EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: DiligentMinds.com on June 13, 2013, 05:49:48 pm
-
.
-
Hello Ken,
I have done a lot of research, tests and measurements about low EMF connections.
https://www.eevblog.com/forum/buysellwanted/diy-low-emf-cable-and-connectors/msg190302/#msg190302 (https://www.eevblog.com/forum/buysellwanted/diy-low-emf-cable-and-connectors/msg190302/#msg190302)
I also looked for an adapter for my Tek DMM 4050 and have only found and bought the 884x-short. If you had the same result, you probably have to build/modify something yourself. Good candidate CuTe binding posts are from Pomona (3770), Multi-Contact (PK4-TS), IET (BP1000) and jswilley (2758). I also tested banana plugs with screws. The Multi-Contact (LS425-AM and LS425-SE/M) are not CuTe, but nevertheless they showed very good results.
Bye
quarks
-
I made a super short pictured here. .625 diameter tellurium copper bar with these (http://www.alliedelec.com/search/productdetail.aspx?SKU=70209777)gold plated beryllium copper banana jacks.
But all the jacks I have found including these only the spring element is beryllium copper, the bodies are gold plated brass. But the total min max span of a 152 hour shorted input stability test on my 8846A using this was only 1µV. There was a 2.4 degree C temperature variation over that time so I don't think thermal emf's of the gold plated brass were a big issue. I think in this case a larger mass short like this is more stable because it is influenced less by short term temperature swings.
(https://www.eevblog.com/forum/beginners/looking-for-goodinexpensive-bench-power-supply/?action=dlattach;attach=23367)
-
for fast equilibrium you have to have as low as possible mass. So the plug comes as fast as possible to the same temp as the connector of the DMM.There is a very good Fluke paper with the Fluke 8508A and I have done my tests exactly how it is shown in this paper.
-
I read an article about a similar thing (thermoelectric error). Higher mass shorts tend to introduce EMF errors, check it out, the relevant article starts on page 20 IIRC. In the article, one of the best shorts they tried was just a small piece of solid copper telephone gage wire.
http://www.callabmag.com/wp-content/uploads/2012/06/apr12.pdf (http://www.callabmag.com/wp-content/uploads/2012/06/apr12.pdf)
-
That is exactly the article I have downloaded from the Fluke site.
http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint (http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint)
att. is a picture of some of the tested connectors
-
I am familiar with that article and it is not stating that a large mass ohmic zero is inherently less accurate but just that it takes a long time to thermally stabilize with the meter. A low mass "Thermal zero" is only good if you have extremely low currents involved. From the article: "Ohmic zeros have very small actual resistance so the measurement value is truly a zero without any unwanted offsets caused by I times R losses based on the current supplied by the meter and the physical resistance of the shorting device. (For example, milliamps of current through milliohms of resistance create microvolts of unwanted measurement voltage.)" In This post (https://www.eevblog.com/forum/testgear/looking-for-goodinexpensive-bench-power-supply/msg105028/#msg105028) and several following it I discovered that the resistance of an "Ohmic zero" is a big deal if high currents are involved as in testing the compliance voltage of a power supply in CC mode. Since time to stabilize is not an issue for me and that once stabilized it is less influenced by any short term temp fluctuations I use it for all testing.
-
I can make solid telurium copper dual or quad binding post to banana adapters. You could then plug this into a flush style meter and use all you low emf spade terminal items or wires in the binding post holes. Having them coupled in a dual or quad arangement will allow tightening of the binding post nuts.
If there is enough interest I can figure out what they would cost based on how many are needed.
-
hello robrenz,
that is a great idea. With my limited machinist skills, I would think of an adapter like this (see att.) and I would leave it always pluged in (unless/until I need the security plugs)
-
Yes like that but much shorter. The banana would stop at a piece of 3mm plastic that tied the posts (2 or 4) together and the binding post shoulder would start right at the other side. That would make it very close coupled so the binding post shoulder would only be about 4mm from the face of the meter. This is important so the banana jacks in the meter are not strained by a large lever arm like the one you showed. This would also be one piece copper construction where there is no separate spring contact on the banana jack. it would be multi finger with an elastomeric expander plug inside the fingers to provide the contact pressure.
