USA calibration club

[bitseeker]

Hmm, that's interesting. Fortunately, the temp trend is the important part.

A few months ago, I saw this comparison of some different hygrometer sensors. It includes the BME280, which seems to be very good.

http://www.kandrsmith.org/RJS/Misc/Hygrometers/calib_many.html

[CalMachine]

Hmm, that's interesting. Fortunately, the temp trend is the important part.

A few months ago, I saw this comparison of some different hygrometer sensors. It includes the BME280, which seems to be very good.

http://www.kandrsmith.org/RJS/Misc/Hygrometers/calib_many.html

This is fantastic!  I wished I would have seen this a few months ago

[bitseeker]

There's so much good stuff out there, you can't hope to see everything. Fortunately, resources such as this forum enable us to benefit from each other's eyes.

[CalMachine]

Here is a log of the weekend!  Should be headed your way within the next day or 2, CatalinaWOW.

[CatalinaWOW]

Timing is perfect.  I am out of town but will be back shortly.  Should be able to get things warmed up a stabilized soon enough to minimize any delay.

[cellularmitosis]

Thanks for posting your graph CalMachine -- I've been doing some logging from a 34401A and have noticed similar issues that appear in your graph.  If I'm away at work or in bed asleep, things are pretty stable.  But then I walk into the room where the meter is, wake up the laptop display, etc, and I see sharp offsets which are large in comparison to the periods of stability.  Humans seem to be problematic to precision data logging efforts!

[CalMachine]

Thanks for posting your graph CalMachine -- I've been doing some logging from a 34401A and have noticed similar issues that appear in your graph.  If I'm away at work or in bed asleep, things are pretty stable.  But then I walk into the room where the meter is, wake up the laptop display, etc, and I see sharp offsets which are large in comparison to the periods of stability.  Humans seem to be problematic to precision data logging efforts!

Ofcourse!  I wish I had the time to get a better test environment set up.  Judging from the graph, the test environment looks pretty noisy...  The 3458A ACAL threshold was set to 0.4 ºC and ACAL'd itself a total of 4 times throughout the weekend @

22/7/17 20:15:21
23/7/17 03:08:26
23/7/17 15:15:25
24/7/17 00:54:05

I wasn't able to match any correlation between the observed noise spikes and ACAL events.  I will be working reducing noise around here!

[Conrad Hoffman]

Noise comes in many forms. Remember that you are an infrared radiator. When I do capacitance measurements on the standards bridge (1615A) or even the old 716C, just holding my hand under the DUT to adjust the knobs will cause a value shift if the DUT tempco is significant. There are also RF sources we never had to deal with like CFLs and LED lamps. They're not usually well filtered. Best solution is a Faraday cage around the lab, then fill the lab with temperature controlled Fluorinert!

[bitseeker]

You guys are scaring me. I probably have noise galore in my lab: PCs, switching power supplies, CFL bulbs, Wi-Fi routers, unshielded ESP modules, etc. Maybe my graphs will have more popcorn than this guy

[CalMachine]

You guys are scaring me. I probably have noise galore in my lab: PCs, switching power supplies, CFL bulbs, Wi-Fi routers, unshielded ESP modules, etc. Maybe my graphs will have more popcorn than this guy

I wouldn't be that worried.  The lab I work at has many large industrial businesses, TV stations, power lines, and cell towers in the vicinity.  Here in the next few months with the new lab coming, I'll have a completely shielded and much tighter environmentally controlled lab.  I'm excited!

[bitseeker]

OK, I'm enhancing my calm. I think I may be next after CatalinaWOW, so I'll find out soon enough.

[CatalinaWOW]

The happy box arrived today.  It is percolating away on its first run as I type.

The photo shows the reference set up in my basement lab.  The arrival of the reference coincided with a historic heat wave here in Southern Oregon so it is a little warmer than usual in the hideout.  This may also affect my plans.  A careful look at the photo will show a TI temperature and pressure measurement dongle.  Normal Oregon temperatures let me get a range of temperatures around normal room ambient just by sticking things outside the window, leaving the measurement system inside.  Trial runs while waiting for the gold standard showed this worked well.  I balanced the materials in the leads and was able to convince myself that thermo-electric effects were negligible at the level of accuracy that I am capable of.  Temperatures ranged from 55 F to 95 F (13 C to 35 C).  Unless the weather bureau lies the range I will see before I send this on will be more like 65 F to 110 F (18 C to 43 C).  Not a nice balanced range around normal lab temperatures.  I will send the temp/humidity dongle along when I send the reference out.  It appears as a serial port so should be easy for anyone to interface.  Before I ship this stuff out I will post the EZGPIB code I am using.  Needs a little polish before public viewing.

