Resistance measurement (4wire)

[spid3rx]

Hello all,

I've been trying to measure the resistance of a pin, which the supplier spec at around 15mohms.

I am using the hp 34401 multimeter with the 4 wire method to do the measurement. Its seems to me that when I probe the tip and the base of the pin, the resistance value is not consistent with multiple tries. The resistance value will somehow drop gradually if I probe it for a longer time. I want to know why?

so I tried connect the total of 6 pins in series so that the resistance value will be 90mohms in theory, I can get a much steady value, but it will drop as time goes by. I want to know why? is it the due to the low resistance nature? or there is some reason behind like thermal affecting it ?

the probing method is by using jig, not by hand to probe (to prevent inconsistency.)

 

[sirius0]

Hi spid3rx,
I have never used that multimeter and the manual is a 9MB download!
could the multimeter be taking a little time to correct for lead resistance? How long does it take to drop? Does it drop to an eventual consistent result?

 

[spid3rx]

Yes, I was suspecting it was trying to correct the resistance value. For the total of 6 pins that I measured, it will drop like from 110mohm to 80+mohm and it takes like 25 mins to reach that.
not too sure how is possible to measure fowith that long period of time just for the resistance...

 

[f95toli]

The 34401A is just a standard benchtop multimeter, nothing fancy.

It could be that what you are seeing is just normal drift; the resistance you are trying to measure is quite small so you need a fair amount of current to get a reasonable amount of voltage. Personally I would never even attemt to do this using a normal multimeter; I would use a good instrument (low drift) where I can control the current OR a custom IV measurement setup.

Anyway, have you checked the specs of the multimeter? What kind of drift do they specify?

 

[0xDEADBEEF]

I think your issue might be more fundamental. There is a known effect with adjustable resistors where a contact is swiping over a resistance wire that these resistances need a "punch through current". This current destroys metal oxides on the surface of the wiper that would produce a false resistance, especially as the resistor gets older. Maybe you have some kind of surfactant effect that gets gradually reduced. Is the pin gold plated? How do you contact the pin? With solder or with strong clamps? The best way to do 4 point measurements is with the sense wires connected further inside the resistor, not attached at the same terminals as the current wires.

More effects could be heating maybe 50ppm/°C, or thermo voltages if the current doesn't flip directions during the measurement and my favorite flicker noise... but people here couldn't tell me much about that.

I almost forgot: Was the multimeter running for at least an hour to warm up?

 

[spid3rx]

thank you all for the comments. I really appreciate the feedbacks.
It seems to me even the fundamental measurement requires a lot of in depth understandings. Yes, I suspected the oxidation and thermal, that's why I make a pcb, soldered all the 6 pins together, and make the sensing + probe on the pcb with the multimeter connector (to eliminate hand probe) in order to improve the consistency.

to f95toli
thanks, I am not sure what drift is, but I will be checking on the voltage and current control. excellent advise.

good point from 0xDEADBEEF
i should stabilize the multimeter by starting up for some time just for warm up. I should try that.
also to answer ur questions, yes the pin is gold plated. I tried before having the sense wires connected directly to the pin, still the resistance will varies. Actually I had tried both multimeter and mili-ohmeter, the trends the same. the resistance will drop through time.

one point to note, the multimeter can be set to constantly measure (fast) or measure at each time interval with averaging (slow). does this consider self cheating ? :P

 

[f95toli]

Drift simply means that the dc-offset or gain of the amplifiers changes over time. The usual reason for this is simply thermal drift which can be significant for measurements of this type. You should at least leave the multimeter on for a few hours before you start the measurement to let its temperature stabilize (For high-accuracy measurements people tend to use "ovens" which stabilises the temperature of the instruments , in many cases this is just a closed 19" rack with a heater and a temperature controller).

(I just spent a day trying to compensate for the the drift in one of my pre-amps, the offset changes by a few mV over a few hours which is bad since it is part of a servo meant to stabilize another piece of kit...Better amp+PID controlled oven seems to be the answer)

There are all sorts of problems that can affect measurements of this type. Do you have access to "proper" measurement equipment? Even something like a Keithley Sourcemeter (e.g. a 2400) would be much better than a multimeter for this type of measurement.
Or, alternatively, an IV measurement setup.

