My DIY 0.005% Resistance Standard. Is there a Point of Measuring with 34465A?

[RoadRunner]

Hi,

Like Dave, I made my self few 10K resistance standards with Y145310K0000V9L Vishay Foil Resistor 0.005% 0.2PPM/C . Pretty high quality Hirschmann thick Gold plated Banana Connector and Hammond Enclosure.

I have two 34465A with me both of them into year from last calibration , As per Specs they are at  (0.0040 + 0.0005 ) ±(% of reading + % of range) .

I am asking my self this question, Is there any point of measuring these with 34465A as they are basically similar as specs.

Any help will be highly Appreciated.

Regards

[TiN]

You'd be best to send it for somebody with calibrated gear to obtain calibration, and then you can use it for your lab.
That TCR you put on a label is not correct however. These resistor are specified for +/-0.6ppm/K for element + your connectors and solder joints will add some more for it.

[Magnificent Bastard]

The TCR is easy enough to get a better idea of it's actual value-- just measure resistance at 0, 25, and 50 degrees-C and calculate the TCR curve from that-- and since this is "relative", the calibration state of the DMM has only a small effect on the answer.  As to an absolute value-- well, I think this will be affected more by humidity than temperature.  I would remove the foam from the inside of the case, as it does nothing useful, and add a large desiccant pack in the case.  Then find someone that has a high-spec meter with recent calibration, and get the absolute value (with some known uncertainty).  Then you have a good "check the DMM to see if it's borked" standard.

[RoadRunner]

You'd be best to send it for somebody with calibrated gear to obtain calibration, and then you can use it for your lab.
That TCR you put on a label is not correct however. These resistor are specified for +/-0.6ppm/K for element + your connectors and solder joints will add some more for it.

I have access to better equipment at my work place so that would not be issue. TCR I got from the resistor spec i did not took into account what how joints will affect along with Banana connector. I wonder how i can calculate TC with all connectors and solder joints all together.

Regards

[Magnificent Bastard]

You'd be best to send it for somebody with calibrated gear to obtain calibration, and then you can use it for your lab.
That TCR you put on a label is not correct however. These resistor are specified for +/-0.6ppm/K for element + your connectors and solder joints will add some more for it.

I have access to better equipment at my work place so that would not be issue. TCR I got from the resistor spec i did not took into account what how joints will affect along with Banana connector. I wonder how i can calculate TC with all connectors and solder joints all together.

Regards

Just put some long banana extension cables on the resistor.  Put it in the refrigerator (along with a thermometer-- so you know the actual temperature).  Wait a long time.  Measure with your DMM.  Then put it a 50C oven (again, with the thermometer), wait a long time, measure that.  Then at room temp (same thing).  There are formulas available at the cost of an Internet search that you can now use to calculate the TCR curve.  Simple as that.

[TiN]

RoadRunner
Resistor spec is not 0.2ppm/K but 0.6ppm/K. You cannot use typical value in this case, because you have only one resistor sample.
I have tested many resistors for TCR before and covered this a little.

[Kosmic]

The TCR is easy enough to get a better idea of it's actual value-- just measure resistance at 0, 25, and 50 degrees-C and calculate the TCR curve from that-- and since this is "relative", the calibration state of the DMM has only a small effect on the answer.  As to an absolute value-- well, I think this will be affected more by humidity than temperature.  I would remove the foam from the inside of the case, as it does nothing useful, and add a large desiccant pack in the case.  Then find someone that has a high-spec meter with recent calibration, and get the absolute value (with some known uncertainty).  Then you have a good "check the DMM to see if it's borked" standard.

I tend to agree with the comment about humidity. It needs to be controlled somehow since the resistor is not hermitically sealed.

Better binding post could also help. Your Hirschmann binding posts are in brass and it's not going to help since they will generate thermal EMF. Pomona 3770 in Tellurium Copper would have been better.

Also since the resistor was mechanically and thermally stressed, it probably need to be calibrated after the assembly. Not sure you can rely on the 0.005% spec from the manufacturer.

