On power up a quick check with the 4395A spectrum analyzer showed that the 1 MHz combs were not present, the 10 and 100 MHz combs were OK. Removing the 1 MHz crystal and checking it on the Crystal Activity Meter showed the crystal as dead. Markings on the crystal are:
71034, BH6A (=HC-6/U holder), 1.0 MHz, 0410-0013C, 7528 (date code) ($150 to buy new)
The Bliley BH6 series are fundamental AT cut with Rseries of about 400 Ohms looks like series mode, hard to find.
13 Apr 2003 - It turns out that the Crystal Activity Meter does not always work. The replacement crystal I got as Jameco part number 14349 also did not work. This prompted me to use the 4395A to measure the crystal equivalent circuit on both the Jameco replacement and on the original Bliley crystal. Both crystals looked OK.
The schematic for the 1 MHz and 10 MHz crystal oscillators are identical in terms of all the DC components, they only differ in the reactive parts, both use a 2N708 transistor. The 1 MHz circuit has about 50 mA of collector current, way too high whereas the 10 MHz circuit has about 5 mA of Ic which is right in the center of it's speced range. The VBE and VBC on the 1 MHz oscillator differ from those on the working 10 MHz oscillator. Replacement of the 2N708 is next.
After replacing Q1, a 2N708, the 1 MHz combs were there but with the trimmer cap fully open the frequency was too low. By installing the Jameco crystal the frequency could be brought right on. Note that the spec on frequency is +/- 0.01% or 50 kHz at 500 MHz and the actual performance is maybe 10 times better, but these are just raw crystal oscillators and are sensitive to temperature and to warm up as the three (1, 10 and 100 MHz) oscillators are switched on. Warm up may take up to an hour each time you switch in a new oscillator. The amplitude of the combs is typically 10 times higher than the spec, so for the 1 MHz combs up to 500 MHz they are around -70 dBm vs. a spec of -80 dBm. The Jameco crystal has a few dB more output than the factory Bliley crystal.
Analog type spectrum analyzers, like the HP 8551, did not have accurate frequency displays and this box could be used to determine the actual frequency displayed. Modern spectrum analyzers, like the HP/Agilent 4395A, are accurate enough to calibrate the 8406A.
HP 4395A Impedance, Network & Spectrum Analyzer
Crystal Activity Meter
Crystal Equivalent Circuits
Page created 1 April 2003.