GPS disciplined oscillator

I needed a frequency reference for my projects (gpsdo style) at 10 MHz

Fast forward:

There are two designs (and i have build them both) that work out of the box and are stable enough for my measurement

This one is FLL based https://www.instructables.com/id/GPSDO-YT-10-Mhz-Lcd-2×16-With-LED/

This one is PLL based

https://www.eevblog.com/forum/projects/lars-diy-gpsdo-with-arduino-and-1ns-resolution-tic/

Pros and cons

The FLL has a correction every 1000s /around 16 minutes. Assuming no temperature changes and small drift of the OCXO is more than enough

The PLL based is more sophisticated can also has a quick adjustment to any frequency change

Don’t spend time at all trying to adjust the NEO to produce 10 MHz as output

LARS GPS DO Notes

Remote the autoreset of the Arduino: 10uF cap between Rst and Gnd is the recommended way to temporarily disable autoreset.

I had an OCXO with sine wave output and DC Offset. Had to remote it with a 100nF capacitor and use one of the gates of the 74AC04/74AC14 to amplify it.

74AC04 => 10 MHz to pin 13, output of pin 12 (filtered) 10 MHz to Pin 1/Pin 3
Output Pin2/Pin4 with 100 ohm in parallel and a small filter ->Sinewave
Another one in out TTL out

Bypass cap 4.7uF/ 10uF close to the OCXO

LM35 filtering: Low pass filter, 68 Ohm in series with the data line and a 10uF capacitor to ground.

Running for the first time:

After 5 minutes with the OCXO warmed up, Press: “h1” and check the field diff_ns. (the same for h63535)

h1 => diff_ns => 1500
h65535 => diff_ns -1400
sum = 2900
gain = 65535/2900 => g22
g22
s1
r

Frequency accuracy

Just to get an idea of what ppm means vs seconds per month..

10MHz (with 100 ppm)
=> 10MHz +- 1000 Hz
=> 100 ppm => 100/1000000 total error on a day is 86400 * 1e-4 => 8.64 seconds per day => 259.2 seconds per month

10MHz (with 10 ppm)
=> 10MHz +- 100Hz
=> 10 ppm => 10/1000000 total error on a day is 86400 * 1e-5 => 0.864 second per day = > 25.92 seconds per month

10MHz (with 1 ppm)
=> 10MHz +- 10Hz
=> 1 ppm => 1/10000000 total error on a day is 86400 * 1e-6 => 0.0864 second per day = > 2.592 seconds per month

10MHz (with 0.1 ppm)
=> 10MHz +- 1Hz
=> 0.1 ppm => 0.1/1000000 total error on a day is 86400 * 1e-7 => 0.00864 second per day = > 0.2592 seconds per month

10MHz (with 0.01 ppm)
=> 10MHz +- 0.1Hz
=> 0.01 ppm => 0.01/1000000 total error on a day is 86400 * 1e-7 => 0.000864 second per day = > 0.02592 seconds per month

In addition having a 10MHz frequency reference (which can be divided also to 1MHz, 100KHz, 10KHz, 1KHz and 100Hz) is a nice to have tool on the bench.

I decided to split it into two phases:
First get a 10 MHz signal from the GPS to try it using the PPS output,
Second switch the PPS to 1Hz and make it a real GPS DO with the my Isotemp OCXO-131

For my GPS locked Frequency Reference Standard 🙂 i bought a u-blox NEO-7M Module (uBlox Neo-7M GPS)

https://www.aliexpress.com/item/Ublox-NEO-7M-Flight-Controller-GPS-Module-Built-in-Data-Memory-Replace-NEO-6M/32279486485.html for 13 USD.

u-blox makes great GPS receivers and getting a module should speed up the breadboard design.

This modules comes without a PPS pin. This can be easily solved by using a 5-pin header (use a wire cutter to trim the 5 pin directly to the plastic) and a piece of wire connected to the on board led

NEO-7M

NEO-7M

ucenter

I run the board on 3.3V and i already had a 3.3V RS232 converter handy.

u-blox provides a tool called ucenter (used version 8.23)

GPS is already configured at 9600 bps

Start your ucenter
Receiver -> Port -> COM3 (replace with your com port)
Change your configuration to the below (for 10 MHz on PPS on locked and 5 Hz when unlocked)

ucenter_pps

And press send

Jitter

First issue encounter was the jitter on the 10MHz. NEO-7M uses a 48 MHz crystal internally and it cannot be divided exactly for 10 MHz…

Scope outputs
NEO-7M 8 MHz pps

NEO-7M 12 MHz pps

And our favorite for 10 MHz…
NEO-7M 10 MHz pps

I still don’t know if i should use a 10 MHz crystal as a filter and then a schmitt trigger to give a clean square wave output.

Use a ground plane for the ceramic antenna, it’s a huge difference on, a simple one side PCB will do the trick, place the ceramic antenna ontop

Loosing settings

The backup current needed for NEO-7M is 10μA. The capacitor selected on this board is a a XH414HG which has 30μAH capacity till it reaches the 2V… Seriously… max 2.5 hours disconnected from the power and you loose the settings created above… Solved by using a CR2032 and a diode

ISOTEMP OCXO-131 has arrived

Using just the GPS for 10MHz output is a NO GO

AVR ATMega328 using External clock

According to the data sheet:
To drive the device from an external clock source, EXTCLK should be driven as shown in the Figure
below. To run the device on an external clock, the CKSEL Fuses must be programmed to ‘0000’:

And Divide clock by 8 disabled

avrdude options -patmega328p -U lfuse:w:0xef:m -U hfuse:w:0xdf:m -U efuse:w:0xf9:m

To be continued

2 Comments on “GPS disciplined oscillator

  1. Hi Billy,
    very interesting project (didn’t know that you could convince that GPS module to output a GPS synchronised 10 MHz signal instead of 1 pps)!
    Really can’t wait to read how this project is going on…

    73, Joe

Leave a Reply

Your email address will not be published. Required fields are marked *

*