Hi,
Have been a lurker here for a while, but thought a fan mod for the Fluke 5200 and other notes might be of interest. Maybe tldr. It's an old design now, with schematics dated 1975, but accurate and stable when working. Problem is keeping them working, as there are issues with the power supply and also, the power amp seems a bit fragile, stretching the state of the 1970's art. A complex, but brilliant design for it's time. There are some 2 watt composition resistors on the power amp, which can go open circuit with heat and age. Easily replaced with metal oxide, but when they or transistors fail, can take out the power supply as well. Power amp now rebuilt, twice, helped by the sockets fitted for some of the transistors. Perhaps they expected it to fail.
Some of power supply rails don't have current limiting, but instead, use low wattage sacrificial resistors that burn out on power amp fault, which can damage the pcb. Insufficient cooling for the psu area leads to a finger burning temperature on the 5 volt regulator. With age, thermal cycling, discolouring of the pcb causes failure of the copper track to substrate adhesive. Heat damaged pcb makes it difficult to replace parts without the tracks or through hole plating coming adrift. For the low value resistors, wirewrap pins fitted to standoff from the pcb, helps fix that, but some parts still running far too hot for reliability.
For the fan, not a great deal of space, but the right psu card support plate looks ok. Used a tank cutter to cut out a hole, to allow air flow from the underside of the pcb, and the fan is spaced off the plate for more air input. Fan is an ex computer server ball bearing item. Power comes from the 5 volt regulator input, around 9.9 volts, so the 12 volt fan is underrun. Result, seems to get the job done, with the 5v regulator case at about 56 C now and the heatsink fins warm to the touch.
There's an interesting 5 pole active filter on the reference supply. What looks like a double section version of the one used in the Fluke 335 voltage standard. The key thing is that the filter op amps are not in the reference signal path, so the reference isn't affected by op amp drift or offset. Can't find any reference on the web for this, but looks like a gyrator design of some sort. Just a handful of parts. May be worth digging into to adapt the idea for other reference applications.
The 5200 is mostly working, frequency and voltages, with just a few frequency setting numbers in error. Seems more or less good right through the frequency range. At one volt, one KHz, it's just 10's of microvolts in error. Not bad for a fourty+ year old bit of kit. A real saga, too many hours, but just need to fix the remaining faults, then have a go at setting it up...
