T.C. measurements on precision resistors

[Vgkid]

Snip*

If you want to double your accuracy, you have to more than double your price
SRL - $1974.00-$4895.00, for the almost 5K unit I would guess it is for the low/high ohms units.
Adjustment to norm/year drift/ 18-25C, with 23.0 nominal
1 k?
±2 ppm /±1 ppm / 3 ppm tot
 
1.9 k?
±2 ppm / ±1 ppm / 2 ppm tot
 
2 k?
±2 ppm / ±1 ppm / 2 ppm tot
 
4 k?
±2 ppm / ±1 ppm / 2 ppm tot
 
10 k?
 
±2 ppm / ±1 ppm / 1.5 ppm tot
 
19 k?
 
±2 ppm / ±1 ppm / 2 ppm tot

[Edwin G. Pettis]

Hello EmmaneulFaure,

Manganin does not have a perfect zero slope TCR at the cardinal temperature, at the very least not consistently, the hyperbolic TCR curve of Manganin is inherent in the alloy, it is 'considered' relatively 'flat' for about ±5°C from the cardinal reference before the TCR starts rapidly increasing.  Zeranin, a relatively new derivative, increases the relatively flat portion of the TCR curve to about ±10°C from the cardinal before taking the same serious change in TCR rate.  I believe there is one or two other derivatives which make small tweaks to the 'flat' portion of the curve and extends the upper range to about 65°C, this not to say the TCR remains as low as at the cardinal point, this extension was made by changing the mix of metals in the alloy slightly, the overall characteristics are still basically the same.

Low ohm resistors and shunts are still made with the Manganin family although it is not generally used for standards any longer being replaced by the Evanohm family alloys, they not only have a very linear TCR but their stability is better and are tough to boot.  Manganin cannot be operated at elevated temperatures, not much above 65°C because they anneal at very low temperatures and that changes them permanently.

Manganin's lack of use is primarily because they have been replaced by superior alloys and better technology in using those alloys.  Manganin was easy to use because it could be soldered (carefully), Evanohm cannot be soldered, it must be welded.

No matter what the alloy used for a low TCR / high stability resistor, the processing of the resistor demands very careful controls and depending on the technology used, those processes are considerably different.  My processes are very different from Vishay's as there are different variables involved.  Another consideration is that current metallurgy cannot control the batch mix of metals tight enough to provide consistent characteristics batch to batch, the fact that we can produce very low TCRs with very long low stabilities these days can be attributed to some 'cherry picking' of the alloy characteristics as every batch of alloy is different.  While the resistor industry can push the alloy providers for ever tighter mix controls there is a limit as to how tight they can get it for a reasonable price.  My most expensive wire runs over $18,000 a pound which depends on the wire size, I have no control over that aspect.  I have a very limited number of alloys which can meet my wire specifications, outside of those, there are none.

Just remember, whether it is a PWW or a metal film/foil resistor, your TCR limits is going to cost more and more as you demand zero TCR because those very low TCRs are in short supply, we really can't create them at will unfortunately.

[MK]

does that mean that for the LTZ1000 temp circuit it would be cheaper to get a matched pair od 1-3ppm resistors than to insist on 0 ppm for both?

[Edwin G. Pettis]

Yes, once you start specifying TCR under 0±3PPM/°C (as an example), asking for TCRs 0±1PPM/°C will immediately increase the price over and above the tolerance cost (tighter tolerance, higher price), then depending on who you are asking, matching (tracking) TCR will involve another price multiplier and if the values are far enough apart, you may encounter yet another price point.  Vishay does supply multiple resistors in a matched configuration, some are even on the same substrate, I'm not sure what their pricing for these 'custom' networks are of late, but as I recall, the price point is fairly high and the delivery is long.  These networks do have very good specs within their limits.