-
that sounds very good to me.
Maybe that is also an idea, look at the att. picture. Inside the plug is a screw that will widen the slotted ends
-
I assume you are talking about the one on the right and it is too out of focus to tell what is going on.
-
That is exactly the article I have downloaded from the Fluke site.
http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint (http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint)
The article is kinda crap showing the sadly common misunderstanding of how thermal emfs are generated.
Thermal emfs are not generated across junctions - junctions are dead shorts. Having any significant temperature difference between two metals forming an electrical connection is equally impossible. The junction can be considered to be almost infinitesimally thin and so have almost infinitesimally small thermal resistance.
-
I assume you are talking about the one on the right and it is too out of focus to tell what is going on.
no they are both the same (just red and black).
They are just the reverse of what we talk about, because they are adapters for binding posts to be able to plug in security bananas. But the mechanisem inside could be also usefull for the wanted adapter.
-
That is exactly the article I have downloaded from the Fluke site.
http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint (http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint)
The article is kinda crap showing the sadly common misunderstanding of how thermal emfs are generated.
Thermal emfs are not generated across junctions - junctions are dead shorts. Having any significant temperature difference between two metals forming an electrical connection is equally impossible. The junction can be considered to be almost infinitesimally thin and so have almost infinitesimally small thermal resistance.
Would you be alluding to that thermal EMF is generated across thermal gradients rather than junctions?
-
That is exactly the article I have downloaded from the Fluke site.
http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint (http://us.flukecal.com/literature/electrical-calibration/watch-out-those-thermoelectric-voltages-cal-lab-journal-reprint)
The article is kinda crap showing the sadly common misunderstanding of how thermal emfs are generated.
Thermal emfs are not generated across junctions - junctions are dead shorts. Having any significant temperature difference between two metals forming an electrical connection is equally impossible. The junction can be considered to be almost infinitesimally thin and so have almost infinitesimally small thermal resistance.
Would you be alluding to that thermal EMF is generated across thermal gradients rather than junctions?
I am of the same opinion, from what I have studied the thermal gradient in a conductor is what causes an emf. the junction itself does not cause the emf. This does not contradict how thermocouples work. In my opinion you can ignore platings because they are so thin that the thermal gradient across the plating would be insignificant. But copper vs copper oxide show that there may be some kind of semiconductor effects of a junction itself. I have been meaning to bring this very topic up to get feedback from those more knowlegeable on the topic.
-
I assume you are talking about the one on the right and it is too out of focus to tell what is going on.
no they are both the same (just red and black).
They are just the reverse of what we talk about, because they are adapters for binding posts to be able to plug in security bananas. But the mechanisem inside could be also usefull for the wanted adapter.
I could make them to do the same thing with the expander screw acessable thru the banana jack of the binding post
-
I haven't studied the physics of the Seebeck effect at all, but my guess is that it has to do with how temperature affects how easy it is for electrons to enter the conductance band. The electrons have to jump to different energy levels as they transition between different metals, and I imagine this may be temperature dependent. If this is true, then even a coating of one atom thick should have an effect. This would also be consistent with the large effect of copper oxide.
You always need two junctions for thermocouples (which use exactly the same physics): the hot junction and the cold junction. Energy flows from the hot junction to the cold junction. The hot junction could for example be the copper-nickel junction at the positive input terminal, and the cold junction could be at the negative input terminal which is 0.5 K colder because it's directly connected to the case (random example). The current flows between the two junctions (terminals), causing an offset voltage.
-
Would you be alluding to that thermal EMF is generated across thermal gradients rather than junctions?
Thermal emf is generated everywhere within a conductor. Proportional to temperature (not linearly) and quantified by the seebeck coefficient of the material.
If one end of a conductor is hotter than the other there will be a net difference of thermal emf between the ends. The trouble is you can't measure it without connecting some more conductors which must also have some temperature difference between their ends and possibly different seebeck coefficients. Connections require junctions but the junctions have nothing to do with generation of the emf.
