T.C. measurements on precision resistors

[RandallMcRee]

No, not groups of four--but how about parallel combinations?

Randy

[Andreas]

Again some results:

this time VH102K - resistors 10K0 with datecode B0641:

all T.C.s have good linearity.

with best regards

Andreas

[TiN]

That's what I like about BMF resistors, usually they don't have curveys and other non-linearity stuff.
More runs:

Resistor	Value	Tempco spec	Number	Measured tempco
Caddock TF020R 4B + 9 + 12	35 kΩ	3S	Calculated by cell : 0.00 ppm/K	-0.19 ppm/K
Caddock TF020R 4A + 10	30 kΩ	2S	Calculated by cell : 0.17 ppm/K	-0.17...+0.30 ppm/K
Caddock TF020R 1A + 1C	200 kΩ	2S	Calculated by cell : 0.02 ppm/K	-0.24 ppm/K
Keithley 2001 R354	100 MΩ			+8 ppm/K ?
Keithley 2001 R366 PTF65 T16	1 MΩ			-2.87 ppm/K
Keithley 2001 R365 PRC HR175N 0.1%	78.7 kΩ			+2.70 ppm/K
Keithley 2001 R358 PRC HR125N	7.15 kΩ			+3.07 ppm/K
Caddock 1737 network element	89.982 Ω			-1.70 ppm/K
Edwin PWW 70K	70.006 kΩ			+1.80 ppm/K

[Andreas]

Hello some further measurements on VHP202Z 19K97 resistors with datecode 1739

It seems that Jason has intermixed his #1 and #2 value within his table.
His single measurements correspond to my #1 and #2 values.

with best regards

Andreas

[cellularmitosis]

Oops . Thanks again Andreas

[branadic]

VHP202Z, 70k, #1, #2, #3

Results: about -0.55ppm/C, -0.6ppm/C, and -0.45ppm/C.

Here I extended each run to multiple ramps, eventually settling on three ramps up and down, at 5min/C, with 30min plateaus at the top and bottom of each run.

Additionally, I added a "Savitsky-Golay" smoothing filter (raw data in light blue, filtered value in dark blue).  Thanks to Conrad Hoffman for mentioning this filter!  https://www.eevblog.com/forum/metrology/data-smoothing-excel/

Savitzky-Golay is fine, but I found the fastsmooth filter implementation is awesome as well and easy to implement in C too:

https://de.mathworks.com/matlabcentral/fileexchange/19998-fast-smoothing-function

-branadic-

[Andreas]

Hello,

to come back to the VHD resistors and the ratiometric measurement by my ADCs.

The math calculation has shown that when the 1K resistor is mounted towards GND and the 12K5 resistor is mounted towards VREF I get around 0.022 ppm/K error for a typical gain drift of 0.02 ppm/K according to the data sheet of the LTC2400.

The other way round (12K5 to ground and 1K to VRef) the same gain drift of 0.02 ppm/K generates 0.27 ppm/K error. That simply due to the fact that I only measure the divider output. To correct for the error I would have to measure also VREF so that the gain drift could be corrected. Instead the calculation with the nominal VREF leads to a large error.

I have made several measurements to find out if the results can be reproduced (and yes the errors are systematic) and a ratiometric measurement on a 10V reference with a HP34401A to find out which result is really true.

It is stated clearly that the measurements with the 1K resistor towards ground are those with the lesser error.
a) they agree with the HP34401A measurements
b) If I calculate the difference of the 12K5 and the 1K tempco of the single measurements I get the values of the HP34401A measurements.

On the HP34401A I first made the error that I did not switch to the high impedance mode in my software for the measurement logging. (so the first measurements were with 10Meg input impedance). But even for this error the relative measurements are robust.

attached the summarized measurements of the VHD resistors.

with best regards

Andreas

[cellularmitosis]

It seems strange that the tempco of ratio #1 is actually better than the individual tempcos of its two resistors, despite the fact that they are both negative (I would have expected a result which was some kind of average of the two)

Edit: and as always, thanks so much for your hard work on this Andreas!

