Current/voltage-drop on the voltage sensing wires of a Four-Terminal mesrmt setup?

Discussion in 'General Electronics Chat' started by SandiegoSD, Nov 27, 2012.

  1. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    Folks,
    I've been experimenting with Four-terminal (Kevin) measurement to measure some low

    resistance. I was trying to recreate what this guy did on a breadborad:

    http://www.kerrywong.com/2011/08/14/...m-measurement/

    Unfortunately in my tests the Device Under Test has a resistance value of under 1 milli ohm (while the 18 gauge copper wires are typically 10milli ohm per two inches roughly, which I use to connect the DUT to the breadboard)
    I guess this is causing troubles for me.
    On the two inner voltage sensing wires, I measure a voltage drop across the wire (which connects the DUT to the breadboard--I'm sticking the probes of my voltage meter onto the breadboard too )

    Does anyone know whether this is because my DUT resistance is too low compared to the wires, that the Four-terminal sensing method just become not applicable for me?? or I'm missing something?

    +++++++++++++++++++++++++++++++
    Update: Ok.. i was measuring this because the measured value for the DUT is higher than what I got with a standalone Keithley instrument. So i was trying to see whether I'm picking up/accidentally measuring the wires or some other contact resistance other than that on the DUT.

    In theory there should be negligible currents on the voltge sensing wires.
     
    Last edited: Nov 30, 2012
  2. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    [​IMG]
    I cropped out the four-terminal portion on the schematic. As shown, red paths are the current wires to the DUT.
    Green paths are connected to the Positive/Negative input of an differential amplifier.

    Now I might be making mistakes here. The question marks above: 1) I've disconnected the Negative input on the amp (-IN) from the GND (pin 3 and 4,etc), this seems to give me better results
    2) Where should I connect the current wire that comes out of the DUT? It seems it's different depending on which GND i'm hooking it up to .the power supply gnd , or the sideline on the breadboard (which is eventually connected to power supply GND)
     
  3. BillB3857

    Senior Member

    Feb 28, 2009
    2,400
    348
    My guess would be that you are also seeing the connection resistance of your breadboard added to the resistance of the DUT. Try connecting your voltmeter directly to the DUT and see if there is a more realistic reading. Also, insure that you have a truly known constant current source.
     
  4. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    yes..i put an amp meter in series with DUT (before it) and get 501mA.

    I tried measure the DUT directly, the result is around the same. 0.3mV indicates 0.6mΩ for the 0.4mOhm resistance.
     
  5. BillB3857

    Senior Member

    Feb 28, 2009
    2,400
    348
    What kind of resistor are you using that should be .4m ohm? Does your meter read 0.000V with the leads shorted?
     
  6. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    electrical connector with 4 small metal pins. I'm suppose to measure the contact resistance across each pin(the 4 are the same). Really small numbers.

    The volt meter is Agilent 34411A. it has a Null function to zero out the offset(? I mean to say clear the readings when shorted)
    although, it might not be so accurate when dealing 0.3 mV. But my amplified value at the differential amp confirm this number.

    I think either my R is too small compared to the wires for the Four-T sensing to work, or it's some grounding issue
     
  7. BillB3857

    Senior Member

    Feb 28, 2009
    2,400
    348
    If it is a connector, I would be surprised if the resistance did not change when connected and disconnected several times. Wiping action, oxide or other contamination, contact tension, etc. are all variables that can cause variation in contact resistance.
     
  8. WBahn

    Moderator

    Mar 31, 2012
    17,777
    4,805
    Keep in mind that you are measuring the resistance of everything between your two probe points (though you should not be measuring too much from the contact resistance of your probes since only negligable current is flowing through those contacts. If possible, you want the probes to be connected well inside of the contacts for the current leads.

    Four-wire measurements are very sensitive and accurate -- while at NBS/NIST we routinely measured nV scale voltages across superconductors using multi-thousand amp current sources. In fact, the current sense resistors that we used to determine what the current was were 0.1 milliohm. Even at that they were large serpentine conductors that you did not want to touch.
     
  9. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    Here's the connector pins I'm using (they're industrial connectors so they are kinda big and, low resistance); the two pieces shall plug together
    [​IMG]
    I'm using white wires for current and blue for voltage.

    The discrepancy I'm been complaining about so far is between the 0.4 mΩ I got by clipping the the aligator clips(two wires of each clip are connected to the current source and volt meter respectively) of a Keithley current-source&voltagemeter combo (using 4-wire measurement setup )directly to the big metal ends on this connecor, and the 0.6/0.7mΩ I got by using the four wires(with the same Keithley current-source&voltagemeter combo)

    So I'm thinking that maybe I shouldn't have soldered the white(current) and blue(voltage sensing) wires together on the connector ends?
    Illustrations about 4-terminal sensing I've seen mostly draw volt-sensing wires on the inner side. but I guess they're eventually connected together, like this resistor
    [​IMG]

    What you think? shall I seprate the soldered wires?
     
    Last edited: Nov 28, 2012
  10. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    well... i just realized that when I twist the connectors, the resistance measured with four-wires actually change quite a bit.
    it doesn't get down to 0.4mΩ though..
     