I do not have the capability to put multiple resistors on a single bobbin mainly due to physical problems, I have given a lot of thought on how to do it, no practical solution as yet.  The PWW solution to multiple resistor networks has always been to encapsulate them in an appropriate box, while this works, there are thermal limitations to it.  It also runs up the cost for the labor and materials to do this.  I have been asked by customers for a 'dual' resistor bobbin design, I am looking into it.

I have supplied resistor networks to customers with the recommendation that they wrap a piece of copper tape snugly around the resistors, side by side and after assembly, encapsulate the entire board in foam to prevent air drafts.  While this arrangement has a bit longer thermal trail, once the board reaches thermal equilibrium, everything tracks tightly and stays in track.  I use certain manufacturing techniques in making these networks, they do cost more than a standard line part but their performance has been reported as exceptional and they have enhanced processing for long term stability.

Sorry I can't divulge the customer's names, they are under NDA so I can't say much about the details either except that tracking has been well under 1 PPM°C.  I can build the networks for anyone else as they are my designs of course.

[ManateeMafia]

Edwin,

Would the use of a thermal epoxy like this http://www.arcticsilver.com/arctic_silver_thermal_adhesive.htm be a good way to thermally couple resistors and have you heard of any issues arising from its use?

[Edwin G. Pettis]

Hi,

I don't foresee any problems using this stuff although it might be more expensive than copper foil tape but it should provide a good thermal conduit between the resistors.  I certainly wouldn't us it on bare foil resistor though. <grinning>  I would still strongly recommend the foam encapsulation of the board, it doesn't have to be too thick and it really works, keeps everybody inside at a pretty even temperature.  It definitely would not bother my resistors to use this adhesive.

[texaspyro]

Even the best thermal expoxies are rather poor conductors of heat.  I make my own out of either silicon carbide or diamond.

Also, when using copper foil to link the two resistors together you should use a thermal transfer medium (i.e. thermal paste) between the copper and the resistors.  Small air gaps between the copper and the resistors have a great effect on the thermal coupling.  And, as with all thermal pastes,  use the bare minimum amount needed to fill the gaps...  thermal pastes also have horrible thermal conductivities.

[Edwin G. Pettis]

While it is true that unfilled epoxy has lousy thermal conduction properties, there are many 'filled' epoxies that do have much improved thermal conduction.  Glass fibers or balls, metallic oxides (for non-conduction) such as alumina, ceramic and in the case of the filled epoxy asked about here, it is filled with silver.  All improve thermal conduction.  Depending on the formulation, the thermal conduction properties can be increased considerably, the drawback is extra cost.

The object here, in the case of thermal equilibrium is not necessarily the lowest thermal conduction possible but keeping the resistors in the network at the same temperature (or close to it, nothing is perfect) and at the same temperature as the assembly is a plus.  Therefor one does not need the best possible thermal bridge between the resistors, after a bit of time has passed, they will reach equilibrium and any apparent drifting will cease.  Copper tape does achieve this and does it quite well at a reasonable cost, thermal paste may or may not be an advantage here, the adhesive on the tape tends to fill in most of the gaps and any micro gaps tend not to be significant.

Perfection sounds great in theory but in reality, it doesn't work that way, there is a point of no return for effort, trying to get that last degree can be very costly and in the end, not significant over all.

[paulie]

I have supplied resistor networks to customers with the recommendation that they wrap a piece of copper tape snugly around the resistors

I believe the concept, known universally as "passive heat-pipe technology", might have first been put into practice by vref pioneer and chief engineer of Calibratory Corp and designer of the highly acclaimed D105. At least held in high esteem here on Eevblog forum in the thread devoted to that product.

[Edwin G. Pettis]

Nope, that copper tape concept has been around for quite a few years, nothing new.

[janaf]

If you are referring to a, by now, conventional Heat Pipe, it's a pipe partially filled with liquid and vacuum, to start with. The pipe is hermetically sealed in both ends. The liquid has low boiling point and boils of in the hot end and condenses in the cold end. It will transfer much more heat than solid copper! But it works best / only with the pipe vertically, hot end down. It may only work with significant temperature differences, I don't really know.

