EEVblog Electronics Community Forum
Electronics => Metrology => Topic started by: voltlog on September 16, 2018, 11:42:42 am
-
I sometimes do multimeter reviews on the youtube channel and would like to have a resistance reference box to use when assessing the accuracy of these meters. They are usually cheap, low resolution meters (up to 4.5 digits) so I don't think I'm in the metrology type measurements area on the same principle where RF guys consider sub Ghz signals DC :-// but still I thought this is the best place to ask. I also have two bench meters Agilent 34401A (6.5 digits) and HP 3478A(5.5 digits) and it would be great if I was able to keep these in check as well with this reference box.
Vishay sent me 3 resistors from their VHP100 series as free samples, they are 1K, 10K and 100K. The plan is to use an aluminium enclosure, drill holes for the binding posts (4-wire arrangement) and connect the terminals using some copper silver wire, by soldering on the ends of the binding posts.
Then I would have the VHP100 resistors, suspended and soldered in between the solid silver copper wires. I would be soldering them at least 10mm away from the can and I would also try to heatsink them while soldering to minimize the thermal shock applied to the resistors.
I've looked for good quality binding posts and I know the Pomona 3770 (https://ro.mouser.com/datasheet/2/159/d3750_60_70_1_01-34263.pdf) (Gold plated Tellurium Copper with polycarbonate insulation) is often referred to on this metrology sub-forum but unfortunately they are too expensive at $12.5 USD + shipping + tax, considering I would be needing 8pcs total. So instead I would like to use the Hirschmann 935 980-811 (http://www.farnell.com/datasheets/1693544.pdf), these are gold plated brass connectors and I got a set of 10 for about $32.
I've read several threads on this forum, talking about the issue of EMF (Thermoelectric voltage) which appear at the junction of dissimilar materials however I would like to know how would this issue affect me with my particular build. What kind of errors am I expecting to introduce into my measurement if I use those Hirschmann gold plated brass binding posts in a 4-wire measurement setup?
-
Hello,
the BIL 20 AU have a very "cheap" look.
Why don't you look at PKI 10A AU?
Regards
Frank
-
the BIL 20 AU have a very "cheap" look.
Why don't you look at PKI 10A AU?
I'm not sure if these have any technical advantage or it's just looks but I can say that I prefer the "slim" type look of the 935 980-811 and the fact that it has that split in the end for soldering connection.
Looking at the two datasheets I spot a difference on the housing material 935 980-811 quotes PF While the PKI 10A AU quotes PA. But once again I don't know if there is any advantage or disadvantage to any of those types of plastic.
-
Given your stated usage, the BIL 20 AU would do just fine (Hirschmann makes decent stuff). :)
-
For normal use the cheaper sockets are Ok. The main point is to avoid a temperature difference between the voltage sensing terminals. If mounted in an more insulating case, the sockets will follow the temperature of the plugs and thus not that much temperature gradient over the brass part anyway. It also depends on the cables used.
For directly connection of just a 4 wire cable one could consider additional srew terminals wired in parallel.
-
You only need low thermal EMF for the voltage sense terminals. Assuming you have a common terminal for the three resistors you only need four low thermal EMF connectors. The current force termnals can be the cheapest and nastiest that look ok.
With 4-W measurements I can't think of any reason why you would need silver plated copprr wires - just about any, cheap, wire will be perfectly ok.
-
Guildline 9330 resistors have copper terminals: https://xdevs.com/doc/Guildline/9330/Guildline9330Datasheet.pdf
My Sullivan T1212 has copper and mercury bath terminals.
-
With 4-W measurements I can't think of any reason why you would need silver plated copprr wires - just about any, cheap, wire will be perfectly ok.
I want to use the silver plated copper wire because it will not oxidize like copper would do after a while.
-
hi
iz it real gold or some titanium oxyde cover?
-
I have not seen titanium nitride used for contacts. It looks golden and is conductive, but can be covered with a thin oxide that makes poor contacts. Resistance is also quite high so it is more like candidate for thin film resistors and AFAIK even used in a few cases.
A very thin gold layer is not that expensive. One sometimes even gets this with cheap transistors or similar.
Silver plated copper wire is not that exotic and it solders really nice.
-
I think the big thing to watch out for with binding posts not designed for low-thermal-emf is that under the relatively thin layer of gold is a thick layer of nickel, and nickel-copper makes quite a thermocouple :/
-
I have not seen titanium nitride used for contacts.
China banana terminals for example
It looks shiny gold but cheap
What about this "gold"?
-
voltlog, you are on the Foil news @ VPG (http://www.vishaypg.com/foil-resistors/) :-DD
-
voltlog, you are on the Foil news @ VPG (http://www.vishaypg.com/foil-resistors/) :-DD
Where? I cant find myself.. I'm lost :palm:
-
(http://)
-
Oh, OK. That's their twitter feed. Cool! Thanks for sharing this.
