Repurposed 1970's Fluke 4216A Programmable Voltage Source

[trobbins]

I've been able to repurpose a freebie 1970's Fluke 4216A Programmable Voltage Source that is quite the boat-anchor given it has no front panel metering, requires a BCD type control via a 50-pin Amphenol connector to generate a single commanded output voltage, and consumes about 20W.  My 4216A is part of the 4200 series of programmable Fluke sources, but precedes GPIB that Fluke appear to have added later.

The heart of the dc voltage source is a Fluke first generation DH80417B zener reference configured as a +10V ref, along with a precision inverting buffer for -10V, and residing on one pcb module along with mains switch and transformer and supply rails (top card slot in photo below).

Conveniently, the +10 and -10V can be picked off and routed to easily added rear panel banana terminals.  I don't presently have any purpose for the remainder of circuitry, including front panel indicators, and ladder attenuator and 100kHz bandwidth output amplifier pcb cards, so that has all been left unpowered to minimise internal temperature rise and noise, but still allow original function to be restored if ever needed.

The unit needed replacement of 2x LM301A (U4, U5) in the main bipolar power rail regulators, likely due to an obviously failed main reservoir e-cap, but otherwise seemed fine as is. 

The U2 DH80417B related circuitry provides separate VREF and sense lines, for each polarity, that could then be selected (Q2,Q3,Q5,Q6) and passed through precision programmable ladder attenuation and a power amplifier to generate a required single output voltage source.  I'm simply connecting to the dual outputs, even though the outputs are sensitive to loading (circa 1.5ppm per 10Meg meter loading).  Supply rails are available to add unity opamp buffering if needed.  Logging with a HP34401A, the +10V showed about a +5ppm/C tempco which was well within 4216A spec tolerance. I added a 50R NTC across the 10V R9 20R trimpot and dialed in a tempco of about +1ppm/C (which may be influenced by the HP34401A and interconnections), and added another separate trim across R16 for the 10V level.

So far this source appears acceptably stable and repeatable, given my 34401A.  I'll try and compare it to my recently aquired Fluke 335D, as that uses an ovenised DH80417B arrangement (https://www.eevblog.com/forum/testgear/fluke-332baf-in-the-slaughterhouse/).