I believe the concept, known universally as "passive heat-pipe technology"

[paulie]

Sorry, that was another poor attempt at sarcastic humor relating to the somewhat horrific feeding frenzy in this thread:

https://www.eevblog.com/forum/testgear/calibratory-d-105-dc-precision-voltage-reference-standard/240/

you came here with a superiority complex....and then made wild, bogus claims about things like "heatpipe" technology....which your device clearly does NOT employ.

We can start with claims about your miraculous "heatpipe" technology.....how about claims of somehow bending the laws of physics, to improve Ti's technology?  Face it, the ref you are selling simply proves the quality of Ti's REF102C package...DESPITE your best efforts to corrupt it's implementation.....

The fellow was a little off his rocker with the vampires and God stuff but the voltage reference itself performed well beyond expectations in multiple reviews there and elsewhere too yet was the object of IMO mostly unjustified technical criticism. Personal attacks from otherwise professional and knowledgeable individuals were despicable.

However I should mention that the teardown (literally) there of Awesomes product was the first time I saw anything like that tape trick and the earliest photos searching the internet were from his implementation. I'm sure he wasn't really the first and true it's not what's commonly called "heatpipe" but still...  Let's drop the subject before this excellent thread gets contaminated any more. I apologize for the diversion.

[janaf]

OK, get it. I followed that thread in the beginning too but dropped out, so yes, please, lets drop it here too.

[IanB]

It is about 4000 times more efficient at moving heat than a similar sized copper bar might be-- and is much lighter as well.

That number of 4000 is very large and rouses the skeptic in me. For instance the linked Wikipedia article has a statement,

The effective thermal conductivity varies with heat pipe length, and can approach 100 kW/(m?K) for long heat pipes, in comparison with approximately 0.4 kW/(m?K) for copper.

--and even that ratio of 100/0.4 = 250 seems itself remarkable.

I suspect this is an area where the implementation details make all the difference and generalized statements are of limited applicability.

[MK]

One company I used to work at used heat pipes, and at a low delta T they do not perform as well as one might hope.

[Andreas]

Hello,

it´s time to come back on the topic.
For those who have forgotten: its T.C. measurements on precision resistors.

After a long journey I picked up some resistors from local customs office today.

5 days from Grand Junction to San Francisco
2 days from San Francisco to East Coast
2 days to Germany
and finally after nearly one week intensive customs treatment I could pick them up.

In the mean time I have rearranged my T.C. setup with the "golden Z201" as reference resistor.
So after having made a suitable fixture for the Ultrohm Plus resistors I will start with T.C. measurements.

With best regards

Andreas

[ltz2000]

This is going to be interesting. Especially the hysteresis (or lack of it).

After a long journey I picked up some resistors from local customs office today.
5 days from Grand Junction to San Francisco
2 days from San Francisco to East Coast
2 days to Germany
and finally after nearly one week intensive customs treatment I could pick them up.

Based on the plastic colour they traveled from the 1970s.

[3roomlab]

@andreas, how much (ball park) did those cost you?

[Andreas]

Hello 3roomlab,

pricing for the 0.1% 3 ppm/K standard tolerance which I ordered is on the LTZ1000 thread:

https://www.eevblog.com/forum/projects/ultra-precision-reference-ltz1000/msg602719/#msg602719

So its comparable to that what you will pay for UPW50 or 8G16 resistors from RS / Farnell in small quantities.
But with the service that you can get every individual resistor value.

Shipment was around +11% for the 33 resistors with USPS.
And import tax (=VAT) adds +19% on the total in germany.

With best regards

Andreas

[Andreas]

Hello,

after measuring the golden Z201#6 I had the hope that the other Z201#7 that I ordered at the same time would be behaving similar.

