T.C. measurements on precision resistors

[plesa]

TaN resistors are for example Vishay NOMCA ,are you sure they are using nichrome ones?
http://www.vishay.com/docs/60117/nomca.pdf

[zlymex]

Thanks @DiligentMinds.com for the reply, seems their claims were not entirely true, who ever 'they' might be.

Also, there is this thing puzzling me:
One Wavetek 7001 specified as 1.8ppm/year for long term drift, consisting of one LTZ1000 plus one set of resistor network(what ever the network may made of).
On that 10mA current source paper, they claimed the annual drift is 1ppm. How a LTZ1000 plus one resistor network of the same type, plus four Z-foils, plus another resistor network for 1mA to 10mA ratio, achieve better drift than a 7001 with fewer variables?
(attached two excerpts from that 10mA article)

[zlymex]

TaN resistors are for example Vishay NOMCA ,are you sure they are using nichrome ones?
http://www.vishay.com/docs/60117/nomca.pdf

Thanks for the info. The only thing I'm sure of is 7001 were using Vishay TDP1603 resistor networks. There are a lot of TDP1603 resistor networks in Fluke 8508A as well, but, Fluke seems using better resistor networks(Vishay 1446) around the reference board:
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/184/

 

[zlymex]

I'm pretty sure that both the 7001 and the CERN reference use the 'A' version of the LTZ...

Nope.

[zlymex]

Well on the CERN reference, there is a 10V output and a 10mA output.  Obviously, the 10mA is derived from the 10V, and so the 10V will have less than 1ppm/year drift, and the voltage to current converter will add a small amount of drift, totaling up to <= 1ppm/year.

The CERN reference was designed much later than the 7001, so I'm assuming that the designer learned some things along the way to make it better. 

And now is much later than the CERN reference, no one has designed a 10V reference with specification better than 1ppm/year.

[quarks]

according to att. CERN schematic, they use a LTZ1000ACH

[quarks]

here is a Fluke comparison spec information

[Kleinstein]

Interesting to see a rather different circuit to do the temperature regulation at the LTZ1000. 

[Andreas]

Hello,

finally I have measured the last 120 Ohm PWW (UP805) resistor.
So its time to show the overview sheet:

The measured values of the UP805 are well within spec (3 ppm/K).
So I can now build a 2nd set of LTZ references with selected sets of UP805 resistors.
This will be mostly LTZ1000 (non A) version and perhaps 1-2 LTZ1000A.

With best regards

Andreas

[Andreas]

Hello,

this article

https://www.eevblog.com/forum/metrology/prototyping-my-at-home-resistance-standards/msg1015157/#msg1015157

brought me to the idea: perhaps there is also a even cheaper possibility?
Those Dale RH25 have a aluminium case. But the sealing is simple epoxy with all probabilities for humidity influence.
What about simple cemented wire wound resistors?

The datasheet looks promising: +/- 10 ppm/K for the Vitrohm KH208-8 resistor.
http://www.vitrohm.com/content/files/vitrohm_series_kh_-_201501.pdf

So I ordered some 1K and 12K resistors (perhaps for a cheap LTZ module?).
I did a quick measurement of T.C. with a peltier.
So perhaps the temperature sensor has a slightly different temperature than the resistor.
But for a first impression this should be sufficient.

First I measured the 12K 10% resistor.
Around 400 ppm change over 23 deg C temperature
-> -17 ppm/K. (ok they say +/- 10 ppm/K in datasheet).

The 1K resistor has 870 ppm over 37 deg C temperature change.
-> -23 ppm/K

So the T.C. is obviously too high for a LTZ reference.
And obviously the resistors also need some thermal treatment before
they could be used as "precision" devices, since the drift is also rather high.

With best regards

Andreas

[Nuno_pt]

Andreas did you find something useful on the VitrOhm website, for resistors to try?

[Andreas]

Hello,

I only took the resistors which where in stock by the local catalog distributor. (Reichelt).
The datasheet of the resistors is linked in previous post.

