Yet another ESI SR104 thread.

[TiN]

I stole nikonoid's thread a little, but probably your guys can tolerate little bit of more spamming of fancy resistors, eh?   

It was a while before I posted any new threads. So here it is, fresh article about ESI SR104 10000 reference standard resistor.



Lid cal report attached to my box:



Little peak inside



I'm not brave enough to go any deeper.



What happened to 4-wire lord Kelvin connection here???

Half-hour of cleaning and removing irrelevant ugly stickers...



Don't forget to clean connectors. Who know what eBay hands held them before.



No business for banana jacks or other non-metrology crap here. Proper TeCu binding posts!



To justify going to my site, I've added realtime calculator for output true resistance, based on lid alpha/beta and Rdev information.
Might be easy for beginner ohm-nuts to join up sub-ppm choochoos. Click on pic to go for it:



There is also offline python app, that does same thing. Here it is, can be also used for any resistor with known alpha/beta/calibrated value, not only SR104.

Result of comparison my voodoo Fluke SL935 ovenized prototype and some SR104 from EEVBlog members for TCR. All TCR curves are relative, not absolute resistance deviation. +23.0C calibrated value taken as a reference.



Jee, that ovenized prototype leaving no mercy to the SR104. Having both resistors pretty much is win-win for the lab now, use FSL935 as working standard and don't ever worry about tempco, and use SR104 as long-term calibration standard. No wonder some other top dogs in metrology have ovenized resistance standards too.

EXCEL file with data and graph.

Calibration check results in table below. Keithley 2002-4 definately need some calibration love, other folks are happily lurking around ~2ppm absolute, which gives me decent confidence for homelab resistance accuracy.

.

More detailed funky plots






Setup with 3458As



Next step is to get my FSL935 back from this month soon-to-be-done second calibration (to establish 6 month stability) from Process Instruments (I hope they don't forget this time to keep unit plugged to mains to keep batteries charged when not in measurement!). I'll use FSL935 to transfer resistance calibration to this SR104.

What would be best way (for lowest uncertainty) to do so, given I don't have ESI 242D? 
I do have F5700A, K6221, HP3245, LTZ-based 10V box with 50mA driving ability and K2182 as a toolkit. 

[TiN]

I was too curious, so I actually digged EEVBlog's Dave Wekomm magical big dollar 10KOhm resistance standard data and added on the TCR chart.



Gives you a good perspective what is the difference between the real resistance standards like ESI SR104 and fancy-overly-optimistic-datasheet Vishay VHA518 10K resistor (which is no cheap deal, think 200-300$ a piece from VPG, I know it first-hand from my 1Meg VHA!) in pretty box. 

[Edwin G. Pettis]

Not too shabby TiN, my SR-104 is slightly better, alpha= -.06 PPM/°C and Beta= -.008 PPM/°C², usually I see less than -0.1PPM to <+0.2PPM variance with room temperature, in all but a very few cases, it is insignificant.  My SR-104 will likely be off for calibration shortly.  That variance is barely larger than the calibration uncertainty of 0.15PPM.

I agree, if your SR-104 is in good working condition, it is best to stay out of it.

I do know of an SR-104 that was calibrated (at a top cal lab) and then read on a freshly calibrated 3458A (full calibration with data), the 3458A was within about 2 PPM, not quite as good as a calibrated ESI 242D but close.  The SR-104 is highly accurate and stable unless some idiot mugged it and unfortunately eBay is a blind in that respect.  I see yours had a Raytheon sticker on it so I would think it was well taken care of while they had it.

Oven stabilization will improve some on the TCR numbers but at the considerable increase of circuitry and power needed, the SR-104 tends to be very stable without it and as long as you know the TCR numbers, the minor compensation is easy to take care of.

I have seen temperature controlled ‘portable’ chambers for putting an SR-104 in but they have an awful time lag between putting the SR-104 into it and the SR-104 stabilizing to that temperature.  They were also rather expensive.

[lowimpedance]

Indeed a nice addition to the LAB. Have you inquired at TR about the past history of this unit ? (an ex Air Force unit from one of the stickers!). VMS int. became TR so given both stickers were on this unit they might have quite a bit of history on it.

[texaspyro]

I put my SR104 in an insulated cardboard box along with a Thunderbolt GPSDO and a 10V reference.  I use Lady Heather's temperature control feature to keep the temperature in the box stabilized close to one of the two values where the temperature is 10.000K.  The Thunderbolt provides heating and a PWMed fan provides cooling.   

[e61_phil]

What would be best way (for lowest uncertainty) to do so, given I don't have ESI 242D? 
I do have F5700A, K6221, HP3245, LTZ-based 10V box with 50mA driving ability and K2182 as a toolkit. 

