Author Topic: DMM Noise comparison testing project  (Read 215739 times)

0 Members and 1 Guest are viewing this topic.

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2252
  • Country: ca
Re: DMM Noise comparison testing project
« Reply #75 on: April 18, 2015, 03:32:06 pm »
To compare the ADC I assume you will need to divide by the used range, also only compare the same number of PLCs, the distribution graphs would then be in ppm. (not nV)
I agree that ppm makes a lot more sense than absolute voltage.

Need to keep in mind too that different meters have different ranges. Agilent usually to 1 V, Keithley 2001 to 2 V, Keithley 199 to 3 V (and multiples). This of course affects ppm of range type data.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #76 on: April 18, 2015, 04:26:14 pm »
Meaning ppm from range?
Representation is easy to modify.
« Last Edit: April 18, 2015, 04:34:23 pm by TiN »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2252
  • Country: ca
Re: DMM Noise comparison testing project
« Reply #77 on: April 18, 2015, 04:39:42 pm »
Meaning ppm from range?
Representation is easy to modify.
Yes, ppm of (nominal) full scale of the range being tested. The range being tested (e.g. 2 VDC) should be including in each graph to give proper context.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #78 on: April 19, 2015, 02:42:01 pm »
        0.01 NPLC0.1 NPLC1 NPLC10 NPLC50 NPLC100 NPLC1000 NPLC

TiN,

I have been busy this past week but I believe I have the 3458A working with all 35 data sets in an automatic acquisition. I hope to modify it with little changes on the newer SCPI meters. I can create all 35 files with one run.

I just need clarification on the data set sizes. I have the script written to acquire based on NPLC size. Would you break down a list of samples / NPLC? Dr Frank mentioned up to 10K samples on the .01NPLC setting and decreasing from there. I just have to plug in the numbers and run the test.

I do have one odd issue that is nagging me. Occasionally, the meter will output a perfect 0.0000000 reading.  It is annoying and I don't think it is a good sample. I will add code to ignore it unless you believe there is reason to keep it. I have attached the file. I will try this on another 3458A to see if this may be part of the controller or software.

 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #79 on: April 19, 2015, 03:33:38 pm »
I was thinking to get fixed timeset for each set. I have three 3 x 2001 gathering data + 2002 now since evening. Can you try 30min?
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #80 on: April 19, 2015, 03:56:25 pm »
I will make the changes and give it a test.
 

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2252
  • Country: ca
Re: DMM Noise comparison testing project
« Reply #81 on: April 19, 2015, 11:59:45 pm »
        0.01 NPLC0.1 NPLC1 NPLC10 NPLC50 NPLC100 NPLC1000 NPLC

TiN,

I have been busy this past week but I believe I have the 3458A working with all 35 data sets in an automatic acquisition. I hope to modify it with little changes on the newer SCPI meters. I can create all 35 files with one run.

I just need clarification on the data set sizes. I have the script written to acquire based on NPLC size. Would you break down a list of samples / NPLC? Dr Frank mentioned up to 10K samples on the .01NPLC setting and decreasing from there. I just have to plug in the numbers and run the test.

I do have one odd issue that is nagging me. Occasionally, the meter will output a perfect 0.0000000 reading.  It is annoying and I don't think it is a good sample. I will add code to ignore it unless you believe there is reason to keep it. I have attached the file. I will try this on another 3458A to see if this may be part of the controller or software.
Why would you think that a 0.00000000 is not a valid sample? It is just as valid as any other!
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #82 on: April 20, 2015, 12:54:56 am »
I got these sometimes too on Keithley's, don't see how that should be a problem.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #83 on: April 20, 2015, 02:40:50 am »
I will leave the data as is and send it in once it is collected. If there is an issue with the samples it should show up in the plots.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #84 on: April 20, 2015, 04:13:01 am »
Here's are view with ppm/range representation.
Combined data from 4 meters, 2V range, 10NPLC.

There are offset correction (align tip to 0.0 xAxis) for all samples to eliminate constant offset from data.

PPM calculation:

Code: [Select]
function parserk(d) {
    d.vmeas = +((d.vk15 / 2) * 1e6)-0.395;
    d.v2002 = +((d.vk2002 / 2) * 1e6)+1.42;
    d.vk13 = +((d.vk13 / 2) * 1e6)+0.74;
    d.vk12 = +((d.vk12 / 2) * 1e6)+0.32;
    return d;
}

This gives histogram like on attached quad_hist2.png screenshot.

It could be easier to split histograms horisontally, so they don't overlap, by manual offsets via chart engine
Then it would look like quad_hist.png

X-scale need to be split by 4 as well, but need to think how to align them with graphs.
Also it could be confusing if ppm-scale would not match between all meters (let's say +/-5ppm vs +/-2ppm), as they will visually look same, if someone not carefully look on scales.

Why I want combine multiple data sets on same graph? Well, this will allow to have multiple ranges and/or multiple NPLC settings shown by one graph on same meter. I don't think comparing 32 graphs for one meter with 32 graphs for another meter would be great idea..

Any suggestions?

P.S. Graphing and processing is done by D3.js script, which is Javascript. I don't know javascript language, just fiddling with values based on google'd examples/charts.
« Last Edit: April 20, 2015, 04:16:05 am by TiN »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline 6thimage

  • Regular Contributor
  • *
  • Posts: 181
  • Country: gb
Re: DMM Noise comparison testing project
« Reply #85 on: April 20, 2015, 07:18:15 pm »
Why I want combine multiple data sets on same graph? