ESI 242D (SP3632) Resistance bridge for SR104 intercomparison

[manganin]

In a number of metrology applications a modern long scale multimeter (like Fluke 8508A) can replace the traditional resistance bridge. But when calibrating the highest grade standard resistors, which deviate only a few ppm from each other, the ESI 242D based system is still the king - unless you can afford a current comparator.





The setup in the photo was originally built for SR104 intercomparison as well as calibrating other standard resistors against those primary standards. Decade transfers were realized with trimmable SR1010 Hamon resistors, allowing the bridge to operate in its 1:1 comfort zone.

The unit shown is a low temperature coefficient version of the 242D, later known as SP3632. It was purchased probably ten years before the special version became a catalogue orderable product and got that new code. The yellow sticker has obviously been added afterwards. The lab also bought a second identical system during the eighties. It has black front panels and SP3632/242D printed on the top name plate instead of the ordinary 242D.

There seems to be no visible difference between the standard and the low TC version, only some additional testing and selection. And knowing the very high build quality, the basic version was probably not much worse anyway.

What draws the attention is the use of Keithley 181 nanovoltmeter instead of the built-in null detector. The goal could have been better sensitivity in low resistance levels. Or maybe it just looked super cool back then.

I remember seeing an old arcticle about SR104 calibrations describing a similar setup and also mentioning the use of a nanovoltmeter. If anyone knows where to find it, please post here.



ESI 240C KELVIN RATIO BRIDGE





Plenty of Evanohm wire resistors inside. 35 of them wound on mica cards (in SR104 style) and the remaining two 900 kohm high resistance types on ceramic bobbins. The cabling is teflon insulated.






ESI 801 DC GENERATOR/DETECTOR





The generator output ranges from 2V/2A up to 600V/6mA max (variable control CW). Enough to kill you. The variable scale is conveniently calibrated in milliwatts dissipated by the resistor. The lever switch turns on the excitation voltage either in positive or negative polarity, which is useful in low resistance measurements where the thermal voltages matter.

The mains wiring of the unused internal null meter has been unsoldered, possibly to minimize leakage. The wire ends can be seen in a transparent sleeve just behind the moving coil instrument. The null meter is actually an HP 419A, but he box was a little bit tricky to open, so no photos were taken this time.





When measuring in the sub ppm region, you can't just use any DC source. The transformer core and the secondary winding have no physical contact: the DC leakage path is cut by several millimeters of air. A copper guard surrounds the whole secondary to prevent capacitive coupling.





The connections to the external nanovoltmeter are quite interesting: the original factory made connector has been substituted by crimped copper tubes and a cover made of cellular plastic. Works very well,  I have succesfully used the same approach with my Keithley 148.

[Conrad Hoffman]

I have a 240C and related stuff, but I seem to remember you still have to calibrate the 240C. Not sure what to use to do that.

[manganin]

I seem to remember you still have to calibrate the 240C. Not sure what to use to do that.

Those adjustments are relative, no absolute value standards required.

[mendip_discovery]

I could see a need to calibrate it. The deviation dial would have to be confirmed and any bias each connection would have to be checked. Possibly some check to ensure that the voltages/current is correct, I know it uses the same for both items but you need to do some checks becuase if its using half the voltage/current your measurments might deviate from others doing the same tests.

Never tested a unit myself so I could be barking up the wrong tree.

[manganin]

I could see a need to calibrate it.

The ratio and the deviation range adjustments are required. There are detailed procedures for both in the very well written manual.

The nature of those adjustments is relative, so you don't need any calibrated (known value) standards to do it.

[Conrad Hoffman]

Yes, but you still need to buy/build/borrow/steal/imagine some very accurate ratios if the instrument is to be as good as it's capable of. If the standards were that easy, you wouldn't need the 240C, would you?

[manganin]

Yes, but you still need to buy/build/borrow/steal/imagine some very accurate ratios if the instrument is to be as good as it's capable of.

No, just a standard resistor and a decade resistor of decent quality.

[manganin]

The second 242D system built especially for the calibration of SR104 and other high level standard resistors. Short rack makes it much more compact than the older unit. Purchased in the late 80s, so there is probably 15+ years in between these two.






This is officially the low temperature coefficient version SP3632 as can be seen in the name plate.  The null meter of the new 801B Generator/detector is a better sensitivity Keihtley 155, so no need for an external nanovoltmeter anymore. The deviation dials are not original, but replaced by the older and more durable solid aluminium machined type. (Wasn't a smart move from ESI to change to diecast.)






Double fused mains connector on the rear.






I don't know what is wrong with my photos, looks prettier in real life.

[Rax]

I don't know what is wrong with my photos, looks prettier in real life.

