Author Topic: T.C. measurements on precision resistors  (Read 271744 times)

0 Members and 4 Guests are viewing this topic.

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #250 on: February 13, 2015, 09:28:55 pm »
Sorry I wasn't clear: what I meant, implications of cycling a whole completed assembled PCB to relief mechanical stress on components?
Thermal cycling is sometimes recommended for removing mechanical stress on sensitive components. Do you have any thought on that based on the measurements?
Heating removes humidity from PCB. So the following thermal cycle will get lower hysteresis.
Thats an old trick to let voltage references with plastic package look better in the datasheet.
But it will not help you in real life applications. (except when you heat up the whole device).

With best regards

Andreas
my2C
Jan
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2430
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #251 on: February 13, 2015, 09:42:48 pm »
Sorry I wasn't clear: what I meant, implications of cycling a whole completed assembled PCB to relief mechanical stress on components?
I fear my English is not good enough to understand what you mean.
But also a pcb stores humidity and gets dry by thermal cycling for the next measurement cycle.

With best regards

Andreas
 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #252 on: February 13, 2015, 09:54:39 pm »
Or it is my English. Not native....

One more try: It is sometimes recommended to thermally cycle a whole assembled board, to get rid of stresses on components, caused by mounting and soldering. If you do that kind of cycling, then what happens to resistors? Is it possible to make a guess, if thermal cycling of a board assembly causes more harm that it does good?


Sorry I wasn't clear: what I meant, implications of cycling a whole completed assembled PCB to relief mechanical stress on components?
I fear my English is not good enough to understand what you mean.
But also a pcb stores humidity and gets dry by thermal cycling for the next measurement cycle.

With best regards

Andreas
my2C
Jan
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2430
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #253 on: February 13, 2015, 10:33:08 pm »
Hello Jan,

interesting question.

Usually thermal cycling or burn in is used to sort out the boards with "early failures" (bathtub curve/Weibull)
so enhancing the reliablity at least of those who survived.
So I would not stress them much more than that what can be expected in later usage.

With best regards

Andreas



 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #254 on: February 13, 2015, 11:18:14 pm »
 :)

About thermal hysteresis and relief by controlled thermal cycling, you might enjoy this : :)

http://www.google.com/patents/US5369245
my2C
Jan
 

Offline Edwin G. Pettis

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: us
  • The plural of anecdote is not data.
Re: T.C. measurements on precision resistors
« Reply #255 on: February 17, 2015, 05:54:49 pm »
Regarding the Z201T and similar physical types, the thermal resistances are not identical from the resistance element to the 'front' or 'back of the molded package.  Since the element is mounted on a ceramic substrate, using the element as the 'front' side, the thermal resistances are slightly different.  The element goes through a slight gap to the molding (which is a pretty good thermal conductor) then to the surrounding air.  From the element to the 'back' side, it goes through the ceramic substrate, through another slight gap, to the molding then to the surrounding air, the ceramic substrate being the main difference in thermal paths between the 'front' and the 'back'.  While standing the resistor on end will provide the most equal path from the sides to the air, the internal paths are not identical.  Vishay is also correct in that the leads are the primary thermal conductors to the outside world being directly connected, as such, to the resistance element.  Laying the resistor on its 'front' or 'back' will cause a larger difference in the thermal paths to the air unless the resistor is mounted to a heatsink, which will provide a lower thermal resistance but will again unbalance the thermal paths.

'Burn-in' has been around for decades, primarily used for the military who demanded the very best possible of the times.  Later it began to extend down into industrial applications where reliability was fast becoming very important as well.  As a general rule, burn-in has not been used for commodity electronics because of the cost and the general attitude that consumers really don't care much about whether the electronics (or mechanics for matter) last for more than a few years.