If you deliberately choose to connect conductors with different seebeck coefficients at the 'cold' end you see an emf due to the different seebeck coefficients which is proportional to the difference in temperature between the ends and you call it a thermocouple.
If you choose to connect conductors with the same seebeck coefficients then the emfs cancel out and you see nothing at the 'cold' end. If you use the same material you could argue there are no junctions but that isn't why you see no emf.
-
What Rufus said is exactly my understanding also.
-
This paper attached is what helped form my opinion. i will post others as soon as I find them.
@ DiligentMinds.com Would you like us to start another thread on this instead of crashing here?
-
I haven't studied the physics of the Seebeck effect at all, but my guess is that it has to do with how temperature affects how easy it is for electrons to enter the conductance band.
My non-physicist never studied it theory would be the hotter the material is the more the electrons move around. The average repulsive force between them is larger because moving around more they get closer to each other. Hot electrons want to occupy more space than cold ones.
-
just saw this
http://translate.google.de/translate?hl=de&sl=sk&tl=en&u=http%3A%2F%2Fwww.elsocz.cz%2Fvodice-hroty-svorky%2F2733-adapter-na-pripojenie-vodicov-do-4mm-zdierok-elso-gw4a.html (http://translate.google.de/translate?hl=de&sl=sk&tl=en&u=http%3A%2F%2Fwww.elsocz.cz%2Fvodice-hroty-svorky%2F2733-adapter-na-pripojenie-vodicov-do-4mm-zdierok-elso-gw4a.html)
-
They look nice but It does not say what the material under the gold plating is. And that goes back to my question of how much thermal EMF effect can the plating have at such thin layers and no one seems interested in discussing that topic :'(
-
can you just bend a piece of thick copper wire to make a U shape?
-
A few empirical points to support my argument that plating matters: High-end references and meters will often use a copper alloy instead of nickel or gold plated binding posts. I also remember a post on one of the test equipment related Yahoo! groups (hp_agilent?) about someone who asked the UK equivalent of NIST about what they used as cables for connecting to voltage standards. They used standard connectors, but send them off to chemically strip the nickel plating before use. This also argues that plating is a big deal for precision measurement. Cleaning copper connectors before use (or air-tight crimps) is important because copper oxide produces a substantial offset voltage. See also the Keithley low level measurements handbook (available for free from their website). You can also find some volt-nuts posts of people complaining about unstable readings with plated connectors.
Is it that plating is thick enough so a thermal gradient is developed that is large enough to produce a significant potential (note that metrology is dealing with µV and nV), or is the Kasap paper oversimplifying the underlying physics? I haven't been able to find anything to support the latter (yet). This book (section 8.08) (http://books.google.nl/books?id=IryMtwHHngIC&lpg=PA527&dq=seebeck%20effect&pg=PA531#v=onepage&q=seebeck%20effect&f=false) appears to do a decent job explaining the Seebeck effect, but with that model I don't see a reason either why copper-other-metal-copper would create any offset voltage if there is no temperature gradient across the other metal. So my conclusion is that there often is a temperature gradient across the plating ;).
Does anyone feel like estimating the temperature gradient that you can expect to occur across a thin layer of coating given the fairly small (a few K) difference in temperature across the total connector? 1 mK difference across nickel plating would result in 10 nV potential across the plating. Which is consistent with this post (http://www.febo.com/pipermail/volt-nuts/2011-October/001369.html). Of course a solid gold connector, or a connector with much thicker plating, may produce more voltage.
-
They look nice but It does not say what the material under the gold plating is. And that goes back to my question of how much thermal EMF effect can the plating have at such thin layers and no one seems interested in discussing that topic :'(
When looking at the offered price, I guess it is probably not a special "low emf" material.
About plating I am still very interested and made some more experiments with pure copper, copper-silver, copper-tin and copper-gold.
I was quite surpised, that tin plating does not seem to be bad at all, therefore I odered some tinned spade lugs and cables. But I have not found any Seebeck numbers on copper-tin, so if anyone finds some, please share.
Just to see how nickel plating will be, I also ordered special high temp cables.
As soon as I get these I can give an update if there is interest.