-
China banana terminals for example
It looks shiny gold but cheap
What about this "gold"?
And more: it´s not copper under the gold(?) - it looks like aluminium and you cannot solder it.
-
the BIL 20 AU have a very "cheap" look.
Why don't you look at PKI 10A AU?
I thought about using the PKI 10 AU for my resistorbox, but they do not have soldering point and only give 2mm for case thickness.
So no good option for me as my chosen plastic case and metalbox (HF type) have around 3mm together.
For now I put an eye on the Keystone 4109 pair (brass, gold plated) for ~10$.
Btw what the hell is the difference between 4108 (20$) and 4109?
-
Practical view:
EMF due to improper match of plug to jack metal is usually of a few μV. It is important on 0.1V and below.
Your 34401A is using 1mA to measure 1kOhm, 100 μA to measure 10kOhms and 10 μA to measure 100kOhm.
All this cases give 1V Voltage drop on measured resistor. That means You will have temporary few microvolts thermal EMF (few minutes, until temperature between plug and jack will equalize). This introduces a few ppm (part per million) temporary error, visible on last digit. This is comparable to floor noise and instability of 34401A (5ppm of the range, according to specification). And few times less than 24h 20ppm stability.
So, if You do not like to wait few minutes until readings will stabilise, go into gold plated. Gold plated brass will be also OK.
If You want to use also Your resistors as Low Voltage divider, go to gold over copper tellurium.
8.5 digit multimeters usually have TrueOhms function (reducing thermal (and any other constant) EMF by current on/off or reversal).
-
Thank you for the practical view, exactly what I wanted to hear and also I learned a few things from this post :-+
-
Re nuts of choosing binding posts...
I would choose well made mechanically: with flat parallel surfaces, allowing for large contact area to spade lugs. Bigger contact area ==> smaller and more repeatable contact resistance, + faster temperature equalizing between spade lug and binding post. Good binding posts should allow keep contact resistance below 0.3 mOhm and repeatability below 0.1mOhm. This is not important in case od 4-wire Ohms measurements, but why not to have good Reference Voltage Divider for the same money..
I would choose Metal case. It can be used as heatsink to resistors, and will equalize differences of temperatures between resistors and between binding posts.
-
I actually used the binding posts which I had, mentioned in my first post above. I used an aluminium enclosure but I did not connect the cans to the outer casing.
Here is the build video if you would like to take a look at how I did it.
https://www.youtube.com/watch?v=4C7xpqP_WBE (https://www.youtube.com/watch?v=4C7xpqP_WBE)
-
Hmmm. You should always heatsink those resistors when soldering them.
Your nice box should have verbiage giving the tolerance as well [emoji41]
Sent from my iPhone using Tapatalk
-
Nice work.
In case of using as Resistor Standards for Your Multimeter You do not have to add heatsink. But some Multimeters use 1mA to measure 10kOhms. Also, if You will use Your set as 1:11 voltage divider or as half of the bridge, and supply it from 10V You may notice influence of selfheating. This power 10V x 1mA => 0.01W may heat Your resistor 1..2deg C, and change resistance 0.3 ..1ppm (depending on TCR of Your resistor).
For Your multimeter it is OK. But if You want to be ready for better multimeter You can add Cu or Al foil to make thermal shortage between resistor can and case.
-
Interesting build. Instead of paying big money for individual binding posts, you could find an old school 6.5 digit meter on ebay for less than $100 and scavenge them. An old Fluke 8502 has 7 posts and you can get them for less than $80 broken. Just a thought. Maybe go a step further and add oil to the inside of your case to ensure long term stability (though that case is rather large). Stoked to see the future results of this project, thanks mate!
-
Interesting build. Instead of paying big money for individual binding posts, you could find an old school 6.5 digit meter on ebay for less than $100 and scavenge them. An old Fluke 8502 has 7 posts and you can get them for less than $80 broken. Just a thought.
80$ for 7 old binding posts is more than 11$/post.
Sounds quite expensive for used brass posts, I do not think they are low EMF type.
You get Pomonas for ~8$ new and they are quite expensive too.
Low EMF TeCu Pomona 3770 are ~15$.
-
$15 is still > $11. ($4 x 7 = $28 saved). Plus you can/will probably resale the cards and/or everything else inside the unit. More importantly, I am imaginative and interested in scavenging what I can, and where I can. I have two Fluke 8000 series units that I am restoring to there previous glory...so its not just about how cheap I am.
-
$15 is still > $11. ($4 x 7 = $28 saved).
But how much of the original gold plating has been lost in the sockets of a well used instrument? I don't know but gold is relatively soft so this has to be a concern.
