Home made standard resistors using Vishay VHP100T and VHP101T

[iisak]

Frank, you must be right. If even in the adored SR104 design the force and sense terminals are just soldered together with a piece of thick copper wire right below the box cover (see the pic in the first message of the teardown thread), it can't be so critical. Although maybe we can't be sure if it is some cadmium solder or something else exotic.

I also have measured many kinds of self made resistor boxes from colleagues and customers which have been stable even if they are just some stable resistor soldered directly to bnc or other heretic connectors...

Maybe I try just simple soldering with good heat clamping in my next version to get mechanically stable sense points.

Adding that thermal mass is now also in my to-do list.

[IanJ]

Hi,

This thread has me interested in constructing a box with 4 resistors in it, looking at Vishay Z-series @ 0.2ppm/degC for the availability/cost and I have the perfect die-cast enclosure for it.
My question is regard to oil filling as has been mentioned, so can I simply completely fill my enclosure with a low dialectric oil?.......any ideas on the oil type to use? Off the bat I see transformer oil to IEC 60296:03.......suitable?

Ian.

[Dr. Frank]

Hi,

This thread has me interested in constructing a box with 4 resistors in it, looking at Vishay Z-series @ 0.2ppm/degC for the availability/cost and I have the perfect die-cast enclosure for it.
My question is regard to oil filling as has been mentioned, so can I simply completely fill my enclosure with a low dialectric oil?.......any ideas on the oil type to use? Off the bat I see transformer oil to IEC 60296:03.......suitable?

Ian.

Ian,
you obviously misunderstood that oil-filling item..
All VHPxxx resistors themselves are oil-filled, and that's all you need.
That provides a direct thermal coupling between resistor element and case.
If you then put that case tightly into a metal mass, maybe with thermal conducting grease, you have a perfect coupling with the temperature sensor and the out case.

You don't need any further oil filling. That's required only, when you want to thermally couple a bunch of resistors, like in the 720A, or when you want thermal coupling and oxygen protection of a naked resistor element, like in the Reichsanstalt or Thompson type Normal Resistors.
In  this case, it would simply be a big mess w/o any additional benefit.

Another thing: the VHP20xZ resistors have a T.C. of about < 2 ppm/°C , guaranteed, typically below 1ppm/°C, but do not at all expect 0.2ppm/°C .. my 5 EA have 0.3, 0.5, 0.7, 0.7 and 1ppm/K..

If you want guaranteed < 0.2ppm/°C, then you need the VHP101.

Frank

[IanJ]

Hi,

This thread has me interested in constructing a box with 4 resistors in it, looking at Vishay Z-series @ 0.2ppm/degC for the availability/cost and I have the perfect die-cast enclosure for it.
My question is regard to oil filling as has been mentioned, so can I simply completely fill my enclosure with a low dialectric oil?.......any ideas on the oil type to use? Off the bat I see transformer oil to IEC 60296:03.......suitable?

Ian.

Ian,
you obviously misunderstood that oil-filling item..
All VHPxxx resistors themselves are oil-filled, and that's all you need.
That provides a direct thermal coupling between resistor element and case.
If you then put that case tightly into a metal mass, maybe with thermal conducting grease, you have a perfect coupling with the temperature sensor and the out case.

You don't need any further oil filling. That's required only, when you want to thermally couple a bunch of resistors, like in the 720A, or when you want thermal coupling and oxygen protection of a naked resistor element, like in the Reichsanstalt or Thompson type Normal Resistors.
In  this case, it would simply be a big mess w/o any additional benefit.

Another thing: the VHP20xZ resistors have a T.C. of about < 2 ppm/°C , guaranteed, typically below 1ppm/°C, but do not at all expect 0.2ppm/°C .. my 5 EA have 0.3, 0.5, 0.7, 0.7 and 1ppm/K..

If you want guaranteed < 0.2ppm/°C, then you need the VHP101.

