T.C. measurements on precision resistors

[zhtoor]

Seems like you US guys do have a serious problem with factual criticism. We could have done a factual analysis on how the failure showed on the resistor and what could be done to prevent such failures in future (one of the things we do for many companies almost every day), but instead you act like your US ego was flawed. Reminds me of your current president, time for a tweet. 
-branadic-

respectfully,

how does Edwin being a "US GUY" enter into this discussion?
are we so gone under our biases to rise above these petty matters?

let us not belittle each other in ANY FORM OR MANNER.

and perhaps we should recognize Mr. Edwin as being one of the top PWW manufacturers with
decades of experience on the subject being discussed and try to learn from him.

in eastern culture, somebody just being elder in years commands respect,
i don't know about the culture being represented here.

best regards and i am sorry if i have offended anyone.

-zia

[Simon]

Can we leave nationality out of this please.

[Edwin G. Pettis]

Branadic.....

As noted earlier, General Resistor has made some minor changes in its resistors, that is not uncommon, I also make changes in my resistors, in both of these cases, the internal basic construction has stayed the same, i.e. the bobbin design and lead assemblies have not changed, what we both have done is to make internal changes in the attempt to improve the resistor's performance.  Since the interplay of materials in resistor construction is complex, sometimes a 'problem' does not immediately show up and can take months to discover, this is not uncommon either.   I did have a problem with epoxy several years ago, that affected only two customers and I replaced both of their orders at no cost, that was one of the problems that did not show up until much later.  Where I am located, the humidity is usually not that high and both of those customers were located in much higher humidity locations, the epoxy, despite what the data sheet claimed, had a less than great reaction to the high humidity over time.  That of course caused a change in material as one would expect.  There has been no further problems.  If there has been a manufacturing defect, I have always replaced the parts as any manufacturer should and those have been very limited in number.

My point about qualifications is that I've got over 4 decades of experience, I already know just about every possible failure mode and most of the ways failure is caused, my examination would be based on that experience, I did not intend to infer that you were incapable of analysis as such, just that you are inexperienced in resistor technology compared to me.

If the loss of these two resistors is such a problem I will be happy to replace them at no cost if you're willing to cover postage.....okay?  Whether or not the failure was caused by handling or defect.

[babysitter]

Goodwill gesture noted!

[branadic]

Got a statement on those 12k resistors:

"...Since an order was in process to manufacture 18K resistors a quantity was utilized to create 12K resistors from the 18K resistor by stripping one side of the bobbin. This practice has been suspended going forward..."

This explains why the 12k are looking like they do. I was also offered a replacement.

-branadic-

[Magnificent Bastard]

Got a statement on those 12k resistors:

"...Since an order was in process to manufacture 18K resistors a quantity was utilized to create 12K resistors from the 18K resistor by stripping one side of the bobbin. This practice has been suspended going forward..."

This explains why the 12k are looking like they do. I was also offered a replacement.

-branadic-

Wow...  I will NOT be doing business with THEM ... !!

[Andreas]

Hello,

to go further on the topic ("measurements") I have the first result of a 8G16 1K resistor.

datecode 1704

Box TC around 12 ppm/K over 30 deg C
Hysteresis as deviation to LMS approximation up to 39 ppm so similar to the 120R resistors.
Drift over 4 days 7.6 ppm where in this case most of the drift (7.1 ppm) was between day 1 and 2 so it is stabilizing.
(could also be some effect from soldering).
Whereas on the 120 R resistors I observed a more constant drift between the days.

with best regards

Andreas

[Andreas]

Hello,

in the mean time I got another 120 Ohms 8G16 with different date code 1722.
You have all seen him already here (naked = x-rayed).

https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg1402107/#msg1402107

but obviously x-raying precision resistors is a no-no.
It seems that the poor candidate got either radiation sick or he feels ashamed.

