ESI Resistance Standard SR1010 and Standard Resistor SR104

[acbern]

I'm confused a bit, which one do you regard as your actual standard? Is it the Vishay VHPxxx?

If you want to have one of your standards calibrated, wouldn't it be better to send the SR104 to the cal lab?

Well, I do have a history with the homebrew one. So I will get that cal'ed to verify. I will then transfer this to the SR104. Since it has changed its value so much, I am hesitant to use it and want to veryfy it has settled. How do I know it does not change again when shipped (I may actually hand cary it there) to the cal lab (always assuming of course that the homebrew low drift rate is confirmed).

[acbern]

+3...4 ppm is typical for an SR104 of that era.

Let the unit cool down to +15...18 C and after that warm back to the room temparature while measuring the main resistor and the thermistor. The tempco should be very close match to the original chart. And there shouln't be any hystereresis left after reaching the original temperature.

So you say since its build date it should have drifted 3-4ppm until today, right? The 7 ppm drift I meaured was within a year or so.
From its build date until today (always provided my current measurement is right and my homebrew reference is not bad) it would have drifted from 10.000,005kOhm to 10.000,032 kOhm, or 2,7ppm, which looks ok.
I will do the measurement you propose soon (currently my thermal chamber is occupied with some other stuff). I am not expecting any deviation though, the current value measured looks nice, and I hope it will be confirmed when I have my standard re-cal'ed. The main problem is, whay was my measurement so much off when done in 2014.  Your explanation re. humidity sounds valid, but how do I proof (if possible at all) and how did it happen?

[ManateeMafia]

@acbern

Since your SR-104 was made before 1990, have you taken into account the +1.69ppm correction in 1990? Not that it is relevant to the apparent drift, but my SR-104 was made in 1969 and I plan on sending it in for its first cal (for me). I was wondering what to expect on its calculated drift since its original calibration value 46 years ago.

[ltz2000]

So you say since its build date it should have drifted 3-4ppm until today, right?

10 kohm (1990) + 3...4 ppm nowadays.

When calculating the total drift you should include the drift of the US legal ohm from the date of manufacture until 1990 plus the 1990 change of definition already mentioned.

[ManateeMafia]

Thanks.

It looks like PTB only adjusted by +0.56 ppm. I was referencing the NIST correction.

[lars]

About the SR1010: I have an SR1010-10LTC and an SR1010-1k from the sixties. I think they are excellent ratio transfer standards between 1:10 and 1:100 but absolute value suffers from seasonal variations. Both my SR1010´s have about 8-10ppm seasonal variations for about 30-40%RH variations. A regression on the 1k series (10k) for the last 5 years give +0.3ppm/%RH +2ppm/C and -0.5ppm/year against a GR1440-10k. The individual resistors in the SR1010-1k is between -1ppm/C and +6ppm/C.

Lars

[ltz2000]

I have an SR1010-10LTC and an SR1010-1k from the sixties. I think they are excellent ratio transfer standards between 1:10 and 1:100 but absolute value suffers from seasonal variations. Both my SR1010´s have about 8-10ppm seasonal variations for about 30-40%RH variations. A regression on the 1k series (10k) for the last 5 years give +0.3ppm/%RH +2ppm/C and -0.5ppm/year against a GR1440-10k. The individual resistors in the SR1010-1k is between -1ppm/C and +6ppm/C.

That is normal. The SR1010 is a transfer standard.

ESI made also an oil filled sister version, which was much better from that point of view.

A proven trick with the dry version is to seal the enclosure and put silica gel bags under a perforated fake bottom.

[zlymex]

That chart of "Drift of 5 SR104 standard resistors at 23°C" was plotted by me more than 8 years ago when I only had two SR104s.
I've been trying to keep track of every SR104 I can find and now I updated the chart to ten SR104s

Edit: update chart to 12 SR104s.

[ManateeMafia]

I hope to get my 1969 era SR-104 sent in for its first cal (by me). I will gladly pass that on to you.

On a side note, do you know if any of the owners of these resistors ever had the alpha and beta constants recalculated? I have read conflicting information that they change as they get older. The SR-104 would be a good candidate to confirm or debunk this theory.

