USA calibration club

[Conrad Hoffman]

I've probably got a platinum RTD I could throw in.

Cabling is important and I'd suggest untinned copper wire with no connectors. Measuring with a meter can be useful, but since everything in my lab is calibrated to my voltage standards anyway, I'm going to measure the difference with a Fluke 845, giving me far more effective digits (though my reference hasn't been measured in forever) than if I used my meter.

Hopefully we'll loop the package around more than once.

[iainwhite]

I'd also like to encourage the creation of simple hand-made references among the group as well. 

cellular,   
Do you think Conrad's reference (linked as the 'simple' one above) is a good starting point to try my hand at building a reference?
I have some 2DW232 & 233 parts ordered thru Technix  - I assume I could use those in this circuit?
I also looked at Julian's thread which looked very promising.

Just trying to find my feet here as I am a noob.

Thanks,  Iain

[Conrad Hoffman]

The reference I posted performs decently but isn't the equal of various others discussed here. IMO, it's still pretty good and the real value of it is to teach you the basics of how most compensated references work, get some experience with zeroing in on the minimum TC point and just get comfortable with the topic. If you want to do it for near zero money, substitute the compensated zener diode with a plain cheap zener in series with a regular diode.

[iainwhite]

to teach you the basics of how most compensated references work

Sounds ideal for me to try, thanks

[cellularmitosis]

Andreas: yes, I was hoping for the simplicity of USB data logging, at least for temperature.

iainwhite: I'm just a noob myself -- I think Conrad's circuit with a 2DW232 would be an ideal place to start!

Conrad: Indeed, cabling is critical, as I am discovering this evening!

My "mailer" cardboard boxes arrived today, so I thought I'd take some measurements with my Keithley 196 to decide which cables I should include with the reference, but my results were disappointingly inconsistent.

I've tried a bunch of cabling variations and am getting different readings.  More importantly, the readings wander around by as much as 150uV, seemingly just from getting up and walking around the room.

I think I have an idea of what the problem is: I stuck a temperature probe directly into the Ohms sense jack and it read over 105F.  Can you say "thermal gradient"?

Across all the various setups I tried, the overall span of readings went from around 10.00104 V to around 10.00150 V.

Setups I tried included:

• Cheap Chinese mini-pincer to banana jack: https://www.amazon.com/gp/product/B00O9WGHQ6/ref=oh_aui_detailpage_o08_s00?ie=UTF8&psc=1
• Mini-hook to banana cables from Frankie's ebay store (short, appear to be bare copper hooks, similar to these: http://www.ebay.com/itm/Silicone-Test-Leads-50cm-with-Sprung-Hooks-and-Banana-Plugs-1-Pair-Red-Black-/200979921195)
• Banana to banana DIY cables, using pomona 6733 silicone 18AWG and cheap ebay gold plated "logico" banana jacks: http://www.ebay.com/itm/LOGICO-10-Pcs-24K-Gold-Plated-Audio-banana-plug-Speaker-Wire-Jack-Connector-/131584318536?hash=item1ea3095c48:g:GEoAAOSwt5hYZG0X
• Banana to mini-hook DIY cables, using pomona 6733 wire, pomona 4555 mini-hooks, and "logico" banana jacks
• Banana to mini-hook DIY cables, using CAT5 (stranded), pomona 4555 mini-hooks, and Cinch solutions 108-0753-001 banana plugs https://www.digikey.com/product-detail/en/cinch-connectivity-solutions-johnson/108-0753-001/J149-ND/5927

I filed down one of the "logico" jacks and surprise!  They do not appear to be copper (the body definitely isn't -- it shines like steel after filing off the plating.  the springs miiiiight be some sort of really light looking copper alloy).

The Cinch banana plugs I salvaged from from a defunct banana DMM breakout project I worked on a while ago.  They are brass with nickel-silver springs: http://www.belfuse.com/resources/Johnson/productinformation/pi-108-0753-001.pdf

I'm definitely open to advice here, but I suspect I'm fighting an uphill battle here -- a 105F to 75F gradient across my cables is just asking for trouble.

