# 10MHz ext ref with fail over to internal 5MHz ref

#### Joester0064

Joined Jun 14, 2023
26
Hi Folks,

I need to make a module that will take a 10MHz GPS ref in, and provide a divided 5MHz output. If the 10MHz is removed then it fails over to an internal free running 5MHz reference.

I plan to build a clock circuit to connect to a DDS which I will program to generate a 5MHz sine. This will take care of the internal 5MHz ref.

My issue is it seems that I am to use a divide by 2 to divide down the 10MHz ref but this will distort the sine wave of the GPS ref. I'm not sure how to do that part. Once I figure that out, I'll work out the fail over switch.

Any thoughts at all are much appreciated. Thanks!

Joe

#### michael8

Joined Jan 11, 2015
414
A square wave is mostly the frequency plus it's odd harmonics, so you can likely filter out the higher harmonics to get
a sine from a square wave.

some ideas: http://techlib.com/files/dividers.pdf

Why the DDS? What is the source of the clock for the DDS?

How about just a 5 Mhz crystal oscillator?

http://techlib.com/electronics/lowdistosc.htm

Anyway, keep in mind that the switch between two 5 Mhz signals will leak.

The output of the switch will be the selected signel along with a weak version of the other signal.
If the two source are not exactly the same frequency, they will go in and out of phase. Depending on the
use of the output this may or may not matter.

Many instruments avoid this effect by always using the internal oscillator but phase locking (or injection locking)
it to the external signal. This way they are the same frequency...

#### Ya’akov

Joined Jan 27, 2019
9,164
Many instruments avoid this effect by always using the internal oscillator but phase locking (or injection locking)
it to the external signal. This way they are the same frequency...
Phase locking an internal reference was my first impulse, it avoids a lot of trouble. If it uses an OCXO as the timebase, it should be very stable and remain accurate even without the GPS discipline.

#### Ian0

Joined Aug 7, 2020
9,817
I can’t think of anything that needs a sinewave at that frequency. Most inputs in the 1 to 10MHz region are clock signals.

#### Ya’akov

Joined Jan 27, 2019
9,164
I can’t think of anything that needs a sinewave at that frequency. Most inputs in the 1 to 10MHz region are clock signals.
Well, there are frequency standards for RF circuits, very commonly a Rubidium disciplined crystal oscillator at 10MHz. They are used for things like microwave transmitters where very high precision is needed.

The Rubidium standards are all over the place.

#### Ian0

Joined Aug 7, 2020
9,817
Well, there are frequency standards for RF circuits, very commonly a Rubidium disciplined crystal oscillator at 10MHz. They are used for things like microwave transmitters where very high precision is needed.

The Rubidium standards are all over the place.
But do they have to be sinewaves? A 10MHz reference for a frequency synthesiser goes into the reference input of a PLL, and the output, divided down, goes into the other input, where they generally drive digital phase comparators.

#### Ya’akov

Joined Jan 27, 2019
9,164
But do they have to be sinewaves? A 10MHz reference for a frequency synthesiser goes into the reference input of a PLL, and the output, divided down, goes into the other input, where they generally drive digital phase comparators.
Well, they do produce sine waves, whether or not they are always used in applications that require them. They are used to produce clock signals in some applications but they are also used for their analog output in non-synthesized RF circuits. This is the output of a typical Rb standard:

#### Joester0064

Joined Jun 14, 2023
26
thanks all,

basically I need to build a module that takes 15vdc in, 10M GPS and has a referenced 5M out, if 10M GPS is removed then it will use an internal 5 MHz ref. Sounds like I wont use a switch for the fail over but some kind of PLL situation. Any further ideas are welcome.

Thanks again!

#### Joester0064

Joined Jun 14, 2023
26
defiantely not easy to find an off the shelf module for this, it is for a U158 Mitec Upconverter.

#### Joester0064

Joined Jun 14, 2023
26
Basically, im thinking the %M output will always be from the VCO its just whether or not it is reference to the divided 10M. Thing is the phase voltage will go out of whack if the 10M is removed.....

#### michael8

Joined Jan 11, 2015
414
On sine waves vs square waves, remember that square waves contain multiple frequencies.
Depending on the length and type of transmission line these may travel at different speeds and thus
not be square at the destination causing the timing point on the waveform to change. Since the
transmission length changes with temperature the phase of the result will drift...

So there are cases where sine waves are better...

And sine wave can be converted back to square:

A Physical Sine-to-Square Converter Noise Model
Attila Kinali