[Andreas]

Hello,

I also recommend reading the first post of my resistors message (especially follow the link with the soldering heat drift)
https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg462296/#msg462296


Some measurement results of 1K Z201 resistors here
https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg462299/#msg462299
So I recommed measuring the actual T.C. instead of relying on a optimistic data sheet spec.

But I have to admit that larger values than 1K (12K5) have also larger hysteresis.
https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg697243/#msg697243
so even at the same temperature it depends from which direction you reach the temperature to get the actual resistance value.
(can also be a humidity effect).

with best regards

Andreas

[RoadRunner]

The TCR is easy enough to get a better idea of it's actual value-- just measure resistance at 0, 25, and 50 degrees-C and calculate the TCR curve from that-- and since this is "relative", the calibration state of the DMM has only a small effect on the answer.  As to an absolute value-- well, I think this will be affected more by humidity than temperature.  I would remove the foam from the inside of the case, as it does nothing useful, and add a large desiccant pack in the case.  Then find someone that has a high-spec meter with recent calibration, and get the absolute value (with some known uncertainty).  Then you have a good "check the DMM to see if it's borked" standard.

I tend to agree with the comment about humidity. It needs to be controlled somehow since the resistor is not hermitically sealed.

Better binding post could also help. Your Hirschmann binding posts are in brass and it's not going to help since they will generate thermal EMF. Pomona 3770 in Tellurium Copper would have been better.

Also since the resistor was mechanically and thermally stressed, it probably need to be calibrated after the assembly. Not sure you can rely on the 0.005% spec from the manufacturer.

Thank you all of informative comments, I am somewhat a beginner  in this all metrlogy stuff.

I was also concern about humidity so i selected a gasket sealed enclosure, and I was also aware benefits of Tellurium Copper Connectors, total cost of them would have been higher than the resistor itself. That is why skipped them.
I will do further reading to see if this steup is good enough for checking my handheld DMMs, though i never targeted it check my 34465As.

Thank you once again.

[hwj-d]

Thats a long time measurement of a 100 ohm oilfilled resistor with my 34461A.
Look at the TC. Ambient temperature, rising from 20°C to 30°C. It is most likely the DMM itself. My DMM6500 measurement from same Resistor is flat. Make the countercheck with your 34465A without acal.

[Jay_Diddy_B]

Hi,
I was inspired by your post enough to build my own, using parts that I had on hand.

I drilled a Hammond 1590B diecast for four binding posts.

I installed four binding posts that came from an ESI ratio transformer.




The resistors I had on hand were 30k 0.1% V53C1 (7818 date code)



I used three in parallel.

The measured value is pretty close to 10k 



Does anybody know the tempco for the V53C1 resistors?

Regards,
Jay_Diddy_B

[grizewald]

A much better way of installing your resistors in a box is to arrange them like this:



Black terminals are force and red are sense. You eliminate the extra wiring over and above the resistor's leads this way and you can extend such a layout as you wish by flipping the arrangement horizontally. The layout also ensures that there are two soldered connections for both sense and force.



The result looks like this:



Calibrated values are on the back:



And internally, it looks just like the the layouts above:



The box is built from a Hammond aluminium case and a reasonable selection of precision resistors with 1% to 0.005% tolerances and as low TCR as I could reasonably buy. The 1Ω resistor is from Bourns, mid values from Vishay, the 1MΩ and 10MΩ from Caddock and the 100MΩ from Ohmite (that's the one with 1% tolerance). When I was at a cal lab earlier this year, I had them measure all of the resistors with their 8.5 digit Fluke meter and then measured them with my 7.5 digit meter just after they had calibrated it. Thankfully, after installing all the resistors in the box, none of them had changed resistance as far as I could see to 7.5 digits.

The box will shortly be filled with silica gel desiccant and sealed with silicone gasket material. I just want to give it a couple more measurements before permanently sealing it. I'll send it off to the same cal lab next year and get a second calibration to update the label on the back.

Like RoadRunner, I also looked at the Pomona 3770 binding posts but they were prohibitively expensive (just the posts would have cost €200!), so I settled on some gold plated brass banana sockets instead. The resistors in the box probably cost around €200 in total and doubling the price for the box seemed excessive. The result is something which is perfectly acceptable for my needs.

