USA calibration club

[Muxr]

I've posted the capture and plotting script on github.. https://github.com/sirmo/muxrplot

All the install and usage instructions are there as well. Let me know if you run into any issues.

edit: I should also add, for the folks on GPIB, you would need to modify this function only to use GPIB instead:

https://github.com/sirmo/muxrplot/blob/master/mplot/logger.py#L94

I can look into adding GPIB functionality as well.

[vindoline]

Thanks Muxr, I'm no programmer, but this looks like I could make it work. The major hurdle for me is going to be on the hardware side.

[Muxr]

Thanks Muxr, I'm no programmer, but this looks like I could make it work. The major hurdle for me is going to be on the hardware side.

No problem vindoline! I actually made a dedicated thread for the script, so feel free to ask questions there, (so we don't derail this topic). https://www.eevblog.com/forum/metrology/mplot-logging-and-plotting-script-for-long-term-measurements/

[Muxr]

I did another 15hour run. I am somehow introducing some significant noise when I am in the office. The grounded Royal Dansk box has helped with big spikes, but there is still a good amount of other noise.

Perhaps I am introducing noise to the Kei2015 or my PSU. Perhaps it's my computer monitors adding noise?

After I ship out the unit I will do some base level measurements of the noise floor with just the two probe leads connected. I also have some BNC couplings on order to improve the shielding.

edit: changed the probes, using the red and black twisted pair that came with the reference. The meter is looking more stable, so hopefully this solves the issue.

edit2:

ok things are looking much much better with the provided red and black twisted pair:


Standard deviation is less than 3 uV, which is really good at this stage. With old probes it was over 10 uV half hour in. Also the mean is looking much closer to where it was calibrated. I will let this one run for about 12 hours and then repeat with the heater (jumper) off.

[cellularmitosis]

Very interesting!  Those red/black probes use high-quality pomona grabbers on one end, but the banana jack ends are just cheap ebay plugs.

In the above graph where there is a quiet section in the middle, was that when you put the reference inside of the tin?  Or was that when you left the room?

Have you done a comparison of inside / outside the tin?

On my end, I finally managed to get a more reasonable result with my Keithley 196 and a few LM399's I've been playing around with.  Previously, if I took three steps to the side, the value would climb by about 40 counts, and then come back down if I moved to stand directly in front of the meter again.

I noticed that the reading jumped if I connected an oscilloscope lead to the board (in fact, just connecting the ground pin changed the value), so I suspected an EMI problem.  However, putting the board inside an aluminum caldero didn't seem to change this 40 count drift at all.

I was surprised to find that simply putting a 0.1uF ceramic cap directly across the (zener) output of the board has reduced that drift to 1 or 2 counts, even when it isn't shielded at all.  (I was surprised because I don't see such a cap in e.g. the Keithley 196 reference schematic!).  That change (combined with removing the lid of my 196, which seems to be running hot), has the output of this board staying within 1 or 2 counts all day today (from a zeroed point this morning).

[Muxr]

Very interesting!  Those red/black probes use high-quality pomona grabbers on one end, but the banana jack ends are just cheap ebay plugs.

In the above graph where there is a quiet section in the middle, was that when you put the reference inside of the tin?  Or was that when you left the room?

That whole graph is in the tin box. The one with a few giant spikes previous to that one was without the box.. so I've been using the tin box with all the subsequent graphs I posted. The quiet period in that graph you're referring to begins like 2-3 hours after I went to bed. This is why I was suspecting maybe one of my monitors was causing interference. Because that's actually how my computer's energy settings are configured:



So it aligns about perfectly. So I am guessing as soon as the monitors (2) went to sleep the noise stopped. And I am assuming the measuring leads from the the DMM and the tin box were picking up interference.

Now that I actually have a way to do long term logging and plotting, I can experiment and eliminate the interference. I will probably setup some of my instruments on the other side of my office at some point, for these types of experiments, and just use a Raspberry Pi to log to. But for now it looks like switching to the twisted pair leads you provided is giving me much better results.

I just checked the current progress of the test, and it looks like I got a big spike again, but overall noise is much lower as reflected by the standard deviation figure as well. 2.7 uV is the best I've gotten so far.