I also made an experiment with making my own crimp spade lugs out of pure copper (but failed so far).
-
1 mK difference across nickel plating would result in 10 nV potential across the plating.
That 10nV potential you speak of "across the plating" is that the potential in the plating only? If so then whilst 10nV sounds large what would be the net voltage generated given that the Seebeck coefficient of the material its attached to is not 0
-
10 nV would be the voltage between the two pieces of copper at both sides of the nickel. I got the number from a table with Seebeck coefficients relative to copper from the NI website (10 uV/K). Seebeck coefficients are usually quoted relative to platina, which is not that useful since platina wires are fairly rare in electronics.
-
Why do you buy low-EMF binding posts but insist on gold plating? Either you subscribe to the theory that plating matters for thermal EMF, in which case gold (over nickel?) plating is bad news. Or you assume that only a significant length of material will have a substantial impact on thermal EMF, in which case something cheap like nickel plated copper will be just fine.
If you go this route and worry about thermal EMF, I would recommend putting the binding posts in good thermal contact to keep the binding posts and plugs all at the same temperature. I'm quite sure this is discussed in the Keithley low level measurements handbook.
-
10 nV would be the voltage between the two pieces of copper at both sides of the nickel. I got the number from a table with Seebeck coefficients relative to copper from the NI website (10 uV/K). Seebeck coefficients are usually quoted relative to platina, which is not that useful since platina wires are fairly rare in electronics.
1 cm^2 of 25um Nickel plating with 1mK across it will be transferring about 1/3rd of a watt. 1/3rd of a watt will warm 10g of copper by 1mK in about 10ms.
Hard to see the plating being significant. The leads would have to be acting as heat sinks for the equipment and you are only going to see effects of differential heat sinking between leads.
-
1 cm^2 of 25um Nickel plating with 1mK across it will be transferring about 1/3rd of a watt. 1/3rd of a watt will warm 10g of copper by 1mK in about 10ms.
I agree that this sounds very unlikely, although I know some instruments where leads were substantially warmer than ambient. Those are not usually precision voltage references and multimeters, however. Still, people observe effects with plated wire and connectors. Is there something else going on that produces a thermal EMF due to the temperature differential between two jacks, even though both are mostly copper with a thin layer of plating?
This is why every high-end ultra precision instrument (that I know of) uses gold-plated CuTe binding posts.
But do these high-end instruments (eg. the Pomona low thermal EMF binding posts) use regular copper-nickel-gold plating, or an alternative that might produce less EMF? Gold over copper is quite good, for example, but the gold gets contaminated by the copper and eventually tarnishes. I'm sure there are other options.
-
All low emf posts/terminals and spade lugs I know of, I belive are direct gold plated or unplated. I do not think/hope there is nickel under the Gold plating.
-
As long as you can find direct gold over CuTe, like the Pomona binding posts presumably are, then I agree that would be a superior solution. But you can pretty much assume that any commercially available gold plated copper that is not expressly marketed as low EMF will have nickel plating in between. And that would mostly defeat the point of buying low thermal EMF binding posts.
If you are not able to find banana plugs without nickel plating, then another option might be chemically stripping nickel plated banana plugs. t involves some strong acids, so I wouldn't want to try it at home, but some people on the volt-nuts experimented with it. You may also find a company to do this for you. This is what the former NWML reportedly did.
-
Keep in mind in your calculations that the "µ" in plating thickness is micro inch not micro meter or micron.
-
Unless its copper plating over FR-4, otherwise known as a PCB ;). Then it's µm for consistency sake.
-
Ok but all the connectors I have checked out the plating thickness is in micro inches.
-
This is why every high-end ultra precision instrument (that I know of) uses gold-plated CuTe binding posts.
But do these high-end instruments (eg. the Pomona low thermal EMF binding posts) use regular copper-nickel-gold plating, or an alternative that might produce less EMF? Gold over copper is quite good, for example, but the gold gets contaminated by the copper and eventually tarnishes. I'm sure there are other options.