-
I am all for re-using so if I had an old instrument and the terminals were in good conditions, I would probably use them in a project. But if you don't already have the instrument I wouldn't buy a used one just to salvage the connectors because you can't possibly know their condition before purchase and the cost would be similar to buying new connectors.
-
the BIL 20 AU have a very "cheap" look.
Why don't you look at PKI 10A AU?
I thought about using the PKI 10 AU for my resistorbox, but they do not have soldering point and only give 2mm for case thickness.
So no good option for me as my chosen plastic case and metalbox (HF type) have around 3mm together.
For now I put an eye on the Keystone 4109 pair (brass, gold plated) for ~10$.
Btw what the hell is the difference between 4108 (20$) and 4109?
Finally I used SEB 2610 F4,8 gold-plated brass from SKS/Hirschmann (1.40€ @ reichelt) for my box - cheap and good enough for this purpose.
The resistors are Vishay Z201 and Ohmite Slim-Mox (1G) soldered on 1,2mm silver-plated copper wire.
Box is RND 455-00169 160x80x55mm IP65, hf-case is 100x64x40mm from junk box.
-
1G looks out of whack in this box. Also 4-wire connection is pointless for high resistance like this. :-//
But insulation and guarding on other side become very important.
-
1G looks out of whack in this box.
Do You mean leackage?
Also 4-wire connection is pointless for high resistance like this. :-//
It is not actually 4W -- double posts only
Autor use it for simple gear connection probably
PS - What about RCA connector? Is it guard?
-
If we assume that the isolation distance of your posts to the metal is app. 2mm and we assume that the Rho of the PVC of the binding posts is app. 1E+13 Ohm-cm and allow 20% max resolution influence then:
- For a 6.5 digit 34401A multimeter you can measure up to 120 K without substantial influence ( 1 ppm resolution)
- For an 8.5 digit 3458A multimeter you can measure up to 12 K without substantial influence (0.1 ppm resolution)
Above these values, you need to put at least the binding posts (and apply guarding for the higher Giga values) onto an isolated PTFE (teflon) area. Teflon has a Rho of app. 1E+17 Ohm-cm.
If we assume two test leads of AWG16 stranded copper wire of 50 cm with a resistance of app. 15 mOhm per meter then:
- For a 6.5 digit 34401A multimeter you can do a 2-wire measurement from higher than 12 K without substantial influence ( 1 ppm resolution)
- For an 8.5 digit 3458A multimeter you can do a 2-wire measurement from higher than 120 K without substantial influence (0.1 ppm resolution)
Below these values you need a 4-wire measurement.
I do like your 4-wire construction, it is a neat job ! :-+
-
1G looks out of whack in this box. Also 4-wire connection is pointless for high resistance like this. :-//
But insulation and guarding on other side become very important.
Yes it is, do not take the 1.21G as serious :-DD
This happens when you have space left on final design and do not want to waste it, so there came 2 more posts and this is the result.
If we assume that the isolation distance of your posts to the metal is app. 2mm and we assume that the Rho of the PVC of the binding posts is app. 1E+13 Ohm-cm and allow 20% max resolution influence
There is ~1mm air gap between outer and inner isolation in parallel with ~1cm height of insulation from nut to conductor, so this is not that bad.
Besides the posts, the leads have major influence and proper way would be using triax cable, connectors and proper measurement system (not affordable).
Nevertheless you are right that this is not the correct way to build precision very high resistance standard, though I would not call the used 1G a very precision one ;)
RCA is for temperature measurement with 10k NTC, guard/shield is blue post.
-
Really nice, clean construction. I like it :-+
-
Now I want one...T.E.A.!
-
1G looks out of whack in this box. Also 4-wire connection is pointless for high resistance like this. :-//
But insulation and guarding on other side become very important.
Yes it is, do not take the 1.21G as serious :-DD
This happens when you have space left on final design and do not want to waste it, so there came 2 more posts and this is the result.
If we assume that the isolation distance of your posts to the metal is app. 2mm and we assume that the Rho of the PVC of the binding posts is app. 1E+13 Ohm-cm and allow 20% max resolution influence
There is ~1mm air gap between outer and inner isolation in parallel with ~1cm height of insulation from nut to conductor, so this is not that bad.
Besides the posts, the leads have major influence and proper way would be using triax cable, connectors and proper measurement system (not affordable).
Nevertheless you are right that this is not the correct way to build precision very high resistance standard, though I would not call the used 1G a very precision one ;)
RCA is for temperature measurement with 10k NTC, guard/shield is blue post.
The 1G resistor should have just two terminals connected to it, this could have done without additional space or costs.
Now you have unnecessary leakage from all other terminals, probably not a problem in your application but my ocd told me I should mention this. :scared:
Otherwise nice construction with the box in a box. :-+
But what about temperature rise with sustained sources and the stagnant air trapped inside the outer box? :scared:
Probably time for medication.