Precision Resistor Standard

[echen1024]

While looking for a good way to test my DMM, I have decided to build a precision resistor circuit. I plan to surround it in mineral oil, with a small resistive heater to the side. Are there any possible improvements I could make?

[EEVblog]

But a $20 0.005% Vishay resistor from digikey and stick in a box with two terminals on it.

[echen1024]

Thank you for your reply Dave! That seems to be the standard circuit here. I doubt that there will be much difference with the heated mineral oil. Oh well. Anyone with a 3458 willing to test it for me?

[SeanB]

Ask Frankie ( Iloveelectronics) to send you his simple RCL test standard. 3 components in a box that have been measured with an accurate meter. Otherwise buy any old board from Ebay from high precision avionics and use the resistors from it.

[EEVblog]

I doubt that there will be much difference with the heated mineral oil.

There isn't, for any sufficiently low tempco resistor at normal room temps, which is what you'd want to use anyway.
With my office aircon I can control the room temp to about +/-0.5C

[AG6QR]

Look at the line of products from http://www.voltagestandard.com/ .   The DMMCheck and DMMCheck plus both combine precision voltage references and precision current references with a few precision resistors of various values.  The advantage over just buying resistors from Digikey is that the actual values of the resistances have been measured by a traceable calibrated meter, and recorded for you.  Plus, if you're that interested in precision resistors, my guess is that you might want the voltage and current references, as well.

Disclaimer: I don't even own one of these products -- all my knowledge of them comes from the website linked above.  It looks like they're reasonably priced, and well more than enough for a hobbyist to make sure a typical multimeter is behaving well.

[echen1024]

I was interested in making my own, but yes, the voltagestandard is a pretty compelling option.

[edpalmer42]

Are you planning on building the resistors themselves or using off the shelf resistors?  If you're building your own, you could use resistance wire that has a low thermal coefficient.  You'd end up with something like these:

http://www.ebay.com/itm/Leeds-Northrup-Cat-8070-Resistor-Standard-/131109714352
http://www.ebay.com/itm/Leeds-Northrup-L-N-4035-1-000-Ohm-Standard-Resistor-/221303915567
http://www.ebay.com/itm/1-Ohm-1000Ohm-0-01-Resistor-Standard-Resistance-an-g-of-Honeywell-Fluke-742A-/230775236712
http://www.ebay.com/itm/Leeds-Northrup-4223-B-0-00-OHm-Precision-Standard-Resistor-/221315161271

The smaller ones are full of mineral oil.  The larger ones with the holes around the top are designed to be immersed in mineral oil during use.  Don't forget to make yours 4 terminal!

Ed

[linux-works]

I just recently bought that DMMcheck device.  good price for what it is, even though its annoying that the chip numbers are scratched off ;(

but I like supporting small businesses and the seller does give good service.

the fact that he measures the vishays with his high end meter and writes down the value is most of the worth, to me.

(one complaint: the 1ma ac feature seems to baffle some meters.  on a 6.5dig HP and also on my keithley 6.5, the 1ma ac test signal almost can't be read.  all the other test outputs are fine and bang-on, with most of my meters.)

[MarkL]

But a $20 0.005% Vishay resistor from digikey and stick in a box with two terminals on it.

Or four terminals for more accurate Kelvin measurements on the smaller values.  Vishay specifies the measurement point as 0.5" from the resistor body, so that would be the point to solder each pair of posts.

[Co6aka]

But a $20 0.005% Vishay resistor...

Why not one of the HZ series (Vishay) resistors?  (Or a few decades of 'em?)

Anyway... What's the consensus on the "best" resistance values for DMM calibration? Without thinking much about it, seems it would depend upon the DMM's Vref, no? And, would one want just a "full scale" value, or that plus a mid-scale value?

Among my precision component junkbox collection I have ten 10K 0.005% resistors in parallel to (in theory) give me an even more precise 1K resistor. Also have some glass encapsulated resistors of less precision but presumably of much greater stability. (Caps too -- made by Corning!) And a funky big red 400M resistor with gold plated leads that's marked 1% tolerance.

...stick in a box with two terminals on it...

Why not four terminals, and wondering if you cleaned them and put a dried-out desiccant pack in the box...?

[ConKbot]

But a $20 0.005% Vishay resistor...

Why not one of the HZ series (Vishay) resistors?  (Or a few decades of 'em?)

Anyway... What's the consensus on the "best" resistance values for DMM calibration? Without thinking much about it, seems it would depend upon the DMM's Vref, no? And, would one want just a "full scale" value, or that plus a mid-scale value?

Among my precision component junkbox collection I have ten 10K 0.005% resistors in parallel to (in theory) give me an even more precise 1K resistor. Also have some glass encapsulated resistors of less precision but presumably of much greater stability. (Caps too -- made by Corning!) And a funky big red 400M resistor with gold plated leads that's marked 1% tolerance.

...stick in a box with two terminals on it...

Why not four terminals, and wondering if you cleaned them and put a dried-out desiccant pack in the box...?

2w/4w measurement really depends on the resistance youre measuring. If youre doing a 10k resistor, you should be able to get less than 0.001% error with your test leads, even with 2-wire ;p

[bronson]

even though it's annoying that the chip numbers are scratched off ;(

Wow, you're right!  How lame.  Can't believe I never noticed that.

