DMM does not stabilize when measuring resistance

[pwnell]

I know this is user error, but why can I not get a stable reading from a 34465A DMM using 2 wire or 4 wire on a 1kohm resistor?  I have 10 PLC, yet the value ranges from 0.982817 to 0.982920 … this seems more than the spec allows for. The resistor is not in circuit.

It takes about 2 minutes to reach some form of stable value.

[gamalot]

What kind of resistance are you measuring? What kind of test leads are you using? How long did you wait for the reading to stabilize?

The test voltage in the 1 kiloohm range of the 34465A is 1 volt. You may need a little time to warm up the resistance under test.

[bob91343]

With that kind of resolution you can't ignore thermal emf.

[pwnell]

It is a 1k 1% carbon resistor. The meter says it uses 1mA on the 1k ohm range.  It feels very cool to the touch. Thermocouple says resistor is 1C below room temp after 5 minutes, yet value still fluctuates.

[gamalot]

I don't have any carbon film resistors in my hand, so I can't repeat your experiment. However, I think your multimeter is worth testing it with a better resistance.

[HighVoltage]

To test your 34465A setup, use your 4 wires and make a short connection in 4W mode to get "0 ohm".
How stable is the reading now?

A picture would be nice, maybe we can spot a problem in your setup?

[TUMEMBER]

Carbon resistors have one of the highest temperature coefficients. No wonder they are "floating". Their TCR can be up to 1000ppm.





The beginning and the end of this movie can be omitted because it is embarrassing in its extremely flashy form.

[joeqsmith]

Sounds like you are having fun with your new meter.   If you want to see the meter's drift in the 1k range, you may want to get some better parts.  This post, you can see a low temp co I used for a similar test on an antique Fluke meter. 

https://www.eevblog.com/forum/testgear/some-old-school-instruments-showing-how-its-done-(hp-3325a-and-fluke-8506a)/msg3463462/#msg3463462

I will often use beach towels to tame the air currents. 

[pwnell]

Assuming a 1k carbon resistor, at a lack of having the specific datasheet I picked any random one (https://www.farnell.com/datasheets/1716725.pdf?_ga=2.43716581.132370299.1498698481-246720399.1496759099) - does that mean the resistance can change by (in this example):

11 Ohm - 99 kOhm = 0 to -450ppm/C

1000 * 450E-6 = -0.45 Ohm per C?

Is my calculation from ppm/C correct? (Assuming of course I stay within the linear region etc).

[pwnell]

I am using 4 wire and PLC of 10, but with the 1k resistor.  Will do the short in a bit.  This is what I get:

[pwnell]

Here is a test of my 100kOhm 1% resistor:


In this case the drift was 29 ohms over 20 minutes, assuming a temperature coefficient of 350ppm/C, it can drift 35 Ohms for each degree C, so I guess this seems reasonable?  Drift = temperature change as was mentioned. 

I will try a Vishay Y10731K00000T9L resistor next... (Need to get one first).

[joeqsmith]

My old HP34401A that I purchased new when they first came out.  It's not been abused outside of getting damaged from a lighting strike. 

Using a 1k0 Vishay S102K +/-1PPM BMF.   Resistor was mounted into a can and placed in a screened box which was grounded.   Meter was set to 10PLC, 4 wire with leads and not allowed to warm up.   Total worse case p-p was 0.003 ohms.   

If I were to allow the meter to warm up for an hour or so, this would tighten up.  It's been a decent meter.       

[HighVoltage]

If I were to allow the meter to warm up for an hour or so, this would tighten up.  It's been a decent meter.     

What Software are you using in the last picture to show the graph?

[joeqsmith]

If I were to allow the meter to warm up for an hour or so, this would tighten up.  It's been a decent meter.     

What Software are you using in the last picture to show the graph?

I've used LabView since the release of 4.0 in 1986.  It was apparent that for T&M, this was going to be a big time saver.  By the time 5.0 was released, I knew enough that I bought it for my own use.   A choice I never regretted.  For the last five years, I've exclusively used LabView 2011 for all of my Windows applications. While LabView is now free to the home user, I have yet to upgrade.   

This particular program is just a quick and dirty read the meter,  plot and record the data.  A few minutes of work.  See attached source code. 

[joeqsmith]

Here are a couple of old carbon composition resistor from my child hood, most likely pulled from old B&W tube TV sets or radios.  Odd to see that gold band.   The resistors were not enclosed as shown during the measurements.  Data was again collected at 10PLC, 4-wire.  Starting with the 1kohm,  I allowed the HP34401A to warmup for roughly 10 minutes, then collected data for another 10 minutes.   P-P is now  0.079 ohms.   I beach towel would help.   

Leaving the meter on (about 20 minutes of warmup time now), I inserted the 100k and let it run for another 20 minutes and noticed about a 100 ohm drift. It was settling out so I restarted the test and let it run for another 40 minutes.     P-P is now 2.6 ohms and we can see the furnace cycle.  Again, a simple beach towel would help filter this. 

[joeqsmith]

Of course, walking into the lab changed the air temp.

[joeqsmith]

Swapping out the 100k carbon comp for a 100k BMF 2PPM part.  The part was left in the air and  the worse case p-p is at 0.35 ohms.   It doesn't take much to create these random spikes in the data.  With winter coming on, the house is getting dry and ESD events are becoming a problem.   I would normally use that shielded box.   

Anyway, I hope this helps provide you some idea on what to expect. 

[pwnell]

Thank you - this has been really insightful.

[joeqsmith]

Glad to help.  Look forward to seeing some data from your sub PPM parts. 

[pwnell]

Finally got my stuff.  So here are some graphs - all same test setup:

Keysight 34465A DMM, BNC cable to a custom made copper tube, inside which the DUT resistor was placed.  I tried not to touch the resistor body as I inserted it into the tube as to mimimise temperature changes.  Tube below:





The four main resistors I tested (there was a 1kΩ and 100kΩ Metal Film - I only show one):



The carbon resistors is from a generic crap brand - they are definitely not 1% as tested:



Test results in next post.

[pwnell]

What is interesting is that the Metal Foil resistor, at 1kΩ, 0.01% and 0.2ppm/ºC turns out to drift more than I expected - I did not measure the temperature change but it could not have been more than 10ºC worst case, which would mean 1000*0.2e-6*10 = 2mΩ drift, yet it drifted 133mΩ (or 43mΩ if you take initial value and final value and ignore the initial bump in resistance) over 21 minutes.

[pwnell]

I saw a similar trend between the short and the Metal Foil resistor - making me think this is more to do with lead resistance than the actual component.  I switched to 4 wire mode and voila:



Delta 1 is for the whole period, if I look at the drift between 4 minutes and 20 minutes I get 0.45mΩ.

[TiN]

Also Z202 max TCR spec is 2.8ppm\K, not 0.2 ppm\K. Don't fall for "typical" specmanship from VPG on these. You cannot expect to have every random resistor meet typical spec. Short term rift of your DMM is also expected to be much bigger than one from resistor itself.

[joeqsmith]

You could ditch the BNC and go to a 4 wire like I show.  Show a picture of  how you connect it.

[pwnell]

Last one is 4 wire