[hwj-d]

This is a very nice optical practically assignment of the sense and force. But does this also make sense from a metrological point of view compared to central soldering?

[maginnovision]

I usually use my 1063 with either kelvin clips or a custom cable with force/sense attached to the same copper crimp connector since only two terminals are connected to each resistor. The larger metal plated with two posts is what you use to 0 the measurement taking the binding posts out of the measurement. Would that be better or worse than multiple solder connections?

[The Soulman]

Another remark regarding connectors for resistor boxes, also to save space and some money, not saying it is better:
I've found it is quite acceptable to use a single binding post for both force and sense connection.
This only works when both connections aren't touching each other inside the binding post, i.e.
use a banana on the inside and a fork or copper wire on the outside.
Any contact resistance is then out of the equation, the only parts that become part of the total resistance path
are the butt-end of the binding post and the solder connection to the resistors leg.
Before I build my r-box with single heavy duty Hirschmann pki10 au I've
tested measured this resistance with two small/used/cheap/crusty binding posts soldered butt to butt together,
just repeated the test today, its still around 0,0003 Ohm and unlikely to change with time and temperature,
the Hirschmann's where even much better but cant remember the exact number and don't have spares at the moment to test.

Ok, worst-case situation, we use our two crusty binding posts for a really nice 1 kilo Ohm resistor,
so instead of 1000 Ohm the total resistance becomes 1000,0003 Ohm, perfectly fine because we calibrate it as a finished product, but after time in a unlikely way the resistance of our binding posts changes with 10%, shifting our resistor value
from 1000,0003 Ohm to 1000,000033 Ohm that's still a 0,03 PPM error worst case,most people can live with that.

For higher resistor value's the error becomes even lower, and single binding posts are also more desirable to help reduce leakage current (less surface area).

For lower value high precision resistors a proper 4 terminal kelvin connection is more suitable, also to take as most of the resistance of the resistor legs out as possible, or even connect to dedicated terminals on the resistor body.

Just my humble 2 cents.

Inside construction:


Separate force (10mA thru spades on the outside) and sense (4mm banana on the inside):


edit: larger pictures

[The Soulman]

I usually use my 1063 with either kelvin clips or a custom cable with force/sense attached to the same copper crimp connector since only two terminals are connected to each resistor. Would that be better or worse than multiple solder connections?

If I was you I'd keep that nice box closed as to avoid damaging and contaminating those nice resistors,
don't ask me how I know how fragile those are. 

See my previous post about single binding post use, also if you have multiple of the same or lower resistors in series
you could apply current (force) from one end of the string to the other and measure the voltage (sense) across each of the resistors individually.
This also helps to allow each resistor to "warm up" the same and let your current source settle.
DO NOT include higher value resistors in this same string as they may be overloaded with to much current.

[Gyro]

Isn't the best way to solder one pair of terminals (typically Force) to the ends of the resistor leads, and then solder the other terminals (typically Sense) to points part way up the resistor leads?

I thought that was the proper way to do true Kelvin sensing. It avoids the force current and sense potential sharing any solder joints.

[maginnovision]

That was a while back when it arrived. I wanted to verify it was all still together and had no signs of damage. I'd also cleaned all of the oxidation from every terminal and cap and had it calibrated. It's been a while and is probably due again. If I have a solid connection I'm not sure why it would matter if force/sense were touching or not. Meaning literally, I don't know.
 So my custom cable has force/sense and guard connections on the meter end(5 total), the other end is guard, + force/sense, and - force/sense(3 total). Would it be potentially more accurate to have force being banana and sense be spades? I don't have low thermal banana connectors right now to test. Need to order more and more wire too. Keep giving cables away to customers.

[The Soulman]

Isn't the best way to solder one pair of terminals (typically Force) to the ends of the resistor leads, and then solder the other terminals (typically Sense) to points part way up the resistor leads?

I thought that was the proper way to do true Kelvin sensing. It avoids the force current and sense potential sharing any solder joints.

Yep, but not absolutely mandatory for higher resistance value's.