The two key instruments in the setup are an HP 3457 and an HP 3456.  Both have unknown calibration histories with most recent signs of calibration more than a decade ago.  I am pleased to see that they are both fairly close.  If I wasn't coming down with the Voltnut disease I would be ecstatic and not go any further.  Unfortunately..... 

I am also attaching a short sample of my test run on the reference.  First column is date and time in MS julian date format.  Next column is temp (C) and the next is relative humidity.  The sensor is not formally calibrated but tracks well with multiple thermometers and thermocouples I have.  I don't have as much information on the humidity sensor, but it tracks a couple of other measurements I have.  Next column is an HP 3478 monitoring the supply voltage.  Currently it is supplying the voltage regulator that Conrad provided to help provide consistency in measurements, but I also plan to to a couple of voltage sensitivity runs bypassing Conrad's regulator.  Next is the HP 3457 which is being operating in 100 PLC mode and using the extra sensitivity register to get 7.5 digit information.  Finally the HP 3456.  Also operating at max PLC, but clearly the extra digits are missing.

[IconicPCB]

Catalina,

Can You please post the calculated temperature coefficients of the three instruments based on the text file.

0.007131491465747643   
9.653462320521867E-05   
0.00020292057647497002

?

[CatalinaWOW]

I will be happy to post TCs once I have a sufficient data set.  The short sample posted covered only a few minutes, far shorter than the thermal time constant of the instruments, and possibly short wrt the voltage reference/sensor system so I don't believe that anything found in this sample is meaningful.  It will be interesting to sort out the instrument TCs from any TC in the voltage reference.  My own references are not particularly stable (I am new on this volt nut trail) but in trial runs I found that the readings from all three instruments tracked each other much more tightly than the movement in my references.  Using Occam's I concluded that the instruments were probably very stable since the likelihood that the three different instruments had the same TC was low.  Not zero since they all are the same manufacturer, but they are of somewhat different eras and market intents.

Also recognize that for temperature differences of the size seen (0.1 degree) the temperature sensed is unreliable relative to the instruments.  The photograph shows that the instruments are stacked and thus likely have a bottom to top gradient, and are separated from the sensor by half a meter or so.

The final issue is not evident from what I have posted.  My HP 3456 has an odd thermal sensitivity.  When it is exposed to a strong negative thermal gradient it shuts off.  Since I normally open these windows in the cool morning hours to bank cool air against the hot afternoon it frequently shuts down.  This had occurred the morning of this short sample and so the instrument had just been turned on and was still stabilizing.  I expect the data from the longer run started yesterday evening will show this.  I am leaving the windows closed this morning so will get a full twenty four hour run in. 

[alm]

The 3456A shutting down does not sound normal. Could be a bad connection that is stressed due to the thermal gradient.

[CatalinaWOW]

The 3456A shutting down does not sound normal. Could be a bad connection that is stressed due to the thermal gradient.

I think it is in the switch itself, perhaps mechanical in the latch mechanism.  I will be diving into it once I send the club reference on.  Maybe I will get lucky and it is just a loose push on terminal.

[CatalinaWOW]

First results are in.  A day late due to a rookie mistake that resulted in erasing the first days data collection.  So now what I have is the run showing temperature sensitivity of the reference.  I can also glean some stuff about my instruments.  Both the 3456 and 3457 are a bit off, not surprising given how long it has been since they were calibrated.  There is also an indication that the 3457 is a bit drifty.  This surprised me as it had not shown up in my trial runs.  Will look more into that in the rerun of day one which is going now.

The real surprise was the temperature sensitivity of the reference.  It is not too bad below 30C, but definitely wanders at higher temperatures.  Fortunately most labs will not have to deal with that.

[Edwin G. Pettis]

It is quite obvious that the heater resistor is too low of a value for the higher temperatures, the LTZ is losing temperature regulation, it needs at least a 13K or higher to operate at higher temperatures.  Remember, it is not the ambient temperature that the LTZ works from but its internal temperature, the heater resistor must be high enough to allow the LTZ to servo its internal temperature at whatever ambient temperature it is in.

