EEVblog® Electronics Community Forum
Electronics => Metrology => Topic started by: bdunham7 on March 13, 2024, 09:27:49 pm
-
I have been asked to administer the remainder of USA Cal Club Round 3 and will likely stay on to do a Round 4 if wanted. I have received most of the kit, but the L&N 10k standard resistor is still enroute. It has an issue with leaking and I'll be trying to correct that before it goes out with the kit. I see no reason to delay continuing the round but if someone really wants that 10k standard, they'll need to defer getting the kit until is is available.
Currently the kit consists of the FX reference, the original PX reference, a GPSDO 10MHz source, a multi-value resistor box, a GPIB adapter, a temperature/humidity measuring device and some miscellaneous leads and whatnot. I'm going to reserve 6 posts below this for each of the major items so that this post isn't so long. I'll cover things here assuming the reader is new to USA Cal Club, but I may not cover every detail about all the items in the kit. Please refer to the first post of USA Cal Club Round 3 for links to instructions for the legacy items.
https://www.eevblog.com/forum/metrology/usa-cal-club-round-3/msg4715543/#msg4715543 (https://www.eevblog.com/forum/metrology/usa-cal-club-round-3/msg4715543/#msg4715543)
I will add a transit temperature logger, some low-EMF Z-plug leads for the FX reference and a 4-wire short. I'm concerned about wear and tear on the FX reference and as much as we like bare copper on binding posts, I think the Z-plugs will spare the FX any further torque-related incidents. If you have extraordinarily good equipment and a light touch, then go ahead and use the bare wire if you want, but I've not been able to tell any difference.
I will send a personal message to the remaining participants shortly. I'm going to loosely try to stick to the order that they were listed by Randall, but my priority is getting through the round in a reasonable time and I may make adjustments to further that goal. The kit is shipped in a large flat-rate box via USPS and I'd like everyone to continue doing exactly that both ways. The cost is $19.90 each way.
The current participant list is:
CatalinaWow
rax
edavid
rockstedy40
SilverSolder
rsjsouza
jjoonathan
bdunham7
conrad hoffman
cscaccetti
uski
AeroETech
Smokey
maxwell3e10
1audio
RoadDog
rhb
ycui7
rockstedy40
mower
zepto
chuckb
CurtisSeizert
Update: Based on the responses so far, I think there needs to be a fundamental shift in the way this is scheduled. Being prepared and able to receive the kit, use it effectively and return it in 10 days or so requires some planning ahead. Lots of people need to defer for various reasons and I had the same issue--when my turn originally came I had a medical issue that prevented me from taking it. I've shipped the kit to the only user who was able to use it immediately so in the next week or so I'll come up with a new plan. If you are on the list above and are not crossed out, please respond to the PM that I sent you.
-
The FX Reference is the central figure in the USA Cal Club kit, provided by TiN from xDevs. You can read some of the details on xDevs.com.
https://xdevs.com/article/usac_2023/ (https://xdevs.com/article/usac_2023/)
What it is and what it does is well documented so I'm just going to concentrate on how to properly set it up and what you can do with it. It takes a bipolar power supply or 2 12V SLA batteries; the +12V supply will draw about 35mA (due to the LTZ1000 heater) and the -12V about 6mA. A high quality linear supply is needed and please be very careful about how you connect it as we don't want to see this valuable resource take any damage. The outputs are binding posts and seem durable, but previous versions have been broken. I have included some low-thermal-emf leads (silver plated z-plugs) that can be plugged into the banana jack part of these terminals to avoid the issue of putting torque on the posts. Please gently install the shorting wire when the FX is not powered on or connected to something.