With best regards

Andreas

[Nuno_pt]

Hi Andreas, yes I see it, I was asking if you look at VitrOhm website to see if there is some resistors worth to try.

I'm looking and only see CR series can be in 0.5% and till 3k9 ~10ppm/K.

You can look here - http://www.vitrohm.com/

Let me know if you find something that is worth to try, to see if I can get them, the Portuguese factory is about 1km from my house.

[Andreas]

Hello Nuno_pt,

I was not shure if it makes sense to use power resistors which are trimmed for cost in a precision application.
In my "lab" (with 18-32 deg C) I will need resistors which have a maximum T.C. of around 1 ppm/K.
So the 0 +/-10 ppm/K spec is already high.
Another thing to consider is the stability spec. 3% for ageing drift is also much more as on precision resistors which are usually specced below 0.05%.
The absolute tolerance for the 12:1 or 13:1 ratio is not so relevant. As long as it stays over temperature and time.

After looking at the catalog: Yes it seems that the CR series would be the best for (semi) precision applications.
It is the only resistor with a welded construction. (The others are crimped).

From size the 3K9 values are rather bulky compared to other precision resistors (8E16, UPW50 or Edwins 805 style).
So the range of resistor values (with CR 254 maximum size) is further limited.

So it was worth a try to test resistors which are a factor of 10-100 cheaper than PWW. But I think it does not make sense for me to do further tests.

with best regards

Andreas

[Nuno_pt]

Andreas, that's what I was thinking also, my workshop also varies alot in temp range, and I can't find anything stable from VitrOhm.
If there was something that was worth to try I could ask a friend of mine that is working there to bring me a couple of them, but with at least the 10ppm/K for the CR series, and like you said with good PWW resistors, it makes no sense to ask him to bring a couple of them.

I can ask him if they have anything better then 10ppm/K.

[Edwin G. Pettis]

Power resistors are not 'precision' by nature or construction.  In the case of these particular power resistors, they made the obvious decision to switch to Evanohm instead of Cupron (most commonly used) for some of the value ranges, they managed to reduce the usually higher TCRs by this simple change of alloy but this does not make a 'precision' out of a power although they have managed to tighten up the tolerance a bit.  Many of the smaller power resistors are/have been welded for many years but the 'weld' isn't the problem here as it is in most precision resistors.  The problem is in the alloy used for the end caps to which the wire is welded, in the case of the CR series (and other similar parts), changing the alloy to Evanohm managed to bring the TCR down somewhat but the end caps have a fairly high TCR, the combination results in a lower TCR but also a nonlinear TCR, notice that the TCR increases with lower values and that they are still using the same old alloy on the lowest values.  In power resistors a crimped joint is not necessarily bad unlike precisions, the conditions present in a 'mechanical joint' do not harm power resistors like they do in true precision resistors.

If you need precision, you have to use precision resistors, they are designed for that use (well at least some are), power resistors are designed just for that...power not precision and unless the circuitry design can tolerate the 'sloppier' performance of powers, stick with precisions to do the job right.  If you are going to insist on PPM performance you aren't going to get it with dirt cheap resistors.

[Kleinstein]

Some of the digital pot might be an option for a relatively stable and cheap resistive divider. Overall resistance is not that good, but the ration can be reasonable good (e.g. 5 ppm/K for the ratio in AD5260) but the choice of higher voltage types is not that large.

[Edwin G. Pettis]

Digital pots are not intended for 'precision' use, while the main divider ratio has a relatively low TCR around 5 PPM/°C typical at half scale only, the TCR varies as it approaches either extreme.  There are no limits on the TCR specifications, only typical and they will vary with temperature.  You will also find that the noise is going to be significantly higher than precision wire wound resistors or foils.....you are wasting your time with parts that will not perform as you need them to.

As I've said before, if you want precision you are going to have to pay for it, there are no 'cheap' short cuts when it comes to precision, you want PPM performance it is going to cost and there is no cheap way around it unless you accept less performance.

