Building a 10MHz Master Clock

Even though I have a GPSDO, and in the process to upgrade it, I still would like to build a 10MHz reference clock.

I have a spare Oscilloquartz 8663-XS OCXO that I wanted to use for my GPSDO, but the lack of room on the PCB stopped me in my tracks.

This OCXO is too good to not use, so I embarked on a little side project to put it to good use.

For a very long time, I already collected information from another user (Gertjan Miedema) that built a frequency reference with that same OCXO, so I borrowed some of his ideas.

Here is the link to that design : https://www.circuitsonline.net/forum/view/126987/1

It's a Dutch forum but I take it you can either follow it, or use Google translate.

There are three areas where I'm going to deviate from that original design. First of all, I'm not going to create a separate power supply for it. The unit will be on 24x7 and I did not want a normal transformer to power the unit. I have a nice 15VDC 1A wall-wart, so I'm going to use that. The stability and precision of the 15V is not that important for this design.

The other area I'm deviating is by using different output transformers. There was a large effort to create very special hand-made transformers, but I did not want to go that far. I'm going to use the commercially available transformers that I already use for several other projects.

Lastly, I'm not going to use the Oscilloquartz reference voltage to power the frequency setting circuit. It seems the logical thing to do, but this so called reference voltage is anything but a stable reference voltage. They are typically created by using a Zener (reference) diode like the TL431 inside the OCXO, but that is simply not good enough as you will see.

During my work with the Reciprocal Counter, described in another post here, I found to my dismay how poor that reference voltage really is. Have a look at that post and search for the following update: "Testing the counter (11-05-2023)" towards the end.

Here is a display of the Vref output that shows what I mean:

This OCXO has been powered for many weeks 24x7 and is in a plastic container isolating it from drafts, but not the room temperature of course. It's easy to see that using the Vref of the OCXO will fall short when you really want a very stable 10MHz, because any variation in the Vref output, as little as micro Volts,  will show-up as a change in the output frequency.

The Oscilloquartz is currently running while using the following circuit, put together after my "discovery". Here it is using a 10V reference I had in my stash already. This prototype is built on protoboard.

My plan is to use the above frequency setting circuit, but with a REF5050, which is a 5V reference.

Here is the schematic I decided on, with values corrected/tuned during the testing.

Another deviation is that I only need two isolated 10MHz sine wave outputs, not 4.

Most of the resistor values are taken from the original design and they work perfectly.

My intent is to put the circuit in a metal enclosure, but put all the electronics in an isolated foam box inside the enclosure so the OCXO will warm-up the inside far enough above the maximum room temperature, hopefully reducing the temperature dependencies of the components.

This is a side-project so may not get a lot of attention while I'm working on the GPSDO version 4 and the Reciprocal Counter.

[update July 2023]

I did manage to create a PCB layout.

The two long multi-turn trimmers top left are from my old part stash, even through they date back to the 70's, they are genuine and of very good quality. The adjustments of these course and fine trimmers will protrude through the front panel, the same as the two SMA output connectors. The length is 125mm and the width of the board is 99.5mm so it will slide in the typical enclosures I use.

[Update November 2023]

I got the PCB's and populated them. I found some issues with the silk screen and some footprint errors that I corrected in a V1a you see below. That is the actual version you can find in the Shared Project section of PCBWay. Later I will add all the files on my Github site.

Based on the specification for the Oscilloquartz 8663, I made a change to the supply. The voltage specification for the OCXO is 12V +/-5%. Originally, I wanted to use a 12V supply but the Schottky protection diode and the series resistor will bring that too much below that specification. The series resistor can be omitted, I left it in there because Gertjan measures the voltage drop and has a circuit that will show the oven activity. I decided to omit that for the moment, because I don't see the practical use when the nit is powered-on all the time. It makes more sense to indicate a stable output after the oven has finished the warm-up period. The additional circuit can be easily added by using a small separate PCB. 

I now changed the raw DC voltage input to 15V and added an LM340 12V regulator that is mounted (isolated) on the side of the enclosure.

There are some footprint refinements and silk screen errors on this V1a that are not reflected on the real PCB V1 below. This PCB was manufactured by PCBWay and as usual, has an excellent quality. One of the giveaway's is to look for holes that are drilled in pads. The hole should be exactly in the center, which they are. The gold plating of the pads makes it very easy to solder the components.

I also designed a Face plate for the enclosure. Make sure you order the black color.

Below is the actual PCB from PCBWay. The black finish with the white silkscreen makes it very professionally looking.