I don't know what you don't like at your photos... this is as handsome as it gets!

[Rax]

Short rack makes it much more compact than the older unit.

How much smaller is this vs. the old one? Can you possibly post the physical dimensions and maybe also the weight?

I had (and probably still have) the opportunity to acquire one of the older units, but the size and weight exclude it from my space - there's just enough to accommodate it. But if one of these intersects, maybe I can opt in.

This unit I could get is, I believe, probably right in between the ones you have - I'm pretty sure it's the early type, but probably a later production of the low-tempco variation (having the SP3632 label).

[donlisms]

A standard 19" rack height "unit" is 1.75 inches.  The 925D resistance standard at the bottom of the stack -- the thing being removed -- is a 4U chassis, so 7" high.  The 240C bridge itself and the 801 generator / detector are both 3U, so 5.25" each.  The total height of the full stack is thus 17.5 inches -- 10U total, or without the decade box, is 6U or 10.5" tall.  Tiny!  It is 19" + overhead wide in either case.  It is as deep in either case.  Unless it's in another case.

I can't see a reason why you couldn't move the 925D out to a separate rack, which you can get anywhere for any price, or just leave it unracked on a shelf somewhere.  Each of the three chassis is shielded without the outer rack, so they can each be used independently, and do not rely on being in that big rack too heavily.  Temperature is probably better with the double casing. There may be some shielding improvements by having the second layer of material; I do not know.  So if your new, smaller rack is made of wood or plastic, I suppose that could be a thing, but I kinda doubt that it actually is.
I do not know the depth of the 801 or 240C offhand; they are probably quite a bit shallower than the full 12.5" depth of the ESI rack.  You can buy racks of various depths.

For the weight...  check the manual?     I don't have a good way to weigh mine.  Nor am I in the mood to pull each device out and weigh it separately.

Without the 925D, the 240C itself is quite happy to compare two resistors that are nearly the same, or nearly the same when one is 10x or 100x the other.  The deviation dial gives you +/- (more than) 50 at a scale of 0.1ppm, 1ppm, 10ppm, or 100ppm.  That establishes the range of comparisons you can make without the 925D or equivalent: good for decade resistors, but not so much full general purpose resistance measurements.

The 925D is quite good in and of itself; calibration might be required, but it's, like... a nice box.

It is handy for substitution, which is a good way to use the bridge overall.  To compare two resistors, you compare each of them to a (not necessarily calibrated) reference, and those two readings reveal the difference between the two.  The 925D is handy for that, but perhaps overkill; a smaller, yet very stable, decade box could fulfill the role, or even a single fixed resistor that happens to be about right.

Depends on what you want to do.  I suspect if you split it into two or three parts, the options for how to store and manage and use it would increase noticably.

[manganin]

I don't know what you don't like at your photos... this is as handsome as it gets!

Looks pale and worn out. I don't belong to the generation trained to take photos which are better than the original.

How much smaller is this vs. the old one? Can you possibly post the physical dimensions and maybe also the weight?

As already mentioned, subtract 4U.
49,5 x 33 x 32 cm without feet.

I don't have a scale available, but it is easy to lift, thanks to the all aluminium construction.

[RolandK]

Maybe the difference is easy to explain:
there are pictures of units with decade resistance standards and some without. That makes a difference in size.

[guenthert]

How is the low thermal coefficient achieved?  Low tempco resistors throughout or is it heated?  And was the purpose to make it more precise or more usable in environments w/o strict control of ambient temperature?

[Rax]

Maybe the difference is easy to explain:
there are pictures of units with decade resistance standards and some without. That makes a difference in size.

True!  I didn't see that at first glance. The current IET Labs manual shows a unit virtually identical to the old, fully featured unit. I assumed they never came any other way, and I've never seen one without that module.

[manganin]

How is the low thermal coefficient achieved?  Low tempco resistors throughout or is it heated?

Resistors selected for low/matching TC.

And was the purpose to make it more precise or more usable in environments w/o strict control of ambient temperature?

Probably to get the best out of the new 0.1 ppm resolution, which was added to the C-version of the 240 bridge. I assume it was at first an in-house modification or developed by some customer using the 242 system for primary level resistance calibrations.


EDIT: Typo corrected. 240 is the bridge, 242 the whole system.

[coppercone2]

THAT COPPER GUARD HAS SEEN BETTER DAYS

[manganin]

THAT COPPER GUARD HAS SEEN BETTER DAYS

The shield is just copper foil wrapped around the odd shape winding. Probably looks the same as when it left the factory

The visible copper shield is actually on the primary winding, not the secondary. Error in my original description.

[coppercone2]

its just funny because everything else is super neat etc. it stands out super hard from the other pictures