ESI SR1 standard resistors

[Rax]

Hi all,
Just looking to stir a bit of conversation over these resistors, which are available very affordably out there for the metrology turbo-hobbyist (the only kind).

I think they have accrued quite a bit of bad rap, from what I'm seeing based upon a critical post or two, and I'm just not sure I'm convinced this reputation is deserved.

I've recently acquired a set from a Tektronix lab that (apparently) closed shop relatively recently and they were measured within a few, very few, ppms of nominal in the easy values. They're also obviously well aged, and btw, aging served them very well if I'm going by the values measured by the lab. Just about uncannily well.

It's also what I'm seeing in my environment - they're within ppm of nominal (up to 20-25, depending on value) - again, almost uncannily so. By spec, tempco is 5ppm or less - again, very good.

I got another one since the original batch (to get a complete set of values) from a large US instrumentation retailer - for a song and a dance, even if carrying a verification sticker by the retailer from 2020 - and it also seems to be pretty much dead on. More nefarious impact from bad reputation, if you ask me. Not that I'd mind getting these for such little dough.

All this is measured with a calibrated 8502A, for anyone wondering. Checked with a "noncalibrated" (but exact, afaict) 8840A which confirms its readings.

Why the bad name for these? Is this what others have seen with these ESIs? I'd be very surprised, but maybe I just got very lucky with my set.

[DavidKo]

The ones, that I have, are quite "used" and only one have both plastic bottom nuts OK. Some of them are only hold by the GND nut. Without measurement I cannot confirm if the temperature PPM is OK or not.

From the mechanical point of view, the nuts are serious flaw of these units.

[Rax]

I actually found a datasheet for the current production ESI SR1 (the one listed on the IET website for $735-$1656) that states it can be sold in "optional esi-style configuration" - which is exactly how the original resistors were sold.
This tells me the original ESIs and the current production ones made by IET are not so radically different from each other.

[bdunham7]

Why the bad name for these? Is this what others have seen with these ESIs? I'd be very surprised, but maybe I just got very lucky with my set.

Well, since you asked, criticisms of these could include:

1) Their small size, which means they're easy to drop on the floor.  I haven't seen all that many of these, but quite a few have been in very poor physical condition.

2) They appear to not be truly kelvin-connected.

3) They are arguably less robust both physically and environmentally because they use a manganin wire wound on a mica card (at least the lower values do) without any hermetic sealing or significant physical support. 

4) Since they came in at least two different paint jobs, if you try to assemble a set with individual eBay purchases you may not get a nice-looking matched set.  The horror!

Of course the L&N NBS-style models are also not truly kelvin-connected and also not hermetically sealed, and although the internals are oil-bathed and pretty well protected if the top is tight, the external terminals are very susceptible to the environment.

I imagine the overall population of these SR1 resistors is probably doing pretty well and it would be a mistake to judge the product line solely by looking at examples that are for sale on eBay or wherever.  The ones that have lived a peaceful life in a dry lab are probably all fine.  Unless the clowns that work there have dropped them on the floor.

[Rax]

2) They appear to not be truly kelvin-connected.

I've seen this stated before, and I'm not sure I understand it. They do have two sets of connectors - top and bottom - and, as far as I understand, a Kelvin connection essentially means two separate leads all the way (internal to the fixture) to the actual points of connection at the resistive element. If the resistive element is connected evenly to the top and bottom of the SR1, would that not qualify as a Kelvin connection?

[bdunham7]

If the resistive element is connected evenly to the top and bottom of the SR1, would that not qualify as a Kelvin connection?

It very well might and I forgot to include my intended disclaimer that I might be wrong about the kelvin issue.  However, this photo shows that the two terminals are not independently connected directly to the manganin, there is a short, thick, presumably copper, conductor in between and then a tinned brass or copper strip on the mica.  Those elements are all common to the source and sense circuits and probably amount to hundreds of µohms.

https://www.eevblog.com/forum/metrology/teardown-standard-resistors/?action=dlattach;attach=207316;image

[Kosmic]

Interesting post from Edwin G. Pettis about the SR-1.

https://www.eevblog.com/forum/metrology/esi-sr1-standard-resistor-current-rating/msg1256635/#msg1256635

The SR-1 series are what are termed 'working' standards, they have the least tightest of specs and of course cost less.  They were intended to calibrate medium accuracy instruments that were two terminal, such as Wheatstone but were also given the additional contacts on the back so they could be stacked in series/parallel and a quasi-four terminal hookup.  You can of course take an more accurate reading of them and use that right away but their short/long term drift is not that great compared to transfer standards.  They were not processed for higher stability like the SR-1010s, operating them at or near their maximum current ratings briefly will have little effect on their stability, running them at or near maximum current for sustained periods will cause them to drift a bit but they will not get hot enough to change their characteristics permanently.