  11. BillB3857

    Senior Member

    Feb 28, 2009
    2,400
    348
    If I were making the connections, I would solder the current source wires into the solder pots and make the voltage sense connections with alligator clips to the outside of the connector. You may be sensing the resistance of the solder between the current source and the actual pins added to the resistance of the pin connection. Like I said in post #7, I'm not surprised that you are seeing variations when moving the connectors.
     
  12. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    I saw from Kerry Wong's page that someone commented this :
    "Are you using alligator clips in which each jaw is insulated from the other jaw? It is important that each probe makes independent contact with the device under test."

    I haven't found a theoretical explanation for this claim yet, but as you suggested, it may make the difference
     
  13. WBahn

    Moderator

    Mar 31, 2012
    17,777
    4,805
    Yes, I suspect this is the problem. The current has to go through the solder joint and you pick up a voltage drop across that joint. If you voltage taps are inside that joint, they won't pick up that voltage drop but, if they are outside that joint, they will pick it up. If the two wires are soldered together in the same joint, then it is almost impossible to determine how much of the joint is "inside" the taps.

    Now, one question you need to consider is what should be part of the resistance you are measuring? Perhaps you SHOULD be including the typical solder joint resistance, or perhaps you shouldn't. but it's almost certain that you don't to be measuring some unknown fraction of it. If you do want to include it, solder the voltage taps on the current feedwire just outside the solder joint to the connector. If you don't, solder the voltage taps onto the connector body directly away from the current feed joint. It doesn't have to be far and it doesn't have to be visibly between the current joints.
     
    SandiegoSD likes this.
  14. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,394
    1,607
    After viewing the image where Mr Wong states "Note that 4-point probes were used for maximum accuracy" and there is no such thing as a 4 point probe method being used I would stop using him as a reference.

    While you can use an alligator clip in a 4 point measurement of a leaded resistor as Mr W does you need FOUR clips to do so. Two for current, two for voltage. You put the current clips at the far ends of the resistor leads, and the voltage clips close to the resistor body. That way the voltage drop on the leads is minimized.

    Soldering the test wires into the pin looks ideal. Two wires lets it be a Kelvin connection where the contact point is within the pin: can't get any closer then that. If you want to try soldering to the outside of the pin body then do so, but notice you've moved the voltage measurement point perpendicular to the current flow, so the voltage drop should be the same.

    Do you have another meter? A current limiting bench supply? Your 34411 has a 100.0000 mV range correct? Use the supply in current limit mode to dump as much current as you have available into the connector. 1A should be good. Use another meter to monitor that current.

    Then take the 34411 and measure the voltage, compute the resistance. If you have a pair of fine point probes you can use them to read various points on the connectors.
     
  15. eblc1388

    Senior Member

    Nov 28, 2008
    1,542
    102
    I would also remove the 1M resistor and the AD8603 Opamp. Here is the circuit which you should try and definitely follows ErnieM's suggestion of using 4-clips Kelvin connection. Note the added resistor at sense Pin#5 of the AD8276 Amp.

    [​IMG]
     
  16. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    Thank you. I see the need to separate the voltage and current clips/jaws.
    If I directly put the volt meter probe on the metal pin(outside of the pin body), the reading is slightly smaller than what's measured at the end of the voltage sensing wire. So very likely the blob of solder joint and some contact resistance are picked up when using the wires like this. I'll try soldering the volt sensing wires on the outside of the metal pins.

    Good observation! I actually did put the extra matching resistor to get the current source to work.
    Yes that 1MegaOhm res doesn't make sense to me .
    Although i believe you'd want the volt sensing wires on the outter side and the current wires on the inner side.
    But for my application I'll definitely try the suggestions above.by soldering the volt sensing wires on the outside of the metal pin and keep the current wires where they are.
     
    Last edited: Nov 29, 2012
  17. eblc1388

    Senior Member

    Nov 28, 2008
    1,542
    102
    Sounds to me you still don't fully understand why four separate connection are required. If you solder them together, you will end up with only two connections to the unknown resistor.

    This is not a Kelvin connection and the contact resistance error will add to your measurement result and NOT be eliminated otherwise.
     
  18. SandiegoSD

    Thread Starter New Member

    Oct 12, 2012
    18
    0
    I mean I'd solder the voltage sensing wires on the green circled areas and keep the current wires where they are. Does that comply with what you're suggesting?[​IMG]
     
  19. eblc1388

    Senior Member

    Nov 28, 2008
    1,542
    102
    I'm sorry no.

    You really would need four separate connections to the unknown resistor, plain and simple. This means that the four connections have to be isolated to each other before any connection to the resistor. You cannot solder two of them together to get the same accuracy.
     
  20. WBahn

    Moderator

    Mar 31, 2012
    17,777
    4,805
    He wouldn't be soldering any of them together. He would have the current leads where they presently are and then move the voltage taps to a separate location on the pins. That should work for him. The idea is to move the voltage taps inside the voltage drop associated with the current injection joints and putting them to the outside of the solder cup will do that. He will then be measuring the resistance from along the pins as well as the pin-to-pin resistance. If he just wants to get the pin-to-pin resistance and exclude as much of the resistance of the pins, then he needs to move the voltage taps to just a couple pin-diameters from the pin-to-pin interface (and not much closer).
     
Loading...