I was looking at a low cost way to build a 10Mhz frequency for my electronics lab. I had a few options that I could pursue, these were…
- GPS Disciplined Crystal Oscillator (GPSDO)
- Rubidium atomic standard (RbXO)
- Caesium atomic Standard
- Oven Controlled Crystal Oscillator (OCXO)
So to make a choice on what I should use I had to come up with design parameters for my frequency standard, these were as follows.
- Had to be low cost
- Had to be portable
- Had to work inside of a building
- Had to be stable, better then +/- 0.5 hertz drift over 2 minutes
The preceding criteria ruled out a GPSDO as that requires an antenna that has a view of the GPS satellites, this would be ok at home but I didn’t want to have to make sure I had a outside view of satellites if I was taking it to someone else’s shack or like the club shack with no windows this would have been impossible to get a GPS lock.
I next looked at atomic standards. The Caesium standards were out of the question due to the cost, second hand you could expect to pay upwards of USD$5000 for one, certainly not low cost by any measure. The rubidium standards were a lot cheaper at around USD$200 so that was an option. This raised the question, did I need the accuracy of a Rubidium or could I get away with a cheaper option ?
This led me to investigate OCXO’s to see if they would suit my needs. First was to see if they met my stability requirements. A typical 10Mhz OCXO has a stability of 5×10-10 This is ±5 mHz drift per second on a 10Mhz signal, well within my requirement of 0.5hz over 2 minutes. The reason I need this stability is for WSPR digital which requires a very stable clock signal.
What about the cost of an OCXO ? Well a quick search of eBay led me to a Double Oven OCXO from a Russian company called Morion. I could get a second hand unit for less than $40 delivered. This particular unit listed stability of better than 2×10-12 over 1 second which is 0.005 mHz and stability of ±5×10-10 per day at 10Mhz. These figures were well within my requirement’s so I ordered two units from eBay.
I now started to put some thought in to the design and construction of the complete unit and what I would need.
I had an old car computer case that I could use for the project so I ripped out the old motherboard and found some rubber feet in my junk box to put on it, this gave me an idea of the size case I had to work with so I grabbed a ruler and measured it up to see if the OXCO would fit. It would fit with heaps of room for an internal 240 to 12 VDC power supply and a battery.
I was thinking about the power requirements for the unit and how I was going to power it. I needed about 1.5 amps for the oven while it is warming up with that dropping to about .5 amps once warm. Wrapping the oven in insulation should drop that even lower. I had an old 4 amp 12VDC power supply from a computer monitor that I could use so I dug that out as well.
At this stage I put the project aside for a couple of weeks while I waited for the OXCO’s to arrive in the post. When they had arrived I put one in the case with the power supply and soon realised I actually had the space for both of the OCXO’s.
I had originally bought two so I had a spare but I quickly decided at the cost of them I could just run both so I would have two units that I could compare against each other to make sure they were still within specification and also It meant I could connect it to more than one device at once. The signal quickly attenuates if you are splitting it.
I then got all the other stuff together to assemble the unit this consisted of the following.
- Trimpot’s to adjust the OXCO
- SLA 12V Battery
- Switches to switch power and outputs
- Veroboard to mount it all on
- LED’s for status indicators
- Volt Meter for battery level
- IEC Socket for Mains input
- DC Barrel plug and socket for 13.8VDC to charge battery
- DC-DC Buck/Boost converter to level out battery voltage
- BNC Sockets for the output
- Current limiting resistors for the LED’s
I soldered the two OCXO’s on to a bit of veroboard and then connected 25 turn 2KΩ trimpot’s with the wiper to the calibration pin with one side of the trimmer to GND and one to the 5VDC reference output on the OCXO, this forms a voltage divider to calibrate the oscillators. I measured the output of the power supply I was going to use and it was 12.3VDC which is within the spec of 12VDC ±5% that they require.
The next problem I would have was to power it while travelling, I didn’t want a huge battery so I used a 1.3AH SLA battery that I would charge off the car while travelling. I needed to keep the battery voltage at 12VDC into the oscillators while I would see between 14.2VDC while charging and 11VDC if the battery was a bit flat.
To get the nice 12VDC I used a cheap $2 buck boost DC-DC converter from eBay. They wont supply the 3 amps I need to warm up but it would handle the .5 amps that I had measured once warm. I had managed to get the 400mA current per OCXO down to around 500mA for both once warm by wrapping the ovens in neoprene foam from a stubby holder.
I adjusted the DC-DC converter to output 12.3VDC so it was the same as the mains power supply, this prevents instability of the 10Mhz signal due to supply differences.
I have each oscillator feeding a DPDT switch, one pole switches the signal the other switches an LED on to show the output state, I can feed either the A or B signal to a common N Type connector or to their own BNC connector. A cheap LED volt meter was added to the front panel to show the battery voltage, these are available on eBay for a couple of dollars.
If you were building the unit with one OCXO and had to buy everything instead of raiding you junk box for parts you would need to invest around 80 to 100 dollars. This includes a case the OCXO and the needed hardware. It took me around three hours to assemble.
So what is it useful for ? Well some of the things you can use it for are as a stable reference to calibrate test equipment like frequency counters and signal generators. With a divider board to generate a one pulse per second signal to sync your PC time. This is useful for the digital mode WSPR that requires no more then ±1Hz drift over two minutes. With a general multiplier/divider DDS you can produce signals from 1 hertz to around 100Mhz from the 10Mhz input. This is useful for calibrating rigs to see if they are on frequency.
All in all I am very happy with the resulting unit and it makes a nice addition to my test bench.