[zlymex]

I hope to get my 1969 era SR-104 sent in for its first cal (by me). I will gladly pass that on to you.

On a side note, do you know if any of the owners of these resistors ever had the alpha and beta constants recalculated? I have read conflicting information that they change as they get older. The SR-104 would be a good candidate to confirm or debunk this theory.

One of my SR104 tested by NBS in Dec 1982, alpha=-0.10ppm/C, beta=-0.028ppm/C2, as shown in green line.
I tested it in Dec 2009, alpha=-0.09ppm/C, beta=-0.033ppm/C2, as shown in red line.
ref: http://bbs.38hot.net/thread-1352-1-1.html

[ManateeMafia]

Thanks for the information.

[jfphp]

I have a 40 years old SR104 and I am very interested by the comparison chart of 10 SR104 but I have 2 questions :
1. Why some SR104 show no drift at all ? Mesurement errors ?
2. Most important, how do you insert in the graph the 1990 change in the ohm definition ?
"The U.S. representation of the ohm based on the quantum Hall effect to be maintained at NIST starting January 1, 1990, using the internationally agreed-upon or conventional value of the von Klitzing constant RK-90 = 25 512.807 ? exactly. From quantized Hall resistance measurements it is calculated that ?(NIST-90) will exceed ?(NBS-48)01/01/90 by 1.69 ppm"
There must be a jump in the various drift lines or did you recalculate the drift without the jump???

[zlymex]

I have a 40 years old SR104 and I am very interested by the comparison chart of 10 SR104 but I have 2 questions :
1. Why some SR104 show no drift at all ? Mesurement errors ?
2. Most important, how do you insert in the graph the 1990 change in the ohm definition ?
"The U.S. representation of the ohm based on the quantum Hall effect to be maintained at NIST starting January 1, 1990, using the internationally agreed-upon or conventional value of the von Klitzing constant RK-90 = 25 512.807 ? exactly. From quantized Hall resistance measurements it is calculated that ?(NIST-90) will exceed ?(NBS-48)01/01/90 by 1.69 ppm"
There must be a jump in the various drift lines or did you recalculate the drift without the jump???

1. I could be the combination of low drift(or right drift) of the resistors with the change in the definition. It is for certain that most good wire-wound resistors are drifting upwards(value increase) if properly made and not constrained.

2. If a very stable resistor is constantly compared to the NBS/NIST standard, there should be a jump in 1990 when graphed.
NBS Ohm was based on the mean value of five Thomas resistors before 1990, and this definition is not constant by nature.
According to an IEEE paper in 1989(http://www.nist.gov/calibrations/upload/eee-38-2.pdf), the drift rate of NBS Ohm was -0.053ppm per year, which means that the five resistors were drifting at +0.053ppm per year before 1990. That change of 1.69ppm in 1990 is not random value in my opinion, rather, it is the correction of NBS Ohm back to the right track.

If a stable resistor was placed into dormancy during 1980 to 1995 for 15 years, it will not aware of all these changes. That is exactly the case for most of the SR104s in my chart. I plotted the chart with dots of data and connect them in straight lines.

Someone has asked me before how I obtain those data on the chart, it is simple, I gather data from the web. Sometime it is already charted like that Japanese SR104(http://www20.tok2.com/home/daisuken/esisr10401.html) but most other times I extract data from calibration stickers(http://lionelectroniclabs.com/esi-sr-104-10k-resistance-standard/) or descriptions(http://www.amplifier.cd/Test_Equipment/other/Widerstands_Normal.htm).

[ap]

@zlymex
Nice work on this overview. What you could do is to take the 1.69 ppm into account in a second version of your chart on the pre-1990 data, so that the real drift would be shown.

[zlymex]

@zlymex
Nice work on this overview. What you could do is to take the 1.69 ppm into account in a second version of your chart on the pre-1990 data, so that the real drift would be shown.

Of course, I should have done that long time ago
Here it is.
The formula is:    =IF(S1>=32874,U1,U1+1.69-(32874-S1)/365.24*0.053)
Where 32874 is the date code of 1-Jan-1990 when the -1.69ppm, U1 is the value calibrated on S1

[ap]

That was fast! Thx!