In any case "best" (highest) T.C. value up to now large hysteresis and also some drift over time.

and a updated 120 Ohms comparison ...

with best regards

Andreas

[TiN]

Ran some numbers on three VHP100T 1K's from eBay 
Measurement setup is same I used before, two 3458s and little TEC box controlled by K2510.

[Andreas]

Hello Illya,

which ramp speed did you use?

with best regards

Andreas

[TiN]

Andreas,

speed was 0.125°C/minute. Time-scale data here and here, including RAW files.

[Andreas]

Hello,

and now the missing 8G16 1K resistors all with date code 1704.

#3 shows a near linear T.C. and the lowest hysteresis from the 3 measured devices.
But all T.C. values somewhat high against the 5 ppm/K max from the data sheet.

And again a overview: this time against the latest 2 measured 1 K resistor types (ZVAR + UP805)

Next will be a PTF56.

with best regards

Andreas

[Andreas]

Hello,

first measurement on a PTF56 (Vishay) metal film resistor. (PTF561K0000BZEB)
http://www.vishay.com/docs/31019/ptf.pdf

Intention was to use them as a better version of my INL adjustment cirquit for my ADCs (currently built with 25ppm/K resistors)
So long term drift is not a issue. But short term drift (T.C.) is annoying with my changes in lab temperature.

Result:
T.C. around 1.3 ppm/K (5 ppm/K according to data sheet) not bad for $1.5 a piece in small quantities.
Hysteresis around +/-2 ppm but most of it seems to be ageing drift in warm cycle.

I will cycle the first resistor for some days to see if the drift stabilizes. (just being curious if they could be used for a LTZ too).

with best regars

Andreas

[Magnificent Bastard]

Hello,

first measurement on a PTF56 (Vishay) metal film resistor. (PTF561K0000BZEB)
http://www.vishay.com/docs/31019/ptf.pdf

Intention was to use them as a better version of my INL adjustment cirquit for my ADCs (currently built with 25ppm/K resistors)
So long term drift is not a issue. But short term drift (T.C.) is annoying with my changes in lab temperature.

Result:
T.C. around 1.3 ppm/K (5 ppm/K according to data sheet) not bad for $1.5 a piece in small quantities.
Hysteresis around +/-2 ppm but most of it seems to be ageing drift in warm cycle.

I will cycle the first resistor for some days to see if the drift stabilizes. (just being curious if they could be used for a LTZ too).

with best regars

Andreas

The Vishay PTF resistors are very good-- they also use some special epoxy that is supposed to be more impervious to water vapor.  Most thin-film/metal-film resistors can be stabilized by a 400-hour bake at 125C-- most of the movement of the absolute value will happen during that bake-- then, at lower temperatures they will be exceedingly stable.

I'm currently looking at the SEI RTAN series of tantalum-nitride resistors.  Same thing: after a 400-hour bake at 125C, the tantalum-pentoxide layer is thickened about as much as it ever will be, resulting in a resistor that will be almost as good as a secondary standard.  These have TCR down to 10ppm/K, but those a are about 4X the price of the 25ppm/K ones-- so buy a 25ppm one with a value that is twice what you need, and then find two (after the bake cycle) that will cancel each other for an almost zero TCR.  Those are very low in cost, but there is a lot of labor in the selection process.  The end result should be good enough for an LTZ reference.  Bonus: these totally ignore humidity; so no humidity effect on their performance at all.   

[branadic]

Nice finding Andreas. I wonder what's the drawback of this resistors... one possible is, that you need to buy a 20k and a 50k resistor to form 70k and that the 50k with 5ppm/K rating is hard to get. RS Components stocks only the 10ppm version, while they have all other values (120R, 1k, 12k) with 5ppm/K. Digikey stockss 80,6k with 5ppm/K, which is as close as you can get.

