what is the deal with DC shorts and offset

[coppercone2]

So I have two shorts I made up for my newly repaired Keithley 2001

One is a double banana jack (pomona) with a piece of 14 AWG copper bare wire (from the wall) soldered across the turret terminals. This one gets close to zero in the meter.

I just made another one that I thought would be better, this time from 12 AWG soft magnet wire that is soldered into two banana jacks to make a U shape. These are the banana jacks that have the screw on plastic bit. I soldered it maybe 1/2 hour ago. The offset from this one is massive (3.6uV instead of 100nV). If I reverse the polarity, the polarity of the offset changes. This makes me think thermocouple. The wire is bent into a U and soldered into the well.

Is this just because the heavy assembly without plastic is sinking heat away from the meter and making a thermocouple? Or is it from the plating type? The solder is the same on both of them.

35 min later:
This one just does not want to settle, even after 30 min. I thought it was better made because I don't know how the pomona solder turrets are mounted on the dual connector. I had them basically "spin" before on a damaged connector, which made me think it might be some mechanical seat (not one piece)....

It was at 3.4 microvolts after settling for 1/2 hour in the meter. I just put the pomona one back in, and it settled to 0.2uV immediately.


Am I correct in thinking that this thing is too good of a heat sink for the front panel that its making a inordinate amount of heat come out of the front of the meter? because there is air flow around the U shaped wire, where with the other one it has a plastic part blocking air flow in the middle of the arc. ?

I am going to try it in the rear connector. I thought to pot it in a container of clear epoxy mixed with fiberglass too (low thermal conductivity), like a matchbox, so all the exposed metal is covered.



What would a good one look like? I thought to bend egg-beater terminals out of fine silver plated spring copper wire, since it carries no current. Would that be good? It should have a high degree of thermal isolation from the banana receptacle body.   



Hmm its doing alot better on the back of the meter. I think its because its a U shape and it was hanigng off the table making alot of air flow. I stuffed a sock in the U and on the back it already dropped down to 1uV. Actally I wonder what would happen if you mix epoxy and like powdered or sanded  styrofoam, like with a rasp. The pink one that is really good.

[coppercone2]

oK SO Like the room temperature was swinging and everything was going wild

What I did is I got a piece of silver plated spring brass (not copper) about the thickness of a good resistor lead, and bent a [ clip out of it with like P's on each end of the [ so it kinda sits in the banana hole (using round pliers and carefully adjusting it until it fit snug but not tearing it up, and I wrapped the exposed part of the [ with a cotton cloth.

High precision mode with filter set to 100 counts (not advanced filter), it gets ~000.0002 with it going as low as 000.00003 sometimes, I can feel room temperature swinging a little because of computer use.


I think I have proper BeCu spring that is silver plated thick some where to make another one thats better then brass of the same kind, and it will get a small styrofoam grip and insulation.


Lol, I have all these different jumpers and I can't tell. It could just be the meter fooling me. My big jumper with the 12awg wire is definately bad, but as for the other ones I can't really tell the difference. Maybe the thin brass one seems like the best... it seems like it jumps around the least, but I thought the spring copper one would be the best but it looks like it jumps more. I guess I would need computer data logging and long time periods and enviromental control and what else to actually see there is no difference between any of them ?


Though, I tried a cigarette lighter test. The BeCu does less offset. But its also thicker. I need the same dimension rod to compare. But it went less even when I held it alot longer then the brass does just for a split second but thats not scientific. At least its fun to stick music note shapes into a meter.

[PascalNE]

Apologies if Im misunderstanding but does your 12AWG one have different banana jacks (non-pomona) so maybe they are causing offsets. And I think you’re right the 12AWG will have a pretty big thermal mass and when it’s acting like a heat sink it’ll cause bigger temperature differences. I think protecting it from air movement as much as possible would be good.

[coppercone2]

I don't know. I think it was pomona. But they are like bulky in comparison.

Actually I think the brass spring might work the best.

this shape out of a spring (but the jumper in the middle part is only 1 wire, not two



I have some really fine BeCu silver plate wire that takes the cake on thickness for all of them, I might try it tommorow. Does it make sense that the really thin wire equilibriates the best at the contact?

Because the current is so small... only temperature matters?

[PascalNE]

Not really my area of expertise but I’d assume that thinner wire will reach a temperature equilibrium much faster than a big piece of copper. Super low thermal EMF cables are often made with thin 22AWG wire. And maybe poking the wire into the hole is making better contact than your banana plugs were reducing resistance and reaching equilibrium faster.