[e61_phil]

No, not groups of four--but how about parallel combinations?

For the TC it doesn't matter if you put the resistors in series or in parallel

[Andreas]

(I would have expected a result which was some kind of average of the two)

Why that: we have small changes on the 2 resistors so in first order we actually get the difference in T.C. for the divider.

And finally the 70K VHP202Z resistors with datecode 1804

All are a bit drifty during warm phase.
to check if it is only hysteresis or permanent drift I repeated the measurement on 70K#3
Result is a average drift of 1.8 ppm during the both measurements.

And finally a overview of the VHP/VH resistors

with best regards

Andreas

[cellularmitosis]

Why that: we have small changes on the 2 resistors so in first order we actually get the difference in T.C. for the divider.

Ah, thank you, I wasn't thinking clearly 

[Andreas]

And again some measurements on 12K0 8G16 resistors from G.R.

These are from the replacements that branadic got on his 18K to 12K stripped resistors.
Datecode is 1807

All resistors show a relative large hysteresis (more than +/-10ppm) over my 10 .. 40 deg C range.
So obviously the epoxy/silicone is interacting with the resistance wire.

T.C. is relative low on all measured devices. (better than the 5ppm/K max specified).

But 3 samples show additionally a significant drift up to nearly 6 ppm (see overview table)
between the measurements of 2 days, which also can be seen as
open hysteresis curve on the single measurements.

with best regards

Andreas

[MisterDiodes]

Just out of curiosity - were those G.R.'s showing hysteresis thermal cycled at all before testing? 

I ask that because we had some G.R. 8E16A 13k's recently that had a bit of that hysteresis before installation - but once the boards were manufactured and stress relieved after soldering (thermal cycled) as per GR's suggested procedure for our application the hysteresis was greatly reduced or just went away.

The need to thermal cycle new resistors after assembly is common among the precision resistor makers - even Vishay VHP's and Z-foils have a recommended PMO (Post Manufacturing Operations - thermal cycle procedure) that varies by the resistor model, expected load condition, temperature range and application.

[Andreas]

Hello,

no there was no PMO from my side (and obviously also from manufacturer).
I also think that the hysteresis will go down to the "normal" values of around 5 ppm in this case over some wear in time.
These resistors are still "too fresh".

For the PMO procedure:
how much does it really help or does it only polish up the measurements for a few hours?

On some (old from datecode) 8G16 resistors I have made the experience that baking at 70 deg C gives
reduced hysteresis for some days which increases over 2 weeks if you have enough humidity (50-60% rH normally in my case).

https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg1374819/#msg1374819

so for me it seems that baking is only a temporary aid, which could perhaps help for new devices but has not much relevance on devices with some wear in. (may of course vary with local air humidity).

with best regards

Andreas

[ap]

PMO for VPG, according to VPG FAEs, is solely helping in reducing future drifts that would occure under load. If a resistor is operated under low load (shelf life conditions) it does not help. PMO effects for these resistors cannot be compared to other resistors. Epoxy resistors have different effects too, as already stated (humidity).

[MisterDiodes]

Actually -   "No Load" would be "No Load", and working with Vishay manufacturing Israel: the PMO reduces 1000hr drift / hysteresis effects at -all- power dissipation levels over 0 mW on a 10K VHP.  We've recently worked with them directly to confirm that fact; and a specific PMO was developed to meet a project drift spec at very low and variable  power levels.   Any resistor operating at true 'shelf life" conditions as defined by Vishay would really have to be at -zero- power, and I can't think of a good use case for that.

And yes - on the GR's we tested the conditioning thermal cycles had long-lasting positive effect (it's part of the board conditioning anyway), and the hysteresis hasn't returned after several months...so I'd say yes that's a positive factor under real load conditions (10V bias on 13k => 8mW), at least on our "real world" test case where the end product runs 24/7/365.

