4 wire measurement with large pre-resistor using Keithley 2400

Thread Starter

Alex_S

Joined Apr 20, 2017
3
Hi,
I'm trying to do a 4 wire measurement. I have the problem that due to a large contact resistance, I cannot measure the test resistance very precicely with a Keithley 2400 Sourcemeter. I can reproduce the problem with this simple circuit, which I have taken from the Keithley manual with the additional pre-resistor R:
circuit.jpg
I'm using a resistance Rs = 9 Ω, which I can easily measure as long as there is no pre-resistor R in the current lead. Also with a pre-resistor R similar to Rs the measurement works.
But with R = 1 kΩ as shown in the figure above and Rs = 9 Ω it doesn't work. Only if I manually chose a suitable current I from 1 mA to 7 mA, I can measure Rs= 9 Ω with the SourceMeter. For other currents, I get these false readings:
7,1 mA, Rs = 9.5 Ω
7,4 mA, Rs = 36 Ω
20 µA, Rs = 9.8 Ω
10 µA, Rs = 10 Ω
1 µA, Rs = 20 Ω
0.1 µA, Rs = 100 Ω

Using another 2400 SourceMeter the errors get even larger, e.g. at I = 1 µA, Rs = -100 Ω (at small currents the resistence suddenly gets negative!). The same problem occurs with a pre-resistor R = 100 Ω and Rs = 9 Ω.

Does anybody have an idea what am I doing wrong? I've always thought that a 4 wire measurement is just the right methode for measuring a small resistance with high contact and lead resistances.
 

ErnieM

Joined Apr 24, 2011
7,991
Even with a R of 100 ohms you have ten times the drop in your current leads (and current source) than you do in the test leads. That may be starving out the current source. I'm not sure how this instrument deals with that (I've never used one).

You can check what it is up to with a separate voltmeter to make sure it is not limited on the current outputs, though if I read the correct part of the manual it should not be.

In any case, inserting a known current means a voltmeter can act as an ohmmeter.
 

Reloadron

Joined Jan 15, 2015
4,463
Hi,
I'm trying to do a 4 wire measurement. I have the problem that due to a large contact resistance, I cannot measure the test resistance very precicely with a Keithley 2400 Sourcemeter.
In your measurement process are you using suitable kelvin measurement clips? For example Mueller Electric Company Model # : BU-78K or similar? I doubt you will have very good results trying to do kelvin type 4 wire resistance measurements using any substandard leads or clips in your measurement plane.For low critical measurements you may want to consider the use of low thermal test leads.

Keithly is literally a few miles down the road from me. Years ago I had a few engineering friends working there.

Anyway, you may need to get better leads for your measurement plane.

Ron
 

ErnieM

Joined Apr 24, 2011
7,991
I do a lot of four point probe measurements to insure metal to metal bonds are less than 0.0025 ohms. As long as there are four physical probes the measurement works just fine. We will make our own probes too using two Pogo style pins to fit into small places.

What does NOT work are probes such as these:


Pomona 6303. Digi-key sells these for $123 and they are pretty useless for very low resistance measurements, as they only have one point per side. They introduce errors on the order of 1 ohm.
 

Thread Starter

Alex_S

Joined Apr 20, 2017
3
Hi,
thanks for all the rapid responses!

@ericgibbs: I've seen this video, but I'm afraid it doesn't help me, because it doesn't tell anything about the 4 wire measurement.

@ErnieM: I've checked with a separate multimeter and measured basically the same current in the current leads as I entered at the Keithley for the source current. And I've checked the voltage drop across the pre-resistor R = 1 kohm, which makes absolutely sence.
And than I also measured the voltage drop across the test resistance Rs = 9 ohm: 135 mV at a source current of 15 mA, which is 100% correct! But the Keithley measures 8,15 V and accordingly 543 ohm. So what is going on with the Keithley??

@Reloadron: I'm not using Kelvin clips. I think, this would not solve the problem. The correct measurement with the separate voltmeter also worked without Kelvin clips. I used probes similar to the ones that ErnieM advised against.
 

Reloadron

Joined Jan 15, 2015
4,463
@Reloadron: I'm not using Kelvin clips. I think, this would not solve the problem. The correct measurement with the separate voltmeter also worked without Kelvin clips. I used probes similar to the ones that ErnieM advised against.
Well my thinking here is pretty simple. You choose the best lead(s) or clip configuration for the job. If Kelvin Clips would not suffice for your application then maybe a set of Kelvin Probes can be purchased or fabricated. Maybe a good picture of exactly what you are facing would help. Keithly does make and market a set of low thermal probes for use with your meter. It should be obvious what you are working with is not adequate leaving you to find a suitable four wire kelvin measurement technique.

Ron
 

Thread Starter

Alex_S

Joined Apr 20, 2017
3
Hi again,
meanwhile I got an answer to my e-mail from Keithley:
There is a limit for the accuracy of a 4 wire measurement with the SourceMeter 2400. Between Sense HI and Input/Output HI the voltage should not exceed 5 V. The same applies for LO. In my case the voltage was already 7 V at I_source = 7 mA for the 1 kOhm pre-resistor. And then the measurement of the SourceMeter is just wrong.
The reason for the inaccuracy at small currents below 10 µA is explainable due to the small voltage drop across the test resistance (9 Ohm) of only 90 µV, but the 2400 has measure specification of ±(0.012% + 300µV) in the 200 mV range. This problem is also reported here: https://forum.tek.com/viewtopic.php?f=13&t=139351&p=282469&hilit=4+wire+measurement#p282469
https://forum.tek.com/viewtopic.php?f=13&t=139351&p=282469&hilit=4+wire+measurement#p282469
To sum up, I would say the Keithley 2400 is just not suited for a 4 wire measurement in my situation. Special probes or clips would not help.

Thanks for all your efforts!
 
Top