[acbern]

Edwin, looking at precision resistors (say in the 5ppm/C, 0.01% range) what is your experience re. ball park figures for the annual drifts per year for these technologies:
-metal film
-metal foil
-wirewound
all related to non-hermetic versions. I am referring to drifts not caused by temperature (so humidity, general aging). This is a broad question, sure, and depends on many influences, but you, or someone else, may have some experience. Seems this is nowhere specified (only load life changes, which are very high; only Caddock has some data with its ceramic type resistors)
All I can contribute is metal foil, non-hermetic, in the 5-10ppm range pa (controlled lab environment; set of 5 resistors tested). Wirewound is certainly better, more material; and no clue about standard metal film resistors.

[Edwin G. Pettis]

Shelf drift, just laying around in a regular room environment, tends to have the lowest drift numbers because it is the easiest environment to live in for a resistor, no power, relatively small temperature range, no on/off cycles.  This figure varies a lot depending on resistor technology and to some degree the value.  For film and foil, they tend to have a bit more effect from humidity than wire wounds due to their construction and very fine circuit paths internally.  Approximate shelf drift for film usually runs in the range of 20-50 PPM/year, sometimes a bit lower depending on the variables.  Foil drift may be as low as 10PPM/year but can also be as high as 35 PPM/year approximately.  Precision wire wounds can vary as well depending on construction, as a general range many are in the 20-35 PPM/year but some can be lower.  As a general rule Ultrohm Plus are spec'd at <10 PPM/year but can achieve much lower drift rates.  I have some 200 ohm resistors that I made for Bob Pease several years ago which I have been keeping an eye on (spares), they have drifted less than 5 PPM over the years (that was about 7 years ago).  This drift can be influenced by value and environment to some small degree, I usually spec a conservative <10 PPM/year in most cases, it is usually less but most of the data I have is just feedback from customers who do track such data.

These are just general figures, the specific resistor manufacturer and type of part/construction determines the actual drift value along with environment, the figures I have given should be taken mainly as guide lines rather than absolutes for any given resistor type and should apply only to shelf 'life', other drift figures are specific to given conditions such as humidity, power cycling and thermal shock for example.

I might add that TCR and tolerance have no effect on drift figures, it is the same resistor no matter what the actual TCR or tolerance is.  TCR is a characteristic of the given alloy and construction; tolerance is merely the limits on the nominal value.

[Dr. Frank]

For Thin Film resistors, Beyschlag created the best (most serious) specifications.

The drift depends on the resistance value, i.e. the lower, the thicker the Thin Film, the more stable, and mainly on  the outer passivation / lacquer, if it's tight, or not.
But it depends lesser / not at all on the T.C.

As a ballpark, humidity and temperature variations create about 100 ..500 ppm of change.
Usually, Thin Film were not tested and specified for Shelf Drift, as this technology makes no sense for longterm stability.

A good example for such a specification of Beyschlags ultra precision series UXA can be found here:

http://www.vishay.com/docs/28726/uxa0204.pdf


Foil resistors have a drift comparable to PWW, I'd guess, and as specified,  as these are real bulk material type resistors, not comparable to thin or thick film technology.


Frank

[acbern]

As a ballpark, humidity and temperature variations create about 100 ..500 ppm of change.
Usually, Thin Film were not tested and specified for Shelf Drift, as this technology makes no sense for longterm stability.

Foil resistors have a drift comparable to PWW, I'd guess, and as specified,  as these are real bulk material type resistors, not comparable to thin or thick film technology.

Thanks, I would think though that foil resistors are also some type of metal film, so as long as they are not hermetic I would expect a similar aging drift behaviour as with metal film. And I would expect wirewound to be lower. Also, I would rather expect 10 to 50ppm/year drift for metal film/foil (shelf life/low power; not temperature drift/extremes), otherwise a 0.05% better resistor would not make much sense. Do you have any specific data supporting 500ppm/year?

[Dr. Frank]

As a ballpark, humidity and temperature variations create about 100 ..500 ppm of change.
Usually, Thin Film were not tested and specified for Shelf Drift, as this technology makes no sense for longterm stability.