After putting it all together and trying to calibrate the output with my GPSDO, I'm pretty impressed.

I first tuned the output frequency to that of my GPSDO with the course adjustment, and then switched to the finer adjustment over longer times.

Here I'm measuring the drift by using the maximum persistency of the DSO, while correcting the drift with the "fine" adjustment.

It's a little hard to see on this screen shot, but the sine wave output is exactly 2Vpp into 50 Ohm as measured by the DSO.

Here are some more pictures on how the inside of the enclosure looks and what I did to insulate the components as much as possible from room temperatures. It does not show the later added LM340 12 V regulator.

Below is the insulation "box" or rather a cover, that I made out of 6mm Styropor. The box rests on top of the PCB. The bottom of the PCB is not insulated, but that is much less affected by drafts etc.

I mounted the OCXO about 1-2 mm above the PCB so that there is minimal heat transfer to the PCB itself. There is also no ground plane below the OCXO again to reduce the heat transfer to the PCB. Obviously, the inside of the box will be heated by the temperature of the OCXO oven, but that will happen pretty uniformly.

There are some holes in the insulation cover, but that is not a major deal I hope. It also prevents overheating of the OCXO as well.

I use a small trip of Styropor on top of the insulation box such that the cover of the enclosure presses the insulation box on top of the PCB.

Initially, I hooked the reference to my DSO, together with the GPSDO, to calibrate it. It's still drifting a bit after having been powered off for a month or so.

At this moment I'm now feeding the output to my counter and will attempt to adjust the frequency, and also register the min-max and avg values.

Running the tests and tune the calibration is a matter of a lot of patience watching the deviation on the counter and making small adjustments to fine-tuning the frequency (;-)).

The latest Gerber files can be downloaded from the PCBWay site here: 

https://www.pcbway.com/project/shareproject/10MHz_OCXO_frequency_reference_1_of_2_ba8a8ef2.html

And from my Github site here:

https://github.com/paulvee/10MHz-frequency-reference/tree/main

Purchasing various parts

I was advised by the support people at PCBWay that my BOM is not complete enough for them to automatically populate the PCB, so I added some more information. The trouble is that I have not figured out how to add this information to the BOM I get out of KiCad.

For FB1, 4 and 5 I used this Digikey partnumber: 240-2551-1-ND
For FB3 I used Digikey pn 240-2391-1-ND 
This is a 2A part, but you cal also use the 1Amp version I used for FB1. 
For FB2 I used Digikey pn 1934-1411-ND
For U1, the REF02, I used Digikey pn 505-REF02CSZ-ND 
For R4, you can use pn CSR1206FKR500CT-ND this is an optional resistor that you can also bridge.

It’s is OK to use a 2.2uF capacitor for C14 and C16, this is not a critical value at all.
For D1 I used general purpose Schottky diodes I have in stock. You can use the S1G with Digikey pn 1655-S1GCT-ND
The Oscilloquartz OCXO is a used version I typically buy from vendors that advertise on eBay or Aliexpress. Just Google for “Oscilloquartz 8663-XS” to get the right version and then select the price. Here is one: https://www.ebay.com/itm/204065244750 
I have very good experiences with seller Queen*s_land https://www.ebay.com/str/bluegirl5

The enclosure I use has partnumber bi0002562 and is described as a "high quality aluminum project box/enclosure case" with the sizes 150x105x55mm. Here is a link that hopefully stays up for a while:

https://www.aliexpress.com/item/32766709803.html

Some assembly tips

Before you solder the two SMA connectors in place, slide them into position on the PCB, they have a pretty tight fit. Then slide the PCB in the bottom enclosure and also mount the front panel to the bottom part of the enclosure. Move the SMA connectors so they are centered and protruding through the holes of the front panel. Tack them in place to the PCB with your solder iron. Now add the two trimmers and make sure they are aligned in front of the two adjustment holes in the front panel. Use some tape to secure them in position, slide the PCB out of the enclosure and solder one of the leads of the trimmers and check the alignment again. Only then solder all connections of the SMA connectors and the trimmers. With these four parts in place and the PCB in position, mark the place for the 3.5mm mounting hole in the bottom enclosure such that the PCB will always be in the right place.

The red power LED will slide in the 3mm hole of the front panel, I use some glue on the back side to keep it there and use leads and a connector to mount it to the PCB.

The four mounting holes on the front panel need to be counter-sinked on the outside to use the screws that come with the enclosure.

Stay tuned for more...

Paul

If you like what you see, please support me by buying me a coffee: https://www.buymeacoffee.com/M9ouLVXBdw