[e61_phil]

Any resistor operating at true 'shelf life" conditions as defined by Vishay would really have to be at -zero- power, and I can't think of a good use case for that.

If you use such a resistor as your (working) standard (private, at home) then it will last most of the time with zero power.

[TiN]

Why is it suddenly zero power , when used as working standard at home? Private households don't apply magic, and resistor still powered by DMM current source, even used alone...
I can definitely see different hysteresis and resistance deviation just from current change from different DMMs, especially on PWW type.

[ap]

VPG in their data sheets (some VHP) defines shelf life identical with less than 10mW load. Aging is driven by environmental effects and by power applied. They consider 10mW low enough to make no difference to shelf storage.

[e61_phil]

Why is it suddenly zero power , when used as working standard at home? Private households don't apply magic, and resistor still powered by DMM current source, even used alone...

I said most of the time. Which should be important for the drift. But, you are right, hysteresis is another story.

PS: I said private, because I don't wan't to start a discussion about if a VHP resistor is a "standard"

[BNElecEng]

Mister Diodes, could you please elaborate on how you burn-in the assembled boards you mentioned? I've tried looking up Post Manufacturing Operations but I'm coming up dry.

[MisterDiodes]

VPG in their data sheets (some VHP) defines shelf life identical with less than 10mW load. Aging is driven by environmental effects and by power applied. They consider 10mW low enough to make no difference to shelf storage.

Not at all.  You have to understand what Vishay datasheets leave out for VHP and Z-foil products, and don't ever take those at face value.

For instance - we just got done developing a a spec for a 10k VHP101 where we have to watch 1000hr life and hysterisis, and we were finding out the standard datasheet specs weren't exactly lining up with our test results.  That's when we started working directly with the source at Vishay Isreal and tracked down some better info.  For instance when they check TCR, that always at lowest possible power or <100uW (to avoid self heating).  Every mW power dissipation is going to affect 1000hr drift by up to 0.5ppm and affect temp by about 0.1C when mounted on a typical 1oz Cu PCB (but depends entirely on the actual mounting method and enclosure, air flow, etc).  The PMO operations are critical to get maximum performance for our application even at power levels of 10uW to 100uW.  Let alone 10mW.

Oh - and that VC they say is essentially zero on the standard datasheets?  In reality you find out that's more like a 3ppm/V on some of their VHP and Z-Foil ratio resistors on values over 1k.  Head's up if you're using these on an application where the resistor sees variable bias - say for the input of an ADC front end or similar.  That's why went back to a sealed PWW resistor for that location - for virtually no VC effect!.

So when you get down into PPM performance:  Every uW power and bias point really matters!

It all depends on the precision BMF product and model number so you have to work with Vishay manufacturing direct on a specific application - then you'll learn more about what's behind the datasheets.

[MisterDiodes]

Mister Diodes, could you please elaborate on how you burn-in the assembled boards you mentioned? I've tried looking up Post Manufacturing Operations but I'm coming up dry.

This was for a client project so I can't really give you an exact step-by-step - but nothing wrong with you contacting Vishay Precision Group yourself with your specific application production target goals.  You'll want to be a customer of VPG (that helps a lot ) and remember every application is different.  You'll need to know temp range range required, bias/power  points, mounting method, an idea of thermal resistance to ambient, target 1000hr, 2000hr and 1 hyr drift rates, maximum hysteresis, etc.  It comes down to thermal cycle procedure after your board has the precision resistors mounted and cleaned - and you'll have to look at your board assembly to find out what parts can be on the board while the resistors are being conditioned / stress relieved at higher / lower temps, and then perhaps you add the remaining parts after resistor conditioning.  Typically you will come up with a heat/ cool procedure with a dwell time at each level.  This will make small changes to the end product, but when you're chasing PPM sometimes this makes all the difference in making your client happy or not.