[kj7e]

It is quite obvious that the heater resistor is too low of a value for the higher temperatures, the LTZ is losing temperature regulation, it needs at least a 13K or higher to operate at higher temperatures.  Remember, it is not the ambient temperature that the LTZ works from but its internal temperature, the heater resistor must be high enough to allow the LTZ to servo its internal temperature at whatever ambient temperature it is in.

Last week I ran test to insure the 13K/1K ratio was not going to be an issue with the LTZ1000A at my intended operating temp of 35 deg C.  My testing showed I was okay up to about 43 deg C, any higher I would run out of headroom.

LTZ1000A heater current at 15v;
23 deg C = 22mA
35 deg C = 15mA
40 deg C = 10mA
43 Deg C = 8mA

[lars]

@CatalinaWOW reply #241

I have to ask about the DUT. Is it an SVR-T or LTZ? Edwin says something about LTZ. If it is an SVR-T I should say it is very good to be an simple NTC compensated AD587LQ with just a few ppm over 17-39C. The NTC just compensates the first and second order and just over a narrow temperature range it is very effective, normally 20-30°. The spec for the SVR.-T also is +-1ppm/C. The third order term found in the AD587 is not compensated at all.

I also guess you just temperature cycled the DUT not the DVMs? Or did all your setup (room) vary 17-39C? How long was the measurement time?

I have to ask what you mean by both instruments are a little bit off? As I see the HP3456 at 23C is within 1ppm that is very close for me. Guess the fluctuation over some days may be more for an HP3456 and also depending on how long it has been on.


@CalMachine reply #228

Is the blue line the SVR-T? If so it seems to vary from 2-7ppm above 10V. Earlier you had closer and less variation. Is the environment the same. In the lab you work? I find this very intersting as it shows how difficult it is to be sure of a result.
As I understand you use a calibrated HP3458 as Dr. Diesel also did (calibrated at your work?). His result was very close to spot  on within 2ppm if I remember correct. Both for the DUT and DMMs. How was the calibration of your 3458?


@-To the owner of the SVR-T (if it is one SVR?) How old is it? Any serial number? Think I saw 911 on some picture? If so I guess it from about 2012-13. Has it been powered most of the time or not?


I think the round robin is really nice but ask for more detailed info about DUT and used setup on every post if possible to easier compare result. Ok, I know I am not better at this and it not easy to know what we as audience needs. Pictures and texts for the DUT and measuring device is nice.

Lars

[cellularmitosis]

Yeah, in a future round we will want to do a star pattern rather than round robin, to detect drift during transit.

[CatalinaWOW]

@lars

I love questions, they focus my ignorance.  I can describe the reference I am testing - owner is cellularmitosis I believe.

Geller Labs Reference - model number is not obvious in the available view.  I don't want to disturb it during a run.
The number 977 is inked on the board.  (Or possibly 911 if written Euro-style, but the characters in question look more North American than the Euro style 1.)  Don't know if this is the serial number.
The key chip is clearly marked AD587LQ with a date code of 0054.  I haven't found a data sheet admitting to the existence of an LQ grade so don't know what that means.

The DUT was exposed to normal diurnal temperature swing (started at approximately peak daytime temperature, and went 14 hours, a little past where temperatures began rising the next day.  NPLC for 3457 was 100, 3456 was at default value of 10 (another mistake that has been corrected for the current run.)  The DVMs were all thermally isolated from these changes in a room whose temperature was 26 C +/- about a degree.  Measurements were taken about every 8 seconds.  The plotted data is the running average of 11 of these measurements. 

My comment about the instruments being off was a lazy reference to the difference in readings and the difference from the stated perfection of the reference.  I have not yet even plotted error bands or the range specified by HP.

[cellularmitosis]

Yeah, the AD587LQ is no longer manufactured.  They do still offer a hermetic CERDIP version (AD587UQ), but is specified over the wider industrial temperature range, rather than the commercial temperature range. 

Andreas has hinted before that the UQ has a less desireable tempco curve, because its was designed to operate over a wider temp range.

[martinr33]

If it could come to the Bay Area, I would love to get my hands on it for a couple of weeks. I have some local friends with well-characterized 732As that would give us some interesting numbers.

[cellularmitosis]

If it could come to the Bay Area, I would love to get my hands on it for a couple of weeks. I have some local friends with well-characterized 732As that would give us some interesting numbers.

Sounds great!  I'll tack you onto the end of the route.  Can you shoot me a PM with your address?