The 34420A nano volt/micro ohm meter uses bare copper connections and how they keep it clean
(https://www.eevblog.com/forum/testgear/lking-for-gold-plated-tellurium-copper-banana-plugs/?action=dlattach;attach=56261)
-
I was not trying to be a nitpicking jerk with that, sorry it came across that way :( It was an interesting example of a reputable manufacturer choosing to not use a plating at all when nV are involved even though it has the disadvantages of possible/probable oxidation issues. It is also my defense that Deoxit is not snake oil.
-
Good candidate CuTe binding posts are from Pomona (3770), Multi-Contact (PK4-TS), IET (BP1000) and jswilley (2758).
I double checked these parts and found statements of direct Gold plating only at
http://jswilley.com/Accessories.html (http://jswilley.com/Accessories.html)
For all others I still believe there is no nickel underneath the Gold. But it is strange that it is not documented.
-
I was not trying to be a nitpicking jerk with that, sorry it came across that way :( It was an interesting example of a reputable manufacturer choosing to not use a plating at all when nV are involved even though it has the disadvantages of possible/probable oxidation issues. It is also my defense that Deoxit is not snake oil.
Anyone know if adding the Deoxit Preservit/Shield to keep the contacts from oxidizing after cleaning would cause a problem (EMF-wise)?
-
Caig technical page (http://store.caig.com/s.nl/sc.18/category.238/.f)read the pdfs linked on the left side of the page. scroll down to the last three charts at the bottom and take a look at them.
-
To keep the thermal EMF's down, I would like to find some banana plugs made of gold-plated ...
If not a secret what exactly are you measuring and what is your setup that make it so important. I believe for certain measurements there are ways you can compensate for EMF if it is important. Bridges, 4, 5 terminal sensing, AC sensing instead of DC etc. How did you concluded that effect of EMF is greater than your measurement uncertainty for whole setup? What are your calculations?
-
Thermal EMF's are a known problem that comes into play in a major way below 100uV (or so). A 6.5 digit DMM is a 1ppm device that can get down to 100nV (or even lower on some models).
...
The *ideal* binding post would be something that ...
As for doing all of the calculations (etc.)-- there is simply no need for this. It is a well known problem that you can read about in various peer-reviewed publications. The solution is also well known, but the components for what I want to do are scarce-- hence, this thread...
Well yes, you jumped right to the solution. Only piece of information I got is that you are measuring VOLTAGE, but what for? Is it a sensor with voltage output? Thermocouple? Pressure sensor? Is there a signal conditioning involved? What is the accuracy of the sensor itself? - this is what I meant by "whole system".
I looked in a user manual of my DMM - it looks like gold is not what I need - see attached image. Table is a bit confusing but I assume it is C above 0 otherwise it would be in Kelvin.
Cadmium-Tin solder must be Cd70Sn30 (70% cadmium, 30% tin, that is by weight I assume).
From all reasonable combinations looks like you should expect 0.3 * 25C = 7.5uV EMF error. That is why I am wondering what kind of measurements are you making where 7.5uV matters?
-
I have made some experiments to build my own high quality low EMF adapters.
Here are two examples (see att. pictures).
The first/easiest is to use POMONA 3770 + 3772 + 4892.
EMF wise, the 4892 is the weak point (Gold plated brass and Gold plated Beryllium Copper spring).
Nevertheless I find the result quite good.
The second one needs some more work and I need to order more parts to finish it.
Baseplate is Teflon for max. isolation resistance, binding posts are Multi-Contact PK4-TS. Weak point is also the plug (Gold plated brass and Gold plated Beryllium Copper spring, but I will probably order goldplated CuTe replacement as soon as I have an idea how I can put them together without soldering (want to crimp or screw it)).
-
Doesn't your multimeter offer something like a short calibration or offset removal? That should be enough to eliminate any parasitic thermocouple effects, as long as the temperature in your lab is constant.
Of course, if you're measuring resistance instead of voltage, it's better to use an LCR meter with 4 wire connections. The AC will eliminate any thermocouple offsets.
BTW, do you have a pic of those split banana plugs? Are they available separately?
-
Nicely done quarks! Next step is a lathe and some tellurium copper bar stock :-+
-
@mos6502,
I do not know if you asked me or DiligentMinds.com, but we both have the same TEK DMM4050 and want to be able to plug in copper wire or spade lugs for lowest possible EMF meassurement, that is why an adapter like this is needed.