In the case of using burn-in, it varies with the purpose, in the case of individual components, it is often used to 'stabilize' the component and in turn, weed out the ones which are not good enough.  In the case of assemblies, burn-in has a similar intention of finding possible premature failures, in the case of high reliability assemblies, most of the components have already been through the stabilization bake, the main purpose of doing it to assemblies is to find bad connections of any type.  This particular type of 'burn-in' does not stress the assembly any higher than the maximum working temperature of the assembly whereas individual components may be stressed above their rated limits already before assembly.

As noted earlier, the type of PCB material can significantly influence board level difficulties, soldering is one of the worst offenders these days.  The TC of most common PCB materials, such as FR-4 and similar types can be quite significant and many are sensitive to humidity, all of which are a bane to SMT components.  The fact that everyone is trying to shrink components even smaller is only contributing to the problems and making it worse.  Leaded components rarely had any problems with the normal board stresses but leaded components are on the hit list, manufacturers are doing their best to drop them from active production.  Burn-in can uncover the faults on a board assembly but it doesn't fix the root cause.
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 1724
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #256 on: February 17, 2015, 07:15:35 pm »
Well, the Vishay BMF resistors are symmetrical concerning the wires, i.e left / right orientation.
There is a theoretical difference in thermal resistance between front/rear orientation, that's right.

But I really think, that's not relevant at all for such measurements and concerning this observed hysteresis, or to say more precisely, temperature lagging.

As the leads transport 90% or more of the heat, there might only be such an effect, if there would be a different coupling over the wires.
Let's say, one would be directly attached to the ceramic body and the resistive foil, and the other would have been bonded..

(That's a well known problem on power semiconductors and on LEDs).

Anyhow, as the Pomonas won't show up before midth of March, I already assembled Andreas Z201 into another aluminium box..

The small alu block takes care, that  front and backside of the resistor is virtually on the same temperature, as is the NTC sensor.

This alu block is a thermal shortcut as the outer alu box, so everything inside will be on the same temperature +/- a few hundredths of a Kelvin, if the measurement is done slowly enough.


Frank
« Last Edit: February 17, 2015, 07:17:09 pm by Dr. Frank »
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 1724
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #257 on: February 17, 2015, 10:38:09 pm »
[...]so everything inside will be on the same temperature +/- a few hundredths of a Kelvin, if the measurement is done slowly enough.

With your measurement septup, it's not likely to happen.

Again : The main thermal conduction path is the leads of the resistor, not the case. You should make a proper thermal contact on the leads, not on the case. Your ugly purple/grey/black/white thick wires are going sink a lot of heat from your resistor, and cause a temperature difference between the resistor element and your measured temperature (on the block) in an order of magnitude which exceed "several milidegrees".

Making a precision thermal system is similar to precision analog electronics. Since you supply the heat from a finite "impedance" source (The conductivity of aluminium between your heater and the resistor), for high precision applications you have to minimize leakage from the resistor to the outside. Make your purple/grey/black/white wire as thin as possible!

Nice thermal contacts via the leads + Small leaks via thin wires and polystyren-like insulation everywhere + A low mass = A fast and precise thermal system.

You got it right, Dude,

these ugly cables are really thin, indeed. So they won't transport that much heat.

Due to the big mass of the alu block and box, there really will be no temp differences inside than more than a few 10 mK..

As the whole alu box is also situated inside an isolation box (for cooling beer), there is a very  even temperature distribution, from the bottom to the top of the (beer) box.
Through this, the cables have to pass also, again reducing possible temperature differences from the cables themselves.

Well, the most perfect setup would be, to "cover" the leads of the resistor also to the alu box, but I had no time to do so..

Anyhow, if there really is a temperature lagging between the resistor and the alu box (i.e. the NTC), this will be visible in the measurement.. If there is  no or a very small measurement loop during fast cool down or fast warm up, then your theory of heat transfer via this cable is busted.

If I get a measurement loop even on slow warm up / cool down, then you 'd be right, and I would have to improve my setup.

Frank
« Last Edit: February 18, 2015, 12:09:18 am by Dr. Frank »
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 1724
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #258 on: February 18, 2015, 11:06:14 am »
these ugly cables are really thin, indeed. So they won't transport that much heat.

