T.C. measurements on precision resistors

[Andreas]

Hello,

first measurement on 150 deg C baked resistor 8G16#4 in comparison with the previous measurement.
The resistor was baked by branadic (thank you for that)
on 14.05.2018 for 7 hours at 150 deg with a previous 1 hour ramp up and a slow cooling down phase.
Snail mail creeped extra slow so the resistors reached me on 18.05.2018.

On evening 18.05.2018 I wrapped one 1K (#4) and one 12K (#4)
with adhesive aluminium tape to reduce humidity exchange with environment.

First measurment on 1K#4 is done on 19.05.2018
old measurement of 24.04.2018 before baking is also shown.
I used the same offset on both measurements so you see that during baking the resistor drifted +100ppm.
Hysteresis is somewhat reduced (+/-7ppm against +/-9ppm)
The next measurements will show if the treat gives permanent change or only temporary.

With best regards

Andreas

[Andreas]

Hello,

further measurements of the Aluminium taped resistor:

I have not the impression that the tape reduces humidity exchange.
Obviously the surface of the not taped front ends is large enough.

On 2nd day (20.05.2018) the average 25 deg resistance value increased slightly by ~5ppm against first day
or ~105 ppm against the value before baking.
Hysteresis is already near the old value before baking.

Further days show that the resistor is stabilizing on the new level.
(And that I have problems to reach the cold temperature because environment temperatures are rising).

So the hope that the hysteresis is from a "fresh" resistor and can be cured by baking the epoxy has not come true (at least for this sample).

with best regards

Andreas

[Andreas]

Another baked sample 8G16#5 this time without aluminium tape.
Now several days between baking and measurement so air humidity could already work.

as reference measurement of 27.04.2018 (before baking).
14.05.2018 baking to 150 deg C.
first measurement on 24.05.2018

Again we can see a shift (this time around +85 ppm due to baking).
Hysteresis after 10 days in air similar to that before baking.
I really have to move in a cooler room now: today 30 deg C in my "lab".

with best regards

Andreas

[3roomlab]

what if you redo the measurement cycle by putting the resistor in a tiny tub of mineral oil (or melt it into wax?)? will that exclude humidity?

[rhb]

In the construction of precision levels of 10 arc seconds and better, the cast iron bodies are subjected to multiple heating and cooling cycles to anneal them and prevent warping.  As these measure 0.0005" in 10.0"  100 ppm distortion would ruin them. The sides where you hold them are insulated to prevent body heat transfer.

It seems to me that an interesting experiment would be to measure  value and tempco for a small batch of  resistors and then subject half of them to  sinusoidal temperature cycles for several days which gradually converged to 25 C starting with the stated OEM temperature range.  Then compare the aging behavior and tempco over a small temperature  range.

Among other things, it might lead to some insight into why there is such a range of tempco behavior in parts from the same batch.

[Andreas]

what if you redo the measurement cycle by putting the resistor in a tiny tub of mineral oil (or melt it into wax?)? will that exclude humidity?

Hello,

I am pretty shure that (pure white) mineral oil will help against humidity.
With wax (which one) I am not shure if it will be really tight.

But I fear I will not do this experiment as I am looking for resistors that I can use afterwards in a LTZ cirquit.
And even the thermal grease to remove after the first measurements was a mess that I do not want to repeat.

with best regards

Andreas

[cellularmitosis]

what if you redo the measurement cycle by putting the resistor in a tiny tub of mineral oil (or melt it into wax?)? will that exclude humidity?

Hello,

I am pretty shure that (pure white) mineral oil will help against humidity.
With wax (which one) I am not shure if it will be really tight.

But I fear I will not do this experiment as I am looking for resistors that I can use afterwards in a LTZ cirquit.
And even the thermal grease to remove after the first measurements was a mess that I do not want to repeat.

with best regards

Andreas

3roomlab, I have plans to do humidity characterization on some precision resistors, I've just been too backlogged to start down that road just yet.  Soon!

[GerryBags]

You could drill two holes in a plastic bottle cap for the resistor leads, put the resistor in the bottom of the cap and then fill it with vermiculite/perlite. Even a little heat from the resistor should drive off any moisture in the material directly adjacent to it and keep the air surrounding it drier than the air outside the cap. Dry the vermiculite out in an oven before hand and you should be able to get a very low humidity around the resistor itself without anything to clean off afterwards but a little dust.

