HP 3466A 4.5 digit DMM repair. An Oldie but goodie. 
YouTube Video (but not yet)
Battery Spreadsheet file
Youtube Video

Why fix a 43 year old DMM?  

HP3466A is a nice instrument, worth putting some effort into. I am retired and have some extra time, and I see that there are very few Youtubes about repairing these. I suspect that there are many of these in the wild, so maybe other owners would benefit from my experience. Besides, I am obsessed; I get a nice dopamine response by fixing stuff.

The HP 3466A is a 4.5 digit 20,000 count bench DMM. It has true RMS AC and AC + DC voltage and current, and wide range 200uA to 2A current range. I wish my HP3478A or 34401A had such low current ranges. Features are:
 hp3466a

This is a fine example of late 70's instrument technology. Dual-slope ADC, DIP IC technology, lots of low-leakage teflon insulators to reduce leakage currents. The core components are three proprietary custom devices: a 40 pin DIP Digital "Controller" IC which is an N-MOS ASIC, an "Input Hybrid" containing the range and function switching JFETs and precision resistors, and an "Integrator Hybrid" containing the FET switches and precision resistors for the dual-slope ADC and auto-zero circuitry. It is very serviceable and I enjoyed fixing this one. Fortunately the three core devices worked. They are no longer available (made of un-obtainium) so your only hope would be to try to scavenge these parts from a donor unit, hoping it doesn't have the same bad parts. Reworking the Integrator hybrid will test your de-soldering and board-cleaning skills to the limit.

Most of the other active parts are either TO-99 (metal can) Op-amps, regulators and transistors, and DIP jellybean parts, mostly 4000 CMOS. They mostly have HP part numbers. It is usually (but not always) easy to find the merchant part numbers. If the merchant PN is not in the BOM, Google is your friend.

The Operator and Service Manual is excellent. It thoroughly describes the operation, circuit functions, and has full schematic, parts placement and detailed BOM documents. The troubleshooting aids are also excellent.

I recently was at a friends house, troubleshooting his garage door opener, and asked if he had a meter. He blew the dust off this one which had not been used in a decade or so. Needless to say the battery was very dead and the unit did not function from AC either. I offered to bring it home and try to repair it. After looking up the specs and service info, I decided it would be worthwhile to spend some time working on it.

First Problem: DIsplay stuck at "OL": Overload

My first reaction was "Hey this thing is heavy." I took the top cover off and discovered why. The cover held a big-old Sealed Lead Acid (SLA) battery and its charging PCB. I plugged it in and saw "OL" display on all ranges and modes. I checked the unregulated power supply voltages, hoping for an easy bad-capacitor fix. No such luck. I took it further apart, removing the shield, saw the controller and 2 hybrids and hoped that the problem wasn't there. I measured the regulated supplies and they were fine: +/- 7V, and -2.5V.  It is an interesting design where most analog parts use +/- 7V, and the digital parts all use +7V. I checked a bunch of the switch contacts, since corrosion on these is a common problem. I randomly probed around, and found that the dual-slope integrator output was pinned negative, which explained the "OL". This doesn't tell you much because a dual slope ADC is a closed loop circuit with many switched inputs. The problem could be anywhere. I found that one JFET analog mux that sets the input to the 2 lowest voltages ranges, had both JFETs ON. One of them should be OFF. These were controlled by a D-type flip-flop on the digital board. Q controlled one JFET and Q-bar controlled the other. It was doing its job, but the level shifter that converts the +7V logic to -7V levels, wasn't. I replaced the defective LM339 quad comparator and that circuit now worked. Still, the "OL" persisted.

I next looked at the ADC comparator, built with two CA3096 transistor arrays.  Each array has 3 NPN and 2 PNP transistors for a total of 10 transistors. They are configured as two, fast, high gain amplifiers, and a simple diff-amp comparator. I  checked a few of the transistors in-circuit with a DMM on the Diode range, and sure enough there were bad transistors in both parts. B-E and B-C junctions should read 0.7V, possibly less in-circuit due to other components in parallel. I measured 0.8 V, 1.4V, all over the map. I un-soldered them and replaced them with IC sockets. This allowed me to plug in 6 2N3904 and 4 2N3906 transistors into the 2 sockets. And it worked! The LED display came to life and measured DC voltages. I was elated and encouraged that the 3 un-obtanium parts all worked fine. CA3096s are hard to find in the USA so I ordered a couple from China.