Sure they will transport some ugly heat! :D

Let's say they're #24 AWG (~0.5mm diameter of pure copper) even if they appear thicker than that. When your chamber is at 50°C, the ambient at 25°C, and there is 30cm of such wire connecting the DUT to something at ambient temp, let's make some simple calculations of what happens :

Thermal conductivity of copper=385W/(mK). Thermal resistance of your two pair of wires : 4 wires * 385 * Pi*0.00025² / 0.3 = 0.001 W/°C <=> 1000°C/W
Thermal resistance of the block-resistor_under_test interface : 2W/°C <=> 0.5°C/W  (Very favorable. It's the typical interface resistance of a well mounted power transistor with silicone pad or grease. Or the minimum available junction to case thermal resistance of a TO220 package)

Thermal flux through the resistor interface and the wires = Delta Temp / Rth = 25°C / (1000 + 0.5) = ~25mW

These 25mW cause through the interface a temperature loss of : Tloss = 25mW * 0.5°C/W = 12.5mK

And it's a very very favorable case.

Don't forget that when you add aluminium mass to improve temperature distribution, you decrease the thermal resistance between two points, but you also add thermal capacity. At some point, the "induced" thermal capacity increases beyong what's acceptable, regarding to the benefits of a lower thermal resistance.

From a dimensional analysis point of view, the thermal resistance is a matter of contact/tranverse area, and the thermal capacity a matter of volume. The volume increases with the cube of a dimension of your system, and the tranverse area only with the square. If the figure of merit of your system is Resistance/Capacity, it varies with x^2/x^3 = 1/x ("x" being a dimensions of your system) : Small systems are naturally favored.

Yes, your approximate calculations seem to be right, and  I also think, they are even on the correct order of magnitude, i.e. not especially favourable parameters chosen.
I used 1,27mm ribbon cable, that should have AWG 28 wires, i.e. 7 x 0.125mm.
This already gives a factor of 2 higher thermal resistance, therefore in the end 6mK temperature difference only.

You also have to consider, that there is some heat transfer between the wires and the surrounding air, and on the first 5 cm, to the temperature of the alu box (it's been taped to the box there) .

The maximum temperature differences are also much smaller, 15°C at most (10°C .. 35°C in the alu box, 20°C room temperature)

So, your calculations even support my estimations which I made for the alu box itself: I.e. less than 1/100 K temperature difference caused by the wires! That's nearly a perfect thermal isolation for that used case, and can be taken out of consideration.

This is also confirmed by the first measurement, I quickly did yesterday night and this morning:
There was no observable temperature difference / shift, due to fast  cooling or heating  ..


Frank
« Last Edit: February 18, 2015, 11:31:28 am by Dr. Frank »
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2430
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #259 on: February 22, 2015, 05:23:48 pm »
Hello,

to find out what created the results of the posts of 08.02.2015 and 13.02.2015 with "rectifier effect" and different hysteresis face down and face up I did further measurements with a new setup which are more towards the "Emmanuel" style.

https://www.eevblog.com/forum/projects/t-c-measurements-on-precision-resistors/msg603949/#msg603949
https://www.eevblog.com/forum/projects/t-c-measurements-on-precision-resistors/msg608306/#msg608306

first Img1866 shows the "face down" setup of the previous measurements.

Img1868 the new setup "face down" which takes more care of the fact that most of the heat is transferred by the wires of the resistor.
2 thin (2 mm) AL-rails with a window for the resistor body and the main NTC.
The wires of the resistor are fixed between heat conductive silicon foils for isolation and heat transfer.
The measurement clips that I used for the last measurements are replaced by solder joints as in the first measurements.
(they would not really fit between the silicon foil).

Result:
1. the dependancy on polarity has gone.
2. the hysteresis is back nearly equally on all measurements.