The FX puts out 9.9999642V with a tempco that is negligible for almost all users. This allows you to adjust or null your homemade 10V references and verify commercial references, DMMs, etc. What I've found it to be useful for is characterizing the tempco, noise and other errors of my various 6.5-digit class meters. I'd like to hear from other users exactly how they have found the FX to be useful. Here is an example of a plot of the FX measured by my Fluke 8846A looking at the difference between having the display on or off. The temperature indicated is the case temperature of the FX, unfortunately I currently only have the means to log one temperature reading to the PC. You'll just have to take my word that the ambient temperature was pretty well controlled to 23.7 +/- 0.5C during this test. The display was turned off for the first half and on for the second and it looks like that made about a 0.2ppm difference. Note that the entire chart is 1ppm from top to bottom and that with the display on (as it would have been for calibration) the meter is off (high) by ~0.33ppm.
(https://www.eevblog.com/forum/metrology/usa-cal-club-round-3-1/?action=dlattach;attach=2076152;image)
The next thing for me to do is to try and characterize the tempco of the meter over a wider range. I haven't finished this process, but I started by trying to look at the tempco of the FX itself. I put it in a place where it was not shielded from my HVAC like the rest of my bench was and recorded overnight. Again, unfortunately, I can't do two temperatures and thus did not log the ambient temp near the meter. However, it would have started about 23.5C and then cooled off to about 21C overnight, only rising again right at the very end. What you can learn from this plot is that the FX truly really does have a very low tempco, just as advertised. And I now know that the 8846A has a slight negative tempco, at least over the small temperature range I've tested here.
(https://www.eevblog.com/forum/metrology/usa-cal-club-round-3-1/?action=dlattach;attach=2076158;image)
-
The PX reference from the original USA Cal Club, provided by cellularmitosis, is a small LTZ1000-based reference without a 10V stage. It's output was last measured at xDevs at 7.0456489V. It should be powered by the 12V regulator assembly provided or perhaps by a 12V battery. There are two resistors, 1K and 10K, built into it and special 4W connectors to interface with it. I've labeled the cables and there is a clear picture of how to set it up in the orginal instructions.
https://drive.google.com/file/d/1AoXDJ-ybEtL40Hh-SGyr5oJMXcGVgf-l/view (https://drive.google.com/file/d/1AoXDJ-ybEtL40Hh-SGyr5oJMXcGVgf-l/view)
A check of the PX vs 8846A indicates that it seems to be accurate and not very susceptible to temperature changes. The temp reading is of the PX case.
(https://www.eevblog.com/forum/metrology/usa-cal-club-round-3-1/?action=dlattach;attach=2076248;image)
-
reserved
-
Can I join the USA Cal Club Round 3.1?
I have a 3458A that has been checked with data in a cal lab and left as is (not recalibrated) a little less than a year ago among other equipment.
I also have a GPIB raspberry PI data logging setup with temperature and humidity ready to go.
-
Yes!! Let's crank this thing up!
It's an amazing opportunity and we should all take advantage of it.
-
I attached a picture of the noise of the FX. I'm using a 10000x LNA and Siglent SDS2104X Plus.
Over about 33.3 minuets, there is a maximum peak to peak noise of 2.7uV.
-
My 3458A was calibrated on Nov 23 2023. That would be about 11 months from when I got the FX and ran this test (10-31-24). The calibration report says 3458A measured 10.0000207V on a 5700A calibrator set to 10V. That would be 2.07PPM above the calibrator at 23.3C air temperature. The standard drift for a 3458A is 8PPM per year. The internal calibration temperature of the 3458A is 40.2C. At that internal calibration temperature, the 3458A measured 9.999984V from the FX with an air temperature 23.4C. The label on the FX says the voltage should be 9.999964V at 20.9C air temperature. That would be 2PPM above the FX. The FX also says there is an uncertainty of plus or minus 0.8PPM. The FX has a temperature stability of -0.0242PPM/C. The 3458A specification gives 0.51ppm/C for the 3458A without ACAL (anomaly calibration).
I did a 2nd run with periodic ACAL. Readings can’t be taken during the ACAL. The 3458A measured 9.999989V from the FX at an internal temperature of 40C and 23.8C air temperature. That would be 2.5PPM above the FX. The 3458A specification gives 0.16ppm/C for the 3458A with ACAL.