[The Soulman]

That was a while back when it arrived. I wanted to verify it was all still together and had no signs of damage. I'd also cleaned all of the oxidation from every terminal and cap and had it calibrated. It's been a while and is probably due again. If I have a solid connection I'm not sure why it would matter if force/sense were touching or not. Meaning literally, I don't know.

In theory the connection is never 100% solid.

So my custom cable has force/sense and guard connections on the meter end(5 total), the other end is guard, + force/sense, and - force/sense(3 total). Would it be potentially more accurate to have force being banana and sense be spades? I don't have low thermal banana connectors right now to test. Need to order more and more wire too. Keep giving cables away to customers.

You have force and sense terminated in the same plug on one side? If so, that's not ideal.
How do I become a customer?

[bill_c]

The box will shortly be filled with silica gel desiccant and sealed with silicone gasket material.

You may already know this, but for those who don't, you can't use just any RTV silicone caulking.  There is a grade just for electrical things that doesn't cause corrosion.

[maginnovision]

That was a while back when it arrived. I wanted to verify it was all still together and had no signs of damage. I'd also cleaned all of the oxidation from every terminal and cap and had it calibrated. It's been a while and is probably due again. If I have a solid connection I'm not sure why it would matter if force/sense were touching or not. Meaning literally, I don't know.

In theory the connection is never 100% solid.

Of course. Nothing is 100%, that's why I use the term solid, not perfect or ideal. 

So my custom cable has force/sense and guard connections on the meter end(5 total), the other end is guard, + force/sense, and - force/sense(3 total). Would it be potentially more accurate to have force being banana and sense be spades? I don't have low thermal banana connectors right now to test. Need to order more and more wire too. Keep giving cables away to customers.

You have force and sense terminated in the same plug on one side? If so, that's not ideal.
How do I become a customer?

Force and sense are crimped into a single spade connection. You could never be a customer. I only work with people locally/in person. If you were you'd need to have some sort of equipment for repair. Most cables I supply are only 2 wire shielded and the 4 wire shielded are 4 terminal each end. This is the only cable I have like this and I made it for the 1063 specifically.

[grizewald]

The box will shortly be filled with silica gel desiccant and sealed with silicone gasket material.

You may already know this, but for those who don't, you can't use just any RTV silicone caulking.  There is a grade just for electrical things that doesn't cause corrosion.

Absolutely! The stuff used for aquariums to bond glass sheets together does not release acetic acid as it cures and is the right stuff for this application.

[grizewald]

This is a very nice optical practically assignment of the sense and force. But does this also make sense from a metrological point of view compared to central soldering?

Probably not, but then again, I wasn't trying to build an SR-1010.

[Gyro]

Isn't the best way to solder one pair of terminals (typically Force) to the ends of the resistor leads, and then solder the other terminals (typically Sense) to points part way up the resistor leads?

I thought that was the proper way to do true Kelvin sensing. It avoids the force current and sense potential sharing any solder joints.

Yep, but not absolutely mandatory for higher resistance value's.

Yes, agreed. You could probably argue that it is a possible improvement the Thermal EMF perspective though. If you swap the force and sense terminals in my description above (so that the, hopefully closer spaced, sense terminals connect to the ends of the resistor leads) then there is only one solder joint on each lead, rather than two.

It would also allow the force terminal connections to be extended with short lengths of wire (from the force terminal to the mid the point on the resistor lead) to cope with the resistor leads being too short to span 4 terminals. Additional Solder joints and wire length in the force circuit are of minimal importance.

Just thinking aloud.

[ConKbot]

The box will shortly be filled with silica gel desiccant and sealed with silicone gasket material.

You may already know this, but for those who don't, you can't use just any RTV silicone caulking.  There is a grade just for electrical things that doesn't cause corrosion.

Absolutely! The stuff used for aquariums to bond glass sheets together does not release acetic acid as it cures and is the right stuff for this application.

Be mindful of the actual silicone though. Ive gotten aquarium silicone that is acetic acid cure. I guess a bit of acetic acid is not so harmful to the fish if the PH is managed, compared to methanol or amines. But the acetic acid will still cause problems for electronics.

[syau]

RTV 3145 is a good material for sealing but no cheap.