So I blame the dependancy on polarity to the measurement clips.
(perhaps some bad contact due to oxidation).
The hysteresis is independant wether a fan is used with open TEKO enclosure or with closed TEKO enclosure.
So there seems to be a larger temperature constant within the resistor.

With best regards

Andreas

Measurements:

15.02.2015 Z201#3 face down normal polarity
16.02.2015 Z201#3 face down reverse polarity

17.02.2015 Z201#3 face down reverse polarity with fan
18.02.2015 Z201#3 face down normal polarity with fan

21.02.2015 Z201#3 face up normal polarity
22.02.2015 Z201#3 face up reverse polarity


« Last Edit: February 22, 2015, 10:26:11 pm by Andreas »
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2430
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #260 on: February 22, 2015, 06:00:15 pm »
Hello,

just to see if the self heating of the resistors creates the hysteresis
I did a measurement with reduced "half bridge" voltage of 2.5V instead of 5V.
Reducing the measurement current from 2.5mA to 1.25 mA.
So I am nearer to the comparative measurements of Frank.
His HP3458A has 1mA measurement currrent for the 1K range.

Edit:

The measurement of 19.02.2015 is with FAN,
can be compared with measurement of 18.02.2015 except 2.5V

The measurement of 20.02.2015 is without fan so corresponds to measurement of 15.02.2015.

Since the power is 1/4 with 2.5V I would expect a significant (factor 4)
change in hysteresis which is not visible.

Of course the ADC noise remains the same so I get a worse signal to noise ratio within the measurements.
The noise appears about doubled with 2.5 V bridge voltage.

With best regards

Andreas
« Last Edit: February 22, 2015, 07:04:27 pm by Andreas »
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2430
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #261 on: February 22, 2015, 07:13:59 pm »
Hello Ken,

2.5 mA in 1K gives 6.25 mW during measurement.
Pulsing is not possible within my setup since this would affect signal to noise ratio.
(I have to average many measurements to achieve around 1uVpp noise).

Temperature slope is 0.12K/minute. So with 7.2 deg C / hour I am not far away from your proposed 5C/hour.
And already this takes a whole day to measure when regarding that I cannot use the first 2 hours
because otherwise I get a "open" hysteresis curve. This fact is a indication that the mold around the resistive
element has some "creeping" effect.

Edit: see attached measurement from 21.02.2015 from 20 deg starting downto 10 deg up to 45 deg then down to 15 deg
instead of using only the part of 10 deg up to 45 and then back to 10 degrees C in the previous post.

Frank already emailed me that he gets similar results with his setup for my Z201#2 and S102#1.
But not with his old (green) Z201 with 49.999K
So for now the best guess is that the changed mold (green to black) of the resistors is guilty for the hysteresis.

With best regards

Andreas

« Last Edit: February 22, 2015, 07:25:49 pm by Andreas »
 

Offline Edwin G. Pettis

  • Frequent Contributor
  • **
  • Posts: 415
  • Country: us
  • The plural of anecdote is not data.
Re: T.C. measurements on precision resistors
« Reply #262 on: February 22, 2015, 07:28:23 pm »
All resistors have hysteresis, even those Thomas 1 ohm standards that have been sitting in constant temperature baths at NIST for decades has hysteresis by virtue of the fact that every time any, and I emphasize any, power is applied, it creates a history of change, the current 'value' of a resistor depends on its current state plus all of its prior states over time that has affected it.  Depending on the type of resistor and how it is made determines just how much reaction a particular resistor will have to these changes in states.  The more stable a resistor's characteristics, generally speaking, the smaller the hysteresis will be to the current stimulus.

