Replacement for HP 8555A Spectrum Analyzer 100V-26.1V Buck Converter

[Jamie_A]

I'm currently in the middle of a repair for my HP 8555A 18GHz spectrum analyzer.




Luckily, I got it with it's operating and service manual, and quickly used it to isolate the problem. The 8555A has a bizarre power distribution design. It takes in 100Vdc, uses a 100V-26.1V buck converter to drop down the voltage, and then linearly regulates that to 20Vdc and 10Vdc. The supplies were all reading about half of their expected voltage values. By removing the linear stage, and substituting it for a constant current dummy load, I saw the 100V-26.1V buck converter drop to 17V output under it's rated load. After putting the linear stages back in and substituting the buck converter for a lab supply, the unit came back to life!

I had no idea switch mode supplies were in used back in the 70s, and it seems they got this module wrong on the first try. My 8555A is an earlier model, and a few years after it's release they swapped out this switch mode supply for a completely different design due to failures (they even recommending replacing it in the manual). Mine has a number of blown ICs and diodes. The caps check out fine with my ESR meter though.



With that said, I'm not going to bother replacing these parts if the design might fail in the future. I'm fine buying a DC/DC module, or designing my own, but I'm having trouble finding options in this weird voltage range. TI has some nice IC like the LM5161, but that has an absolute maximum input of 100V. There's also the AD LTC7138, which maxes out at 140V.

Does anyone have recommendations on how to approach this? I've done DC/DC designs in the past, but never something this high voltage. Any off the shelf options I'm missing?