Teardown: Standard Resistors

[zlymex]

I've been fascinated by standard resistors of their performance, cannot stop thinking of what's on earth inside.

Before teardown, let's look at some of the important factors of a standard resistor, such as specified by esi SR104:


 - Stability
Specified as ppm/year, It is a long time factor that most important for a standard resistor.
Every Chinese and Japanese resistor has a CLASS, they classify them according to annual drift.

 - Temperature Coefficient, alpha23
This is the slop of the temperature chart at 23 deg C, in ppm/C. Because the temperature chart is curved for most resistors, alpha varies with temperature. That's why it has a subscript specifying at what temperature the alpha is measured. Also, for standard resistors that often used in lab environment, they don't specify or test in large temperature range. Normally 18 to 28 degree C is suffice.

 - Temperature Coefficient, beta
The unit of beta is ppm/C2(per sq deg C), not ppm/C as above spec. The temperature chart can be roughly represented by a quadratic curve, this is especially true for standard resistors where the temperature range is limited. This beta is the coefficient(constant) of the second order equation. The larger the beta, the more the curvature. Most of the metals and resistors have negative beta(parabolic shaped). If a WW made of manganin, the beta is usually -0.6ppm/C2, If made of evanohm the beta usually around -0.028ppm/C2. Foil resistors often have even smaller beta.

 - Power Coefficient
Represent in ppm/Watt. This parameter depends on how heat dissipated and also related to tempco.

 - Adjustment to nominal value, ppm
Its the deviation from nominal value at the time of being made, or may call it initial error, initial accuracy.

One thing also worth noting is that when people measure a standard resistor by a DMM, the standard resistor is superior in most of the cases even the DMM is a 8.5 digits. Most of the imperfection/noise/instability in the test results are probably from the DMM rather than the standard resistor.

ADVANTEST 10k, find no spec, 1 hermetically sealed WW resistor inside.



Chinese BZ3, 0.001 Ohm, class 0.01%, manganin.




Remarks: This is one of the nine BZ3 resistor set, 0.001 Ohm, max current 33A, everything is heavy from binding post, thick manganin stripe. There is a small compensation resistor in parallel with the main.


esi SR104, 10k, stability 0.5ppm/a. Oil filled, hermetically sealed evanohm WW in a stainless steel container.





What can I say, the best 10k resistor ever made. Pity that no further teardown.

[zlymex]

esi SR1, 100 ohm, 50ppm/year


Normal wire on mica, the favorite way of esi.


Fluke 742A, 10k







Stability 4ppm/year, tempco is very small. Four 39992 Ohm hermetically sealed evanohm WW resistors in parallel, a 32 ohm adjust resistor is in series with one before parallel. Fluke is probably the largest hermetic WW manufacturer.


Fluke 742A, 1 Ohm



Stability specified as 8ppm/year. Twenty 20.01 Ohm hermetically sealed evanohm WW resistors in parallel, parallel another small adjust resistor of about 2k.
The inside photo was by Leicamr of 38hot.net

[zlymex]

IET SRL 1 Ohm
Same specification as Fluke 742A 1 Ohm but different internal construction.



The small resistor on the top is probably the compensation resistor.

IET SRX, 1Ok
I bought this one from IET 11.5 years ago, specified as 10ppm/year, 1ppm/C. One hermetically sealed WW resistor with some compensation/adjustment





MI 9331, 100 Ohm, Specified as 5ppm/year, 0.1ppm/C




Transmille RS3000


http://www.transmille.com/references-and-standards/precision-references-and-standards/3000rs

Although this is not an actual teardown but I know what's inside.
When this UK standard was first released in 2009, I saw something interesting on their website that must be mistakenly put by someone the manuscript of the resistors they used in the standard. A few days later it's gone but I saved it.



Another interesting thing about the script is that they described the tempco of  VSC232Z as 0.05ppm/C, apparently mislead by Visahy's datasheet. Later, they modified this to 0.5ppm/C

[zlymex]

Wekomm RS9010A, 10k




More photo and info see Dave's blog #834 and xdevs.com
Specified as 1ppm/year, 0.3ppm/C, one hermetically sealed Vishay foil resistor


I must say this is not up to German standard as the actual tempco of -0.47ppm/C is larger than specified. Wekomm may have misinterpret the datasheet from Vishay like most of us did, and they didn't throw away or compensate ones with large tempco.
Also they are contradictory on those max voltage and max current specs. How can you apply 10mA max current on a 10k without exceed the 20V max voltage. Well, they may say they are on 'which is smaller' basis, but why 10mA max for 10k in the first place?
Another thing is that the resistor is off by as much as 7.72ppm, but the spec says it should be within 2ppm.

Inside VHA518-7, there are seven foil resistor elements.



Yokogawa, 1k, class 50ppm, not hermetically sealed






Lastly, let's teardown some hermetically sealed WW and see how they are made.