[coppercone2]

Well, its very hard to see any difference between the dual banana pomona and the wire notes. But my DIY dual banana is attrocious.

I will throw it in the ultrasonic maybe, the contacts are a little oxidized on that one. But it measures OK.

I think the fire test might point to the copper one being the best material.

The tiny spring copper I have is like something less then 32AWG but it should be springy enough for a good contact if you bend it right.

It's called a instrument, putting notes into it just makes sense 

[Kleinstein]

Surface layers or plating have very little effect. There is just not much thermal gradient over the thin layer. HIghly oxidized contacts could still be bad from acting as a rectifier for EMI.

There are 2 possible ways to get thermal EMF errors: one a thermal gradient over the connector part and the other it drawing heat away from the usually warmer meter and than causing a change in the temperature gradient inside the meter, e.g. at the sockets.
For the 2nd effect it can help to cover the connector part with some cloth for extra thermal isolation.

[zlymex]

I use these for many years.

[coppercone2]

Hmm those look really thick.

Could it really be a corrosion rectifier then...


I did not want to try to stick thick bent wire in there because I thought it might scratch things. If you bend thin wire like a 'note' then only polished rounded wire is being pushed into the banana receptacle. I will just try a clean shiney 14 awg piece of copper, lightly bent into a simple U , without the connectors 

[tggzzz]

Have a look at this conversation...
https://www.eevblog.com/forum/testgear/why-rotary-tip-on-4mm-banana-plugs/msg5634873/#msg5634873

[coppercone2]

The contact area increases thermal transfer though, unless you have thinner contact leafs. To me it seems like you should be using something different for no current measurements. If you solder them it will eventually break.

What I had in mind is to make egg beaters out of the 34 or whatever gauge spring wire, then silver braze two of them together in a [ shape. But I am not good enough with wire yet to make that. The contact should be good and the thermal transfer should be neglegible.


That's why I thought brass too, because the thermal transfer is worse. Same with BeCu. But there is thermocouples then.

Stainless also has very low thermal conductivity, with titanium being around there too.


Other idea  was to make mini stainless egg beaters and dip them in my crucible 'silverer' but it would loose spring.... it would need silver electroplating (which I don't want to get into). 


But I have seen those banana connectors before and they probobly do have very low thermal mass. The pipe ones



Another option is to have a helix leaf spring with a 'thing' in the middle, so you can hold it, "wind it" (and pull on it), insert it into the connector and have it expand in there. When you twist a helix leaf spring it should shrink a little in diameter . This one would need some kind of 'key' for installation and removal though. And it would have a single fine wire inside bonded to the spring, so the spring is the contact, but the contact weld is deep inside of the spring (near the depth of the bore)

Kind of like a reusable heli-coil


Low mass design gets ridiculous and annoying


I am sure the metrology guys would love to have a micro sized 'utilities key' for installing DC shorts.
https://inspectoroutlet.com/products/sewer-cap-wrench (this one is a meter long)


https://www.yamamoto-ms.co.jp/en/product/b-72-p14/ - spiral contact for testing electroplating (inspired the idea)
https://specialtytest.com/product/spiral-contractometer-includes-two-calibration-weights-0-5oz-one-teflon-coated-spiral-instructional-dvdextra-wire-repair-kit/ - device
- instructions




Oh yeah and the entire assembly would be two of these special micro leaf helix springs, connected by something like a 30AWG wire that has two styrofoam plugs on it (which you insert into the banana connector when you insert it), and then a tiny two piece Styrofoam clam shell that you put over the tiny wire loop after its installed (with micro magnets), and you would need to wait 30 minutes for it to equilibrate before you measure zero. I think it would make employees happy and calm to use this robust assembly.  Just thinking about it makes me want to reject technology and go to a tibetan monestary

[Simmed]

found it
there is this thread about zero ohm


https://www.eevblog.com/forum/metrology/zero-ohm-diy-4-wire-standard/?all

[coppercone2]

Well I did find a electroplating recipe for mu-metal that you can deposit on the styrofoam insulator that protects the wire 


but realistically you should have a heat flux meter built into the DMM front panel so you can calibrate it for the expected test leads that will be used 

For those interested, the pink hard styrofoam in home depot is one of the best insulators you can get


I actually had to park my ass in front of the Fluke MFC before for a few days to measure things, and I assure you the laughter is purely sadistic with thoughts about the poor operator that has to do this stuff. The thermocouple calibration stuff was the worst.