Foil resistors can be quite 'stable' but can also have relatively high hysteresis because of their construction, the foil has very little mass and will suffer greater hysteresis to the same stimulus than a wire wound resistor might.  It is not solely the function of mass that bears on hysteresis but it can have a significant effect on the out come.  Hysteresis can also be difficult to isolate from other temperature effects, for example, the starting temperature of the resistor should be very accurately known because after the stimulus, the temperature should be returned to the original value to accurately assess hysteresis.  Any errors in temperature will manifest as an apparent hysteresis.  Another source of error is the measuring instrument, just how stable is it over the given time period in which the hysteresis is to be measured, when attempting to measure very small changes, every little detail in the chain becomes very important.  You can't measure PPM or sub-PPM measurements accurately unless you can control all of the variables sufficiently to minimize the error sources.  Are thermal EMFs known and under control?  The entire measurement chain is part of the measurement and result.

If you have analyzed your circuit needs accurately and have correctly picked the right resistors for the job, then hysteresis will likely play little if any role in the long term operation of your circuit.  In the case of these LTZ1000/A reference circuits, hysteresis, unless unusually large, will play an insignificant role in the performance and long term stability.  Under the normal operating conditions of the LTZ (and that includes fairly harsh environments) the noise floor is the limiting characteristic, most of the DC performance is buried in it, for these commercially available versions, real drift on the order of 1-2 PPM/year is as good, on average, as it is going to get.  Very long term drift may even get a bit better with time but like all Vref chips, their performance varies a bit chip to chip, someone is going to get a really terrific sub-PPM drifting chip (after some long period of time), others wouldn't be that good even after a few years.  Trying to change the chip's inherent characteristics by tweaking outside components or internal currents is pretty much a waste of time.  This chip was designed for best performance with the given information from Linear Tech.

The folks at Linear Technology knows as much as anybody about how to get the best performance out of them, if better performance is possible (commercially) then I'm sure Linear Tech would announce it, as it is, the LTZ is the best chip you can get your hands on and the best thing you can do to keep that performance is to use very stable, low noise resistors around it (and that eliminates a lot of resistors, including most of Vishay's).
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 1724
  • Country: de
T.C. measurements on 5 precision resistors
« Reply #263 on: February 24, 2015, 08:56:46 pm »
I measured R(T) / T.C. for 5 different precision resistor technologies, all with the same setup, i.e. using the 3458A with 4W and Offset Compensation, a precision NTC, in the same thermal box, with an aluminium DUT holder, and with equal temperature profile.

The leftmost, black  ones are 2 bulk metal foil resistors from Andreas, Z201 #2, Z-foil, TC < 1ppm/K and S102JT #1, C-foil, TC ~ 2ppm/K both have 1.0000k, 0.01%, Vishay brand.

The 3rd, green resistor is also a Z201, 49k99, 0.01%, but in an older packaging from about 2004.

The 4th, blue resistor is a bulk metal foil from Alpha Electronics, type FLCY,  TC typ. 2.5ppm/K, 25k, 0.1%

The 5th is an econistor from G.R., 1k, 0.1%, TC < 5ppm/K

The 1k resistors were measured at 1mA, the 25k and 49k99 at 100µA

Stability over time of the 3458A was monitored by an external 10.00000k reference resistor, and its deviation after the hours of experiment was < 0.5ppm in all cases.

The thermal box was inserted in a small beer cooling box, so that a stable air column was formed inside, which gives a continuous temperature layering, from cold at the bottom to room temperature at the top of the box. (picture shows measurement on another box)

By this 3 fold thermal encapsulation, and the thermal inertia of aluminium block and box, NTC sensor and DUT were on the same temperature during the slow temperature changes.

Inserting a cooling pad quickly cooled the assembly down to 10°C, and an incandescent lamp, by its IR emission, quickly heated to about 40°C.


The results show a pronounced hysteresis figure of both black BMF resistors, especially as the end values do not match the  starting values.
Also, a virgin curve can be observed.

After the thermal cycle, both resistors were held at a constant temperature of about  26°C.
A creeping of the (hysteretic) resistance value, directing towards the initial value, could be observed, in both cases about 1ppm/30min.
The Z201 shows a very bulged / distorted hysteresis loop, probably due to the creeping behaviour.

