"War story..." I have had two Rb cavities with short circuits. The first one was repaired using Pelican wire. That one developed an open circuit in the magfield wiring. Got a second cavity that developed a short during operation but this time I had over current protection in place which salvaged the electronics inside. The first cavity was scrapped and I managed to rescue the pelican wire and could reuse it for the second cavity.  A friend found two 780nm filters from Midwest Optical and gave me one as a thank you for help in the past. After some metal work, the filter fitted almost perfectly in the inner hole in the Rb TX assembly where a small transparent glass disk was removed and modified the outer plate  (with the many small holes removed) No other modifications were made. The filter is supposed to withstand several hundred degrees celsius. I have photos of this but AFAIK time-nutsdoes not allow attachments. Having experienced multiple deiscrete semiconductor failures as well as power outages I shut down my 5065 years ago but after the 780nm filter implementation, the unit locked after about one hour. I also added a high temperature braker at the Rb TX base plate. Previously, I have modified the power supply circuitry by removing the original transformer which gave a recified DC of more than 36 volts which almost fried the series pass transistor and instead used a modern 24V SMPS with ample output filtering. The A7 input amplifier had been replaced with a recreated HP version using better 2nd harmonic detection as well as modern OPAmps.  I also made a redesign of the +/- 20V regulator board/magfield circuitry. Output noise and stability of that design is better than HP's original design. The old board was in a bad shape after leaky capacitors and had to be replaced anyway. The original 5 MHz oscillator have been replaced with a moden 10811 and with that a  10 MHz sniffer tap as well as a regenerative 10/5 MHz divider board that is a  direct drop-in replacement for the HP TTL-divider board. (Ideas from Luciano Paramitthiotti IZ5JHJ and John Miles KE5FX). Uncertain if that design reduced the phase/jitter  noise though but the principle works. The only thing I have not done is to replace the old DTL synthesizer with either a TTL-version if I can find/build one or recreate that design with a microprocessor controlledsynthesizer. At least my 5065 is more of an eternal "project" than something useful as there are several GNSS diciplined products that can be built or purchased via the internet. I could use a few meters of the Pelican wire as spare but not at $500 a reel. Have not had any sucess in finding any other well insulated product that could be used as replacement. Ulf Kylenfall

-------- Ulf Kylenfall via time-nuts writes: > The only thing I have not done is to replace the old DTL synthesizer with either > a TTL-version if I can find/build one or recreate that design with a microprocessor controlledsynthesizer. Any tiny microcontroller with a PWM output and 5V tolerance will do the job. I used a 8-pin DIL ARM chip when the unobtainium delay-chip in the divider got flakey. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.