Here are my two home-brew transistor arrays consisting of 2N3904 and 2N3906 transistors plugged into machine-contact IC sockets. 6 of the transistors plugged right in, the others required a bit of lead-bending. 

array

Next: Digit 4 goes dead

Unfortunately about 10 seconds after power ON, the 4th digit went blank. I suspected the 7-segment LED itself, but that wasn't it. Turns out that digit is controlled differently than the rest, it has its own 7-segment decoder and drivers. The decoder has a display blank input which is driven by a D-Flip-flop. I checked the inputs and outputs of the '4013B flip flop and the inputs were fine, but the output was bad. I replaced it and the digit came back to life.

I checked all the ranges using my DIY-SMU, and they not only worked, but the instrument was pretty much still in calibration. Wow!

Battery Upgrade to Lithium?

The battery pack is a 6V, 2.5 amp-hour, Gates SLA battery made of three 2.1V cells. Date code is 1979! I could just replace it or rebuild the battery pack, but could not find an exact replacement. The cells are about $25 each, or $75 for 3. How about replacing it with two 18650 Lithium cells: smaller, lighter, more available (I have some), and much cheaper? I looked at the complicated charging and dc-dc circuit and decided that increasing the charge voltage from 7.4V to about 8.4V for two Lithium cells could work. I'll use 18650 cells with built-in protect circuits to keep them from being over or under-charged. The instrument draws about 250mA, so should operate 8-10 hours from a 2.5AH Li battery. It charges the battery at 300 to 500 mA, fine for Lithium.

Here is the original battery charge circuit. It's a bit complex and took me a while to understand. During battery charging, it is pretty standard: a constant-voltage circuit with power transistor output and current-limit override. +7V into R806 is the reference voltage. R806, R818(trimpot), R805, and R827 / CR806-810 form a reference voltage divider. R804, CR811, and R807 form a feedback voltage divider.

During charge, Q803 is OFF. Diodes CR806 to CR810 cause the input voltage to increase to the charge voltage.

During battery operation (discharge), Q803 is ON, and the circuit is similar. R287 causes the input voltage to decrease to the discharge threshold voltage and makes the circuit into a lower voltage comparator. When the output drops below this lower threshold, the sneaky 1Meg resistor R817, resets flip-flop U803A, shutting down the DC-DC converter, which shuts down the instrument.

battcircuit

I wrote an Excel spreadsheet (below) to analyze the charge and discharge voltages, and so I could figure out how to modify the circuit to apply the correct voltages to a LiPo battery. Columns B and C show the discharge and charge voltages for the original Lead-Acid battery.  I first calculate the current through the reference divider circuit, then the voltage. Then do the same for the feedback divider to derive the Battery Voltage. Diode voltages are measured values for CR811 (0.45V) and CR806 - CR810 (2.2V).

Columns D and E show the Lithium cell voltages. I chose about 4.15V, less than the usual 4.2V,  for the lithium cell maximum, to protect the cells from cell voltage imbalance. If I used a 2 cell circuit with proper cell balancing, The voltage could be safely increased to 8.4V.

The final circuit changes required are:

Initially I only changed R806 to change the charge Voltage to 8.2-8.4V, but the discharge voltage was about 7.1V so the instrument would turn off prematurely. It should be closer to 6.0V, 3.0V per cell. The difference between Lithium charge (8.4V) and discharge (6.0V) is higher than the circuit provides. This difference is caused by the difference in voltage across the 5 diode chain vs. the drop across R827,11.5K. The diodes can be thought of as the coarse voltage difference, and the resistor provides the fine voltage difference. To increase this difference, reduce R287, and increase the diode voltage. To increase the voltage of the 5 diode string which is about 2.3V (0.46V per diode), either add a sixth diode (2.7V) or replace it with a zener or shunt regulator. The current is only about 25uA, so LM385-2.5 is a good candidate: its minimum current is 20uA.  I picked the extra diode. Quick and easy rework, and it allows 2.7V difference vs 2.5V for the LM385.

LM385 would also reduce the temperature coefficient (tempco) of the charge voltage. 5 diodes have 5 * -2mV/C = -10mV/C tempco. LM385 has about +/-50ppm * 2.5V = +/- 0.25mV/C. SLA batteries like the negative tempco.  Lithium's prefer a stable charge voltage. I'm pretty sure:-).

analysis
 
Here is the new Lithium batteries installed along with the 2 resistor mods. The battery holders are held down with foam tape. The old SLA battery is also shown. I will add single pin connectors to the new battery. I get > 2 hours operation from these old (and unknown capacity) cells. New cells are on order.

li batts


   Dave's Home Page
Last Updated: 12/28/2021