-
I connected the ground of the FX to ground of my Fluke 732A and measured the voltage difference between the two using a 34420A nano volt meter. The graph shows the FX 9.9999642V calibration voltage minus the voltage difference measured by the 34420A. The Fluke 732A has a temperature coefficient specification of 0.05PPM/C which is close to the FX. This the first time the Fluke 732A has been calibrated.
-
My oscilloscope shows exactly 10MHz with the GPSDO.
-
The calibration report says my 3458A measured 9999.594 ohms on a 5700A calibrator set to 9999.657 ohms. That would be 6.3PPM below the calibrator at 23.3C air temperature.
While the FX has a temperature coefficient of -0.0242PPM/C, the L&N 4040 10K has a temperature coefficient of 8.461PPM/C at 23C. The 3458A has a 10K resistance temperature coefficient of 3.1PPM/C without ACAL and 1.1PPM with ACAL. In other words with the FX, temperature has a greater influence on the meter. With the L&N, temperature has a greater influence on the resistor. However, the L&N temperature coefficient decreases at higher temperatures. According to xdevs, the temperature coefficient becomes zero at 30.3C and doesn’t go above 10000.4 ohms. However, my 3458A is measuring a resistance above 10000.4 ohms at a temperature below 30.3C. The drift for this resistor is -3.7PPM/year so the resistance should be going down with time. In this experiment I am measuring the air temperature and not using a thermometer touching the resistor.
-
Since measuring temperature accurately is important for the L&N, I'm testing a TMP117 and a 100 ohm 4 wire RTD (Resistance Temperature Detector) in a bag in water and ice bath to see if they get close to 0C. Fluctuations is from me mixing the thermometers, water and ice.
-
The standard drift for a 3458A is 8PPM per year.
How old is your 3458A? Is it truly an HP branded, or an Agilent, or Keysight? If the former, I'd expect the drift at this age to be far less than the 8ppm they spec'ed at the beginning of the life of that product (unless the reference module was recently replaced, which would reset the clock, obviously). It's likely in the fractional ppms, if even that. Effectively, I expect option 002 to not make a difference with units past a certain age.
The other comment I have is also a bit tangential to the topic of US Cal Club and your empirical results from using the kit, but it's an observation I made when having the FX here alongside an HP 3458A with relatively fresh calibration, where the ACAL was critical to any relevant measurement (bdunham7 was also here to witness it). It may depend on many factors - age, [long] time spent in storage, etc. - but based upon that experience I highly recommend doing a fresh ACAL with the 3458A any time one executes a critical voltage measurement.
Good work with the kit!
-
How old is your 3458A?
I'm not the original owner but if I stick my serial number here:
https://support.keysight.com/s/service-and-warranty-status?language=en_US (https://support.keysight.com/s/service-and-warranty-status?language=en_US)
Shipped Date:
2/17/2008
Is it truly an HP branded, or an Agilent, or Keysight?
It says Agilent on the front.
-
I would like to join this round too if its not too late. I'm based in CA.
I have a recently calibrated 3458a and DMM7510 and a "secondary standard" 10K resistor from Texas Components.
This way I'll be able to verify how the meter is doing and will be able to get more accurate readings for my REF80 experiments.
-
When temperature is stable, I was able to get a resistance reading of 10000.095 ohms at 23C. Using the 10000.0577 ohms values (assuming no drift since January 10, 2023), that would be 3.7PPM difference. Drift of -3.7PPM/year would add an additional 6.87PPM or 10.5PPM difference.
