The main voltage reference used in the PS5004 is a National Semiconductor LM399. These references are thermally stabilized, and after about 6 months of aging through normal use, have a stability which rivals standard cell primary voltage references. However, the IC process did not support laser trimming, so the only downside to these otherwise great parts is they have about a 5 % initial voltage tolerance – a challenged to adjust to the precision required for an instrument with target accuracy of 0.01%. This is way beyond the acceptable adjust resolution of a 20 turn trimmer pot. I ended up with a bank of binary weighted resistors in the leg of a voltage divider to pre-select the voltage to be in range that can be adjusted by the 20 turn trimmers. Initially these were to be controlled with a DIP switch, but this would require the user to correctly enter the 4 bit binary code for each “try and miss” step during calibration. I found the cam operated DIP switch we ended up using, which simplifies this step by doing the proper binary coding for each increment/decrement.
As the manual reads, you would only need to perform this step when replacing the reference IC, as any aging drift should be adjustable with the range of the 20 turn trimmer. To set this coarse pre-selector switch, you need to adjust the span trimmers to the center of their range, which is a lot of turns for a 20 turn pot. The trimmers don’t have positive stops at the end, they just keep turning and a small “click” is felt in your screwdriver when you get to the end. So centering a 20 turn trimmer first requires turning to the end more than 20 turns (to assure you are at the end, then backing up 10 turns. This is why you see these span trimmers being adjusted so many times in the Cal procedure.
If this is quoted directly from the manual (I don’t have access to my copy right now), it looks like there is a missing word that would help.
“Connect the PS 5004 OUTPUT to the digital voltmeter input using a BNC cable and a pair of BNC-to-banana plug adapters. The shield side of the BNC is connected to the PS 5004 OUTPUT terminal, and the voltmeter Low input.”
Connect the PS 5004 OUTPUT to the digital voltmeter input using a BNC cable and a pair of BNC-to-banana plug adapters. The shield side of the BNC is connected to the PS 5004 - OUTPUT terminal, and the voltmeter Low input.” (add a minus sign between “PS4004” and “OUTPUT” - these should be Bold type face to indicated these are the connection point names)
A challenge in manual writing is that everyone, the engineers, technicians, and even manual writers themselves, are “too close” to the equipment we are writing about, and procedures are the same as we have all used over and over. So it is easy to assume the reader already knows something that we all know – but only because we are “insiders”. What this step is trying to say is to use a shielded BNC cable to measure these precision DC voltages, and make sure the shield is connected to the PS5004 negative output terminal, and the DMM low / guard terminal. It is important to drive the DMM Guard input and not leave it floating.
Third question: These are 20 turn trimmers, so needing to two resolutions is possible, especially if the instrument is being calibrated the first time after many years of storage. I think the need for the coarse adjustment is described above – it is necessary to get the range covered by the 20 turn span trimmers close enough to account for a 5% initial tolerance in the initial voltage reverence.
On this topic – if you are calibrating a PS5004 or any high precision instrument (Fluke DMM etc.) that has been not used for a decade or more, you should probably first power it up and let it run for a week or more. Large drifts from aging may be moisture absorption in components and circuit boards.
Final question – there is interaction between the span and zero adjustments. This results because Zero volts output is not actually zero. Studying the schematic, see the output has an active load on it that can sink small currents below 0 V. So the supply can be adjusted to negative output. In fact, the original design would allow precision adjustment of the output to minus 100 mV, but the supply can only sink a few mA in the negative direction, as it is not q four quadrant supply (only single pass transistor). The application would be measuring input offset in DC amplifiers, etc. I thought it would be quite useful, We did some customer trials with an engineering prototype, and the users found the feature to be confusing because of the current limitation when the output went negative. So we pulled it out. This feature is implemented only in the firmware – no hardware changes were made. Thus the supply actually has some output voltage when adjusted to Zero, which is removed by other offsets in the circuit. Thus the gain action of the coarse and fine span trimmers will move the zero output setting, which requires an iterative adjustment procedure.