ALSO with how SLOW it goes, even with the helper programs, what fluke should do on the next one, is make the machine quiet. Fucking NO ONE wants to be near that thing. Seriously the people that are actually interested in nano volts and stuff are NOT people that like machine fan noises. I really think that maybe they think companies are hiring former blacksmiths or heavy machine gun operators for doing calibrations ?? ( I always imagined that the guy that approved it might have been involved with using the 16 inch cannons on the USS Missouri). I am not even joking, I feel like I see ex-navy people around this stuff ALOT. Like ok, I know why you are questioning my ability to tolerate noise, its because you are half deaf from working on the flight deck on an aircraft carrier.


I did kind of like seeing what the meters are really measuring, but the noise is not tolerable.


I did put a noctua fan in my k2001 BTW

[TUMEMBER]

I use these for many years.

This is an economical and common sense solution. I opted for the "clover", it worked for "zeroOhm" and "zeroVolt". The photo shows how easy it was to prove what a "shit" the "leading product of the world's leading industrial hegemon" turned out to be at some point. For those who are visually impaired, the range is 100 mV, slow, full short circuit. And several hundred thermocouples were measured with something like that.
It is enough to "bathe" copper wire in citric acid from time to time and rinse it in distilled water, and it will remain without an oxidized layer.

[TUMEMBER]

There are 2 possible ways to get thermal EMF errors: one a thermal gradient over the connector part and the other it drawing heat away from the usually warmer meter and than causing a change in the temperature gradient inside the meter, e.g. at the sockets.
For the 2nd effect it can help to cover the connector part with some cloth for extra thermal isolation.

You can try to measure the socket temperature using a PT100 sensor. The diameter must be matched closely to the socket diameter, i.e. 4 mm, and at the same time it must be a "short" sensor, i.e. the length of the socket channel, to prevent "extraction" of heat through the "jacket" of the temperature sensor. Unfortunately, each meter will have a different time constant and it will change with the change of the ambient temperature in the laboratory. But the order of magnitude itself can be determined. In my case, the SOLARTRON 7062 is a task for Herakles, it has an exceptionally "thought-out FISHER connector" - their ass should swell for this choice.

[coppercone2]

are you sure those are good the fluke one is a acrylic block on a circuit board with tiny traces for copper.  I think that your copper wire flower is losing alot of heat. I feel the heat on the side vent of the keithley and the top

I put two notes in the sense and source terminals and hung a third note in the middle (so its just hanging on the bits that are hanging out, and the offset for 4W  is 80 micro ohms. I hung a transistor off it to weigh down the contact and its staying mostly the same. (0.75 inches of fine wire and 2 silver contacts). That is really bootleg but its low thermal mass, im not going to solder today, but I think I might try to make a single piece one with a dot of solder in the middle to try to make a real kelvin short that is also very thin (like a tripod with 4 legs). I can use silver to bond it in the middle instead of solder too.

like this, but with loops on the legs so it stays in place



IDK if it makes sense to compare 4 wire ohms shunts, 2 wire ohm shunts and 2 wire voltage shunts? (depending on current, but volts is really low). Well, I doubt you can get the stuff thin enough to matter 

[coppercone2]

how does your flower clover shunt do in another meter?

[coppercone2]

I got a piece of 14 awg zinc wire that I just bent into a simple U and shoved into the banana jacks because it came in the mail

still does better then the banana connector one. Like alot. The zinc wire, copper wire, brass wire, pomona plug are all very similar.

The meter has been on a long time, and I think that the copper one is the best


Lol, this banana plug could troll someone to hell. its basically a power source. I wonder if it can charge a capacitor
its free energy machine

[mawyatt]

Here's a 4 Wire Short/Shunt with #12 Solid Cu Wire we use, and a Precision Resistor with a Thermistor attached by heat shrink tubing.

We found that with our KS34465A DMMs show a temperature rise of ~2C on the thermistor, other meters such as DMM6500 showed less rise. This indicates the KS DMM Banana Sockets are at least 2C above ambient, likely 3~4C at the socket inside the DMM.

Anyway, they are for use with our multiple 6 1/2 digit meters.

Best 

[coppercone2]

I put a small leatherman in there and it works way better then the bad banana jumper

ACtually it drifts more, so the banana plug is better

ima stop before I have to buy new sockets

someone should not plug a leatherman into a 3458a


(I put two steel rods into the front panel and lightly wedged a leatherman in there to act as the DUT)

Offset is 5mV. bad banana is like 4mV, copper is 0.01uV. Leathermans down to 3mV so its like the bad bananas


Man the leatherman with the keithley electrometer range switch bolts is beating out my copper short. I thought this short can handle like 30 amps.

How the fuck did I manage to build something so impressively bad. I was proud of this thing! And god damn zinc wire too.