Perhaps, in a way like this:?
https://www.yumpu.com/en/document/view/11827423/a-wheatstone-bridge-for-the-computer-age-les-huntley-fluke

Or simpler: What about putting both resistors in series and let 1mA flow through both of them. Measuring the voltage on both with a 3458A. Reverse the current and measure again. This should end with something around four times the 10V transfer accuracy of the 3458A (or even better if RSS is used). Wouldn't it?

[ap]

Yes, using a 3458A is a good way. What needs to be taken into account is that the voltage source may have short term drifts and there may be a residual EMF. Needs to be figured in.
And the weighing factors for the related uncertainties of the differentials in the RSS calculation need to be considered. I ended up with a little less than a conservative 1ppm total.

[ramon]

That is too simple. He is probably building his own automated guarded Warshawsky bridge and "Jake" switch system clone. 

Tin, no need to wait for your FSL935. Can use your current TEC system to make your own TCR chart between 18 and 28C? With the same setup (if possible) as all your previous resistor TCR test.

[quarks]

very nice

I tried the py script and have problems with the output format



does anyone have a hint what needs to be changed in the script to get the output csv file with , instead of . for the decimal points

[Echo88]

Next thing to get is a Guildline DCC Bridge to really see that sweet stability of the SR104.  Very nice TiN!

[ManateeMafia]

TiN needs to buy one and send it to me for safe-keeping. He won't budge... yet.

[TiN]

I tried the py script and have problems with the output format
does anyone have a hint what needs to be changed in the script to get the output csv file with , instead of . for the decimal points

Please post resulted CSV, I'll try add support for commas.

[quarks]

Please post resulted CSV, I'll try add support for commas.

thanks a lot

[Pipelie]

TiN needs to buy one and send it to me for safe-keeping. He won't budge... yet.

TiN need to buy one of this box for safe-keeping.

[Pipelie]

Looks like the Resistor had been adjusted on 30 Jul 3 03, there is a black resistor 0.38 ohm inside, next to the light green resistor.

[Vgkid]

That looks like it is connected to the thermistor.

[chuckb]

What would be best way (for lowest uncertainty) to do so, given I don't have ESI 242D? 
I do have F5700A, K6221, HP3245, LTZ-based 10V box with 50mA driving ability and K2182 as a toolkit. 

Perhaps, in a way like this:?
https://www.yumpu.com/en/document/view/11827423/a-wheatstone-bridge-for-the-computer-age-les-huntley-fluke

Or simpler: What about putting both resistors in series and let 1mA flow through both of them. Measuring the voltage on both with a 3458A. Reverse the current and measure again. This should end with something around four times the 10V transfer accuracy of the 3458A (or even better if RSS is used). Wouldn't it?

An easy way for TiN to develop that 1ma of test current is to use just the output stage of one of his FX PCBs. The positive power supply voltage would need to be in the 25v range. Reference attached pdf.
A stable 10v reference (he probably has some laying around) and the FX board would establish the voltage drop across the incredibly stable SL935. With the SR104 inline with the 1ma current test current you may have the basis for a good voltage drop comparison.

There are a few more details to work out yet...

[3roomlab]

I was too curious, so I actually digged EEVBlog's Dave Wekomm magical big dollar 10KOhm resistance standard data and added on the TCR chart.

Gives you a good perspective what is the difference between the real resistance standards like ESI SR104 and fancy-overly-optimistic-datasheet Vishay VHA518 10K resistor (which is no cheap deal, think 200-300$ a piece from VPG, I know it first-hand from my 1Meg VHA!) in pretty box. 

i cant rem where, but didnt zlymex made his own spin of a vishay mix and have a remarkably good result? i wonder if you could add his to the graph? if that beats the wekomm ... ?

[ManateeMafia]

I like the one Dr. Frank did using a VHP202Z here https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/25/

He probably has a lot of data collected on it by now.

[TiN]

Well to add into table I need resistor's alpha and beta values at +23C.

[Echo88]

https://www.eevblog.com/forum/metrology/teardown-standard-resistors/msg892700/#msg892700 This one? I think Alpha and Beta-coefficient are noted for 23°C.

[TiN]

What is better then one magical ESI SR104?



Yes, four ESI SR104 are better! 

Thought it could be interesting of someone, inter-comparison between four different 10000 Ohm standards everything to everything, using MIL 6000 binary voltage divider-based bridge system. Full round of four hops resulted in deviation just 7 ppb, while worst resistor shown 12 ppb peak to peak deviation. Good enough for any practical use.