[babysitter]

You could try some fluoro-chemical stuff, like everybodys favourite vapour phase solder medium Galden, or Novec 7100, or Flourinert.  Some datasheet-rading ahead but at least its great for thermal coupling but I think water would not be solved in it but form a separated layer on top or on bottom, depending on density.

[3roomlab]

according to my compulsive FEMM simulation disorder, the metal sleeves appears to improve temperature uniformity inside a resistor, the difference between a naked (UPW like) resistor is about 2 to 3 fold (but this is owing to a 1C pin to pin temp difference wrt the naked UPW). if the "lip" of the "canister" pulls around near to the leads, the advantage jumps to maybe 5+ fold? im not sure what exact advantage this would hold  . but if we look at those fluke precision resistors which are heated, they do take alot of effort to drive everything at the same temperature. its always a temperature problem somewhere ... always ! bloody celcius and kelvins!

flour-inert? they seem like insulation? max 0.06W/m.k?
but i did run into something interesting, alumina-powder oil mix. im not sure about its homogeneous/colloidal ability, maybe the alumina powder have to be very very fine, 1000 mesh? 2000 mesh? i think a 13:87 ratio = about 1 W/m.k. which by comparison with other non conductive fluids, i think is strangely fantastic. by coincidence i read about them when looking at milling fluids. individually (if google is to be trusted), the resistivity should be around 10T 100T ohms?  plausible use? maybe? again im not sure.

not sure if it is worth mentioning, articles about sulphur in oils
https://www.doble.com/wp-content/uploads/2008-DBDS_Destruction_Lewand_and_Reed.pdf
https://patents.google.com/patent/US2177343

inhibitors
http://dsiventures.com/specialty-cooling/sulfur-inhibitor-fluid/

[SvanGool]

according to my compulsive FEMM simulation disorder

Thanks for the simulation, very interesting ! 

In order to exploit your disorder, I have some questions:

• Can you explain in detail what your picture shows?
• What are the blocks on the right side?
• Which are the leads with associated leakage (I see branches on both the left and right sides, or is this an axial resistor)?
• What are the materials you assumed for the sleeves, encapsulation, leads and resistor body?
• What thermal sources/loads did you assume?

[3roomlab]

Thanks for the simulation, very interesting ! 

In order to exploit your disorder, I have some questions:

• Can you explain in detail what your picture shows?
• What are the blocks on the right side?
• Which are the leads with associated leakage (I see branches on both the left and right sides, or is this an axial resistor)?
• What are the materials you assumed for the sleeves, encapsulation, leads and resistor body?
• What thermal sources/loads did you assume?

the model of the resistor is done based upon the Xray photos posted by branadic on the WW resistors. 3 major materials here is air, plastic and metal.
the blocks on the right try to mimic some random load after the resistor.
so yes its a axial WW resistor, the left end is injected with 50mW. the gif is flipping in 2 frames, (for the top resistor, it is with a closing lip, and 1 w/o). but what i dont know here is, do fluke resistor have leads going deep into the resistor body?

the materials property in FEMM are to some scale, metals:plastics:air >18000:21:2. i sort of quantize them in blocks to mimic a depth, FEMM assumes a very deep solid material (1inch thick FR4? 12inch thick? i think? for air it might be ok, but i think WW resistors are not 12 inch thick). i do this by roughly converting from W/m.k to C/Watt or Watt/C and back to W/m.k. and then, inject some heat into some fixed length in FEMM to check/rescale the temperature. this chain of re-adjustment no doubt have many sources of error, but at least i know the values and result are trying to reflect say a 1.6mm thick FR4 with 1oz copper on top and not 12inches of copper which is a huge false reading. so it is just a reduction of error from aspects of thickness, but whether other errors will dominate? im not so sure.