The main thing is to get the current up to some reasonable value and be able to flip the polarity to eliminate drift.

"Slow" measurement just increases the integration time; it will only help against noise, not drift.

 

[sirius0]

Just curious.
Why do you need to measure the resistance of these pins?

 

[spid3rx]

f95toli
Thanks again, I slowly get the idea of the "drift" and should be aware that those thermal is really a nuisance in affecting the result if I am measuring really low resistance with accuracy.
The IV measuremetn setup that you mentioned, is it that you are referring that we should have some sort of power supply to pump in the current and measure the voltage in order to have a better control and accuracy? or is it this can be obtained through an equipment like the 2400?

I now take the initiative to read the manuals of the 34401 multimeter. I know that for 4 wire measurement, the multimeter will generate the current and the sense wire will measure the voltage at the load. However, I seems not able to find any method to control how much current to supply. Is this true? correct me if I am wrong.

Sirius0
slight resistance value can affect the output of my load, for worst, after certain usage, resistance might increased. Hence I need to make sure and measure if the resistance cannot reach to the not recommended state.

 

[0xDEADBEEF]

The source meter can source current or voltage up to 200V. You can set limits for your resistance measurement. I don't know about the HP but usually they have certain current values for different resistance ranges. If you attach and detach the pins do you get consistent readings? Does the resistance decrease over time for other wires as well?

I just checked the temperature coefficient for copper, it looks like 1°C will produce half a percent difference in the reading make sure they don't get warm...

 

[spid3rx]

0xDEADBEEF
No, as of I measured today and tomorrow, I don't get consistent data. I am thinking its due to the equipment thermal. I guess I should have let the equipment to be on for a while to let it stablize the thermal effect.

If I attach (measure) and detach and re-attach (remeasure) , the readings will start from the recent lowest values onwards. for e.g. if initial measured 100mohm, after some time, it drops to 90mohms, detach and reattach, the reading will start from 90mohms onward. if it is not stable, it will continue to drop from 90mohms to maybe 80mohms. I tried to left it measure overnight. The next day I observed is very stable, but I don't think its a feasible way to measure like that due to long timeline :P

 

[Phrak]

For the experts here: Is there any chance of either thermo coupling or oxide layer rectification in these sorts of measurements?

 

[0xDEADBEEF]

If I attach (measure) and detach and re-attach (remeasure) , the readings will start from the recent lowest values onwards. for e.g. if initial measured 100mohm, after some time, it drops to 90mohms, detach and reattach, the reading will start from 90mohms onward. if it is not stable, it will continue to drop from 90mohms to maybe 80mohms.

That sounds like too much deviation. You could check your meter. If the theory of the instruments temperature is correct, it should always give the same readings after leaving it on for a day and always very different readings right after turning it on. To check it you should put in some wire instead of the pin when you let it warm up for a day. I would also be interested in the measurement signal. It is very useful if the current direction switches, to cancel offsets. You could attach an oscilloscope in parallel to check if that is the case.

I tried to left it measure overnight. The next day I observed is very stable, but I don't think its a feasible way to measure like that due to long timeline :P

It might be the only way do do it this way, but there are other techniques for resistances that are this small. Thomson-Bridge (aka Kelvin bridge) or a Carey Foster bridge but it is probably best to buy a better meter, if you find that it is really the limiting component.

 

[Altrepair]

You MUST use Kelvin Clips if you wish to have any luck at accurate measurement of low resistances. Also thermal voltages will effect the reading you get. So try to keep the temps stable and remember the current flowing through your device under test will heat it up which will generate thermal voltages.

 

[sophiecentaur]

I seem to remember, from way back, that it is a good idea to use an AC signal source in conjunction with your bridge in order to eliminate the possible effects of polarisation across contacts.
I would not reckon that a multimeter would be a suitable method for getting reliable measurements of such a subtle quantity. You need to ramp up your experimental setup.