Here I describe my implementation and show photos of the equipment.
Background
I've built two systems; my first was based on the inexpensive Toyocom TCO-613 5 MHz
ovenised oscillator that appeared on the UK surplus market for typically
£20 ($35) in the Pye/Philips HS400 unit. I'd wholeheartedly recommend that
approach; for peanuts you got a strong enclosed chassis, mains PSU, an
excellent crystal oscillator and plenty of empty space to incorporate the GPS
receiver and Shera controller. If you cocoon the oscillator in 20-25mm
polystyrene sheet, performance is more than adequate for most needs.
Nevertheless, I wondered just what could be achieved with a top class oscillator such as a rubidium resonance or a Hewlett Packard 10811 crystal unit. I didn't really want to wear out a rubidium standard, since it would be on continuously, so opted for the crystal oscillator.
Oscillator
Research showed that a double-oven version of the HP 10811 was inside the HP
Z3801A GPS Time/Frequency standard. This unit was (around 2001) readily available
from the advertisements in QST and on eBay for typically $250.
Z3801A details.
A unit was acquired, the HP oscillator and associated PSU board removed and the remainder discarded. The PSU board takes in 54V DC (it's from telephone equipment) and generates +/-15V for the inner oven and oscillator with plenty to spare for the Shera controller and there's lots of 5V for the GPS engine. Also on-board is the outer oven control system. Perfect!
GPS Engine
I've been using the Rockwell 12-parallel channel Jupiter GPS engines for some
years now, and opted to use the -T version which is optimised for static users
in time/frequency applications. Rockwell Semiconductor Systems became Conexant
Systems Inc. in 1999, and the GPS marketing interests are presently (2002)
owned by Navman. The Jupiter-T
is pin and software compatible with the Motorola UT+ engine.
Jupiter engines also have a precise 10 kHz output synchronised to GPS time which can form the basis of a really simple 10 MHz standard.
Controller and PIC
The controller board was obtained from A&A
Engineering (PCB #217), whilst PIC and 18-bit DAC were supplied by Brooks Shera
(d. 2013 Mar 16, Santa Fe, USA) whose support of this project was outstanding.
Manipulating the DIP switches inside the enclosure was found to be a tiresome business on my first controller, so for the present system the filter "N" settings were brought to the front panel via a push-button 'thumbwheel' switch.
Divider
On my first system only the basic frequency was output; 10 MHz to the front
panel. Almost immediately I realised that sub-frequencies would considerably
enhance the system's utility so for the present system a PCB containing several
74HC390 decade dividers was built. 10 MHz, 5 MHz and 1 MHz have dedicated BNC
sockets on the front panel, but 1 Hz to 100 kHz are selected singly via a 6-pole
thumbwheel switch.
[2020 Nov 13: I have qty 4 spare PCBs 0081-001; please email quoting this part number].Another option would be this PIC-based divider from Tom van Baak.
Enclosure
I wanted a compact system, as space is at a premium here, and selected a
simple, neat aluminium cabinet that measures 260W x 90H x 250D (10.25" x 3.5" x
10"). It's OKW Enclosures Ltd
Unicase 2, part M5502110, available in the UK from RS Components 222-115 [this is not
a RadioShack number] and in the USA from Allied Electronics P/N 272-0119.
There's just sufficient headroom for the oscillator and its mounting standoffs;
about 0.5 mm clearance. All PSU components and the oscillator are mounted on
the base; GPS engine, controller and divider PCBs are mounted on an internal
shelf. See photos.
Performance
The double-ovened HP 10811 is very stable with respect to environmental
temperature, and a filter setting of 5 is normal. The DAC control voltage shows
little change from day to day; a millivolt or two. Without absolute test
equipment such as Caesium frequency standard it's difficult, or at least, very
time consuming, to characterise the fine structure of phase and frequency of a
standard such as this. Work continues, [Update 2008 - I now have the test equipment ...]
but suffice to say that with frequency accuracy and stability useful to 10 GHz or
more, this system is outstanding. Thanks again, Brooks.
Cost
Beware! Constructing this HP 10811 based unit to such a high standard came close
to $1000; my first system, more sensibly engineered, cost $250 - about the same
as a Z3801A. Yes, I know what you're thinking; please don't ask ...
Documentation
I like to document my projects properly; saves much trouble later. Manual
in PDF format, 63kb.
Photographs
Here are some pictures of the system taken before the front panel lettering was
applied. Click on the images for high resolution 1600x1200 versions, typically
300 kb.
Last updated: 2020 Nov 13