New to measuring resistance standards with my 3456a

[valley001]

Hi all,

Having fun here with my old test gear.  I have an ancient 10 ohm resistance standard purchased on ebay and I would like to measure it with as much precision as my 3456a can provide.  Below is my setup.

display 6 digits
100 power line cycles
4 wire
OC Ohms is on (does this cancel the lead resistance?)

I am using copper wire between the standard and the meter, fastened directly to the 3456a inputs and the scree terminals of the standard. 

I am getting a reading  09.9974  the last digit bounces a bit between 3-5 or so.  Last calibration on the resistance standard (if the sticker is to be believed) was in 1972 and indicated 9.9928 ohms. 

Does all this sound reasonable?  I am a total noob at precision resistance measuring.   

[CatalinaWOW]

Assuming you used the 100 ohm scale and had 6 digit precision set, the spec allows 6.4 mOhm error 90 days after calibration (.004% of reading +24 counts of 100 microohms), and an additional 0.0004% for each additional 90 days since calibration.  If yours have been far longer than that, like mine, I would say your readings are entirely reasonable.  In fact, assuming the calibration standard is right, you are still within the 90 day spec. 

Of course you have no idea how close the standard is since it is so long since it has been calibrated and you have no idea of the intervening history.  It would be a coincidence if both your meter and the standard are out in the same direction and magnitude, but it is certainly possible.  Welcome to volt nuttery.

[valley001]

Thanks for your input, it sounds like I am on the right track with this.  I was using kelvin clips but could not get a stable reading clipping them to the awkward screw terminals on the standard, so I elected for the copper wire with bare ends.  Seems much more stable. 

The 3456a is my best voltage/ohms measuring instrument but it has not been calibrated since 2006.  I have thought of having it calibrated and using it as my "standard" but just haven't looked into what that would cost or entail. 


 

[CatalinaWOW]

Your meter seems to be good enough for most normal uses.  Unless you have a specific need for high accuracy, or are doing it for personal entertainment there is little reason to go to the trouble and risk of calibration.  The risk comes because it almost always necessary to ship you meter to the cal lab and back.

[alm]

Adjustment of the 3456A is via trimmers, unlike the digital calibration of almost every modern DMM. Innovation back then was that the most common trimmers are behind a panel on the front panel, so you don't have to disassemble the meter. But still have to turn trimmers instead of having a computer doing everything. So expect to pay more for adjusting a 3456A than for example a 34401A.

The 3456A has excellent stability. It would not surprise me if it was still well within 1 year specs. Of course that does not give you traceability, if you care about that.

Four-wire measurement cancels the lead resistance, because it separates current-carrying wires (which will drop voltage over the leads due to the I*R drop) from the voltage-sesnsing leads. (almost) no current through the sense leads means almost no errors due to lead resistance. Look up Kelvin sensing.

OC (offset compensated) Ohms is to correct offset voltages (as opposed to resistances) in the circuit under test. The most common source of this is the Seebeck/thermoelectric effect causes small voltages at temperature gradients at dissimilar metal junctions. So if you'd have brass banana plugs connected to copper binding posts, this could produce some microvolts of offset voltage, which can disturb the resistance measurement. Offset compensated Ohms first measures the voltage with the current source turned on, then the voltage with the current source off to measure offset voltages, and subtracts the two values.

[dietert1]

Let's also mention ambient temperature. That may affect both the resistance standard and the meter. One effect is the temperature coefficient of each device, another one is the thermal EMF at the wire connections.

Other recommendations when trying to get precision from a previously unused HP 3456A:

- The mechanical front/back switch needs movement every now and then, otherwise it may cause trouble. Like audio equipment: Instead of buying expensive new devices, one should move all those cheap RCA connectors once a year and everything will be fine.

- Open the top covers and check voltages of the in-guard power supply. There are some electrolytic capacitors prone to fail (by design).

In the case of a HP 3456A a calibration is undesirable, since it will dilute traceability. Better write down your check results in the lab notebook and use the info to apply corrections by hand in case of precision measurements.

Regards, Dieter

[CatalinaWOW]

I agree with Dieters comments with one minor exception.  Calibration has minor confusion over its meaning.  To some it means compare your instrument to a traceable standard and report results.  To most it includes adjustment of the instrument to meet original specs.  If you desire traceability, and want not to disturb stability as Dieter recommends any competent calibration lab will follow your instructions. Other options would include joining a calibration club which shares traceable standards, or purchasing such standards.  An example would be to have your resistor calibrated, which would be substantially less expensive than calibrating your 3456.

