10 MHz Reference clock distribution

10 MHz reference clock distribution box

As I have some RF test equipment, various transceivers for my Ham radio hobby, I have a need for a precise 10 MHz reference for all this. Normally you use a "GPSDO" device for this. A GPSDO locks an 10 MHz oscillator to the GPS timing signals, and as the GPS timing is extremely precise the result is a very accurate 10 MHz (you can find more details of the principles behind GPSDO on the internet).


Choices behind the design, main objectives:

  1. Compact solution. Only one single box with GPSDO and distribution.
  2. Quick solution. Use "of the shelf" GPSDO to cut development time (I spent only 1 single day for everything)
  3. Sinus outputs, all my equipment (incl. DF9NP injection board for IC-9700) supports that.
  4. Galvanic isolation. I don't want the distribution box to interfere with normal grounding.
  5. Price was not too important.



There are plenty of GPSDO devices you can buy, just to mention two: 


For injection of LO into my ICOM IC-9700 I use this board also from DF9NP: 49.152 MHz injection board


The typical GPSDO devices deliver a single (or up to 4) outputs with 10 MHz signal. As I wanted 8 ports, and also at the same time would like to have galvanic isolation, I decided to make my own "distribution box". Some of the design is inspired by Dieter DF9NP's devices, some is from my own imagination :)


As I was in a bit of a hurry, I did not want to design, implement and test GPSDO designs, so I made a shortcut. The distribution box I designed, is simply an amplifier with eight galvanic isolated outputs. It relies on "some kind of" GPSDO to deliver its 10 MHz input signal. The distribution board/box takes this reference signal, amplifies it, and drives eight outputs with the (sinus) signal (approximately 6 dBm in 50 ohm). 


All the eight outputs are galvanic isolated using Mini-Circuits ADT1-1+ transformers. I wanted galvanic isolation as I did not want the distribution box "taking part" in the grounding of instruments and radios, I would like to view each of the eight outputs as completely isolated, this also means there is no risk of groundloops, grounding of equipment that I did not want to have ground on etc. The transformer has no spec on isolation voltage (as far as I could find), so don't expect it to withstand large voltages!


The amplifier/clock driver is a Renesas 74FCT3807 that I used in other projects (parallel DSP projects etc). Also Dieter DF9NP uses this in some of his projects. Its a nice device that will work up to 166 MHz. 

Each of the eight outputs has filtering so the square wave output ends up as a sinus wave before being fed to the output transformers. 


In my box, I use a GPSDO from Dieter, DF9NP. I made a large cutout (118 x 76 mm) in the distribution board so the DF9NP GPSDO would fit there, there is plenty of room to put pretty much any GPSDO module you might find. Only requirement to the GPSDO used is that it must deliver a 10 MHz GPS locked signal at around 0 to +10 dBm. If higher levels are available there is provisions on the board to install an attenuator on the input. 

The board has a SMA female connector, this receives the 10 MHz from the GPSDO. There is also switched 12VDC available for the GPSDO as well as a "lock" LED if the GPSDO has an output for that (you need to  adjust the series resistor for your particular GPSDO). On the back of the distribution box there is a SMA female bulkhead connector, this connects to the GPS antenna input on the GPSDO. This allows you to connect an GPS antenna to the backside of the box, and the signal will then be fed to the GPSDO via this bulkhead connector/cable.


All of the above results in a nice and compact device with everything inside. Only connections needed to the outside is 12VDC (the box uses linear voltage regulator to avoid any switching noise), the GPS antenna. In addition to this, 8 BNC connectors are available each delivering 10 MHz at approximately 6 dBm (50 ohm).


The distribution board, front and backplates are all made of 1.6mm PCB (ordered from JLCPCB) and fits directly in a Bud 1402D box.


The complete cost of the components for the board is €154,- including everything (except a GPSDO) if ordered from Mouser (see below). Add to that €12,- for the 3 PCB's (mainboard, front and backplate).


More than 40% of the cost is the "Bud 1402D" box. Each of the eight outputs costs around €8,-, the transformer is the most expensive part in each output, price around €4,- each, the rest are the filter components and the BNC connector.


I designed the board so it would fit in that as I already have a number of "DIY" projects in this series of boxes (f.ex my URC - Rotorcontroller and some other, not yet documented, projects).


Below you will find schematics, BOM file (with order codes for Mouser and Farnell) and Gerber files for the board, front (both with "LOCK" and "NOLOCK" text) and backplate. With this, you will be able to make your own board (use at your own risk!).

Please note that there are two versions of the front plate. One has the text "LOCK" and the other "NO LOCK". Select the one you need depending on the GPSDO you are going to use. The DF9NP I use has a "No lock" LED, other GPSDO has a "LOCK" LED. Also note that the series resistor (R16) to the "LOCK/NO LOCK" LED might have to be changed, in the schematics/BOM below it is listed a 0 Ohm.



Bill of materials PDF file

(You need to add a few passive resistors/capacitors to the BOM list to make it complete (unless you have these at hand already), please check schematic.


Gerber files for PCB board

Front plate with "LOCK" text for GPSDO, Gerber files Front plate

Front plate with "NO LOCK" text for GPSDO, Gerber files Front plate

Gerber files for Back plate


The gerber files can be uploaded "as is" to JLCPCB. I selected 1.6mm thickness and black silkscreen for all three boards.


Below is the complete BOM list for the project from Mouser, with prices for each item.

Add to that the €12,- for PCB's (plus some shipping) and f.ex €74,- for the DF9NP GPSDO.

The design of the front and backplate was done in Autodesk Inventor (MCAD). The process is very simple, there is no need to do a lot of measurement in the electronic CAD program, Altium Designer (ECAD) and move these to the mechanical part of the project.


When doing the PCB board, it is VERY important to have accurate 3D models of all components. If that is the case, it is just a matter of exporting the board (including components) as a STEP file and then import that in the MCAD program. The Bud 1402D box also has a accurate STEP file, this is also brought into the MCAD program.

The box and the PCB are then put in an "assembly" in the MCAD with the correct constraints applied.


After this, it is just a matter of making two new parts that both are derived from this assembly. All the connectors on the back and the LED, switch etc. on the front are then extruded from these parts. Both the front and back are then converted to "sheet metal" parts and exported as DXF files. 

The two resulting DXF files are then imported into the ECAD system where they will define the outline and cutouts of two new PCB boards. I exported the front and back PCBs as STEP files also and imported them into MCAD, this allows me to check that everything fits and looks "nice".


Its then just a matter of generating Gerbers for the front and back, and order these from the PCB manufacturer.

I get most of my prototype PCB's from JLCPCB, the cost for the main PCB is €9,- the front and back are both €1.50,- so total €12,-


Below are some screenshots of the MCAD process.

More information coming, measurement results etc