-
I just returned the kit. It was interesting to explore. For the most part really boring (in a good way). I fired up the FTX reference with a Leader bench supply, gave it an hour to stabilize and connected it to my 732A with a QuantAsylum QA350 https://quantasylum.com/products/qa350-microvolt-dc-volt-meter as a nulll meter. The QA350 is really good for this. Its USB powered, fully isolated (10G isolation) 1 uV resolution with decent PC software. Its been replaced with a QA351 https://quantasylum.com/collections/frontpage/products/qa351-uv-dc-voltmeter with two inputs. I modified mine by adding banana connections to work around the thermals of a BNC.
And then things got boring. maybe 3 uV change between the standards (+65 uV difference) ove the several weeks of testing. Ambient temerature changed as much as 25F with no obvious difference. Basicaly they tracked. Good. no obvious drift over 2 weeks. The changes are essentially noise.
Next issue the marked voltage on the top vs. the voltage vs. temp chart on the bottom and the readings from the Xdev's web site. I decided to ignore the confusors and just use the marked voltage as "accurate". My uncertanties are way larger than any of these issues.
The other curiosity- The 732A I believe has never been adjusted since it was made. I got it as surplus from UC Berkeley and never adjusted it. These are supposed to drift positive but The measurements are all below 10V. A friend checked his against mine in 2019 and we determined a voltage then. Which is lower than the current value. Again all less than my uncertanties. Recors in the attached photo.
I checked the 10K reference resistor against my 10K's and made note of the different numbers on 2 Fluke 8506a's. I may try recalibrating however I have a high probability of blowing it completely so I may just leave it. The differences were small on my scale. I contributed a 10 Ohm Rubicon (similar to the 10K L&N) that I don't really need which may prove useful to someone.
I have 2 GPSDO's so another would not contribute a lot of info for the overhead of installing it.
I really do appreciate the opportunity to check my references.
-
I currently have the FX reference only and want to share my results so far. The error between my DMM and the FX is 1.58 ppm 8)
My 3458A has a certificate of calibration from Agilent dated October 20, 2011. That is its only official calibration. I performed a calibration using the FX reference about 2 years ago and at that time the meter was about 6 ppm high. The Agilent certificate lists the 10 volt dc uncertainty at 3.64 ppm (option 002). The meter was purchased as new/used and came with all documentation and accessories and was packaged in the original Agilent container.
It appears that the total drift of the meter, from the time of manufacture, is less than 10 ppm ... remarkable!
Many, many thanks to TiN for the contribution of this trusted reference to the USA Cal Club.
-
If i'm not mistaken, the FX reference will be shipped to us next. We will characterize it with our 732 reference bank and tag it with the latest value. Our reference bank was calibrated using a 732c sent to Keysight's Standards Calibration Laboratory in California. It will be interesting to compare it to the xdevs value.
TomG.
-
I forgot to say that we have an ISO17025 accredited laboratory, and we're going to donate this calibration/characterization to the group so you'll have an accredited calibration (minus the certificate) going forward. For anyone interested, our uncertainty value for this is under 0.2ppm.
TomG.
-
Is the FX back yet?
TomG.
-
Should be. Tracking info says it was delivered March 31, 2025, 11:38 am.
-
It's on its way. I'll post some numbers here when we get it.
TomG.
-
It's Alive!! Initial readings -> 70.49 degrees F, 31.49% rh.
3458a -> +9.9999542
2002 -> +9.9999805
Zeroed leads, ACAL ALL on 3458a, stable reading over 30min.
Voltage bank testing script is running now.
TomG.
-
That is miraculous USPS performance...
I'll be interested to see how closely you can measure its warmup performance. 30 minutes will be almost there, but it seems to take 12-24 hours to squeeze out those last few ppbs.
-
First Voltage Reference Bank FX assigned value: +9.999956709. Less than 24 hours powered up.
Setup conditions: +/-12Vdc by 4x6Vdc AGM SLA batteries for lowest noise. Actual voltage was 12.8Vdc at start.
70.49 degrees F, 31.49% RH
Hope this is useful to the group. Batteries should last 4-5 days, so I'll update with daily readings and send it back by week's end.