[doktor pyta]

Sad things also need to be posted. Below pictures of my DIY shorting bars after over 2 years. Bars were used maybe 10 times and were always touched by the edges.
I think the gold and silver layers were too thin.
Also there is interesting .pdf in the attachment. Guildline 'borrowed' a picture from my website and put it into their brochure.
That's not elegant for such company. Haven't they resources to take their own photo ?

[MK]

...
Also there is interesting .pdf in the attachment. Guildline 'borrowed' a picture from my website and put it into their brochure.
That's not elegant for such company. Haven't they resources to take their own photo ?

Why not send them the fee for using your photo?

[dietert1]

Today we received a SR1010 with 12x 1K from ebay. Internally it looks similar to the 100 Ohm unit shown above. It has the original calibration dated 1969-09-29 attached and perfectly readable.
My first measurements using a HP 3456A resulted in the data below.
Resistance measurements are 17 ppm off on average, a mixture of DVM misalignment and drift of the resistors. Somebody determined 0,07 ppm/year, that would be 3.5 ppm after 50 years. So probably DVM misalignment is the bigger part of the difference.
The original deviation pattern is clearly present and subtracting the common shift, only about 3 ppm remain as standard deviation of a single resistor. Great after 50 years.

Regards, Dieter

[0.01C]

Here are some pics.

And for comparison a later unit manufactured by IET labs.

It is so sad what has happened to many classic ESI and General Radio products after adopted by IET.

Your picture does not show the most important resistors.

[dietert1]

The recommendation to put desiccant into a normal "dry" SR1010 worked for me.
When using our SR1010 in a resistance measurement bridge (3 resistors as voltage divider and 1 resistor as current source) i found a TC of -1.9 ppm/K at room temperature and a 8 ppm drift after inserting about 1 cm of silica gel at the bottom. The drying process started at 44 % HR and took about 18 hours. I mounted a PT1000 sensor on the inside of the SR1010 top plate in order to correct ambient temperature variations numerically. There is a heat conduction delay of about 4 to 5 minutes from temperature variations as seen by the temperature sensor to TC effects on the bridge. Now i got stable results at a +/- 0,05 ppm level over several days.

Another exercise will be looking at the copper wires between the SR1010 and the DUT (Alpha Electronics HCZ500R00T).

Regards, Dieter

[dietert1]

About thermal EMF with the SR1010: I did some measurements, that yielded a 0,34 uV/K slope of my setup under ambient temperature changes. And i noticed that a copper spade on one of the binding posts yielded up to 3 uV when heating it between two fingers, so the binding posts don't seem to be copper.
Before i start scratching parts - maybe somebody did that before.

Another question with respect to desiccant: It was recommended to "seal" the SR1010. Anybody with an image?

Regards, Dieter

[SvanGool]

...
Another question with respect to desiccant: It was recommended to "seal" the SR1010. Anybody with an image?
...

If you can compensate the leakage by the amount of desiccant, as is mentioned in this report, you maybe don't need to seal at all. But you do need to find a way of roughly estimating the leakage.

[dietert1]

Yes, and nowadays it's quite simple to supervise humidity using integrated sensors. Since i have a lot of desiccant in that SR1010, i won't touch it any more, but observe humidity inside.
I made a closed box around it including 20 mm of styrofoam thermal insulation to reduce leakage and temperature changes. Hope to reduce the binding post thermal EMF by keeping everything at the same temperature, thus avoiding expenses for another 26 low thermal binding posts. Now the SR1010 temperature sensor indicates a thermal time constant of about 1 hour - much better than before. Thermals seem to be down by a factor of 10 in comparison to before. I guess oil may be necessary when running the precision resistors at varying loads, but it does not help with the thermals.

Regards, Dieter

[guenthert]

Yes, and nowadays it's quite simple to supervise humidity using integrated sensors. Since i have a lot of desiccant in that SR1010, i won't touch it any more, but observe humidity inside.
[..]

  If you don't intend to replace the "consumed" (or "spoiled") desiccant, but just observe the internal humidity, why use a non-zero amount of desiccant in the first place?