Frank

I understand the oil filled VHP series, but I was looking at the Z-series, which are not oil filled...........Just thought it might help keep them more thermally stable and removal of air would help against humidity changes......poor mans hermetically sealed.
But I guess have them in a closed enclosure and insulated will be enough.............

Ian.

[branadic]

If you then put that case tightly into a metal mass, maybe with thermal conducting grease, you have a perfect coupling with the temperature sensor and the out case.

Using a small oil film instead of thermal grease makes a better thermal contact. This is at least what temperature sensor companies will tell you.

-branadic-

[Dr. Frank]

I understand the oil filled VHP series, but I was looking at the Z-series, which are not oil filled...........Just thought it might help keep them more thermally stable and removal of air would help against humidity changes......poor mans hermetically sealed.
But I guess have them in a closed enclosure and insulated will be enough.............

Ian.

The Z series denotes a different resistor alloy only, there's C, K and Z alloy, with different nominal / typical T.C.s, i.e. +2ppmm/K, -1pp/K and +/-0.2ppm/K, respectively.

The VHP101 contains a series combination of two resistor elements, with C and K alloy, to get a compensated T.C., below 0.3ppm/K

And you can as well have the Z alloy inside a hermetic, oil filled case, I have used the VHP202Z at that time, and they perform extremely stable over time, apart from the mediocre T.C.

So you know that also, and probably mean that you had a look on a molded Z-resistor.
That's also no good idea, if you buy cheap, you will end up with a cheap result.

The mold compound itself is a thermal resistor, and the oil will not do any good to improve that.
Also it would not protect the resistor element as it would inside a hermetically  sealed case.

That would not make any sense at all, but a big mess only.

Frank

[mzzj]

VPG resistors are NOT specified at 0.x ppm/C. You will have to buy lots and use your equipment to sort them.

Have you seen any statistical data how good they really are? It would be most interesting! I now have some preliminary data from 8 individuals and they seem to be 0.3 ppm/C or better. I will do more accurate measurements after they are stabilized. After some months maybe.

I have measurement results for 4x 100 ohms and 4x 25 ohms VHP202Z resistors from last  several years. Long term stability has been reasonably good but 25 ohm resistors have horrible tempco around room temperature ( looked like close to 2ppm/cel)
Compiling the measurements to one clean excel file is another project to do...

Vishay VHA -4Z series with 4-wire connection has been disappointing in long-term stability and 100 ohm resistor also had "larger than hoped for" self heating effect.

BTW: If I'm guessing right who you are look for calibration reports xxxx12E001 and  xxxx12E002 

--
Matti

[mzzj]

[mzzj]

For comparison our 100ohm  Tinsley 5685A total drift has been smaller than the 0.5ppm calibration uncertainty for last 10 years so I would say you get what you pay for... 

[Echo88]

Did you compare the said resistors against the Tinsley 5658A and was the Tinsley calibrated periodically against a better standard, like a QHR?

[mzzj]

Did you compare the said resistors against the Tinsley 5658A and was the Tinsley calibrated periodically against a better standard, like a QHR?

Exactly.
The 25 ohm and 100 ohm  resistors are all connected in series and I just use "the box"  to periodically verify our Hart Scientific/Fluke 1590 "super thermometer" (or resistance bridge if you wish) linearity. For that purpose I don't need spectacular stability or tempco so VHP202z's work reasonably well.

[splin]

mzzj, your results are a bit dissapointing to say the least. The 25R results especially look terrible - Edwin Pettis claims his wirewound resistors typically drift less than 3ppm/year whereas yours appeared to take 5 years or so to settle down to < 3ppm/year.

Your results are well adrift (pun intended) from Vishay's claims. According to figure 11:

http://www.vishaypg.com/docs/49789/vfrguide.pdf

Both a 103R and 157R VHP101 moved less than 1 ppm over 8 years, in a random walk rather than the monotonic curves you measured. Perhaps more importantly, Vishay specify drift at typically less than 2ppm in 6 years in their datasheets.