At this point, I have to apologize Andreas, for casting doubts on his measurements!
His measurements setup, using a much lesser costly equipment, gave qualitatively same or very similar results, and they were nearly on the same order of noise figures.
Well done, Andreas!

Compare here: https://www.eevblog.com/forum/projects/t-c-measurements-on-precision-resistors/msg462299/#msg462299
and here: https://www.eevblog.com/forum/projects/t-c-measurements-on-precision-resistors/msg462303/#msg462303



The three latter resistors do not show hysteresis, identifiable by the fact that the initial resistance value at ~ 26°C always closely matches the value after the temperature cycle.

Therefore, only a slight temperature lag between sensor and DUT can be observed, due to fast cooling/heating.


For the black BMFs, I can hardly assign a T.C., due to the hysteresis.
The Z201 changes about 5ppm in a window of 30°C, that's about  0.2ppm/K average TC.

The S102J is practically useless as a stable resistor, as the big hysteresis spoils the 2ppm/K specification.

The old Z201, 49k99, has a linear TC of +0.5ppm/K, w/o hysteresis.. very good.

The AE FLCY, 25k has a slightly hyperbolic characteristic; the linear TC would be around -0.9ppm/K.

The econistor has a relatively high, linear TC of about 4.6ppm/K.
I've already seen a better sample, of about +2ppm/K, recently.
Anyhow, I will have to measure all the econistors, 5EA of 1k and 12k, for my LTZ1000 , to match these pairs for best overall TC of the reference circuitry


There's one question left, for discussion:

I really wonder, why these newer, black BMF from Vishay have this hysteresis and creeping effects, in comparison to the oder Z201, and also the VHP202Z (in AT51 XTAL type can) I also have measured.
Maybe, that the smaller mold case causes this effect, or they may have simplified the inner construction, maybe omitting a mechanical buffer layer over the resistance element.
 
« Last Edit: February 25, 2015, 06:53:41 am by Dr. Frank »
 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on 5 precision resistors
« Reply #264 on: February 25, 2015, 10:27:22 am »
Good to see Dr Frank's results match Andreas's!

I'm not that surprised that Andreas is getting useful and accurate results. The setup uses differential measurements all the way and the signal changes are small, ie ADC DNL, noise, resolution and short term stability means more than INL and absolute accuracy. However hard to quantify....

It also seems the "thermal side" of his setup has also developed a lot compared to the first tests.

Yes, Well done Andreas :-+

At this point, I have to apologize Andreas, for casting doubts on his measurements!
His measurements setup, using a much lesser costly equipment, gave qualitatively same or very similar results, and they were nearly on the same order of noise figures.
Well done, Andreas!
my2C
Jan
 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #265 on: February 25, 2015, 10:52:11 am »
The lead material of the S10XX series is solder coated copper while the Vishay hermetrics are Kovar. I did not look up the lead materials of the other resistors.

The EMF of Kovar - Copper is 40uV/C while solder - copper is somewhere near 1uV/C

The 1uV/C doesn't make me worried but 40uV/C does. Any thoughts on that? Something seen on measurements?
my2C
Jan
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 1724
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #266 on: February 25, 2015, 11:10:20 am »
The lead material of the S10XX series is solder coated copper while the Vishay hermetrics are Kovar. I did not look up the lead materials of the other resistors.

The EMF of Kovar - Copper is 40uV/C while solder - copper is somewhere near 1uV/C

The 1uV/C doesn't make me worried but 40uV/C does. Any thoughts on that? Something seen on measurements?

The 3458A Ohm function cancels all such emf voltages, by its Offset Compensation feature.

Frank
 

Offline quarks

  • Frequent Contributor
  • **
  • Posts: 821
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #267 on: February 25, 2015, 11:14:42 am »
The lead material of the S10XX series is solder coated copper while the Vishay hermetrics are Kovar.