[KK6IL]

Adjustment of the 3456A is via trimmers, unlike the digital calibration of almost every modern DMM. Innovation back then was that the most common trimmers are behind a panel on the front panel, so you don't have to disassemble the meter. But still have to turn trimmers instead of having a computer doing everything. So expect to pay more for adjusting a 3456A than for example a 34401A.

I like that my 3456A's have trimmers. I don't have to worry about a battery failure or software glitch destroying the calibration data, and I can calibrate a single range without stepping through a multi-level menu system as on the 34401A.

John

[The Soulman]

Adjustment of the 3456A is via trimmers, unlike the digital calibration of almost every modern DMM. Innovation back then was that the most common trimmers are behind a panel on the front panel, so you don't have to disassemble the meter. But still have to turn trimmers instead of having a computer doing everything. So expect to pay more for adjusting a 3456A than for example a 34401A.

I like that my 3456A's have trimmers. I don't have to worry about a battery failure or software glitch destroying the calibration data, and I can calibrate a single range without stepping through a multi-level menu system as on the 34401A.

John

*adjust 

[valley001]

Some pictures for fun. 

Here is the 3456a, the reading is lower this morning.  It has only been on a couple hours compared to yesterday it was on all day. And its a bit colder in the room.   Unrelated: The 3450a is taking 100mv from my 332D, I have been working on repairing it and it is now under evaluation.



Here is a reading on the resistance standard with the 3450a.  The pic was a lucky catch it was dithering between 5-6. 

 

[dietert1]

Congrats to that nixie DVM. Appears to be working very well.

Please note that the HP 3456A manual recommends a 100 Ohm standard for calibrating the 120 Ohm range.

The HP 3456A isn't very strong on measuring small resistors. Its biggest test current for resistors is 1 mA, so there will be a precision limit around 1 milliOhm (about 1 uV). It is very easy to have some extra uV from thermal EMF in the wiring, yet i don't know how far the 3456A autozero method takes care of that. Other specialized micro-ohm-meters work with larger test currents and everything gets easier.

Regards, Dieter

[GigaJoe]

he did measure with O.C. compensation , that offset parasitic  " extra uV from thermal EMF in the wiring" in theory

[bob91343]

I recently obtained a couple of GR decade resistance boxes.  Measuring them with my 3456A yields pretty much exact readings through all settings.  And deviation can be explained by residual resistance of the decade boxes.

I am thrilled that all is so close.  And I am sure it's no accident.

I got some Chinese 'Kelvin clips' and use them on my 3456A with good success.  And their dimensions are such that I can clip them to the GR boxes securely to make my measurements.

While I am sure more precise measurements are possible, I feel that both the boxes and the ohmmeter are doing the job as well as anyone could expect.

[valley001]

The 3456A is the best $100 I have ever spent.  I have the room for it so the extra ordinary size isn't too much of an issue. 


Bob- I have the cheap kelvin clips also.  They developed a problem with intermittent contact at the spring tension piece on the banana jack. I soldered the tips of the banana jacks (fusing the spring piece to the inner portion) and that has solved the issue, but now I assume there would be thermal EMF issues with them. 

I would really like to get my hands on a Fluke 720a, but the price on those is just too high for me.  I have been looking at lesser units like the Biddle Precision divider on the auction site right now, I think it is two orders of magnitude less linear compared with the fluke though.  +/-10ppm vs the fluke at .1ppm.

[tggzzz]

The 3456A is the best $100 I have ever spent.  I have the room for it so the extra ordinary size isn't too much of an issue. 


Bob- I have the cheap kelvin clips also.  They developed a problem with intermittent contact at the spring tension piece on the banana jack. I soldered the tips of the banana jacks (fusing the spring piece to the inner portion) and that has solved the issue, but now I assume there would be thermal EMF issues with them. 

I would really like to get my hands on a Fluke 720a, but the price on those is just too high for me.  I have been looking at lesser units like the Biddle Precision divider on the auction site right now, I think it is two orders of magnitude less linear compared with the fluke though.  +/-10ppm vs the fluke at .1ppm.

There are other Kelvin Varley dividers; I have a Julie Research VDR 106/7 which is 1ppm. Very heavy, so watch shipping costs

[srb1954]

OC Ohms is on (does this cancel the lead resistance?)
   

OC ohms stands for Offset-compensated Ohms measurement. It doesn't cancel the effect of lead resistance (the 4-wire connection does this) but it compensated for errors caused by thermal EMFs by measuring the thermal EMF voltage before doing the resistance measurement and using this measured thermal EMF voltage to correct the measured resistance value.

[bob91343]

Maybe there should be a new group called Ohm Nuts.