On my end, I finally managed to get a more reasonable result with my Keithley 196 and a few LM399's I've been playing around with.  Previously, if I took three steps to the side, the value would climb by about 40 counts, and then come back down if I moved to stand directly in front of the meter again.

I noticed that the reading jumped if I connected an oscilloscope lead to the board (in fact, just connecting the ground pin changed the value), so I suspected an EMI problem.  However, putting the board inside an aluminum caldero didn't seem to change this 40 count drift at all.

I was surprised to find that simply putting a 0.1uF ceramic cap directly across the (zener) output of the board has reduced that drift to 1 or 2 counts, even when it isn't shielded at all.  (I was surprised because I don't see such a cap in e.g. the Keithley 196 reference schematic!).  That change (combined with removing the lid of my 196, which seems to be running hot), has the output of this board staying within 1 or 2 counts all day today (from a zeroed point this morning).

Interesting. It will be good to see other folks do some measurements with this traveling reference and compare our findings.

[Muxr]

Just compared those spikes before and after the tin box and they are much lower. When you used the aluminum caldero did you ground it?

I have both my lid and the bottom part of the cookie tin box grounded with alligator clip jumpers from the tin box to one of my oscilloscope BNC connectors grounds (it's conveniently close).

Also the other thing I should mention. I have the negative port of my PSU also jumpered to earth's ground. So the actual DUT/voltage reference is grounded as well. I noticed you were powering yours of batteries, perhaps try grounding the negative lead as well.

[Muxr]

@technogeeky she's on her way to you.

So the final two tests (both about 17 hours long). First one is with the temperature compensation jumper on:


And this one is with temperature compensation off (I even tried using slow measurement mode):


You can see the number of pops or spikes is much greater. Haven't isolated those yet.. could be my instrument, environmental interference or the reference itself.. will be interesting to see more results.

[Alex Nikitin]

@technogeeky she's on her way to you.

So the final two tests (both about 17 hours long).
You can see the number of pops or spikes is much greater. Haven't isolated those yet.. could be my instrument, environmental interference or the reference itself.. will be interesting to see more results.

Spikes excluded, the drift is not much different on these two graphs.

Cheers

Alex

[nikonoid]

If I am not too late, I also would like to participate.

I got a calibrated K2000 as well as K2015, 34401a, DMM check plus, two dial-a-volt standards and now repairing K2002.

I am in Northern New Jersey. Feel free to PM me.




Sent from my iPhone using Tapatalk

[bitseeker]

@cellularmitosis: On this first go-round, is the primary goal to get lots of samples made via the members to see how the standard behaves and holds up? Is calibration of the standard taking place as well or will that happen on a future trip around the loop?

[technogeeky]

@technogeeky she's on her way to you.

Roger.

[vindoline]

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

I just had a quick look at this project. Looks great! If you get this working, I'll definitely make a batch! Please keep us updated. 

[cellularmitosis]

nikonoid: we can probably work something out -- the reference will be working its way through jersey.  I can email out a couple of updated shipping labels to stitch you in the loop.

bitseeker: well, the goals of this first round are more about getting people involved and working the kinks out of the system.  Getting us all to go through the motions -- from that will emerge a set of scripts, practices, hardware, knowledge, etc. which are known-good which we can use in future rounds, with other references (hopefully many references from many forum members!).  But yes, it is also about establishing the beginnings of a known history for this particular reference.  My ultimate vision here is something like a small fleet of references which circle around the country every so often, which establishes a "relative" history or drift, and then occasionally sending them to someone like CalMachine to have a (paid-for) professional spot check, which ties that drift to a known value.

[cellularmitosis]

I just had a quick look at this project. Looks great! If you get this working, I'll definitely make a batch! Please keep us updated.

vindoline, the boards are in!  Now time to get serious about that firmware...

(I'd be happy to mail you a board for free -- I ordered like 10 of them through dirtypcb's).

[Muxr]

I just had a quick look at this project. Looks great! If you get this working, I'll definitely make a batch! Please keep us updated.

vindoline, the boards are in!  Now time to get serious about that firmware...

(I'd be happy to mail you a board for free -- I ordered like 10 of them through dirtypcb's).

heh, I use dirtypcb's too Good to see their blue looks good, I usually order red. Digging the project by the way! Most of my gear has rs232 or ethernet support but I do have a few that could really use a GPIB adapter.