-branadic-

[mimmus78]

I'm currently looking at the SEI RTAN series of tantalum-nitride resistors.  Same thing: after a 400-hour bake at 125C, the tantalum-pentoxide layer is thickened about as much as it ever will be, resulting in a resistor that will be almost as good as a secondary standard.  These have TCR down to 10ppm/K, but those a are about 4X the price of the 25ppm/K ones-- so buy a 25ppm one with a value that is twice what you need, and then find two (after the bake cycle) that will cancel each other for an almost zero TCR.  Those are very low in cost, but there is a lot of labor in the selection process.  The end result should be good enough for an LTZ reference.  Bonus: these totally ignore humidity; so no humidity effect on their performance at all.   

What happens due to phisical stress to those resistors?

[Magnificent Bastard]

I'm currently looking at the SEI RTAN series of tantalum-nitride resistors.  Same thing: after a 400-hour bake at 125C, the tantalum-pentoxide layer is thickened about as much as it ever will be, resulting in a resistor that will be almost as good as a secondary standard.  These have TCR down to 10ppm/K, but those a are about 4X the price of the 25ppm/K ones-- so buy a 25ppm one with a value that is twice what you need, and then find two (after the bake cycle) that will cancel each other for an almost zero TCR.  Those are very low in cost, but there is a lot of labor in the selection process.  The end result should be good enough for an LTZ reference.  Bonus: these totally ignore humidity; so no humidity effect on their performance at all.   

What happens due to phisical stress to those resistors?

You mean with board stress?  It's G10/FR4 epoxy-glass vs. ceramic substrate; so, nothing really.  These are cheap enough you could prove me wrong.  The data sheet does not give typical values for long term stability; but viewing their engineering data, I was able to determine that the long term load-life is about +200ppm (125C for 1000 hours).  That is 0.02%, and that is competitive with foil resistors in epoxy packages.  Also, as already stated, most of that +200ppm is in the first 400 hours of that 1000-hour burn-in, with very little change after that.

I have been unable to determine how SEI is getting such low TCR for these resistors.  Ta₂N film resistors are typically -125 (+/-25) ppm/K, and SEI now has these down to 10ppm/K (spec, so typical will be better).  In the scientific literature, I found one paper where a noble metal is simultaneously sputtered along with the tantalum in a 5% nitrogen + 95% argon atmosphere.  The high TCR of the noble metal (probably gold) compensates for near zero TCR (if they "hit it" just right).  They might also be using a sputtering target that has had a small amount of gold alloyed with the tantalum.

These are some pretty amazing resistors for the price!  US$0.21 each for a 1000-piece reel.  They are NOT as good as (say) a VPG FRSM, that has guaranteed TCR of <0.5ppm/K (typically 0.2ppm/K) between 0^oC and 60^oC.  But, the FRSM is about US$7.00 in 1000-piece quantities.

[ManateeMafia]

Is this the paper?

https://www.researchgate.net/publication/234454148_Stability_of_Tantalum_Nitrides_Thin_Film_Resistors

[Magnificent Bastard]

Is this the paper?

https://www.researchgate.net/publication/234454148_Stability_of_Tantalum_Nitrides_Thin_Film_Resistors

Nope, but thanks for the link-- I will read that paper too.  The SEI RTAN series uses gold contacts to the resistors, so it looks like they are addressing some of the long term contact stability issues with that.

Digikey is hosting a training module on these resistors:
https://www.digikey.com/en/ptm/s/stackpole-electronics-inc/rtan-tantalum-nitride-thin-film-chip-resistor/tutorial

And here is a video:
https://www.digikey.com/api/videos/videoplayer/smallplayer/5355134798001

[RandallMcRee]

Hmmm, not impressed.

The susumu RG series that Diligent Minds pointed us to has TCR of <=5ppm and the samples I have live up to that. Prices are reasonable, too, TCR on the RTAN is 25-50ppm.

https://www.mouser.com/Search/Refine.aspx?Keyword=susumu+rg

[amspire]

Typical annoying data sheets that don't mention anything that they do not want to talk about.

