small, very accurate shunt resistors?

Try Isotek for their 4-wire parts; but your specs are tough.

John

You need a 4-terminal shunt. I happen to have one 0.1 ohm shunt that was used for current measurement. The accuracy is .02%. I also have a matched set of 0.1, 1.0, 10, 100.1 and 1010 ohms. They were designed to be loaded by a 100K input impedance. Let me know by e-mail if this is what you want.

Norm Strong

An 0.1% 50 milliohm resistor must be accurate to within 0.05 milliohms. If you really need that kind of accuracy, then I think that trace resistance will be a catastrophic problem. Is your circuit designed for a Kelvin (four-wire) connection? I have very little knowledge in this area, but I am sure that it would be necessary and I doubt that it would be sufficient.

Also, you implied in a previous posting that you were trying to use a smaller current sense resistor so that your circuit would produce 0 to 2.048 volts instead of 0 to 4.096 volts. Clearly there is more to your circuit than just the resistor; perhaps if you posted that then someone with experience in this area could help you.

Jonathan

Keep in mind that when you get to such low value resistors, you need 4 connections to them to get an accurate voltage proportional to the current through them. The best you may be able to do is to have the current carrying traces going into two of the outside corners of the resistor and the voltage connections coming out from the opposite inside corners of the pads. Very nearly as good is current into one side of the resistor, voltage from the other side. The idea is to include as little trace and pad drop as possible in the measurement.

And the problem with a paper clip is....?

--
Many thanks,

Don Lancaster
Synergetics   3860 West First Street  Box 809  Thatcher, AZ 85552
voice: (928)428-4073 email: don@tinaja.com

Please visit my GURU\'s LAIR web site at http://www.tinaja.com

TC.

John

Shunt resistors 0.1 Ohm and lower are mostly 4-terminal even in through hole mount, and anything below 10 Ohms is rare in SMD. The parts you think you want may be custom (interpret as $$$$). You could make your own from manganin sheet cut a few percent low calibrated with software or dremel...

I can't tell from the description if these are anything near an 0805, but Digikey does sell some 0.5%, 50 milliohm 4 terminal current sense resistors. Part # RW1S0CKR005D

These are about 3 times the size you asked for and I don't know who stocks them but they look like fine devices.

By the way, have you figured out how you are going to amplify and digitize the signals from your shunts in the presence of trace drops, while keeping the accuracy any where near 0.1%? I have managed 2% or

3% without heroic efforts, but not much better.

--
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

If you use Zeranin paperclips rather than the plated steel ones from Wal*mart it would be better, but you still would have the tolerance issue (unless you can handle the calibration issues). ;-)

Tempco of steel is about +0.5%/K, worse even than Cu, so the desired

0.1% accuracy is lost with tenths of a degree change in temperature.

Can you use a less-than-spec accurate (but stable) Kelvin-connection shunts that other have mentioned and calibrate? As others have mentioned, you really don't want hardly any of the +0.39%/K copper traces in the measurement path.

Do you need a shunt made with four terminals? If the copper is to contribute, say, 0.5%, then you need an equivalent length of about

3mils of 12 mil wide 1-oz traces on each end. Maybe this is achievable by running traces off of either side of an SMT pad for a four-wire connection, but it sounds a bit dodgy. I've done something similar for an instrument that measured standard CT output...

Bare (low tempco alloy) shunts are made in high volume for multimeter use (both a jumper style for low accuracy meters, and IIRC, some

4-terminal types) but generally not nearly that accurate.

Then there are possibilities like this:

For an LVS3 (0.5" x 0.25" SMT) 0.05 ohm 0.1% with 15ppm/K tempco they want $8.30 each, which is a bit pricey, but maybe tolerable for you. They're obviously a lot bigger than 0805, but it meets all your other specs (and appear to be in stock).

Here are some in 2516 (0.25" x 0.16") which are offered in your specs as well:

(watch the tempco and maximum current rating)

And some in 2512 (0.25" x 0.125") which also seem to be offered in your specs:

(UHS 4-2512)

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

A major problem with all of the SMD resistors, is that there are *NO* kelvin contacts! So, assume you have a 50 milliohm SMD resistor that you determined was within 0.1% of that value. Now solder it to your PCB, and measure the resistance on the traces close to the resistor. Hmmm...seems to me that you will not be any where close to 50 milliohms + 500 MICRO-OHMS!!! The damn solder joints will have far larger resistance, and one might ask if you remembered to attempt kelvin connections??? And..where does one measure these low-R SM units? at the top, close to the "J" bend? half way down from the top? at the bottom of the "J" bend where it would connect to the PCB trace? Furthermore, what about thermo-electric effects? I presume one could go on about the problems...

Maybe the paper clip is still attached at the top of his Word document, and will not motorcycle away...

I stand corrected about Kelvin SMD resistors. I see that Vishay complains about tin/lead oxidation and about poor reliability of conductive epoxy. BUT in the same breath they seem to recommend !gold! which has a history of seperating from solder at the interface...

Go have a look at:

or

That ought to get you started.