Voltage Bank: 4x732A, 4x732b, 1x732c (characterized at Keysight's Standards Calibration Laboratory) with 34420a using NIST measurement methodology running custom data acquisition and mathematical regression script.
PS - Kudos to my friend Ilya at xDevs, truly remarkable performance!!!
TomG.
-
Ok, the day 2 number is: +9.999955289Vdc. Temp and humidity within +/-1 of yesterday's values. Next update will be tomorrow evening.
TomG.
-
I guess I should also point out that the difference between these numbers is nearing the noise threshold of all the devices involved.
TomG.
-
OK, day 3 last reading: +9.999955144, day's average: +9.999954989.
TomG.
-
OK, day 3 last reading: +9.999955144, day's average: +9.999954989.
That's a lot of digits! Is the final result going to be 9.999955 +/- 2µV? That puts it at -0.9ppm from the previous calibration over almost 18 months and dozens of cross-country trips via USPS.
One question about this--what is your educated opinion on how close your 10V is to xDevs 10V?
-
OK, day 3 last reading: +9.999955144, day's average: +9.999954989.
That's a lot of digits! Is the final result going to be 9.999955 +/- 2µV?
The digits result from averaging samples over the hours, so we keep the digits for the math. Yes, the value looks like it's going to be +9.999955, well within +/-2µV.
TomG.
-
One question about this--what is your educated opinion on how close your 10V is to xDevs 10V?
I just noticed your question. xDevs' and our 10V values are extremely close, probably +/-1ppm or less. The master 732c in our lab was compared to Keysight's primary JVS less than a year ago, so our confidence level is very high.
TomG.
-
Ok, day 4 reading: +9.999955408. That's the average. Variations pretty small today.
TomG.
-
You guys wanted to know how close I was to Ilia's value. Now you're going to have your answer because I dropped the FX reference off with him and he will characterize it. And then we'll see the difference.
TomG.
-
So Ilya's first measurement of our 732b transfer standard is less than 0.18ppm away from our value. So very high level of confidence in our measurement setup as well as his. Hope this is useful information to you guys.
TomG.
-
Thats really reassuring, even though its well below anything I could confidently measure. I now have some confidence in my 732A measured value.
-
Second 1audio's point. Many thanks, TomG, all this data is in spot on agreement with everything I've been able to assess across my bench standards (to a scary level of agreement, candidly).
-
A request was made for final value and measurement uncertainty for this calibration/characterization. Here it is:
+9.9999554Vdc, uncertainty for this measurement is 0.2ppm.
TomG.
-
Hello Forum Members,
I have an Agilent 3458A which has lost its calibration data. CatalinaWOW encouraged me to join the "club" here in the hope that an excellent voltage reference, etc., might come to its aid. So please sign me up or let me know how I can help. I'm in So-Cal if that matters.
Thank you very much!
-
I have an Agilent 3428A
3458A maybe?... Otherwise 3478A?
Welcome to the club!
-
Ah yes, that was a typo. It's an 3458A, 8.5 digit DVM (fixed).
Thank you! :)
-
Ah yes, that was a typo. It's an 3458A, 8.5 digit DVM (fixed).
Sent you a PM.
What condition is it in and has the reason for losing the cal data been addressed? When these get old enough to have problems they typically need heroic levels of TLC to get them back into shape.
USA Cal Club Round 4 is coming in a bit, I've been a bit preoccupied and that may be a number of weeks off. I want to get it reorganized to be as useful as possible to as many people as possible.
-
Heyo :D
It appears my volt nut disease is flaring up again after being in remission for years.
I was curious if the px ref could make its way to my place?
(edit: bdunham7 shot me a DM, thanks!)
-
Yes, I've also been curious if the standards might make another round.
-
Hit me up if you want it characterized before sending it to the members here.
TomG.
-
Hit me up if you want it characterized before sending it to the members here.
TomG.
I surely do! Thanks so much!
-
I would be up for a recheck as well.