Like the Vishay datasheets all say:  Contact Vishay Precision Group for details of PMO procedure....It will all be different depending on what model resistor you're getting, what value, what tolerance,  etc.

[Edwin G. Pettis]

To some degree, the PMO (or thermal shock) varies depending on the particular resistor technology and in the case of Vishay parts, the particular model.  PMO is done without power generally although some specs do call for it in certain instances.  The temperature extremes used also varies depending on the part or technology, it is different for PWW resistors than film/foil.  The military thermal shock specifies -55°C to +125°C or +150°C (again depending on the resistor) and sometimes the lower temperature is higher than -55°C.  Initially, thermal shock was developed to try and weed out so-called defective resistors, those that may go out of spec at some premature time or may fail due to possible manufacturing flaws.  Later, it was applied to resistors for the purpose of 'speeding up' the aging process as such, an attempt to reduce early stage drift (1st year).  This 'speeding up' using thermal techniques does not produce substantially predictive results, it does reduce initial aging but not within an accurately predictable range.  The number of cycles also can vary, 5 tend to be the usual number but it can be more or less.

Resistor drift as a group is generally predictable based on history of measurements.  For example when a data sheet says that a particular resistor type will change by X% because of a certain prescribed condition (such as power, temperature, soldering, ect.), you can be pretty certain that as a group most or nearly all resistors will stay within that percentage.  This specification is by actual testing, not mathematical prediction.  The problem is that resistor drift is essentially an uncontrollable characteristic, each resistor is going to behave a bit differently than the next when drifting, it is not possible to accurately predict the drift of any one resistor within a group except by chance.  Drift as a whole is only specified on the basis of past performance of fairly large quantities of parts.  In the case of the so-called 'cheap' resistor, it is not designed to have a low drift characteristic, merely examine the drift specifications in the resistor data sheet and you will see very significant drifting for the same or similar conditions that precision resistors will exhibit much lower drift for.  This also means that 'cheap' resistors will generally not develop much lower drift rates with time like the better precisions do.

Trying to use 'cheap' resistors for a precision application is like trying to change the proverbial sow's ear into a silk purse.

One other note, cheap resistors are going to have significantly higher noise levels than the best precisions, PWW resistors do not exhibit 1/f noise for example, virtually all other resistor types do to some degree.  PWW and BMF have the lowest overall noise as well and that is another very important characteristic, especially if you're making a high precision Vref.

[splin]

Every mW power dissipation is going to affect 1000hr drift by up to 0.5ppm and affect temp by about 0.1C

Did Vishay suggest that the additional drift is (mostly) a result of the 0.1C temperature elevation alone or due to some ageing mechanism triggered by the current flow, independant of temperature? What might that be - electromigration? That would, with my very limited knowledge in this area, be surprising given the much lower current density (j) compared to thin film resistors (approx 100x thicker). My understanding is that you'd expect electromigration rates around j^2 or 10,000x higher for thin film (but that also significantly depends on the composition of the resistive element) and I don't believe that thin film resistors are anywhere near that bad.

There's no way that 0.1C alone could cause that much drift given that Lymex and Dr Frank have seen long term drifts better than the datasheet (typical?) of < 2ppm/6 years and there's no way they could have maintained the temperature to << 0.1C over long periods! (Of course you did say 'up to 0.5ppm' which may apply to resistance values significantly different to their's).

The datasheet power derating graph from 70C to 125C implies .37C/mW which is significantly different to the above. Is the 0.1C the temperature rise of the BMF element itself or the metal can and thus measurable without dismantling the part?

Oh - and that VC they say is essentially zero on the standard datasheets?  In reality you find out that's more like a 3ppm/V on some of their VHP and Z-Foil ratio resistors on values over 1k.

Shock! Surely you're not suggesting that Vishay's datasheet (< 0.1ppm/V) is anything but 'Legal, decent, honest and truthful' (the UK Advertising Standards Authority's mission statement)? I expect the 'values over 1K' exception is clearly detailed in the microdot at the end of the disclaimer section.