Build a 10MHz OCXO for the shack for less than $40.00

VK2RK

Active member
Oven Controlled Crystal Oscillators can be sourced for as little as $5.00 and as much as $20.00 depending on type
These units are surplus disposals, I think from the mobile phone industries, offering very good long term stability that for the price one should build one for the shack. Can be used as a reference signal at 10MHz to calibrate radios and measuring equipment, even as an external reference for instrument that use non ovenised oscillators that have poor short term stability.
These modules provide short term stability of +/- 5ppb (0.05Hz @10MHz) and long term stability of 100ppb per year equating to 1Hz @ 10MHz long term stability, this drift if the oscillator is run for one year continuously, the short term stability figure is the changes in the environment temperature with swings from -10 to 70 C. in reality the frequency inaccuracy will be a lot less the specified stated values.

Modules used
CTI OSC5A2B02 5Volts Square Wave Output (External voltage reference)
CTS 196 5Volts Sine wave output (Internal voltage reference)

The design of the PCB is to accommodate either module, provision is made for required voltage reference components on the CTI module that are not fitted to the board if the CTS module is used.

There is also on the PCB provision for a 50 Ohms to 50 Ohms attenuation is so required

The operating voltage can accommodate from 6 volts to 24 volts, using a low cost buck regulator ($1.50)
Note that at voltages less that 13.8 V the input current will increase, its recommended to run the unit at 12 to 14.2 volts

The PCB is not yet ready for production as I am waiting for some components to confirm dimensions and to also provide for an alternate buck regulator, when I am happy with the PCB I will go into production.

The prototype has been running now for several weeks and the stability is excellent exceeding specifications, the warm up time from first switch on is 3.5 minutes.

Note these modules new sell for many hundreds of dollars...


PCB.jpg

R1, Z1, D2, C2 Not required if using CTS Module (Reference Voltage Provision)

R2, R3, R4 Used to provide attenuation of output signal

IMAG3742~2-20210630-133231392.jpg
 
Last edited:

VK3ZYZ

Moderator
That looks pretty good. I'm waiting for mine to arrive!
Those regulator boards a quite cheap, and I have used them in some things I've made to play with.
But, in my commercial designs, I use the Traco power TSR 1-2450 (or similar ) terminal regs that replace 78xx chips.

TracoPower TRS-1.png



No heat sink required.
There are a few brands of these available now. A great update for a lot of devices to reduce heating and power consumption as these are switch mode regulators that drop straight in the original regulator's place.

This is a power supply mod I intend to do for my frequency counter, along with the ref frequ locking mod mentioned elsewhere. The power saved by getting rid of the linear reg will help offset the oscillator module's needs.
 

Attachments

Last edited:

VK2RK

Active member
Those look good, the maximum current of 1 Amp and the cost is what limits me from choosing them.
The ones I have selected cost less than $2.00
Thanks for the heads up, will come in handy some time or other.
 

VK2RK

Active member
I found this one...
View attachment 723

$3.98 so a lot less than the "real"ones.
You are looking at replacement for 7805 and pay a premium for the pin placements, in my case I have no such restriction and use a more generic pin layout.
There are several types to be had, all use the same chip, I have chosen two types based on availability and provisioned the PCB to accept either, the price is less than $1.50 each.
Tests show the noise figure from either type to be very low and less than 30mV, under load even less.

Images below shows the types I am using and have tested.

IMAG3742~2-20210630-133231392.jpgs-l500.png
 
Last edited:

VK3ZYZ

Moderator
Yes, those are very handy. I do use those like on your board.
And some versions are preset without the trim pot.
The ones shown above would be fairly neatly modded as a 3 terminal replacement if needed too.
It is amazing how cheeeeep these things are now!

Oh, and just for interest, replacing yhe 7805 in my counter with a 3 terminal switcher, dropped the secondary current from a bit over 400mA to 270mA.
 

VK2RK

Active member
Yes, those are very handy. I do use those like on your board.
And some versions are preset without the trim pot.
The ones shown above would be fairly neatly modded as a 3 terminal replacement if needed too.
It is amazing how cheeeeep these things are now!

Oh, and just for interest, replacing yhe 7805 in my counter with a 3 terminal switcher, dropped the secondary current from a bit over 400mA to 270mA.
Not surprised as the voltage drop across the device that is consumed as heat is lost energy.
The ones on the right of my post are programmable and can be set to a fixed 5V

s-l1600.jpg
 

VK2RK

Active member
In the quest to achieve better stabilty verses temperature changes I have selected a LM4040 - 4.1V constant voltage referance device.
The original PCB design can be used with the new device, the new board has provision for the T92 foot print as the image shows.OCXO_V2.jpg
 

VK3ZYZ

Moderator
Photo on 13-07-2021 at 3.02 pm.jpg
I have just grown one into my counter.
This has an AD585 5V reference and a 10K multi turn pot for adjustment.
I hope the AD586 is ok. It turns out I have a lot of them in stock.
All that remains is to let it sit running for some time then calibrate it.
 

Attachments

Last edited:

VK3ZYZ

Moderator
Ray and Josh helped me to set the frequency.
That proceeded a lot better after I soldered the gnd end on the calibration pot!!!!
Now it is at home on my UPS and it will be of interest to see how it settles in.
 

VK2RK

Active member
View attachment 741
I have just grown one into my counter.
This has an AD585 5V reference and a 10K multi turn pot for adjustment.
I hope the AD586 is ok. It turns out I have a lot of them in stock.
All that remains is to let it sit running for some time then calibrate it.
What one did you use, J or A or K or L and B the drift error varies from 25 PPM to 5 PPM ?
Assuming the J those cost just over $10.00 each

The one I selected with postal charges is $2.50

I am trying to get great stability at the lowest possible cost, in your case if your have them use them, well done.
 

VK2RK

Active member
Project Diagram.

The addition of a low pass filter will change the square wave type module to a sinosoidal output, something to do later on.

OCXO-V2.jpg
 

VK3ZYZ

Moderator
Precision Freq Source.png
This is what I have.
It looks like the AD586 is the "L" version, so 15ppm/1000hr
 
Last edited:

VK2RK

Active member
I thought I had finished but I had not...

Some changes have taken place aside the use of a constant voltage source.
Component values had to be adjusted to operate with in the best least temperature drift specifications of the constant voltage source, the ideal current into this device is 1mA, the Control voltage of the module specify that the impedance is 10K, with this the new value for R1 was calculated to be 495 Ohms, the closest in 1% tolerance was to use two 1K metal film resistors. Not that critical as long as the current into the LM4040 does not exceed 1mA thus using two 1K resistors is a good solution.

Then in the calibration of the oscillator, it became clear that this is best done when the instrument has been allowed to temperature stabilise within the housing, in wanting the very best achievable result, removing the top off the instrument to calibrate was not the best way to achieve the required result.
An externally accessible control was an obvious solution, at the same time why not make this a fine adjustment utilizing two potentiometers, one internal for coarse set, and the external accessible one for the final fine calibration.

If requiring a finer control the value of the 150 Ohm resistor can be increased, but note in doing so the resistor temperature coefficient becomes more noticeable, the lower the R value the least will be the effect.
In these things all are a compromise and the achieved result is the sum of all drift components.

The HP counter with this new Oven is now 5% better than the original manufacturer specification with a short term error of better that 1 x10^-7 that is below the count error of +/- one digit.
The accuracy achieved on the HP counter is now 5 x 10^-9
Note that on the Ovenized HP5385A there is no Oscillator Adjust, just a hole, this is for the non compensated reference oscillator in the cheaper model, the hole was a convenient place to have my fine adjust pot.

ocxo.jpg






IMAG3805-20210721-142220501.jpgIMAG3806-20210721-142219915.jpg
 
Last edited:

VK2RK

Active member
PCB boards arrived from manufacturer, built two units using the alternate OCXO's and regulator boards, both OCXO require the external reference that now uses a constant voltage source (LM4040) The stability is very good with a drift error of +/- 0.001 Hz More testing to be done, along with a low pass filter to be designed for the OCXO that outputs a square wave signal to then provide a sine wave. the user then can make the choice of square wave or sine wave output. The filter will reside in place of the attenuator resistors. Got to do a bit more testing as the modules behave differently, I think the small square one could be the better of the two.


IMAG3816~2-20210805-232715206.jpg
 

VK2RK

Active member
The testing of the board that used the CTI module produced a new set of problems that took some time to figure out what was going on.

The problem was that the board was sensitive to differing loads causing a frequency shift of +/-0.1Hz This was not acceptable and I puzzled on the reasons causing this result.
The other board with the CTS module did not exhibit this effect, so I thought that the CTI module required a buffer to isolate the load from the module. Bummer I thought, this will increase the component count and the associated cost of the project.

Had a look at what other did on the net, some used a buffer for providing more than one output and others did not use a buffer.
Well either they are not aware of the problem I was experiencing or something is wrong with the design of the board.

I noted that if I touched the metal case of the module the VC voltage would change, interesting I thought, noise could be effecting the VC but why?
A good look at the data sheets showed that the VC impedence on the CTI module is 100 K ohms and the CTS module is 10 K ohms, there is the answer.
The CTI module would be subject to common mode noise issues were CTS module not as much, this revealed an issue with grounding of the PCB.

I tend not to bond as many points to a common ground clad as this makes soldering difficult if the PCB is not preheated, in this case I have no choice but to do so, you would not thing that a very small value of resistance in a ground track would be a problem yet in this case it was any resistance no matter how small in the earth tracks is a path for common mode noise to get in, I could measure voltage drops (Very small in the micro volts) across the tracks in question.

New version of boards on order ( Almost done with project ) I said this already...

As a side not I noted a design that the author used a voltage reference that had a very low output impedence in the range of 100 Ohms, I now understand why he did, but in my case I am trying to keep the cost very low, so far I am achieving the require results, just that I now have a few boards that will end up in the waste bin... Such is life LOL

PCBa.jpgPCBb.jpg
 

VK3YNV

Moderator
Nice work. It's good to find such problems early on in the process, rather than have to try and fix them later after boards have gone out.
 

BillC

Active member
A very interesting problem Rob, on some boards we do see masses of ground plain through stitching , I suppose so as to produce isolated fields etc. I have often noticed tracks on boards have rows of grounding via's on either side of a track, I suppose that would control the field within the track itself . Shame you can't modify your beautiful boards. Good luck.
 
Top