I think I'll go ahead and mail out the reference with some solid CAT5, the pomona mini-grabbers, and the screw-banana plugs.

(I really, really need to get GPIB up and running.  Trying to describe the behavior of a DMM using sentences just doesn't cut it.)

[cellularmitosis]

(argh, forgot the forum doesn't honor EXIF rotation tags on photos, d'oh!)

[cellularmitosis]

Placing an ice cube on one of the banana plugs brings the reading down to 10.00084 V (about a 400uV drop).

(edit: oops, attached the wrong photo)

[Vgkid]

I think you need to open up your Keithley dmm. I cannot think of anything that should cause it to get that warm. Unless they use the front panel as a heat sink

[Conrad Hoffman]

Yeah, that's way too warm for precision work. I've had very consistent results with un-tinned solid copper phone wire. The stuff they sell here as "bell wire" is similar and works well. Every time I add a connector of any type, including copper alligator clips, errors start building up. Thus, I just clean the ends of the wires with Scotchbrite and screw them under the banana jacks. If the meter only has push-in banana jacks, you'll have to attach something and keep them at the same temperature. IMO, confidence at the 1 ppm level isn't going to happen with plated connectors.

[cellularmitosis]

Thanks for the feedback, guys.  This morning (after the 196 was turned off all night, and lab temp down to 71.0F), I removed the top cover and powered it up.  Initial reading was 10.00031, then climbed to around 35 to 45 within the first minute.  That's 1,000uV down from where I was last night!  20 minutes later it is wandering around 54 to 62 (edit: just checked on my lunch break and it is around 10.00072 to 78 V at 73.1F lab temp).

I'll get the SVR-T shipped out within the next day or two, and while it is in transit I'll start another thread for investigation / repair of the 196.  Hopefully, by the time the SVR-T returns to Austin, I'll have a more stable and GPIB enabled setup

[bitseeker]

That's a very peculiar temp gradient, indeed. Looking forward to the repair and what may be causing the issue.

[cellularmitosis]

Coming your way, Muxr!

[Muxr]

Good stuff, thank you cellularmitosis!

I can measure with my 2015 Kei and my 8845A Fluke and report the figures, and send it along. As far as I know neither of them have been calibrated any time recently, however at least they are both surprisingly close to each other when compared to my AD584KH reference (TempCo 15 ppm/C, Long Term Stability 25 ppm/1000 Hrs), either dead on or off by a few uV depending on the measured voltage of the reference.

Will report here when I get the package.

[Muxr]

Finally got around to writing a script to data log and plot my measurements on the Keithley 2015. I can include this for the traveling reference as well.

This is an example of it running for about 7 hours. Summarizes max/min readings, mean and the standard deviation.


Also looking to datalog temperature in the future as well. Not sure how to go about it. Perhaps I just build an arduino module around SI7051 and query it over UART, that should be easy enough. If anyone has any other suggestions they are welcome.

Also I am using a very simple Python script to log data over rs232, which just saves it to a cvs file. And then another script which uses Python's Pandas and Seaborn to draw the graph. If anyone is interested I can share it (needs some cleanup still).

[Conrad Hoffman]

I can do GPIB to some of my equipment. What's a good interval between measurements?

[Muxr]

I can do GPIB to some of my equipment. What's a good interval between measurements?

I am experimenting with polling once a second.

[RandallMcRee]

Interested in the script. Also have a 2015 and GPIB but scpi is foreign to me. Some guidance would be helpful.

Thanks,
Randy

[Muxr]

Interested in the script. Also have a 2015 and GPIB but scpi is foreign to me. Some guidance would be helpful.

Thanks,
Randy

Cool, I'll clean it up and post it on github and link here (tonight or tomorrow).

I have it broken into two scripts. Logging and plotting.

Logging: is super simple.. I have one of those rs232 USB adapters.. so I just connect to it via /dev/stty device that the system recognised (this is on Mac OS, but should work on Linux just the same, perhaps even on a raspberry pi, you can also do it in Windows I am sure but I wouldn't know how). As long as you can read GPIB from Python you should be able to do the same thing.