About the split banana plugs, there are Fluke TL2x4W-PT II I (2x4-Wire Ohms Tweezers) Test Leads, which I do not have.
@robrenz,
I need to find a good way to put the binding posts and the bananas together. Without having your skills I need to crimp or maybe screw them together.
-
By any chance does the set screw thread in the banana plug match the stud thread of the banana jack? If so you could remove most of the banana plug on the right leaving just enough threads to use it as the nut to hold the jack in the Teflon block. That would also require making the stud length of the jack just the right length so the Teflon gets grabbed just before the jack threads bottom out in the plug threads. Since the bearing area of the plug is very narrow it would easily extrude the Teflon so you would want to use a washer to get a larger bearing area on the Teflon.
-
The MC PK4-TS have 4mm thread and the banana LS410-TS CuTe 4mm plugs (see att. pict.) I want to order have no thread (but solder Connection, which I do not like). If I cut it as you suggest, drill a hole and make a 4mm thread inside, there is 0.5mm wall thickness left, which should be stable enough, but the Gold plating is gone.
If I switch to Pomona 3770 with the imperial thread, it is just a little bit larger than 4mm, maybe a pressfit is then just stable enough, what do you think?
-
I would not trust a press fit with that thin of a wall in copper.
-
thanks robrenz,
when I stack 4mm banana plugs, it is the same wall thickness and I had the feeling it is stable enough, that is why I guessed it should work.
My other option with the 4mm thread idea, will than have the same problem and even envolves my imprecise handwork. So this should be even worse.
What would you do to solve this?
-
Get a rolled thread tap instead of a cutting thread tap and roll the threads into the material. You do not remove any material and the rolling work hardens the copper making it stiffer.
-
Hello SeanB,
I have not thought of that, but it sounds like a very good idea and I even can disassemble it if neccessary. The only thing is I can not do it myself and have to find a shop who can/will do it for my few plugs.
Thanks
quarks
-
Get a rolled thread tap instead of a cutting thread tap and roll the threads into the material. You do not remove any material and the rolling work hardens the copper making it stiffer.
Those parts would need to be supported in a collet while using a thread forming tap. The forces are high and would just expand the thin wall part if not supported. I know you know this already but the work hardening does not make the part stiffer it just raises the yield point which makes it seem stiffer. The modulus if elasticity (real stiffness) is not significantly altered by heat treatments or work hardening. The classic example of this is the question, which is stiffer a 66 RC high speed steel tool bit. or the same size piece in low carbon steel. Many machinists first answer is the tool bit and would prove it to you by putting both in a vise and trying to bend them with an adjustable wrench. The low carbon bends easily and the tool bit seemingly resists bending at all for the same pressure applied. In reality the deflection per unit load is basically identical on both (same modulus of elasticity) but the plastic yield point of the low carbon piece is much lower than the high speed tool bit and is reached very easily with the wrench. FWIW
-
I notice the IET SR1 series standard resistors have the jacks you're looking for. I wonder if they could sell just the connectors? Or maybe hook you up with their suppliers. If not, you could try to get lucky with some ebay units but it seems like most are missing the bottom posts. I found http://myworld.ebay.com/bage9299/ (http://myworld.ebay.com/bage9299/) selling a couple with the posts still; shame to tear them apart just for the banana parts!
-
@robrenz,
probably my english is not good enough, but does this mean a rolled thread is the wrong way to go?
Please let me know what you would try.
@Galaxyrise,
I have tested almost all products I could find so far. Actually I do even have the ESI bananas (see att. picture lower left).
As far as I can judge it, they are also Gold plated brass with Gold plated Beryllium Copper banana and therefore no improvement.
-
@robrenz,
probably my english is not good enough, but does this mean a rolled thread is the wrong way to go?
Please let me know what you would try.
There is nothing wrong with the formed thread but it exerts very high forces on the material in the process of forming the threads.
I would machine a proper one piece unit from tellurium copper and just use deoxit to keep copper oxide from being an issue. :-[