David Wrigley's Microwave Website, Manchester, England

Brooks Shera High stability Frequency Standard

The Brooks Shera GPS system is basically a digital PLL with long time averaging. See his web site for more info. (Brooks Shera website). His system takes a one second pulse from the GPS receiver and uses this as the reference input to the system. The phase of the local crystal oscillator is compared with the average of 30 seconds worth of GPS input and this difference is used to control the local crystal oscillator. The assembly of the PCB is no problem - I used sockets for all of the IC's. What may require some effort is to locate the DAC devices - Brooks Shera has a few in stock, he can also supply the programmed PIC's. (I decided that I would like to be able to program PIC16C73's and I built my own programmer to do it - this took a little more time.) The system can be used with any one second pulse, whether GPS, MSF or BBC1 derived. The Shera control PCB has been contructed and mounted within a HS400 unit to control the built-in 5MHz reference oscillator (which had to be modified for the purpose). This is now operational and has been fitted with a Motorola UT+ GPS Receiver although the other sources are also being constructed and it will be interesting to see how much difference there is between them.

Picture overview of the HS400 modifications

The front view of the HS400 which has been modified with the addition of display of the temperature of the reference oscillator and also a display showing the of the amount of correction in parts per 10^11. The three LED's from the controller PCB are also on the front panel.

The top view shows clearly that the central sceened box has been removed and the controller PCB just fits in nicely. Above this, mounted on a bracket from the rear panel is the GPS receiver. This allows the antenna connector to be accessed directly via a hole in the rear panel.

On the rear panel can be seen, from left to right, The original reference oscillator adjustment access- covered by a large rubber plug, the rectangular hole left when the unused connector was removed, the new BNC connector giving 5MHz buffered output, Two 9pin D connectors for Controller logging and GPS receiver communication, The hole for the GPS antenna connector. and then the rest of the original connectors for 24V DC input/fuse and 230V 50Hz input.

HS400 modifications - constructional details

5MHz buffered output. This information was originally written for "Scatterpoint" the magazine of the UK microwave group. Now available as a Zip file. HS400-buffer.zip

DOWNLOAD HS400-buffer.zip

Reference Oscillator -addition of EFC. This was published in "Scatterpoint" No 5 - June 2001. This is now available as a zip file. HS400-EFC.zip - New info 7 Oct 2001

DOWNLOAD HS400-EFC.zip

Negative 9V PSU - tiny charge pump working from a stabilized +12V provides more than enough for the DAC

The assembled PCB shows the conventional components mounted SMD style (less drilling). The reverse side is unetched copper bonded thru at the ground connections, with clearance holes for the input and output pins. Tests show that the unit will provide 9V at the required 13mA with 130mVp/p ripple, but it will go beyond this to well over 30mA at 8V and 250mV ripple. After the on-board 7905 the ripple is less than 2mVp/p (about the same as the +5V supply - only higher frequency). The PCB has since been conformally coated and fixed by means of double sided adhesive tape to the rear of the front panel. A zip file with assy, artwork and BOM is available for download.

DOWNLOAD neg9vpsu.zip

Mechanical aspects of fitting the GPS controller system. Will be available as a zip file. HS400-GPS.zip

Bill of Materials/sources. Will be available as a zip file.HS400-BOM.zip

The controller correction voltage which is applied to the varicap in the reference oscillator, was plotted and found to be a good guide as to how the unit was behaving. It also pointed the way to improving it still further. The time constant is set by switches giving N=1 to 7.

Note the initial effects of bringing the oscillator up to the right frequency and afterwards the drift down due to the warm-up of internal temperature controller. Within an hour it’s in the right ball park (towards 1pp10^10). Using N=3 is helpful in having a relatively short time constant thus correcting quickly, yet being long enough to smooth out much of the granularity of the GPS measurements.

N=4 and 5 Plot

Most of the first hour was with N=4, then it was switched to N=5. This plot shows the effect of a larger time constant in smoothing the variations in GPS measurements. Notice also the drift caused by the system compensating for the temperature coefficient of the quartz crystal oscillator. It should be mentioned that the room is thermostatically controlled and the temperature changed by just over one degree C. The temperature coefficient is about 5pp10^10/degreeC of ambient. Whilst this is corrected within the system, it would be better to reduce this effect to increase the fundamental stability of the system. With long time constants the correction takes time during which there is uncertainty of the precise value. After the required compensation has been applied the graph levels out again and then short term variations are of the order of 2 pp10^11.

Overall Progress

BBC1 - using receiver modules from an old VCR, a Sync/sep and a PIC divider. This unit is now working to within about 5 parts in 10^10, but isn't as stable as the GPS system.(2000Dec20)

GPS - based on a Motorola UT+ receiver - a system which is now working to within a few parts in 10^11.(2001Jan19). However after much logging it has since been realized that the temperature sensitivity of the crystal causes a drift rate which is hard to follow at N=5. After advice from Brooks Shera, it has now been reset to N=3 which provides for a more rapid correction of temperature drift but with an increased granularity of around 1pp10^10. This will provide an accuracy and stability of around 1pp10^10 or 1Hz at 10GHz - plenty good enough for the beacon. Work to reduce the temperature effect is now reaching completion. Initial trials using a Peltier effect heat pump to control the external case of the reference oscillator, have been completely successful, and have resulted in a frequency stability and accuracy between +/-4 to +/- 1 part in 10^11 - depending on the time constant setting. Details will follow.New info 12 July 2001 A graph of the error control voltage using settings between N=4 and N=7 is shown below:-

Last updated - 2005 December 19