So either:

a) VHP201Z drift performance is not as good as the VHP101 (I'd be surprised if there is much difference)

b) Your devices suffered some environmental conditions (shock, temperature extremes etc) either before, or after you received them, which caused large shifts in values requiring a few years to recover. What temperature ranges have they experienced during your tests?

c) You were unlucky (8 times or twice from a batch POV),

d) Vishay got lucky with their test samples.

e) Vishay published the results for the best performers out of the batch of many on test (surely not!)

f) Vishay's results are for parts that had been around for a year or three before they started the test

g) Vishay's test had been run for rather longer than 8 years and they only showed the last 8 years measurments (surely not!)

h) Some combination of the above or something else.

To be fair to Vishay, Dr Frank and zlymex had very good results with VPR hermetics WRT drift. zlymex had drift << 1ppm over several years, but they may have been well aged parts - see reply #30:

https://www.eevblog.com/forum/metrology/teardown-standard-resistors/25/

The above resistors are also much higher resistance and it may well be that the extremes of values, high and low, may drift rather more than those in between. It may well *not* be the case of course, in the absence of more data.

Since long term drift performance of Vishay hermetic foil resistors is a key characteristic and selling point I assume they will have hundreds, if not many hundreds of parts on long term tests including each of the different models and parts from different batches. Given the importance of the drift characteristics I'm surprised that they seem to have been *very* economical on the amount of data they have made public. Perhaps I have been looking in the wrong place but I can't recall seeing results for more than a handful of individual specimens.

It would be marvellous if they would publish much more drift data but it clearly ain't gonna happen - commercially sensitive or embarassing perhaps.

Big customers may get sight of the data but surely small to medium customers will account for a significant percentage of their production given that these parts are  expensive and relatively specialised. I wonder how many of the latter customers take Vishay's claims on trust and how many run their own long term and expensive qualification testing?

[EDIT] added comments on 25R performance.

[mzzj]

So either:

a) VHP201Z drift performance is not as good as the VHP101 (I'd be surprised if there is much difference)

b) Your devices suffered some environmental conditions (shock, temperature extremes etc) either before, or after you received them, which caused large shifts in values requiring a few years to recover. What temperature ranges have they experienced during your tests?

d) Vishay got lucky with their test samples.

a) could be that the VHP101 goes trough some special "PMO"
b) impossible to say what happens between factory and our lab but it was custom order (not NOS or recycled..) Ever since soldering the resistors in the box they have been sitting in room temperature 23+-2c calibration lab.
did my best during construction not to shock the parts. ie heatsink the leads with flat pliers during soldering and not stressing the resistor body during bending the leads (no idea if its needed but tried to stay on safe side)
d) I usually read the component datasheet "typical" values as "lies, lies, bigger lies"  Vishay is especially good with their tempco specmanship   

[iisak]

mzzj: interesting data! And some of it really was availlable closer to me than I expected...

You are getting the same -1...-2 ppm/year for 100 ohm Vishay resistors than we have since 2012. At least for us the predictable dirft at that level is not a big problem. High TC would be more problematic if and when these are being used outside the most stable air baths.

[mzzj]

Related rant: Just got information that Fluke* can't sell Tinsley standard resistors to europe because of ROHS 
!"#%"#¤%#"¤ Dammit!

*) Yes apparently Fluke has grabbed also tinsley's standard resistor business 

They tried to offer Fluke 742 as an alternative, and even that one is close to 5000 eur/usd 

[mzzj]

To be fair to Vishay, Dr Frank and zlymex had very good results with VPR hermetics WRT drift. zlymex had drift << 1ppm over several years, but they may have been well aged parts - see reply #30:

https://www.eevblog.com/forum/metrology/teardown-standard-resistors/25/

The above resistors are also much higher resistance and it may well be that the extremes of values, high and low, may drift rather more than those in between. It may well *not* be the case of course, in the absence of more data.

Sounds like one plausible explanation.  http://www.vishaypg.com/docs/63171/TN104.pdf

Vishay also says: ". An additional mechanism concerns ohmic values below
100 . The ohmic value of internal connections can
change with time and temperature by a few milliohms — a
drift equivalent to 10 ppm in a 100  resistor and of 1000
ppm in a 1  resistor. Very low-value foil resistors require
special construction of their internal connections. For
values below 0.5  the “Metal Strip” style of resistors
includes very robust internal connections and an option of
a four terminal external connection

---

"Correction factors for drifts in lower values and thermal
resistance data can be obtained from the Applications
Engineering Department"



Haven't really paid attention earlier but now noticed this tidbit:

"ZERO TCR OF FOIL RESISTORS
Ultimately, the TCR of the foil resistor is affected by two
opposing physical phenomena, which depend both on the
resistive element on its own, and its relationship to the
substrate to which it is bonded.
Resistivity of the free foil changes directly with temperature.
After bonding, the difference of the temperature coefficient of
expansion (TCE) between the foil and the substrate will
cause compression or expansion strains on the much thinner
foil, directly affecting the resistance change with temperature
(strain gage effect).
These two effects occur simultaneously on the resistor with
temperature changes, and can be detrimental to the
performance of the resistor. They can however, also be used
to negate each other toward improving the overall
temperature characteristics of the resistor."

I have to say that I don't necessarily like the idea 

[meandeev]

Hello Frank,

the construction of your setup makes sense to me, so I will "copy" it.
which binding Posts did you use in your setup? Is the resistor glued in the alu-block, clamped in it or held by the thermal grease (and the wires)?
Where did you get the box (or which dimension did you use)?

Ronny

...

My setup can be found here: https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg464413/#msg464413
...

Frank

[splin]

Sounds like one plausible explanation.  http://www.vishaypg.com/docs/63171/TN104.pdf

Vishay also says: ". An additional mechanism concerns ohmic values below
100 . The ohmic value of internal connections can
change with time and temperature by a few milliohms — a
drift equivalent to 10 ppm in a 100  resistor and of 1000
ppm in a 1  resistor. Very low-value foil resistors require
special construction of their internal connections. For
values below 0.5  the “Metal Strip” style of resistors
includes very robust internal connections and an option of
a four terminal external connection

---

"Correction factors for drifts in lower values and thermal
resistance data can be obtained from the Applications
Engineering Department"

Interesting - I'd missed that. So if the early years drift rate of low value parts is rather higher than middle value parts then why doesn't the datasheet say so - they do it for the TCs after all? They show differing drift rates in Vishay Beyschlag precision thin film datasheets for different resistor values for MMA0204 and UXB0207 resistors.

Assuming that your results are reasonably typical that means the "typically < 2ppm drift in 6 years" claim in the D/S is extremely misleading given your 25R's 30ppm/6 year drift. I guess they don't state which 6 years they mean.

I've just found another Vishay document which includes the same 157R and 100R VHP101 drift graphs I linked earlier (with < 1ppm total drift over 7 years) but this time they are attributed to a Swedish customer, Metrima AB.

http://www.vishaypg.com/docs/49568/purtol.pdf

Look at the title (page 4);

"Documented performance from a customer for hermetically sealed VHP101 Foil resistors which have been on test for over 10 years."

But the graph only shows 7 years data!.  There could be an innocent explanation such as 'the final three years data were invalidated due to some experimental flaw' but the cynic in me says there has to be a strong possibility that they are cherry picking by leaving out large drifts in the first three years.

Why publish customer data rather than their own of which they must have plenty? Perhaps they think it may carry more weight because of a semblence of independance and that people don't trust their own very much? 

[mzzj]

"Documented performance from a customer for hermetically sealed VHP101 Foil resistors which have been on test for over 10 years."

But the graph only shows 7 years data!.  There could be an innocent explanation such as 'the final three years data were invalidated due to some experimental flaw' but the cynic in me says there has to be a strong possibility that they are cherry picking by leaving out large drifts in the first three years.