[vindoline]

Looking good!

[technogeeky]

I received the kit this weekend. I'll try to make some measurements on Monday and Tuesday and have it on its way by Wednesday.

Let me know if you want me to insert some more shipping labels, or change the next hop!

[bitseeker]

bitseeker: well, the goals of this first round are more about getting people involved and working the kinks out of the system.  Getting us all to go through the motions -- from that will emerge a set of scripts, practices, hardware, knowledge, etc. which are known-good which we can use in future rounds, with other references
...

Sounds good! Thanks again for getting this started.

vindoline, the boards are in!  Now time to get serious about that firmware...

(I'd be happy to mail you a board for free -- I ordered like 10 of them through dirtypcb's).

You might give the firmware from Emanuele's Arduino GPIB project a try. At least it could be a good starting point.

[orin]

I did another 15hour run. I am somehow introducing some significant noise when I am in the office. The grounded Royal Dansk box has helped with big spikes, but there is still a good amount of other noise.

Perhaps I am introducing noise to the Kei2015 or my PSU. Perhaps it's my computer monitors adding noise?

After I ship out the unit I will do some base level measurements of the noise floor with just the two probe leads connected. I also have some BNC couplings on order to improve the shielding.

edit: changed the probes, using the red and black twisted pair that came with the reference. The meter is looking more stable, so hopefully this solves the issue.

edit2:

ok things are looking much much better with the provided red and black twisted pair:


Standard deviation is less than 3 uV, which is really good at this stage. With old probes it was over 10 uV half hour in. Also the mean is looking much closer to where it was calibrated. I will let this one run for about 12 hours and then repeat with the heater (jumper) off.

Does that SVR-T have a capacitor on its output?  I had to use about 0.22uF to prevent an HP 3455A reading differently when the leads were swapped... and they are susceptible to RFI from cellphones too.  A nice metal box, cap on the output and twisted pair from the SVR to meter should get rid of most of the noise and hopefully, the spikes.

[Muxr]

Does that SVR-T have a capacitor on its output?  I had to use about 0.22uF to prevent an HP 3455A reading differently when the leads were swapped... and they are susceptible to RFI from cellphones too.  A nice metal box, cap on the output and twisted pair from the SVR to meter should get rid of most of the noise and hopefully, the spikes.

I don't think it has a capacitor at the output. Lars here posted the schematic: https://www.eevblog.com/forum/metrology/best-out-of-the-box-10v-reference/msg1162989/#msg1162989  I didn't investigate it closely. I did eventually move to the twisted pair which helped with general noise (see the latest results: https://www.eevblog.com/forum/metrology/usa-calibration-club/msg1222185/#msg1222185). The spikes were largely gone with the temperature compensation jumper on as well.

I've been characterising some AD584 references since, also an LT1021 based DMM Check Plus, and I am not getting those spikes. Overall noise and tempco was better on SVR-T though (not counting the spikes).

[RandallMcRee]

I posted in the Kx thread, but thought I would post here as well--
if anyone needs a spare Kx Ref board, I have one. $17 + shipping....from OSH
https://xdevs.com/article/kx-ref/

Hope to contruct one, run it a while and then send it around in the future.

Thanks,
Randall

[nikonoid]

If you still have it, I will take it.


Sent from my iPhone using Tapatalk

[nikonoid]

I recently came across this development board for temperature and humidity readings.

http://www.mouser.com/ProductDetail/Texas-Instruments/HDC1010EVM/?qs=KuuZdrM3jLxa5c3Zf%2fmjNA%3d%3d

While not the most precise instrument out there, it has OK specs and ease of USB port. This could be a temp/humidity sensor to travel with voltage standard.

Does anyone have direct experience with these?

[RandallMcRee]

A good suggestion.

I don't have experience with that one but note that TI has several with similar specs. This one, the 1080, has identical specs with a wider temperature range, same price:
http://www.ti.com/tool/hdc1080evm

It's also $29 at digikey or the TI store. Datasheet mentions exporting data as a CSV which is universal, but primitive. The problem then might be aligning the time axis, i.e. making sure that the temperature time reading is correctly aligned with your GPIB time, if that makes sense.

Regardless, the suggestion of putting a (small) usb circuit in the mailing for measuring temp&humidity (T&H?) is a good one.