[Andreas]

some further results on 120R 8G16 resistors (baked at 70 deg C a little longer ago)

and the overview over all 8G16_120R resistors.
So the replacement parts that branadic got are now all within T.C. spec (5 ppm/K)
Some are even below 2 ppm/K.
But hysteresis and drift (most probably due to humidity) are still rather large.

with best regards

Andreas

[Andreas]

Hello,

some further results on the 150 deg C baked 1K 8G16 resistors.
All now measured some days after baking. -> all with large hysteresis.

When I look at the overview table there is some correlation
between the drift measured on the 10-40 deg C temperature cycles
(before baking) and the shift due to baking.

Resistors with large drift before baking also have a large drift due to baking.
So I am wondering if there is also a correlation between baking drift to shelf life ageing.
Unfortunately my reference resistor is not stable enough for long term tests with only a few ppm drift over a year.

with best regards

Andreas

[EmmanuelFaure]

Unfortunately my reference resistor is not stable enough for long term tests with only a few ppm drift over a year.

It's time to upgrade!

[Andreas]

Hello,

here some Z202 resistors. 1K Datecode B1708-

http://www.vishaypg.com/docs/63131/z202.pdf

spec is ± 0.2 ± 1.8 ppm/K or ± 0.05 ppm/°C "typical"

reality near room temperature is more somewhat "<1ppm/K"
The 2 samples have opposite sign of T.C.

and a overview (together with the last measured 8G16)

with best regards

Andreas

[Andreas]

Hello,

another comparison before and after baking at 150 deg C.
this time the 12K0 8G16 resistors (some weeks after baking).

In average there is a 20-30 ppm drift due to baking (with one exception).
I have also a larger humidity (especially on the last measurement) than on the measurements before baking.
So the larger hysteresis follows the increased humidity.

And again a comparison table.

with best regards

Andreas

[TiN]

Doing some matching/sorting of recent batch of BMF's. All these are brand new 2018 mfg, no PMO or other special order requirements.
Delivery time was 181 days.

Results so far:


Non-hermetic S102KT kinda funny wobbly due to major humidity/hysteresis effects. VHP's on opposite are steady and linear for vast majority.

Boxes with labels:



Still have about 10 of 120 ohms and 10 dividers to go thru.

[guenthert]

  Now while a wee bit too large to be placed onto a PCB, those "Thomas style" (? the construction seems very similar, although at 10kOhm the used material will be quite different from those super-stable 1Ohm resistors) certainly count as precision resistor.  This soviet made P331 is of the .01% class (don't know over which temperature range that applies) and well aged.   I received it only recently (via Fleebay from Ukraine) and can't offer any long term drift data.  I don't have a climate chamber either, so I put the DUT out on the balcony in the morning and after about an hour brought it back in and measured it while it warmed up to room temperature (which is currently fairly high).

  Neither my Datron-Wavetek 1271 (which I used to take the measurements for the plots below) nor my HP34401 are recently calibrated (if ever), but they eerily agree about the value of this resistor (so much even, that I fear some mind-over-matter mechanism is in play 

  The plots are from two sessions a week apart.  Each consists of multiple series with data points roughly distributed equal in time.

  The near-zero TC point seems to be close to 300K, which puzzles me.  Is this by accident or intention?  Did its zero-TC point perhaps drift over the years or was the resistor meant to be used in a warm air or oil bath?

  With some good will, one recognizes a parabola there, which is a bit deformed in the left leg (early measurements) of the first graph.  Not sure, what caused that, but I'd guess, that there the temperature of the DUT was rising too quickly, so that the temperature as reported by the NTC in the thermometer well didn't well reflect the actual temperature of the resistance wire.

[ManateeMafia]

guenthert,

Is it possible to determine the type of metal used in the binding posts? I have wondered about those resistors and what was used to make them. Are the posts silver plated or possibly ferrous?

[MegaVolt]

This soviet made P331 is of the .01% class (don't know over which temperature range that applies) and well aged.

0.01% at 20 degrees. Works in oil or other silicone fluid.

Photo of interiors http://zapadpribor.com/r331/

[MegaVolt]

Thermostat for Р331: https://www.iztech.ru/termostaty/precizionniy_suhoblochniy_termostat_temp-2/

[MegaVolt]

Instruction P331

[MegaVolt]

Instruction P331

[guenthert]

Thanks a bunch.  Wished I'd be able to read Russian though.