[TUMEMBER]

It doesn't work because such a solution is not applicable in my private meter. K2700 was a service meter at work - I don't work there anymore and it's not my problem. Explanation in pictures. What's the harm in trying the clover? The cost is small, a bit of work with pliers to bend it into the right shape to fit the meter sockets. K2700 had an advantage - it didn't have a fan, although it's only 6 1/2 digits.

[coppercone2]

I don't think you understood what i said. I just wanted to know if it worked in another meter

what happens if you plug it into another meter?

[TUMEMBER]

I don't think you understood what i said. I just wanted to know if it worked in another meter

what happens if you plug it into another meter?

It seems that I wrote it quite "legibly". Not everyone is an American and has 10 DMM at home for their own needs. These are expensive toys, in Poland you have to consider whether to spend it on a used car or a gadget. My next meter does not meet the criterion of accuracy and reproducibility in this type of measurement, because it is SOLARTRON 7150. Find its metrological parameters and draw conclusions. I am "registering from the topic".

[coppercone2]

I just want to know if that clover is actually a good 4-wire shunt or not. I left that meter on with my bad banana shunt for many hours now and its just not gonna settle. I thought the shunt would be great because it looks beefy, well soldered, low impedance. If I made a judgement on the K2001 with that bad shunt I would be wrong (because I tried it with good shunts). At most I can say that maybe the keithley has a thermal problem where it does not like measuring voltage with heavy gauge wires. (which is a interesting observation because naturally people might just got for 12AWG test leads for everything and not realize there is a potential benefit in some meters for say having a different pair of thin gauge leads for measuring voltage).


If you actually put that shunt in another meter that is similarly accurate and it worked it would at least tell us something about it. All I know right now is that for sure it does not agree with that meter, not if the meter is bad, because the design is unorthodox and contrary to things I hear (and saw with purchased calibration artifacts) about low thermal mass being the 'norm'

My 34401A reads the bad shunt that the keithley disagrees with as good (2uV offset) ,and if I plug in my low thermal mass shunts I also get 2uV offsets. Something is different about it, but I need to let it warm up for 24h before I make any conclusions. Things are slow to settle but its peculiar that the same shunt reads say 5uV in the K2001 and only 500nV in the 34401A (on power up 15 min after being off for a while). I am curious to see how things will look after an hour,. I tried to set the filters so they update similar intervals. I see it started negative and its moving up. For all I know in a few hours it will drift up to +5uV and be just as bad as the keithley, then I really know that its not a great shunt if it causes problems in two different DMM. But This thing looks BOSS for say shorting out a power supply



The difference is that the 34401A has a soldered or brazed copper pipework for the front panel, and the K2001 has little panel mount connectors with small pins (2mm IIRC) that go to a teflon wire. The k2001 is totally different then 34401A (which is basically a fucking heat exchanger for a nuclear reactor)




It looks like because HP used heavy hardware for the front panel instead of little crap, it looks more stable. (and this is a hallmark of keithley where they won't make any parts for themselves that are bad ass). Maybe keithley needs to buy a 500W iron for their factory. Or maybe there is something more to it

https://www.ftm-technologies.com/produit/600w-heavy-duty-soldering-station/?lang=en

Though, given that the 34470A also uses heavy pipe work for the front panel, I think keithley is just cheap. 

[coppercone2]

Hmm after a good half hour now, the little note jumper reads as 300nV on 34401A and the bad banana reads as 3uV

its bad even with the big pipes. its just bad 

some how it looks like I have soldered together evil

And 5uV. I pulled it out and put in the note and it fell down to 500nV

I don't have a third manufacturers meter but I will just say that this jumper is... bad unless you are using it to short a DC power supply or something.


I think I will give the leaf a going over with some polish to see if it improves, and then try the ultrasonic cleaner. Just scrubbing the bananas with IPA had no effect. Maybe I can try some kind of acid too (I need to see which acid works for what I am guessing is nickel), I don';t mind pickling the whole part to see if it improves it.

A rather substantial amount of crud came off the banana from rotary polishing with a buff and compound. Maybe offset went down a little. I wonder if there is crud on the leaf pole interface. Ultrasonic pickling should take care of that. Now its hanging out at 2.8uV. The towel is on it but the room is getting drafty from a outside wind rain

The copper wire was from a big ass renco inductor BTW


Nope, its back to 4uV.  and the note is 1uV , maybe the meter drifted a little its really starting to get to what people might say is stable zone for 34401A


I gave the bad one 15 minutes in the ultrasonic with alconox. Still the same. I am out of citric acid to try to pickle it