Pesision resistance calibrator design

[tplayne]

Hi all,
I recently came into a GenRad 1689 RLC bridge for a steal off ebay. It seems to work fine but needs more accurate calibration resistors then I had laying around. GenRad sells both cal service and a cal kit, but both are exorbitantly expensive. I figured that for the cost of 4 Vishay z-foil resistors + pcbs + a jumper as 4 wire short I could fabricate something very similar to the cal kit listed here. Then I could measure these references at work and get pretty accurate R and Q values to cal my bridge from. However, this is my first undertaking that potentially involves impedance controlled pbc design as well as working with such high accuracy parts.

I know some of the basic stuff,

• 2U gold plated PCB fingers for longevity
• Preheat PCB prior to soldering
• Avoid touching resistor leads with bare hands
• Keep iron temp around 295c during soldering
• Sink heat from resistor body during soldering (perhaps use wet sponge around resistor body?)

In addition to the above, is there anything else I should know about/ take into consideration when fabricating for high precision? Any advice is appreciated.

[fcb]

I'd give the resistor a little more width, makes it easier to clip on temporary heatsinks.

Also you probably don't need the copper fill and a bazillion vias.  This will increase unwanted capacitance between the resistor legs, probably not enough to matter, but..

I wouldn't use plated thru-hole for the resistor either, plating will affect the 4 wire measurement you want.

[tplayne]

The big copper fill is meant to be a guard region and would be connected with wire to the guard input on the 1689. Not sure if that changes your evaluation, but that was the plan.

As for the plated through holes for the resistor, how else would I connect the resistor to all 4 "leads" of the kelvin slot if not using pth?

[Kleinstein]

Shilding may work, especially for capacitors. However it would not need via stitching. So just 1 via should be enough.

For the resistor one could still solder to both side of the resistor, if the hole is not plated through. A plated hole would not really make a big difference, unless the solder joint is really bad, or the hole way too large. Worst case it would shift the effective legth of the resistor leg by a small fraction of a milimeter and even that shift would stay constant. So for normal purposes not an issue.

Not plated holes may be a point if one has more alternative footprints, e.g. to also allow for a capacitor or different shape resistor.

The 4-wire short would also want at least 1 of the holes not plated through. One could still use a drill to remove the plating, or cut a trace at a suitable point.

[mendip_discovery]

My comment is if you have space for Measured and Q, why not add a spot for the last calibrated date.

Might be worth looking at some test points so you can measure the resistance with other meters.

[donlisms]

The resistor body will get hot due to heat traveling in through the leads. Consider... by the time the heat comes back out through the body, it's already done the damage.

Maybe think more about sinking the heat from the leads before it has a chance to enter.

(Also... I've never heard anyone get nervous about touching leads with bare hands... is that a Thing?)

[tplayne]

You're probably right, a proper heat sink clip is cheap enough that I should do it the right way. And I've only ever really worried about touching resistors when they have ultra high resistance values, such that the oils on my hands transferred to the board might be a preferential current path, and bypass the resistor I care about. Probably not relevant for the resistance values that project will concern though.

[SoundTech-LG]

I love my 1689. You stole yours off ebay. Mine was in a pile of other great test equipment being thrown out by my former employer. Lots of rescued gear that day.
If you make your calibrator board, you might be able to sell some here.
I would certainly be interested.