Apart from usual bobbin/teflon insulation etc, the special thing about the hermetic resistor is the seal. They have those pre-made glass/ceramic disk with a small tube in the middle and thin layer of metal ring outside to be later soldered to bass tube case.
Although brass has different temp expansion co-efficient with glass, it's flexible and will expand and extract with the hard glass as temperature vary. If the casing is thick, they cut the edge part to make thinner. The inner metal on the hand must be Kovar which expand the same with glass, otherwise, the not-so-easy-compressed 'rod' will soon destroy the seal or even break the glass especially in high temperature when soldering the leads. But, Kovar is notorious for its large thermal EMF that should not be used in precision WW. They solved this dilemma by first implant a Kovar tube(instead of Kovar wire) that bond with glass well, and pass the copper lead through the hole and fill the gap by soldering. Solder is soft that act as cushion if temperature varies and still keep the seal.

The end of the teardown. Sorry for taking too long and split, I had difficulty in uploading images that often stopped in the middle.

[Vgkid]

Looking forward to the rest. I have a 1k SR1 resistor(I tore it down too).

[TiN]

Very nice summary Hope will can update it as it evolve.

[quarks]

Great post

here are some pictures from my Leeds & Northrup 4210 "Thomas Type One Ohm Standard" resistor.
This was advertised by L&N as the "Prime standard for use in maintaining value of the ohm".

[quantumvolt]

Very interesting.

I would like to know how the 4 resistors and the trimpot on the compensation board of IET SRX, 1Ok are connected and what values they have (at least for me, they cannot all be seen clearly). Maybe it would be possible to infer if it is all resistance value adjustment, or there also are some temperature coefficient elimination (reduction).

Thanks for a very informative post.

[nidlaX]

Wow! Nice stuff you have access to!

[daqq]

Thanks a lot! Great images!

[tszaboo]

Very good stuff.

Another interesting thing about the script is that they described the tempco of  VSC232Z as 0.05ppm/C, apparently mislead by Visahy's datasheet. Later, they modified this to 0.5ppm/C

Yes, the VishayPG datasheets are very very misleading. They claim insane typical drift on the top of the page, and then you scroll down, and see that while the typical is 0.0005 PPM the actual worst case drift is 0.0005+/-2 ppm. And since it is in the datasheet, it is not even a lie.

If you want to see something stupid, order a few thousand of their precision resistor. Once I saw that they delivered a big order, where the SMD resistors were in trays, not reels. I guess the bending would destroy the accuracy. But that is not the funny part. The tray was tiny, containing 25 resistor. Way to go, production ready!

[manganin]

The Chinese standard resistor is a direct copy of the Soviet version down to the smallest details.

Of course the Soviet design wasn't their own either...

Chinese BZ3, 0.001 Ohm, class 0.01%, manganin.

[zlymex]

The Chinese standard resistor is a direct copy of the Soviet version down to the smallest details.

Of course the Soviet design wasn't their own either...

You are right, I have a few standards of the Soviet as well.

[zlymex]

Very good stuff.

Another interesting thing about the script is that they described the tempco of  VSC232Z as 0.05ppm/C, apparently mislead by Visahy's datasheet. Later, they modified this to 0.5ppm/C

Yes, the VishayPG datasheets are very very misleading. They claim insane typical drift on the top of the page, and then you scroll down, and see that while the typical is 0.0005 PPM the actual worst case drift is 0.0005+/-2 ppm. And since it is in the datasheet, it is not even a lie.

If you want to see something stupid, order a few thousand of their precision resistor. Once I saw that they delivered a big order, where the SMD resistors were in trays, not reels. I guess the bending would destroy the accuracy. But that is not the funny part. The tray was tiny, containing 25 resistor. Way to go, production ready!

I knew it's true, they have mislead me before. Although didn't  order a few thousand, but a few hundred is more than enough. When my first order of 50 VHP101 came, it arrived in one plastic bag.

[zlymex]

Very interesting.

I would like to know how the 4 resistors and the trimpot on the compensation board of IET SRX, 1Ok are connected and what values they have (at least for me, they cannot all be seen clearly). Maybe it would be possible to infer if it is all resistance value adjustment, or there also are some temperature coefficient elimination (reduction).

Thanks for a very informative post.

I've never care about the adjustment part before since I think it is the usual way of trim a resistor for exact value like those in Fluke 720A.
Since you ask, I just teardown it again and it is. There is no tempco related compensation what so ever. If you need something for temco compensation, I myself developed a way of doing that: compensate the alpha tempco and trim the value at the same time. Also, there are ways to compensate the beta as well.
I've update it with the circuit above, there is a 39.2 Ohm resistor on the board but shorted.

[manganin]

IET SRL 1 Ohm. Same specification as Fluke 742A 1 Ohm but different internal construction.

The internal construction is actually the same. Multiple resistors connected in parallel with the copper equalizing plates. But in the IET version the whole package was resin potted which allows to use much cheaper non-hermetic resistors. The solution is not perfect but another layer of protection seems to make the humidity time constant long enough.

