the datasheet indeed it's discontinued
http://www.epsontoyocom.co.jp/discon/epsondiscon/2000catalog/spg_8640series_e.pdf [Broken]
but it didn't go to 10mhz.

if they still use same oscillator in a more modern product i think you'll be surprised at its performance.

If you need the precision of an oven look for a surplus HP 10544A oven contolled crystal oscillator Ebay lists several this morniing. But, it outputs 1volt sinewave you'll need a buffer behind it.
http://www.ebay.com/itm/HP-Agilent-10-Mhz-Crystal-Oscillator-10544A-/400165955341

Last edited by a moderator: May 6, 2017
6. Jul 23, 2012

### vk6kro

I have one of these:

http://www.ebay.com.au/itm/10MHZ-BL...651?pt=LH_DefaultDomain_0&hash=item27ae11f7b3

which will give 0 to 5 volt pulses out, but probably not into 50 ohms. Ideal for driving counter chips though.

There is a control pin which can be connected to the wiper of a 10 turn pot, with 5 volts across it, so that the frequency can be set to exactly 10 000 000 Hz. Note that is 8 figure accuracy.

I have a Rubidium standard to calibrate against and I use the excellent PC program Spectran to do the calibration. Once warmed up it will hold calibration within 1 Hz.

So, if this is used with a DDS signal generator, I can set a frequency to within 3 Hz or so from 0 to 30 MHz.

It gets better if I use the Rubidium standard.

This is equipment that would have cost thousands a few years ago but is now available to anyone.