5.5D vintage Systron Donner 7205 DMM with 0.5% FS error

[trobbins]

I am trying to clarify why i get a 0.5% low FS displayed measurement in a vintage Systron Donner 7205 DMM.  This DMM has a front-end analog section which appears to be fine, as it sends an accurate 1V FS signal for voltage, current and resistance processed levels onto a subsequent analog section that prepares signals for a final dual-slope integration based measurement/display digital section that uses a Fairchild 3814 IC.

The relevant schematics of the front end sections are on pages 83-4 (of 89) of the manual at https://www.dalmura.com.au/static/Systron-Donner%207205%205.5%20Digit%20Digital%20Multimeter.pdf, and the digital section schematic is on p.80.

Before starting a FS cal check, I was able to confirm ok levels for shorted inputs on all ranges, and confirmed supply rails are ok, and exercised the range selection switches (with judiciously applied contact cleaner spray where appropriate), to remove the typical switch contact related gremlins, as well as removing the BCD option board (which was a contributor to noisy readings). 

During FS cal checking, the front end switching and dc zeroing circuitry on p.83 generates a 1V FS level at the 'Amp Out' TP1 node, and I have confirmed that for voltage, current and resistance references of various decades applied to the DMM inputs, I do get an accurate and processed 1.0000 Vdc level being passed on to the next section (p.84 of that doc). 

But that is where my understanding of the analog and subsequent digital processing circuitry functions stop as to identifying why and where there may be a 0.5% low FS error subsequently coming in.  I have confirmed the +10Vref  and -10Vref levels generated on the p.84 schematic are acceptable (error and balance within 2mV). 

I'd appreciate any guidance on what to assess on the p.84 analog circuitry, as I haven't yet got the scope out to confirm digital control lines are appropriately applying the +10 and -10V refs and the zeroing, or the operation of the integrator and zero detector analog circuitry, prior to being passed to the digital board.

Ciao, Tim

[Kleinstein]

A bit strange for a dual slope ADC the integrator uses different resistors for the input and reference currents. So the gain may drift with R88 and R89, though this should be pretty accurate ad good resistors.
The integrator is also unusual in that the signal is applied to the positive input and the integration resistors towards ground or the references.

An easy point to check could be if there is leakage from VR6,VR7 , that may attenuate the signal before it reaches Q13.

Is the 0.5% FS error the same for the positive and negative side ?  Another point to check could be the gain at some 50% of FS, as the error at full scale could also be a linearity issue.

[trobbins]

I appreciate the effort to respond - ta.

The 0.5% error is the same for pos and neg polarity inputs of the same reference.  The error is linear with amplitude, so circa 0.25% error at 50% of FS.

I can see your concern about VR6,VR7 - I will have a look at the pcb and parts and see if I can easily lift a leg on one of those diodes (transistors).

I wasn't confident to try and change the +10Vref and -10Vref levels to see if that was a quick fix, or made sense as a fix for starters 

[trobbins]

I removed VR7 and confirmed no change to 0.5% FS error, so returned it.  I then checked if it was feasible to adjust the +10Vref and -10Vref levels as a way of calibrating the display value to 10V FS, and that seemed to require the internal ref levels to be lowered to circa +/-9.987 (ie. abt. -0.13% of anticipated).  The equipment instruction manual only describes calibration of the various range trims in the first analog section, effectively assuming the +/- 10Vref levels in the following analog integrator are correct.

Adjusting the internal +/-10Vref levels seems to be the simplest path forward, so I will warm everything up and calibrate the meter across the ranges to confirm everything displays as anticipated.

Ciao, Tim