The voltage reference is obviously a TI REF 5050 in a SOIC package.   Mine is marked "5050 K" "23CPFV'.  Interesting that I don't see a "K" variant on the datasheet: http://www.ti.com/lit/ds/symlink/ref5050.pdf

I also have difficulty at directly measuring current (DMMs: 8840A, 3457A).  When I shunted the current output across a 1.50k metal film resistor I got 1.50V, which divided out to be 1.00024A.  That's exactly what it says on the included sheet, so I'm suspecting I have a few blown fuses to replace in my DMMs...  I'll investigate later.

The only other IC on this board is U2, which generates the positive voltage for the current source.

• Pins 1 and 5 are physically cut off.
• Pins 7 and 8 are Vcc (8.8V: 9V batt minus a diode drop), Pin 4 is ground
• Pin 6 is the output to the DMM (then through a 5K 1% resistor to ground).
• Pin 2 samples the 5K resistor's voltage.
• Pin 3 receives the +5.0000V reference voltage.

I haven't used anything with pinouts that look like that but it doesn't seem too crazy... not worth scraping the numbers off anyway.

Just for fun, check out the torturous route C2's ground takes. 

[rexxar]

Just for fun, check out the torturous route C2's ground takes. 

That's amusing. Surely they could have moved the trace for pin 2 over a few mil 

[bronson]

You'd certainly think so.

Dave's uCurrent gold shows the current at 1.0001 mA on the uA range.     I'm guessing the fault is in the amp ranges of my multimeters.  Lovely.  Might spend a few hours this weekend figuring out what obscure fuses I need.

Kinda surprised they didn't wash the flux off the DMMCheck before calibration...  I'll have to take a bunch of readings, scrub it, and see if it makes any difference.  Good thing I'm still in my 2 year re-cal period.

[saturation]

How much accuracy do you need?  You can check for old model decade boxes from GenRad or similar via eBay or some new versions:

http://www.precisionresistor.com/MC-7-Calibrator.html

Although these resistors are 0.02% OOB you can calibrate them against a reference ohmmeter to define their actual resistances, check their tempcos , and finally monitor them over a year to see if the specs are up to par.

[echen1024]

Well, a 5K 0.005% is $30.42 on DigiKey.

[Galaxyrise]

(one complaint: the 1ma ac feature seems to baffle some meters.  on a 6.5dig HP and also on my keithley 6.5, the 1ma ac test signal almost can't be read.  all the other test outputs are fine and bang-on, with most of my meters.)

Odd, I don't recall any problems on my 34401a or my keithley 2000.  Try explicitly putting it into the low range.

[Vgkid]

For the prc the 1/10 ohm are 200ppm, while the rest are 100 ppm, you can ask for more accurate resistors(probaly not for the calibration set). They were selling 250 ohm .005% 5ppm/c resistors for around $7 ish over the summer.

[echen1024]

I will be using a Hammond Manufacturing watertight enclosure, as well as a simple 555 thermistor type control unit. Shouldn't be too complex. This will allow for relatively high tempco resistors to be used, but for the sake of producing something VERY accurate, I might use a 00.01 or 0.0025% Vishay resistor.

[larry42]

While looking for a good way to test my DMM, I have decided to build a precision resistor circuit. I plan to surround it in mineral oil, with a small resistive heater to the side. Are there any possible improvements I could make?

Sounds like a bad idea to put things in oil (you never know about the contaminants). Moreover, what is the tempco of these resistors, or are you working in the arctic?

BTW how much precision do all you people need? Is everyone on this forum working at a NIST cal lab?

Buy a few 0.01% resistors from DK or Farnell (or Y145310K0000V9L , Vishay Zfoil if you want to get really fancy)- I've worked as lead engineer for 15 years, designing stuff from subsea to millimeter-waves and have I have never needed to know resistances to better than 0.1%.

[linux-works]

BTW how much precision do all you people need? Is everyone on this forum working at a NIST cal lab?

its a mental disease many of us have.  the technical term is 'precision envy'.

"sometimes, a probe is just a probe.  but sometimes, not"

lol

[Fsck]

How much accuracy do you need? 

need? not that much really. I doubt more than 5% of the projects done here (not at place of employment!, just hobbies) require a 3 sigma error of less than 50ppm.
want? to have our 5.5-7.5d meters dead on.

[echen1024]

While looking for a good way to test my DMM, I have decided to build a precision resistor circuit. I plan to surround it in mineral oil, with a small resistive heater to the side. Are there any possible improvements I could make?

Sounds like a bad idea to put things in oil (you never know about the contaminants). Moreover, what is the tempco of these resistors, or are you working in the arctic?

BTW how much precision do all you people need? Is everyone on this forum working at a NIST cal lab?

Buy a few 0.01% resistors from DK or Farnell (or Y145310K0000V9L , Vishay Zfoil if you want to get really fancy)- I've worked as lead engineer for 15 years, designing stuff from subsea to millimeter-waves and have I have never needed to know resistances to better than 0.1%.

Precision is a big you WILL catch.

[larry42]

Precision? I prefer accuracy.

And then you can start to worry about stability and traceability.

Methinks that one would do well to listen to the advice of the more senior people in this thread...

The accuracy of components available off the shelf today is incredible compared to just a decade ago.





Sent from my One S using Tapatalk