Why publish customer data rather than their own of which they must have plenty? Perhaps they think it may carry more weight because of a semblence of independance and that people don't trust their own very much? 

So that they can say "it's just typical case and we even didn't measure it by ourselves" if there is liabilities. 

VHP101 "window method" tempco is just another show of specmanship if you ask me.. no guaranteed max of absolute tcr anywhere. Could be S-shaped curve with  2ppm/c slope around room temperature and still fit inside the window 
Graph says "typical" 0,3ppm/C (that would be worse than "typical" for VHP202Z)

[mzzj]

Did some digging on lab:

Vishay 100 ohm VHA516-4Z (Z-foil 4-wire hermetic resistors) 6 year total drift
sn 1: -23ppm (this one has been transported around and has probably seen some temperature swings during transit)
sn 2: -9ppm
sn 3: -13ppm

One of them is pretty much spot on at the moment:

 

[mzzj]

And some other values from VHA516-4Z set:

200 ohm -16 ppm /6y
300 ohm -5 ppm  /6y

All of "my" vishay resistors seem to drift down. 10, 25, and 50 ohm resistors have been also going down by various amounts.

[Dr. Frank]

Hello Frank,

the construction of your setup makes sense to me, so I will "copy" it.
which binding Posts did you use in your setup? Is the resistor glued in the alu-block, clamped in it or held by the thermal grease (and the wires)?
Where did you get the box (or which dimension did you use)?

Ronny

At best, you would use CuTe binding posts, from Pomona.
Most important is the high isolation of the plastic, because other manufacturer (multi contact, now obsolete)  used lower grade plastic, which caused  ~1ppm error @ 10kOhm.

CuTe has low thermo couple voltage, important only for DMM w/o Offset Compensation.

The holes in the aluminium block (or copper, if you have access) are hand-made, so it's more a loose fit.

I have used the usual thermal grease for transistors. Better would be the low temperature physics thermal grease, called Apiezon.
Either grease should be good enough for thermal contact and mechanical fix.

As some of by boreholes were too wide, I had to use one or two layers of aluminium foil to fix the resistor.
The temperature sensor is also attached with thermal grease.

I used Hammond aluminium boxes, 92 x 38 x 31 (mm), maybe mod. 1590A, because I also needed very good thermal contact to the aluminum block (again thermal grease used between block and case), and a very good thermal conduction material around the whole assembly. So that serves three purposes, first a thermal short all around the resistor and the aluminium block, additional thermal mass for thermal inertia, but also good thermal conduction, to bring the whole box to a controlled temperature.. I simply place these boxes on top of the 3458A, so you can stabilize the temperature to 25°C, depending on the location on the top case.

I would use bigger aluminium blocks, and also bigger aluminium boxes, to have even more thermal inertia.
I would also use Pt100 sensors, instead of the precision NTCs.

Frank

[Conrad Hoffman]

Remember that thermal grease isn't a very good thermal conductor. It only works well for transistors because we use a very thin layer, hopefully no more than a few ten thousandths of an inch. Loose fits with a gob of grease won't couple as well as close fits with a small amount.

[IanJ]

Hi all,

I've started a wee build, hopefully I will get some VHP100/101 resistors eventually, but for now I am using Vishay Z-series (0.2ppm/degC).

Biggest problem I found was getting stable banana sockets, i.e. ones that didn't require D-shape holes, didn't break the bank and also didn't physically have issues that affect the contact thus resistance........and in the end I opted for what you see in the pics below.

Once I get the VHP100 series resistors I'll solder them directly to the posts, but for now the z-series are fitted to the tabs. I'll also solder jumpers from main post to sense post instead of using the tabs.

Enclosure is a Hammond 10758PSLA (die-cast)

PS. Still waiting on the 100k, and still to characterize the ones fitted.

Ian.

[usagi]

similar to build I did a while back.

reminds me, i need to get them checked again.