Thanks to a tip from EEVBlog forum member retiredcaps, I recently picked up a Fluke 8020B DMM for $5 + postage from the usual auction site. This model dates from somewhere in the early to mid-80s and is an updated version of the 8020A, Fluke's first handheld DMM. The seller's description said this one "would not power up". However, when I installed a battery, it worked perfectly. Overall, the unit is in very good cosmetic shape, especially after a thorough cleaning. Unfortunately the button for the AC/DC selector switch is missing.
So for those who may be interested in 30 year-old DMM technology, let's take it apart!
The case is held together by 3 self-tapping screws. I deem myself qualified to remove the screws.
Accessible (barely) from the battery compartment is a 2A glass fuse, with a spare thoughtfully provided in a plastic clip. The battery cover fits tightly and is a major pain to remove and re-install.
Flexible shielding material wraps around most of the circuit board assembly, bonded by a small screw to a metal stand-off. The shield is permanently attached to the case bottom half with melted-over plastic studs. The 3A HRC fuse is not accessible until the meter is disassembled to this point. Also accessible is a DC cal adjustment, AC cal adjustment, and a "high-frequency" compensation adjustment. The meter is spec'ed to 5KHz bandwidth with reduced accuracy from 1kHz to 5kHz.
The circuit board layout is dominated by the push-button switches. The daughter board on the right contains the continuity beeper circuitry and piezo noisemaker. The fuse block is quite firmly attached at that angle.
Double-sided board. All through-hole components. Wrinkly solder mask, which is common for PCBs of this era. The PCB artwork is dated 1981.
The LM358 op-amp on the main board is the AC converter and is factory-selected for 600kHz GBW. The continuity piezo noisemaker is on the little daughter board, triggered by an LF442 op-amp comparator and a 74C00 quad nand gate as oscillator and driver. The JFET-input LF442 is selected for 100pA input bias current. The continuity buzzer is supposed to be used with the 2K-ohm range which results in a 115-ohm threshold. But since the buzzer is driven by a comparator from the voltage divider, the buzzer also operates in other resistance ranges at non-useful thresholds.
The 3A fiber tube HRC fuse is in series with the 2A glass fuse, providing 10kA interrupting capability in case of a serious fault. Series varistors, a thermistor(obscured) and a 1K 2W fusible resistor provide over-voltage protection. The two beefy 2A diodes limit voltage across the current measurement shunt.
All through-hole components, no SMD in sight. The large gray-black tapped resistor is the current shunt for measuring up to 2A. The blue resistor network behind it is the primary voltage divider. The input thermistor is fully visible in this shot beside the 1W 100K resistor.
External power input jack is under the display. Not usually found on modern DMMs due to the safety hazard. The power rails are not isolated from the meter's input jacks, and thus could raise the 9V input to hazardous voltages.
The 4030 quad exclusive-OR gate is used to drive the LCD decimal points and lo-bat annunciator, gated with the LCD backplane clock. The rectangular orange components are resistor arrays.
Those are some sturdy-looking switches. They consume a lot of space, but there are no extra components hidden underneath.
This meter is 3-1/2 digit, 2000-count. The display is a fairly generic "1888" with decimal points, leading minus sign and a "BT" low-battery annunciator. There are no bar graphs or units indication.
The black elastomeric connector is tall and thin and held in place with a removeable plastic piece. The protective transparent lens is 2.4mm thick Perspex.
The A/D converter/LCD driver is a 40-pin DIP Intersil chip marked with Fluke p/n 429100. Probably very similar to the ubiquitous ICL7106. Looks like a 1985 date code. The device in the TO-18 metal case below the trim-pot in the upper left is a 1.22V bandgap voltage reference. The clock crystal is 3.2MHz.
Measuring an 806-ohm 0.25% precision resistor mounted on a dual banana plug. The tilt stand is a bit short, but the display is already angled, so the net result is a good viewing angle. Note the external power jack at top right.
Measuring a 5.000V reference voltage. I never touched the trim-pots. It arrived this way. Not knowing this unit's history, I have no idea if it was ever in a calibration program. But in any case it's still perfectly useable.
Thanks for looking.
Edit 1-July-2017: Photobucket sucks.