But first surprise: the datecode was B1315- (and not B1305- as on the "golden" resistor)

Second surprise: T.C. + hysteresis is much larger.

The results of Z201#7 Datecode B1315- on 21.03.2015
T.C. with box method: 0.818591506 ppm/K

3rd order LMS interpolation referenced to 25 deg

A 0 = -1.43422451376099E-0001
A 1 =  9.64007179982438E-0001
A 2 = -1.26483309879412E-0002
A 3 = -3.34940837358552E-0004

So T.C. at 25 deg is 0.96 ppm/K

max deviation from LMS 2.57933723463504E+0000 ppm (hysteresis + noise)

In the mean time I am in doubt what the data sheet terms
"typ. TCR" and "max spread" really  mean.
"typ." seems to be the average value over a large number of batches and samples.
And with "max spread" I am not shure wether the 1 sigma or the 3 sigma limits are meant.
Up to now I had expected that max spread means something with 3 sigma. But I have already
2 samples which are above +/- 0.2 +0.6 ppm/K on Z201 at least for the slope at 25 deg C.
But the data basis is still not sufficient for any statistics.

I have also attached the overview and included Franks results for the 1K00 resistors.
The hysteris cannot be compared directly since the noise
on Franks HP3458A is about a factor 5 lower than my setup
and is determined "graphically" at 25 degrees.

The test fixture for Edwins Ultrohm Plus 805 style resistors (I will refer to them as UP805) is nearly ready.
So I will most probably start with first measurement on 1K resistors tomorrow.

With best regards

Andreas

[janaf]

Would appreciate feedback on this idea:
https://www.eevblog.com/forum/testgear/temperature-test-box-for-component-characterization/msg635672/#msg635672

[3roomlab]

Hello 3roomlab,

pricing for the 0.1% 3 ppm/K standard tolerance which I ordered is on the LTZ1000 thread:

https://www.eevblog.com/forum/projects/ultra-precision-reference-ltz1000/msg602719/#msg602719

So its comparable to that what you will pay for UPW50 or 8G16 resistors from RS / Farnell in small quantities.
But with the service that you can get every individual resistor value.

Shipment was around +11% for the 33 resistors with USPS.
And import tax (=VAT) adds +19% on the total in germany.

With best regards

Andreas

thanks for that. they look rather affordable 

[janaf]

Andreas, I don't want to hijack your thread, but I'm doing my first measurements right now, showing results "real time" in thread
https://www.eevblog.com/forum/testgear/temperature-test-box-for-component-characterization/msg636219/#msg636219

[branadic]

Edwin, please let me ask, why are your resistors look like they do? Is this (ugly) yellow color some sort of corporate design? My guess is some black jacket would have been more effective in advertising.

[Edwin G. Pettis]

Well, despite the fact that yellow is my favorite color, I didn't pick the color, my colleague picked it and it is the original color.  For one thing, it does stand out.  In the many instruments our resistors went into, you couldn't miss them.  We weren't the only one, one other resistor house here in the States actually had different colors for each of their different series including red, blue and green.  I must admit, that green was kind of sick looking, matter of fact, my bobbin supplier actually insisted that I take some bobbins that he had made up in this 'green' color, I've still got some of them in stock.  Print will show up quite nicely on the yellow, black particularly well, white will also work.  Some of the other resistor houses do use a black shell but they really don't 'stand out' and trying to wind wire on a black bobbin....doesn't work very well at all.  For the time being, they will have to stay this color, particularly as long as the inventory holds out but even then, I do not think I will go for black.  Ultronix used a relatively dark blue but then the supplier stopped making it, they ended up having to use a sickly green color as there was only one source for the stuff.  Ultronix did make a line of shelled resistors which were black, I still have some of them.

Besides, yellow has been used many times, Sprague Atom electrolytics for example and there are still some capacitor manufacturers that are still using yellow so there is old precedent for the color.

I must say, you are the first one to say they are ugly <grinning>