What is missing is the voltage coefficient. It can be a big factor in film resistors.

Vishay sell a mil-spec Tantalum Nitride resistors that are 0.5ppm/V. That is still a 5ppm shift at 10V - definitely an issue for metrology.  This is where wirewound are superior.

If voltage coefficient is not defined for the RTAN resistors, then there is a good chance it is worse. I did see a Welwyn graph that was showing that for two different types of film they tested, film resistors with gold terminations had a much bigger voltage coefficient then lead-silver connections.

[Magnificent Bastard]

Typical annoying data sheets that don't mention anything that they do not want to talk about.

What is missing is the voltage coefficient. It can be a big factor in film resistors.

Vishay sell a mil-spec Tantalum Nitride resistors that are 0.5ppm/V. That is still a 5ppm shift at 10V - definitely an issue for metrology.  This is where wirewound are superior.

If voltage coefficient is not defined for the RTAN resistors, then there is a good chance it is worse. I did see a Welwyn graph that was showing that for two different types of film they tested, film resistors with gold terminations had a much bigger voltage coefficient then lead-silver connections.

I agree except that NiCr resistors, no matter how good the passivation system, will degrade quickly in the presence of water vapor.  Ta₂N resistors simply ignore water vapor, and even oxygen after a thick enough tantalum-pentoxide layer has built up.  I think I stated that you would need to do some selection and also a burn-in process to get the results that I wrote about, as well as combining two (or more) resistors to get the low TCR needed.  Yes-- you can't buy a single Ta₂N resistor that will be as good as a PWW or foil-- but with some labor added, you can end up with a resistor that has very good temporal stability.

As far as the voltage coefficient goes, thin-film resistors are going to be the same as metal-film, and those are pretty good.  But you are right; this should be in the data sheet.  In addition, this is (in general) a fixed constant is it not?  If it is fixed, then does it really matter, since a voltage reference is operating at a fixed DC value?  I can see where it might be a problem for an audio signal.

IRC/TT Electronics also sells the PFC series, but I think the RTAN is higher performance in terms of temporal stability.

[amspire]

As far as the voltage coefficient goes, thin-film resistors are going to be the same as metal-film, and those are pretty good.

I gather thin film resistors have a typical voltage coefficient less then 2ppm/C which is probably good enough to ignore most of the time, but for precision resistors, it would be handy to know whether it is 2ppm/V or 0.5ppm/V like the Vishay resistors.

[TiN]

Moar resistor tests.

Fluke 2464674 HV resistor network from Fluke 8846A, pin1-2, 9.88 MR : -8.3 ppm/K, linear
Fluke 2464674 HV resistor network from Fluke 8846A, pin3-4, 99.86KR : -5 ppm/K, linear
Fluke 2464674 HV resistor network from Fluke 8846A, pin8-9, 8.98KR : -1.8 ppm/K, linear
Fluke 2464674 HV resistor network from Fluke 8846A, pin9-10, 89.9KR : -1.93 ppm/K, linear
MPX 1.8MR, 0.5%, WW : -9.4 ppm/K, linear
MPX 1.8MR, 0.5%, WW : -1.3 ppm/K, little curvy
PTF56 75KR : -5.7 ppm/K, linear
PTF56 86.7KR : -2 ppm/K, linear
Riedon BMF 15K from Digikey, 15KR : -0.24 ... +0.18 ppm/K curved, limited +20..+41c
VPG BMF custom VHP/VHA 95KR : -0.3 ppm/K, curved, limited range +20 to +38c
Wirewound 1W 800R : -14.8 ppm/K flat
Edwin PWW, 120R : +1 ... -2 ppm/K curvy
Edwin PWW, 70KR : +2.1 ... -3 ppm/K curvy

[Andreas]

so the xray really did change the ppm drift? or it is a coincidence?

I will ask dumb questions, this specification is 100% certified

nope, just kidding