When I have a look at all my VPG Z-Foil resistors, they are all tinned.
I wonder where and why VPG should use Kovar.
Can you share where you saw the Kovar information?
 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #268 on: February 25, 2015, 11:20:41 am »
Still worried abouth the 40uV/C in real life applications. A change in 1/40C between the two pins of a resistor would cause 1uV error even if this can be cancelled with current reversal in measurements.
my2C
Jan
 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #269 on: February 25, 2015, 11:22:16 am »
I may be completely wrong but I was assuming the hermetic / metal cans use Kovar?

The lead material of the S10XX series is solder coated copper while the Vishay hermetrics are Kovar.

When I have a look at all my VPG Z-Foil resistors, they are all tinned.
I wonder where and why VPG should use Kovar.
Can you share where you saw the Kovar information?
my2C
Jan
 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #270 on: February 25, 2015, 11:26:02 am »
Wrong thread, I know, but it's the same problem with the LTZ1000ACH.

From the LTZ1000ACH datasheet:

Quote
The kovar input leads of the TO-5 package...
my2C
Jan
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 30130
  • Country: au
    • EEVblog
Re: T.C. measurements on precision resistors
« Reply #271 on: February 25, 2015, 12:14:00 pm »
Hi Dave,
This one, obviously:
http://www.vishaypg.com/foil-resistors/case-studies/study/wekomm_1/

Yep, that's the one.

Quote
But not good for the described big temperature range, as the Vishay foil resistors show hysteresis.

These are not standard Vishay parts, they are:
Quote
custom specific made part, together with a special packaging.

Quote
And 0.001ppm stability or uncertainty?
Nope, never... This German engineering GmbH would need a Klitzing Hall standard.
Even the best secondary standard, the ESI SR 104 (which  practically has no hysteresis) does not approach that level..

They claim to be working with  PTB (the german national standard institute) to achieve that tolerance on demand.

My resistor has been tested and is about 0.5ppm / °C
and is about 7.7ppm off nominal 10K.
They expect that to drift by up to 0.1ppm by the time I get it.

 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 30130
  • Country: au
    • EEVblog
Re: T.C. measurements on precision resistors
« Reply #272 on: February 25, 2015, 12:20:59 pm »
For the tolerance, you really meant 0.001% [10ppm], not "0.001ppm", right?
I don't think there is any way to even *measure* 0.001ppm-- well, maybe with a QHR and cryogenic comparator, but I think the leads connecting to the QHR inside the liquid helium Dewar will prevent you from getting to 0.001ppm [because the leads are not superconductors].
So, a "typo" right?

No, that is what they claim they are working on in conjunction with PTB
http://www.ptb.de/index_en.html
I have no idea if it is feasible or not.
 

Offline janaf

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: se
Re: T.C. measurements on precision resistors
« Reply #273 on: February 26, 2015, 12:35:38 am »
I looked up a few Vishay metal can hermetics like VHP100 series and the DON't use Kovar! Relieved!
Quote
Lead Material #22 AWG (0.025 Dia) Solder Coated Copper
my2C
Jan
 

Offline barnacle2k

  • Regular Contributor
  • *
  • Posts: 52
  • Country: de
Re: T.C. measurements on precision resistors
« Reply #274 on: February 26, 2015, 01:42:34 am »
I looked up a few Vishay metal can hermetics like VHP100 series and the DON't use Kovar! Relieved!
Quote
Lead Material #22 AWG (0.025 Dia) Solder Coated Copper


I don't think that info is correct.
Because copper to glass seals don't work and break due to dissimilar thermal expansion coefficients.

Edit: Ok thanks to diligent for the more detailed info on that.

Anyways: the results for the Z201 Resistors are ALARMING and i think at this point it we should contact Vishay about this.

I am now thinking about ordering those Audiofoolery naked Zfoils for my Project, since the whole circuit will be hermetically sealed and under transformer-oil anyways.
(Much cheaper then the hermetic zfoils)
« Last Edit: February 26, 2015, 03:39:18 am by barnacle2k »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf