Offset compensated Ohms adaptor (precision low Ohms measurement)

[robrenz]

I finaly figured out that I could do offset compensated ohms on my Fluke 8846A even though it does not have that feature. I called Fluke tech support to ask them if my technique was valid. I had it right except I did not realize that I needed to short the source leads to each other not just disconnect one source lead.

This is a video on details of low ohms measurement and construction of an Offset Ohms adaptor.  This adaptor should work on most 4 wire meters.

[free_electron]

Cool video. One little problem.... Silicone wire ? For low level measurements ? Ehhhh. Go read the chapter in the red keithley book on triboelectric effects.....

[robrenz]

Cool video. One little problem.... Silicone wire ? For low level measurements ? Ehhhh. Go read the chapter in the red keithley book on triboelectric effects.....

Thanks, I know silicone is on the far end of the triboelectric scale.  I just happened to have those kelvin clips with the silicone wire from the current reversal adapter that I showed in the video. I checked what kind of readings a very vigorous massaging of the leads would produce and it is only 50 or 60 uOhm.  In use since this must zeroed out with each change of connections and the leads will be still during measurement I don't think it will be an issue.  I am no expert and am happy to be corrected.

[codeboy2k]

Great video, thanks!

Wouldn't you always want to measure the voltage across the resistor at a point close to the body of the resistor, and not at the ends of the leads?

It's obvious that the 2 leads together add about 2.7mOhm to the measurement. I suppose there is a chance that the points near the body are hotter than the points near the ends of the leads, due to resistor self-heating, but the error due to the thermoelectric effect there would certainly be less than the error due to lead resistance when measured from the end.  If that was a legitimate concern, you could move the measurement point slightly away from the resistor body, say about half a body's length away from each side, so your measurement point is not right at the body, but also not very far away from it either.

Do you agree, or have I missed something ? 

[robrenz]

Great video, thanks!

Wouldn't you always want to measure the voltage across the resistor at a point close to the body of the resistor, and not at the ends of the leads?

Not necessarily, at very low levels the lead resistance can be a significant portion of your reading so you want to measure so that it represents the actual application.  if you have a half inch of leads on each side in use then measure it that way.

It's obvious that the 2 leads together add about 2.7mOhm to the measurement. I suppose there is a chance that the points near the body are hotter than the points near the ends of the leads, due to resistor self-heating, but the error due to the thermoelectric effect there would certainly be less than the error due to lead resistance when measured from the end.  If that was a legitimate concern, you could move the measurement point slightly away from the resistor body, say about half a body's length away from each side, so your measurement point is not right at the body, but also not very far away from it either.

Do you agree, or have I missed something ?

If a very low resistance needs to be very accurate then it would typically be a 4 lead resistor. 2 leads for the current and 2 leads for the voltage sense.  That controls where you are measuring the voltage drop of the resistor.  Also you often want to know the effective resistance at operating temperature. So you would actually take a resistance reading with the resistor warmed up by running the anticipated current through it until the temp stabilizes and then measure the resistance.

[codeboy2k]

OK thanks.

I was looking at your video and talking about the out of circuit , raw measurement of the resistance.  i.e. measure close to the leads and maybe a bit away if you're concerned about thermal effects.  Yes, it makes sense to measure at a point along the leads that represents how it would actually be used in circuit, and to bring it up to operating temperature.  Actually, if I am making something that precise and I need to know, I would often try to know how it behaves at the two operating temperature extremes.  Then I can design properly for that range of variations.

I'm aware of the heating effects, and until you've actually seen it yourself, it's almost something you don't ever think about, and then once you've seen it, or it's messed you up one time, then you can't stop thinking about it.   I've seen resistances change when a person walks by in the room and the draft blows in.   that was a shocker to me when I saw it for the first time.  Whether that matters or not depends on the application.

If a very low resistance needs to be very accurate then it would typically be a 4 lead resistor. 2 leads for the current and 2 leads for the voltage sense.  That controls where you are measuring the voltage drop of the resistor.  Also you often want to know the effective resistance at operating temperature. So you would actually take a resistance reading with the resistor warmed up by running the anticipated current through it until the temp stabilizes and then measure the resistance.

Yes, good points to make. thanks!

[T4P]

I'm aware of the heating effects, and until you've actually seen it yourself, it's almost something you don't ever think about, and then once you've seen it, or it's messed you up one time, then you can't stop thinking about it.   I've seen resistances change when a person walks by in the room and the draft blows in.   that was a shocker to me when I saw it for the first time.  Whether that matters or not depends on the application.

Must have been a carbon composition resistor 

[robrenz]

A very generous EEVblog forum member is donating some low loss cable to minimize the triboelectric effects. When I get it I will compare its performance to the silicone insulated leads and post the results.

[SeanB]

You never worked with precision balances, even breathing near them will produce an offset. Some are so sensitive just holding your finger near the mass pan will show due to radiated IR from your body heat. You know that they are sensitive when you have large massive stone tiles bases to hold them on a table, which is firmly bolted to the floor and the wall. I have a check mass for them, it is 1.0000g, with an error on the verification of 50 milligrams from that value. I do not touch it at all, handling with tweezers only. I was asked if I wanted the smaller masspieces verified, but decided that it would be not effective to verify a 100mg, 10 and a 5mg foil mass. I use them on the 200g masspiece during calibration on one massmeter as the masspiece is 130mg low.

[MBY]

Excellent video, robrenz! Now I know what to build when my PD2005 heater project is finished!

[robrenz]

Thanks MBY,  I have many more in the works.

EDIT:
BOM of what I used
Hammond Manufacturing 1551PBK  Enclosure; Box-Lid; Handheld; ABS, UL94HB; Black; 1.58x1.58x0.79In; 1551Series

Abbatron / HH Smith 425AA  Plug, Banana; 15; Brass; Beryllium Copper; GOLD plated
 
C&K / ITT / Cannon  8168SHZQE2  Switch, PNL-MNT, Pushbutton, SPDT, ON-MOM., 6A@120VAC OR 28VDC; 3A@250VAC

ebay kelvin clips http://www.ebay.com/itm/NEW-Mueller-BU-75K-10-amp-electric-kelvin-test-clip-/160507941763?pt=LH_DefaultDomain_0&hash=item255f04d383 These are actual Meuller brand. not a chinese copy.

[saturation]

Great work, robrenz, when done, please keep us posted with the new cable performance.

[DavidDLC]

Great video, thanks!

Wouldn't you always want to measure the voltage across the resistor at a point close to the body of the resistor, and not at the ends of the leads?

It's obvious that the 2 leads together add about 2.7mOhm to the measurement. I suppose there is a chance that the points near the body are hotter than the points near the ends of the leads, due to resistor self-heating, but the error due to the thermoelectric effect there would certainly be less than the error due to lead resistance when measured from the end.  If that was a legitimate concern, you could move the measurement point slightly away from the resistor body, say about half a body's length away from each side, so your measurement point is not right at the body, but also not very far away from it either.

Do you agree, or have I missed something ?

I will measure it on a point in the leads close where I will solder it !

[free_electron]

resistors can be sensitive to all kinds of stuff..
try repairing a keithley electrometer.. you'll be tearing your hair out ...

i had one fith a fried input. the dual j-fet in there is solid unobtainium... ( i have the last three ones in existence , so if anyone needs one  )
and the resistors are all glass fused .. there is one 100 Gigaohm resistor with 1% tolerance. get a greasy finger on its outer body and kiss it goodbye ...
you can't even clean that sucker with alcohol...

If you have to solder anything in this circuit you need to wait 4 hours before doing the calibration to make sure there is no more thermal gradient...

the service manual of these machines is actually funny to read...

required equipment :- tera-ohm meter ... whathefu ?
do not touch the body of the resistor to avoid contaminating it....
when you adjust the zero balance ater touching the trimpot you need to wait several minutes ( suggested : 15 minutes ) to let the input settle. the fact you poked a trimmer key ( one of those plasitc trimmy pot turners ) through the calibration hole and touched the trimmer has created enough charge differential to upset the input stage ... this needs to disperse first ...

they are an absolute pain to repair...
oh , and keep one hand in your pocket ... the whole input stage floats between -200 and +200 volts.... no touchy !

i have one . maybe i'll do a teardown. it is very interesting equipment. this thing is so sensitive it can actually count electrons...


And capacitors... those suckers also have very funny properties... microphony for example : tap on a ceramic cap and it produces voltage. becasue its dielectricum is actually piezo material.... send ac current trhought them and you can hear it ....
solder them so the dielectricum goes above curie point ( 147 degrees C for the barium based dielectricum in classic x7r style caps ) and the capacitance 'reset's. it will drift to a rate of 1 to 2% per year ....

So forget about using these in a precision oscillator or filter. the fact that you soldered it in means you can wait 4 or 5 years for the sucker to stabilize ...
so it;s funny to see people trying to make superstable oscillators and use class-ii caps as part of the timing element...

[codeboy2k]

i have one . maybe i'll do a teardown. it is very interesting equipment. this thing is so sensitive it can actually count electrons...

but then... will you F- it up if you do    I would love to see it but only tear down a broken one

And capacitors... those suckers also have very funny properties... microphony for example : tap on a ceramic cap and it produces voltage. becasue its dielectricum is actually piezo material.... send ac current trhought them and you can hear it ....
solder them so the dielectricum goes above curie point ( 147 degrees C for the barium based dielectricum in classic x7r style caps ) and the capacitance 'reset's. it will drift to a rate of 1 to 2% per year ....

So forget about using these in a precision oscillator or filter. the fact that you soldered it in means you can wait 4 or 5 years for the sucker to stabilize ...
so it;s funny to see people trying to make superstable oscillators and use class-ii caps as part of the timing element...

I know about some of those properties of capacitors, but not everything. For example, since you mentioned it, I do want to build a precision oscillator one day. I want digital precision to input the exact frequency, and with a true analog sine output, not DDS.  I want a real analog VCO, with feed back to a digital counter to maintain the precision of the loop. I haven't started anything, and I don't know what to expect from it, but I am pretty sure there would only be silver mica caps. At least that was my first choice, not sure (as I haven't designed anything yet) if it's actually still the best choice. And there are better (more expensive) caps.

I'm no expect in super stable, ultra-precision electronics, but one of the best things about Dave's blog is that there are engineers here that are experts in this field and that they share their knowledge and experience in their respective fields.  And I love to read it.  I've been doing analog electronics, digital electronics,  RF (but not gigahertz stuff) , antenna design, embedded systems design, FPGAs, software design, etc. for over 25 years now.  And I still see things on this board that I have not come across before, and I learn from it all.

Cheers!

[alm]

i have one . maybe i'll do a teardown. it is very interesting equipment. this thing is so sensitive it can actually count electrons...

Calibrating them is also fun. Please set your resistance source to 1 Tohm +/- 0.5%. Now take your current source and set it for 10 pA +/- 0.5%. Please feed it 624.2 million electrons and check for -100.0 pC reading.

[free_electron]

exactly ...
Those machines are totally bonkers.

I'm not afraid of breaking it. i know how to repair them . mine was a freebie... becasue it was shot. i spent more than a year trying to find the damn dual jfet in the input stage...
the store that had them had 4 left so i bought all 4  .
taking the cover off and the shield plate doesn't upset it. touching things or bending a wire does upset it.

The machines backplate actually has big warning labels on it not to make certain connections or leave certain switches open as it can fry the input.... and that is how they die.
Electrometers are not equipment to be handed to junior engineers or people that are not in a lab frequently enough... you need to know what you are doing.


It's like the bolometer RF power sensor that HP has. They frequently show up for 10$ on ebay and people try to repair them  ... most frequently a little gold wire between the hot side and the input stage burns out because of power overload.
1) You can't solder that... tin/lead solder creates thermal noise and this thing measures thermal delta ! so you need a wire bonder...
2) if you solder it i t needs 4 hours to thermally stabilise before you can try to align it ( i have the service manual. the 4 hours is mentioned in there ... )
3) The circuit board uses very stiff wires to connect to the output connector. Calibration of the thing involves bending the wires until it reads correctly....

So the first mistake people make is to take the cover off and unnmount the pcb from the case ... by desoldering the wires one by one and bending all the wires out of the way... you may as well throw it away then. No chance you ever get it to be close where it needs to be ...

[robrenz]

So how exactly am I going to make a meaningfull comparison of the silicone leads to the low noise cables?
Other than massaging the cables after zeroing and seeing what readings I get on the 8846A all things that come to mind in my limited knowledge are going to be full of other noise that I wont be able to separate out.  All ideas/input is greatly appreciated.

Edit:
I found these two articles that after a brief scan lead me to believe the " massaging the cables after zeroing and seeing what readings I get" method may be the only practical method at my disposal. Since I am more interested in its effects on the measurement value than quantifying the amount of charge developed.

This shows homemade coulomb meter attachment. http://www.dicks-website.eu/coulombmeter/enindex.htm

[robrenz]

Bump.