You only really need to execute one command: serial.write(':FETCh?\r\n'), wait a bit (I wait a second) and fetch the data.

I just simply write this alongside the timestamp into a comma separated file. That's it. The added benefit is FETCh command works for any measurement you have selected on your instrument. So no need to change the script if you do Ohms, VDC, VDI.. it should all work the same. Also I believe this command is the same for all Keithley instruments and Agilent and Fluke should have similar if not identical.

The trickier part at least for me was plotting this data. But basically the second script just takes this CSV file that's been recorded or is being recorded to by the first script (can do it in real time), crunches it using Pandas statistical time series framework library and plots it with Seaborn to make it look "pretty". The end result is a .png file. This was a lot of trial and error to get the graph to exactly where I want it, but that above is the end result. I think I am just going to also add the duration delta between start data point and end datapoint to my summary before I post it.

[Muxr]

Oh look what just arrived.


I am feeding it 15VDC per instructions on Geller's website and letting it warm up for an hour:


Then I will measure it for 24 hours or so. Post office is closed Monday, Tuesday I have meetings all day long but will be shipping it out to the next person (technogeeky) wednesday.

[cellularmitosis]

THIS IS SO COOL!!!

[vindoline]

Looks great! I would love to be able to log data like that from my Keithley 196 but it only has a GPIB connection. All of the solutions I've seen have been hopelessly complex and expensive.

[cellularmitosis]

vindoline, when I get GPIB working on my 196, I'll share the solution with you.  Since GPIB is a 5V bus, I should be able to bit bang a solution, which means the whole solution should be less than $10.

I'm currently waiting on these boards to arrive: https://github.com/pepaslabs/atmega-gpib

Then it is just a matter of writing the firmware

[cellularmitosis]

Muxr, I forgot to mention, if you have time, please also take some data with the temperature compensation jumper removed.  Might as well!

Also, I just realized I forgot to secure the trimpot screw before mailing it out.  If you have a dab of hot snot or nail polish handy, feel free to do so!

I'm also working on a simple lab temperature / humidity logging system.  I have a BME280 and Si7012 breakout boards and my plan is to log to PC via a serial to USB adapter.  It sounds like we are all working on similar problems!  This thread will be a great place to share solutions.

[kj7e]

For environmental data logging I use this;
http://www.onsetcomp.com/products/data-loggers/mx1101

Not the cheapest solution but very high resolution, some really cool logging features and supper easy export of data in various formats.  Native excel export can be used to overlay plots from other data.

[Muxr]

This is exciting, thanks for suggestions everyone:

Muxr, I forgot to mention, if you have time, please also take some data with the temperature compensation jumper removed.  Might as well!

Good idea, will do!

Also, I just realized I forgot to secure the trimpot screw before mailing it out.  If you have a dab of hot snot or nail polish handy, feel free to do so!

10-4

I'm also working on a simple lab temperature / humidity logging system.  I have a BME280 and Si7012 breakout boards and my plan is to log to PC via a serial to USB adapter.  It sounds like we are all working on similar problems!  This thread will be a great place to share solutions.

Good call, humidity is also important.

As to the results so far. Looks like I am influencing the results somehow.

The reference is on the other side of my desk where I use my computer. I went to bed about 3am.. you can see a slight rise and a pretty uniform trend after that.. so my presence looks to be interfering somehow. Also I have no clue what those spikes are about. That was just around the time I returned back to my office. Perhaps EMI interference.

The unit is powered of my linear E3611A PSU.. I am going to go out and buy me a box of Royal Danish.. empty the cookies and ground it.. place the unit inside of it (after I apply nail polish to the trimpot) and repeat the test. See if I can minimize the environmental interference.

In either case, this thing is looking much better than my cheapo AD584KH reference.. and it's great to be able to log and plot this data (the script is coming today.. I just finished adding a few more features as you see in the summary section).