T.C. measurements on precision resistors

[EEVblog]

When I get a little time, I'll be happy to post some, and yes, my 40+ years does qualify me as an expert whether you accept it or not, I do accept your opinion, you are welcome to it but on the flip side, what qualifies you to say that I am not what I say I am or my opinions are valid or not?

I was not doubting your qualifications. In fact I did some research and you do seem legit.
But once again, I go with the evidence, not because someone just declares so on a forum because they are an expert.
You've made a bold claim that Vishay's scientific papers are "laughable and inept" so you are the one that gets to prove that, or else don't be surprised that some people might be skeptical of that claim.
That's how it works.
This is not about doubt of your experience in the field, nor is it meant in any way personal.
It has nothing to do with whether or not your opinion are "valid" or not. Your opinion is just that, an opinion. Sure an experts opinion might be worth more at face value than some anonymous forum user, but ultimately it always comes to evidence to say who's opinion is right. This isn't philosophy we are talking about, this is engineering where there is a right and wrong answer based on  the evidence.
Even if Bob Pease came on here and declared Vishay's scientific papers "laughable and inept", I'd still ask for the same proof. I'm a hard arse like that 

[EEVblog]

The cal sheet for my Wekomm standard
http://www.eevblog.com/files/WekommResistanceStandardCalSheet.pdf

Looks like they use this bridge:
http://www.mintl.com/media/pdfs/accubridge.pdf
and a Genrad 1444A standard

[Edwin G. Pettis]

Hi Dave,

Thank you for posting the calibration document, interesting.  A bit of a clarification here, I did not say that all of Vishay's papers were laughable or inept, just some of them and yes, they are somewhat self-serving in nature, that doesn't necessarily mean they are all junk.  If one has a good understanding of the subject matter, reading their papers or watching their videos supports my conjecture without a doubt.  Unfortunately, most people watching or reading this stuff does not have the necessary knowledge to separate the wheat from the chaffe. 

The Measurements International is a DCC ( direct comparison current) bridge, hence it has a bit of an advantage over my ESI 242D at ?10K, the 242D has the advantage above 10K.

DiligentMinds is correct in that Vishay's papers are a mix of scientific and opinion which is questionable at times, I do not know of any of the Vishay 'papers' being peer reviewed as such, for that matter I constitute a peer and as such review them now and then.  There are some well written 'white papers' (a more correct term than scientific) that Vishay has put out, it is just difficult for most people to know the difference between the flowers and the manure, that is kind of where I come in.  I, in no way, thought your comments were 'personal', there was really nothing personal in your manner nor are my comments meant to be.

<chuckling> In the case of Bob Pease, if he were here, no doubt he would tell you in no certain terms that he didn't give a ____ but Bob usually didn't pull any punches no matter what the object of his comments thought.

You must agree that qualifications and opinions go hand in hand, one who is qualified on the subject matter is also qualified to make opinions and those are usually well informed.  Of course, the corollary of this is that some people who think they are qualified also think highly of their misinformed opinions and don't know any better.

I do thank you for your voicing your concerns, I never expect everyone to agree.

[Vgkid]

The GR1444A was introduced in 1970($600.00), and was not listed in the 1978 catalog. It appears to be an analog to the ESI SR104 resistor. The earliest mentionof the SR104 is in a 1971 ESI242D manual(I didn't try very hard). It is easier to find GenRad infformation than ESI. I wish IET Labs hosted some of ESI's documentation.
 http://www.ietlabs.com/pdf/GR_Experimenters/1970/GenRad_Experimenter_March-June_1970.pdf
Reading further I see H.P. Hall was used in the creation of the article .

[3roomlab]

These are not standard Vishay parts, they are:

custom specific made part, together with a special packaging.

http://www.vishaypg.com/foil-resistors/case-studies/study/wekomm_1/

Wekomm engineering paired the VHA518-7 resistor with carefully selected components to form a transfer standard product.

VHA518-7 (= 7 resistors in series)



VHA518 long term drift measurements against the Quantum Hall by the Dutch Metrology Institute:

http://www.vishaypg.com/docs/63620/63620.pdf

VHA518-11 (= 11 resistors in series)

more noob questions 

when its so many resistors in series, does it mean they use the "+" and "-" ppm drifts to cancel out each other, after knowing the characters? and that is the primary thermo/electrical "pairing" criteria? and probably sorting thru thousands of resistors, arrange a series that give a linear tempco?

what is this "dutch quantum hall" being mentioned?

edit : this 1 could be a tough question. why did they choose 23^oC as the zero deviation centre temp? maybe dave ... ask KCB this question?

[babysitter]

Have a look at the data what KCB (which is not the PTB!) is accredited for. I use my advantage of being a native german speaker... or at least almost, as I am coming from a part of germany where they speak something I wouldn't necessarily call german

(translation mode on)Resistance: 0,1*10E-6 * R best reportable uncertainity.(translation mode off)

But why do they give a tempco in their test report?

Neither do they list a device which was used to realize the temperature slope or what it is referenced to, nor does the DUT itself provide a temperature measurement like the SR104 which could be specified together with the main resistor.
They can use whatever they want to heat up their unknown reference sensor, what does it tell us about resistor temperature, especially in still air?

(They are not even accredited for temperature calibration.)

I think that is a fact they should note.

[MK]

One troubling thing about the vishay VHA518-11 is the 6.5 K version is not showing root hours settling down, it was drifting faster after 2000 days then it was at the beginning! so that means it does need constant recals to trust its behavior in my eyes.

[Vgkid]

That is a good question. I have some American equiptments datasheets dating from the 60's that specifies the cal temp of 23degrees C, while my Solartron is specced at 20deg.

[ltz2000]

this 1 could be a tough question. why did they choose 23^oC as the zero deviation centre temp?

That is a good question. I have some American equiptments datasheets dating from the 60's that specifies the cal temp of 23degrees C, while my Solartron is specced at 20deg.

The early British metrology equipment, like standard resistors, had 17 degrees Celcius calibration temperature. Not sure when the change to 20 C took place, could be in the 1950s. I quess they have 23 C nowadays like the rest of the world.

Back then cooling was not available, heating only. The calibration temperature needed to be higher than the room temperature. Quite cold in those Victorian buildlings.

[ltz2000]

Well, the GenRad 1444-A was the precursor to the ESI/Tegam/IET-Labs SR-104 10K resistor.

The GR1444A was introduced in 1970($600.00), and was not listed in the 1978 catalog. It appears to be an analog to the ESI SR104 resistor. The earliest mention of the SR104 is in a 1971 ESI242D manual(I didn't try very hard).

The SR104 tested against the computable capacitor in NPL Australia was manufactured in 1965. The earliest commercial SR104 I have seen dates back to 1967.

The GR 1444-A was introduced later as a competitor to the SR104 but turned out to be less stable. The GR unit wasn't silicone oil filled like the SR104 but nitrogen instead and constructed from only two mica card resistors.

[Dr. Frank]

Have people published detailed research papers in rebuttal?
I have not researched any of this of course, but putting my skeptical hat on, I find it hard to believe that one of the leaders in the field of precision resistors for many decades produces "laughable and inept" scientific papers.
Mistakes? I'm sure that happens, but "laughable and inept"?
Proof please.
And BTW, proof is not "go read the professional forums" and "Unlike you folks, I've been in the resistor industry for over 4 decades, I have the experience and knowledge which no one else in this forum has an apparent claim to."
/skeptical hat off

Hi Dave,

I'm also a singulary voice only, as I agree in parts what Edwin Pettis is saying.

But I'm also very hard-arsed, concerning reputable specification for components, as myself, I was creating automotive component specifications for a long time.. and have some experience on Vishay, including many of the companies they took over, all were our  suppliers..
Beyschlag once was one of the best manufacturers of Thin Film resistors, with the most reputable specs.. until Vishay swallowed them..


So I can really assess the quality of the current Vishay specs, and find the ones about BMF technology exaggerated quite often.

It is very obvious for every engineer, if you only look carefully on the data sheets, that the typical data are much too optimistic, partly advertised in big letters, compared to the upper limits they specify.

For example, they always claim a typical T.C. of 0.05ppm/K, on top of the spec, even in the headline, but on the left lower part of the page, they specify a T.C. of +/- 0.2ppm/K +/-2ppm/K maximum.

Only that latter specification parameter is reasonable and serious, so they really do not screw the customer, in the end.


My personal experimental experience with my 5 VHP202Z resistors was exactly, what you doubt, that such a reputable manufacturer tries to lead the user astray by these advertising methods.

These components in reality (according to my own measurements, which I also posted in the EEVBLOG forum) all showed a T.C. between -0.3 to -1.0 ppm/K, no sign of a positive parabola shaped R(T), with a typical 25°C minimum point, as the Z-foil technology would imply, according to several scientific papers from Vishay, where even the honourable Dr. Felix Zandman had signed.

After my complaint about this bad performance of the parts, the Vishay representative had to admit that one can not rely on the typical data and on the intended technology characteristics so far, as advertised...and that there is no guarantee from Vishay about these typical characteristics...
So the new standard from Weekom should be observed very critical also..


In the end, the BMF technology is still very good, and beats PWW components in some aspects, but the production variance is much too big.. an indicator, that Vishay is not really able to control this technology in last consequence.

My impression also is, that with the early death of Dr. Zandman, the respectability of Vishay decreased...as the bean counters took over instead, maybe.

I disagree to Mr. Pettis, that all the papers from Vishay are toast..
Some are very interesting, and several claims really reflect reality, for example the long term stability of these oil filled, hermetically sealed types..again from my own measurements on these 5 resistors.

These have drifted less than 1ppm apart in the last 4 years.. so 2ppm/6years as specified TYPICALLY seems to be reasonable.

Well, that's a singulary finding also, but every bits and pieces sum up to a reliable picture.. in negative and in positive aspects.

I append one of the key documents from Vishay about their BMF technology, regarding the characteristics of the different foil types, C, K and Z. (Z like Zandman)
These characteristic shapes also show up in the BMF specification, but Vishay measures the T.C. not by the physics method, T.C. = dR / dT, but by 3 temperature points only, i.e. the butterfly method.
Also some sort of misleading..

Frank

[EEVblog]

After my complaint about this bad performance of the parts, the Vishay representative had to admit that one can not rely on the typical data and on the intended technology characteristics so far, as advertised...and that there is no guarantee from Vishay about these typical characteristics...

If they are "typical" characteristics, then, well, there is no guarantee! Either a spec is typical or it is guaranteed.
If it's typical and doesn't meet spec, bad day for you.
If it's guaranteed and doesn't meet spec, bad day for Vishay!

So the new standard from Weekom should be observed very critical also..

FWIW I've been assured that the resistor in the Wekomm standard is a custom part and hand selected. So in theory, no matter how bad Vishay's standard production spread is, it's possible to pick the good ones.

[Dr. Frank]

If they are "typical" characteristics, then, well, there is no guarantee! Either a spec is typical or it is guaranteed.
If it's typical and doesn't meet spec, bad day for you.
If it's guaranteed and doesn't meet spec, bad day for Vishay!

FWIW I've been assured that the resistor in the Wekomm standard is a custom part and hand selected. So in theory, no matter how bad Vishay's standard production spread is, it's possible to pick the good ones.

Dave, you missed my point.. I was fully aware of the maximum limits..

BUT if you specify typical values, these should be met for a reasonably big quantity, if you buy several.. if you understand typical parameters correctly, in the sense of a statistical distribution.. and the typical value representing the median.

And also, a clear characteristics of the Z-foil technology is the parabola shaped R(T) behaviour (see appended document) .. in other words this R(T) shape is strictly required, if the compensation technique between metal foil and ceramic substrate is correctly applied.

So, if you don't see this parabola, even not over a wider temperature range (and the Vishay representative told me, they could also not guarantee this shape), then the very logical reverse conclusion is, that this component has no Z-foil characteristics, and that this Z-foil compensation method did not work for the DUTs.
And as a last conclusion, that also explains the 10..20 times higher T.C. than typical...


And for the Weekom standard:
Correct, you may pick the resistor with the lowest T.C., that won't change over time, probably.
In fact, 0.45ppm/K is not stellar.. comically on the same order than my single chip VHP202Z types.

But you cannot pick for best long term stability, as accelerated testing is not possible, and only time will tell, as Mr. Pettis correctly stated.
Frank 

[ltz2000]

RS9010 tempco from the calibration certificate compared with a good SR104

[Dr. Frank]

OK, here I have another very interesting CERN document .. in several aspects.

It describes the construction of an ultra precise / stable 10mA current source for the calibration of the SC coils of the accelerator.

One author is John Pickering, godfather of the LTZ1000 volt reference Datron / Fluke 7000, cofounder of Datron (1271 / 1281 DMMs => now Fluke 8508A, 4910 reference (?)). He also invented the hysteresis cancellation on the LTZ1000, and the statistical TaN divider.

In this paper, the current source relies also on a 10k Z foil resistor, which is needed for ultra low T.C.
The current source shows an overall  T.C. of typ 0.07.. max 0.14 ppm/K for the output current, that also means that the 10k Z-foil also has to have even better characteristics...

Therefore, there really exist Z-foil resistors with that ultra low T.C., so Vishay may be able to control the technology sometimes..

Or how many resistors did they need to sort out for getting some really good ones?


Therefore follows another argument: If the yield of these "good ones" is high, then the company really has control over the technology.
If the yield is low, or this special feature shows up randomly only, then the company fails to control this technology..

For latter case, in the end, you get some quite reasonably good resistors, but which are not remarkably better than PWW technology.

You may decide by your own.

Frank

[janaf]

Dr Frank, it may also be that the "good ones" are cherry picked for important customers while us mortals end up without those golden ones

[Dr. Frank]

Dr Frank, it may also be that the "good ones" are cherry picked for important customers while us mortals end up without those golden ones

That means, that there would be the possibility that Vishay regularly offered T.C. selected parts, and they would have the possibility to make a 100% screening of production parts.

I asked the Vishay representative about that, exactly, and he responded 2x "NO".

The only possibility would be to buy a 100 fold quantity and select on your own.

Or they offer other parts with guaranteed lower T.C.
The trick with that special type is, that Z-foil is NOT used at all, but one C and one K foil element in series!
C has positive parabola, K a negative one.. obviously more predictable than Z foil..

I think, it was VHD100.., or so. VHP101.. maximum 10ppm change over 30°C window, gives 0.3ppm/K average.


So I really wondered, what they offered to Mr. Pickering, they did not want to offer to me..
His 1281 was full of Vishay BMF, so he had a better standing, obviously

Frank

[acbern]

Besides the fact that the Wekomm part in my opinion is way overpriced (even when considering its calibration), one should put things in perspective. The SR104 is a true primary standard (historically at least, today it is rather Quantum Hall of course). A resistor standard based on VPG hermetic foil resistors is both pretty stable (I can confirm Dr Franks stability data based on cal data) and has a low TC (0.2ppm/K is not bad, I have seen SR104s with similar and higher TCs). Compare this to the not really cheap Fluke standard resitors, and you will see that this is a very good alternative to the SR104s and Flukes out there, given its price. Nobody considering to buy a SR104 would seriously consider the Wekomm anyways, it is adressing another market in my opinion, more a competiton to Fluke.
How Wekomm came to the conclusion to charge 5k for it is beyond me though; even the most expensive VPG hermetic 4 terminal 10k resistor I have been offered was a few hundred bucks only, plus say 500 for a good cal if they cannot do it themselves.

[ltz2000]

Besides the fact that the Wekomm part in my opinion is way overpriced (even when considering its calibration), one should put things in perspective. The SR104 is a true primary standard (historically at least, today it is rather Quantum Hall of course). A resistor standard based on VPG hermetic foil resistors is both pretty stable (I can confirm Dr Franks stability data based on cal data) and has a low TC (0.2ppm/K is not bad, I have seen SR104s with similar and higher TCs). Compare this to the not really cheap Fluke standard resitors, and you will see that this is a very good alternative to the SR104s and Flukes out there, given its price. Nobody considering to buy a SR104 would seriously consider the Wekomm anyways, it is adressing another market in my opinion, more a competiton to Fluke.

Yes, it is not fair to compare a Volkswagen with a Ferrari. But it would be, if they had the same price tag. And with the Wekomm resistor that is actually the case. It costs the same as a new SR104 (or 4-5 used). First I suspected a decimal error though...

For comparison the VHA518-7 resistor which is the heart of the unit was approximately $50 when I last asked. And based on the graph I draw from the calibration data, there seems to be no real improvement in tempco to the off-the-shelf VHA518.

Even if there was, why pay $5500 for a tempco selected resistor when you can buy a basic VHA518 for hundred times less and ovenize it. With an aluminium block and a relatively simple thermostat the tempco will be excellent . And you also get rid of the possible irriversible or long time constant errors caused by the temperature variations. The power comsumption of the oven will be minimal because the tiny resistor is easy to insulate very well.

The drift is another issue. 1 ppm/year vs. SR104 <0.1 ppm/year typical. The two resistors measured by the VSL showed drift rates of 0.4 and 1.9 ppm/year. Most importantly it is not (yet) predictable. Accelerating like some other Vishay resistors?

I know this a tough business and there are not too many new products introduced outside the big players. So we should treat the new company nicely and understand that their first product may not be a direct hit. But I still feel uncomfortable with their "marketing first" approach which reminds me a little of the high-end audio business.

[Edwin G. Pettis]

To Dr. Frank, reply #367, to clarify again, I did not say all of Vishay's papers were laughable and inept, only some of them are and those particular ones were reviewed by some other people (Bob Pease was one of them) and agreed that they were what they were.

It was my understanding that Wekomm was a young independent Dutch company like so many others, bought out by Vishay.  As I've said many times in the past, Vishay does not like competition, they will either buy them up or try to belittle them if they believe they are any kind of a threat to them, that is a legacy of Felix.  Vishay has destroyed many reputable resistor companies over the years with no qualms about it.  I agree that Vishay does make some very good resistors but the customer must be fully aware of Vishay's 'tricks' in advertising and data sheet shenanigans or else you run the risk of not getting what you thought you were.

The wonky TCR curve noted in the Wekomm calibration sheet is indicative of multiple resistors being 'matched' together for a given low TCR, while this can work exceedingly well for a narrow temperature range, it does not work well over wider ranges unless the TCR curves are linear and not one single Vishay part has completely linear curves over temperature.  The curves are still wavey even if they are of low TCR values and this may not be an issue in a given circuit if the TCR only needs to stay within a given range but most designers would much rather have a linear TCR that is predictable over range.

As to Dr. Franks speculation on Vishay's cherry picking resistors, if they are doing it and that is highly likely in some cases, I rather doubt they would publicly acknowledge it as they would consider it a blight on their reputation that they could not control their manufacturing processes close enough and in some instances, Vishay has 'quietly' acknowledged this fact off the record.  I very seriously doubt you will ever get that comment on record from Vishay.

Given that the Wekomm part is made up of so many resistors, the time and labor involved to cherry pick all these resistors for one unit is probably is sufficient enough in their eyes to warrant a $5,500 price tag, however, until this standard can consistently demonstrate characteristics long term to warrant that kind of money, I think they are over pricing themselves out of the market for the time being.  While I have not been terribly impressed with the Fluke standards construction (they cherry pick too), they at least do have something of a track record for reasonable long term stability but I also think they are a bit over priced as well for what is in there vs performance.

[MK]

One thing that concerns me about that Vishay foil "std" resistor, no thermal well for the thermometer, how can you confirm it is at a known temperature without having a very very steady temp in the cal lab...

[EEVblog]

I think the Wekomm resistor is meant as a "secondary transfer standard", to compete with the likes of the Fluke 742A series [which they are discontinuing].

Of course it's a transfer standard. Not sure why anyone would think it's a primary standard?

[ltz2000]

The good old ESI SR1 nowadays manufactured by the IETlabs. Manganin (Zeranin) on mica card. Stability specification 2 ppm/year, much less in practice. Tempco 1 ppm/C.

The 10 kohm version costs $560 according to the 2012 price list.

http://www.ietlabs.com/decaderes/resistance-standard/esi-sr1-calibration-resistor.html

[texaspyro]

I have a box with an SR-104,  10V reference,  and GPSDO in it.   The thermal control temperature is set at the upper point where the SR-104 TC curve crosses 10,000 ohms.  Self-heating from the GPSDO / Vref / PWM'd fan mixing in external air does the temperature control.  The components are baffled/isolated from air currents from the fan.  RMS variation of the internal temperature is in the millidegree range.

[ltz2000]

Question : Manganin alloy naturally offers "parabola" shaped resistance/temperature curve, with a perfect zero slope near 25°C. Produced since May, 1893 (Wikipedia said). Some of you pointed out that at this time it's impossible to find a resistor with such characteristics, even from the best manufacturers like Vishay. Since the material is available for more than a century, I don't understand why this performance isn't achievable today. Is it about manganin manufacturing related issues (Soldering)? Did I miss something?

Manganin and its relatives are still widely used (even in bulk metal foil resistors). It is easy to solder, you just need to be fast.

But there are even better alloys available (stability, tempco, resistivity, corrosion). Edwin is the expert.