Repairing an HP 3478A Multimeter with a Hacksaw
tomverbeure.github.io
tomverbeure.github.io
That looks like an early version of what is now called a "Zebra connector".[1][2][3] Those are a stack of alternating conductive and insulating layers, and look striped, hence the name. These are used like gaskets between two PC boards. There are several conductive layers per pin, so you get a connection without carefully lining everything up. They're vibration-resistant, so they tend to be used with keypads and in automotive dashboards.
I had no idea those were still around. I saw them mentioned once in some article about early digital watches.
This would be a useful technology for things that stack together mix-and-match like Arduino shields, but you want something less bulky and more reliable than 0.10 header pins.
[1] https://www.zebraconnectorsmfg.com/product/elastomeric-conne...
[2] https://www.fujipoly.com/usa/products/zebra-elastomeric-conn...
Zebra connectors are still extremely common for interconnects between PCBs and passive-matrix LCD glass.
The Achilles heel of a bunch of PCB alternatives (conductive ink, conductive plastics, plastic laminates like zebra connectors) is resistance, which will be hundreds of times worse than copper. Fine for reflective LCDs or capacitive keypads, bad for anything that uses more power. For the carbon connectors in the Fujipoly catalog https://www.fujipoly.com/usa/assets/files/Fujipoly%202015%20... they list a current rating for a 0.040"x0.040" pad as 0.005 amps. That's 3.125 amps per square inch. A single 0.10" header pin can do something like 4 amps, and you can fit a hundred of them in a square inch.
Non-ideal for shield stacks that want to move current between layers, not just data, like wifi shields or battery shields. (Really high current devices like relay shields or bigger motor controller shields often have their own dedicated power connectors)
The gold and silver connectors have better current capacities, but my psychic powers tell me they cost a lot more.
The 3478A is a pretty decent and versatile meter. It has three weak points though:
1. The aforementioned RIFA caps
2. A lot on the analog side is done by a hybrid, including driving relays, input switching and a bunch of trimmed reference resistors (current source gain and parts of the input divider iirc). Also, the mounting of the hybrid is also what holds the front/back input switch in place. Stuff's flexing a fair bit when using that switch! If that hybrid is shot, the whole thing is pretty dead.
3. The calibration is stored in a battery-backed SRAM on the outguard side of things. I don't think you can extract the SRAM contents, so battery swaps need to be done live. If you send one of these in for calibration, it's going to be about as expensive as another 3478A (n.b. the circuit allows you to do stupid things like e.g. shorting the battery for a short time, because the SRAM is fed by the battery and the power supply through or diodes, and there's a 1uF or something like that tantalum near the SRAM. I don't know how long it survives off of that cap)
4. Don't feed mains into it when in ohms mode.
You can read and write the SRAM contents over GPIB. See the references section at the bottom of the article.
I should see if I can talk to it directly, it's probably HID or CDC.
You can also recalibrate it from the front panel easily if you have another meter to compare it to with similar or better resolution.
Guess who screwed one up :)
One time I used a hacksaw to remove part of a 4-lane PCIe connector on a server motherboard so I could plug in a 32-lane PCIe video card. It worked great.
Another time I wanted to plug in a full-height USB 2.0 PCI card into a (Soekris) embedded SBC low-profile slot (limited by the enclosure), so I sawed off the upper two USB jacks. I had to do some minor rework to the card (reconnecting some of the broken traces with mod wires), but it worked great afterward.
You can discharge a capacitor by shorting it with a screwdriver (with an insulated handle), or by clipping a resistor to the capacitor leads for a less abrupt discharge. I think people who do a lot of appliance repairs learn to look carefully for capacitors when opening appliances, and make sure that any moderately large ones (or ones that there was reason to believe could have been charged to a high voltage) are discharged before beginning work.
I don't know how common large capacitors are in multimeters. I think there's a much bigger risk when opening a CRT monitor or a microwave.
Most multimeters are battery powered - this one is the exception, this one is a bench meter and is supplied by an IEC cable on the rear, but so long as you unplug in and let it sit for 30 seconds or so before opening it up, you're just fine.
You generally only have to worry about capacitors being charged still on tube-type gear (anything with high voltage circuits), anything with solid state parts, will self discharge within moments.
Obviously, do not have the meter hooked up to a circuit under test while working on it.