Foil resistors have a drift comparable to PWW, I'd guess, and as specified,  as these are real bulk material type resistors, not comparable to thin or thick film technology.

Thanks, I would think though that foil resistors are also some type of metal film, so as long as they are not hermetic I would expect a similar aging drift behaviour as with metal film. And I would expect wirewound to be lower. Also, I would rather expect 10 to 50ppm/year drift for metal film/foil (shelf life/low power; not temperature drift/extremes), otherwise a 0.05% better resistor would not make much sense. Do you have any specific data supporting 500ppm/year?

Usually, metal film is identical to Thin Film technology, with film thickness on the order of several hundred nm.

Metal Foil has several µm thickness.

I made no statement about the annual drift of TF resistors, as I've never seen any shelf stability data.

These 100..500ppm are absolute drifts after humidity and temperature / power dissipation stress.

It's correct, anyhow, that < 0.1% trimming accuracy for TF makes no big sense, as the stress induced drifts cause changes of the same order of magnitude.

Frank

[Alex Nikitin]

Shelf drift, just laying around in a regular room environment, tends to have the lowest drift numbers because it is the easiest environment to live in for a resistor, no power, relatively small temperature range, no on/off cycles. ... Precision wire wounds can vary as well depending on construction, as a general range many are in the 20-35 PPM/year but some can be lower.  As a general rule Ultrohm Plus are spec'd at <10 PPM/year but can achieve much lower drift rates.  I have some 200 ohm resistors that I made for Bob Pease several years ago which I have been keeping an eye on (spares), they have drifted less than 5 PPM over the years (that was about 7 years ago).  This drift can be influenced by value and environment to some small degree, I usually spec a conservative <10 PPM/year in most cases, it is usually less but most of the data I have is just feedback from customers who do track such data.

As an example, I've bought a set of LT450C PWW 0.002% tolerance 3ppm/C max resistors, made in 1989. All of these drifted up less than 50ppm from the nominal value (so less than 70ppm max in 27 years). Two 10K resistors I've measured came up as +30ppm one (I'm using it now as a reference, after trimming it down with 300M in parallel, so far it is almost scary stable) and +15ppm another.

Cheers

Alex

[mimmus78]

Hi everybody.

I come out with my solution to do TC of resistors.
My method to TC of resistors basically consists in heating this resistors to 25°C, keep this temperature for some minutes, than heat up to 55°C and calculate TC.
I actually do no do negative temperature sweep and I'm thinking in doing some more complicated thermal profiles ...

The "thermal chamber" is realised by few simple components:

  - 2 power wirewound resistors
  - the resistor/dut
  - 2 pieces of Teflon tubes where the resistor leads are inserted
  - a temperature sensor based on the DS18B20
  - some aluminium foil

I found this setup very convenient, easy to realise and with very good thermal properties.
Despite having a big chunk of metal, aluminium foil still has good thermal conductivity but very low thermal capacity.
This means it react fast to the heat produced by the power resistor, and power is immediately transferred to the DUT without having the thermal inertia of the big chunk of metal.
Thermal transfer is also optimal because the resistor in encapsulated in the albumin foil leaving no space around it, even 70% of the leads are encapsulated into this aluminium foil.

The temperature sensor is inside the aluminium foil too, just near the resistor. This means the sensor is thermally in contact with the resistor ... with some more aluminium foil acting as thermal dumper.
Resistor and temperature sensor can have different temperature, but not too much because everything inside the assembly have equalised temperature.
All this assembly if fully encapsulated inside other aluminium foil that works as temperature equaliser/dumper.

The tricky thing was to develop a software that is able to counterbalance heat loss with heat produced by the power resistor to keep a relatively stable temperature ... but this is another episode.

I attach some photos of the assembly so if someone want's to replicate what I did will be more easier to do it.

PS: I added also a thermal switch that cut off power if temperature goes over 70°C ... don't want to set on fire my lab