The "wings" in the equalizing plates are a clever idea, because they allow soldering the adjustment resistor after the potting.

[zlymex]

Great post

here are some pictures from my Leeds & Northrup 4210 "Thomas Type One Ohm Standard" resistor.
This was advertised by L&N as the "Prime standard for use in maintaining value of the ohm".

Thanks quarks, 4210 is a great resistor. Is that on the bottom red paint? Or there is solder underneath for the double wall to hermetically seal?

[quarks]

Although didn't  order a few thousand, but a few hundred is more than enough. When my first order of 50 VHP101 came, it arrived in one plastic bag.

I did order even less than you and received individually packaged resistors, some even in plastic cases (see pic.)

[zlymex]

IET SRL 1 Ohm. Same specification as Fluke 742A 1 Ohm but different internal construction.

The internal construction is actually the same. Multiple resistors connected in parallel with the copper equalizing plates. But in the IET version the whole package was resin potted which allows to use much cheaper non-hermetic resistors. The solution is not perfect but another layer of protection seems to make the humidity time constant long enough.

The "wings" in the equalizing plates are a clever idea, because they allow soldering the adjustment resistor after the potting.

Great info. I'm curiosity about why the spec of the old IET SRL series are exactly the same as Fluke 742A.

[manganin]

Is that on the bottom red paint? Or there is solder underneath for the double wall to hermetically seal?

Solder which is just "tamper sealed" by the red paint.

The construction is very simple. The edges of the inner cylinder are bent so that it tighly fits inside the outer cylinder. The coil is wound on the inner cylinder. After the heat treatment the inner cylinder is slided inside the outer one and soldered together at both ends.

[zlymex]

Is that on the bottom red paint? Or there is solder underneath for the double wall to hermetically seal?

Solder which is just "tamper sealed" by the red paint.

The construction is very simple. The edges of the inner cylinder are bent so that it tighly fits inside the outer cylinder. The coil is wound on the inner cylinder. After the heat treatment the inner cylinder is slided inside the outer one and soldered together at both ends.

Got it. I'm asking this because I saw another teardown of a Cambridge standard where there is those red paint as well, but there is no solder inside the paint.

[manganin]

The construction is very simple. The edges of the inner cylinder are bent so that it tighly fits inside the outer cylinder. The coil is wound on the inner cylinder. After the heat treatment the inner cylinder is slided inside the outer one and soldered together at both ends.

By the way, the Soviet 1 ohm (and higher) standard resistors share the same contruction. Except that the the edge of the outer cylinder bended at the bottom and the inner cylinder at the top. The Chinese version is probably similar.

Unfortunately the stability is not even close to the Thomas. The mechanical construction was easy to copy, but not the art of the heat treatment.

I saw another teardown of a Cambridge standard where there is those red paint as well, but there is no solder inside the paint.

The Cambridge was also sealed by soldering but there is a separate bottom plate between the inner and outer cylinders. Please see the picture attached.

[zlymex]

The construction is very simple. The edges of the inner cylinder are bent so that it tighly fits inside the outer cylinder. The coil is wound on the inner cylinder. After the heat treatment the inner cylinder is slided inside the outer one and soldered together at both ends.

By the way, the Soviet 1 ohm (and higher) standard resistors share the same contruction. Except that the the edge of the outer cylinder bended at the bottom and the inner cylinder at the top. The Chinese version is probably similar.

Unfortunately the stability is not even close to the Thomas. The mechanical construction was easy to copy, but not the art of the heat treatment.

I saw another teardown of a Cambridge standard where there is those red paint as well, but there is no solder inside the paint.

The Cambridge was also sealed by soldering but there is a separate bottom plate between the inner and outer cylinders. Please see the picture attached.

I also teardown the Chinese 1 Ohm, top and bottom all soldered. However, there were many factors that they did not do it correctly as I have pointed out at 38hot times ago, so I did not include the teardown here. As for the average aging of those Chinese resistors, it varies. Here are the distribution of 24 samples, some are good but others not.


The bottom of the Cambridge is soldered, what about the top?

[manganin]

The bottom of the Cambridge is soldered, what about the top?

The top is soldered. But the connection bar seals are far from hermetic. The Cambridge type is oil filled, not air, so there is also an oil screw at the top plate.

But the ceramic feedthrough of the Soviet type is not perfect either. Many resistors which have spent decades in an oil bath have changed from air filled to oil filled...

As for the average aging of those Chinese resistors, it varies. Here are the distribution of 24 samples, some are good but others not.

Very interesting data of the Chinese BZ3. Large variation (poor quality control) is typical for the Soviet type too.

How did you eliminate the temperature coefficient in your measurements? The TC of the Soviet type is so high that an oil bath is a necessity. And the hysteresis is remarkable which can lead to wrong assumptions if the resistors are not stored in constant temperature between the measurements.

[doktor pyta]

Good job, zlymex, Thanks!