When these were originally designed, instruments like the 3456A or 3458A did not exist nor are they designed to 'plug in' to them, they have the standard 0.75" C-C spacing which has been around for decades but that doesn't mean they were meant for something that didn't exist yet.  They are not really good enough to cal a 3456A even with the pseudo four terminal hookup, they were meant for instruments no better than 0.05% at best, something on the order of 4.5 or possibly 5.5 digit resolution with 2-wire resistance measurement.  The third terminal is not so much a guard terminal as a ground terminal for the case to minimize noise pickup.  Remember the minimum accuracy rule, the reference should be at least 5 times more accurate than the instrument's accuracy, 10 is even better.

Never mind the market hype, these are not one off copies of SR-1010 resistors even if they look just like them, they are not processed the same.  I have a copy of the BOM price list for the SR-1 series, the resistor is one of the cheaper parts in it.

And you can find the definition of a working standard over here:
https://encyclopedia2.thefreedictionary.com/Working+Standards

Working Standards
standards used for practical measurements in scientific research, production, commerce, and other fields. They differ from base standards, which are only used to calibrate other standards. Working standards are classified as measures, measuring instruments, measuring transducers, measuring assemblies, and measuring systems. The last two categories represent various combinations of the first three categories with additional auxiliary equipment needed for measuring, including current sources, switching and control devices, communications lines, stabilizers, thermostats, and specialized computing systems.

So a lot of those working standard had a rough life going all around a manufacture for example used to calibrate and adjust equipments and instruments.

Personally I think SR-1 are, most of the time, overpriced for what they are. I had better luck collecting SR1010, which is a transfert standard.

[alligatorblues]

I have (2) SR104s. Those are pretty good. I had one calibrated by Keysight, and it had drfted <3ppm in 47 years. 

[Rax]

I have (2) SR104s. Those are pretty good. I had one calibrated by Keysight, and it had drfted <3ppm in 47 years.

This is about SR1s, not SR104s... But even so (or even more so), one of my resistors drifted 2ppm in a few decades too, another one less than 4ppm (going by lab calibration sticker). I'm cursorily measuring them at even less delta from nominal than that (the entire range of resistors - 1 ohm through 10 Mohms), depending on the fluctuations I'm getting. I do have to watch the temperature, but that probably throws all my instruments out of whack when not proper. Given what I paid for them - which is quite a bit less than the typical ask you may see on the usual marketplaces - I think they're a steal.

I'm not making them into primary lab standards, but for a hobbyist bench, they may be exactly what one needs and honestly much more.

[bdunham7]

Personally I think SR-1 are, most of the time, overpriced for what they are. I had better luck collecting SR1010, which is a transfert standard.

I've looked at those, but it adds upt as you need a lot of SR1010s and all the accessories to cover a wide range.  OTOH, you can get all the intermediate values like a Dekabox only less convenient and more precise.  xDevs did a writeup on the SR1010 and the one he evaluated didn't do any better than the SR1s that I've seen. YMMV, but I think an unsmashed SR1 is probably close to as good as you get short of a primary standard like the SR104.  What else is there?  As far as overpriced, the trick is not to pay too much!

[Kosmic]

YMMV, but I think an unsmashed SR1 is probably close to as good as you get short of a primary standard like the SR104.  What else is there?  As far as overpriced, the trick is not to pay too much!

I don't think we should talk about the SR104 here since it's clearly in another league.

These days, putting a good Vishay resistor in a box would probably give better results than an old SR-1.

[bdunham7]

These days, putting a good Vishay resistor in a box would probably give better results than an old SR-1.

Yes, that's an idea and one I'm going to do for a 1R shunt/standard but I'm using a different resistor.  The problems with going the Vishay Z-foil + box are that you have to make sure you get one that is environmentally protected (not the audiophool versions), you end up with lower current limits and it ends up costing more for a $35 resistor, four $15 binding posts and a box than for a decent SR1 on eBay.  And then it still isn't truly kelvin-connected and that may matter for the 1R to 100R values.

[Rax]

These days, putting a good Vishay resistor in a box would probably give better results than an old SR-1.

My point is that you wouldn't (I mean cost vs. performance, at least). You'd need to go to a VHP101 which, by itself is - at least in the US - about $120. It's also a 50ppm part (from nominal). By the time you put it in a box (aluminum? some thoughtful temperature conditioning provisions?), get four or five good quality, low EMF binding posts, do proper wiring inside, you're well outside what the SR1s are going for (actual sale price - one can be had for less than $50). I am intently not including temperature probing provisions in this, as obviously the SR1s don't have that either.

And for that cost you'd still be pretty far from the 2-4ppm from nominal than (at least some of) these seem to be at (the unabused ones). Tempco may add to this, but I think that's more of a non-issue this day and age.

[Rax]

https://www.eevblog.com/forum/metrology/teardown-standard-resistors/?action=dlattach;attach=207316;image

On their mechanical robustness - I think the image above may indicate why, although working standards and therefore prone to abuse and manhandling, they seem to have mostly withstood that with panache.

The mica card + the unifilar winding likely have very little weight which together with the dimensional rotational arm would create relatively little angular momentum stressing the assembly in case of impact (against a very thick conductor that has a very strong mechanical commandment over the resistive assembly, opposing the angular momentum).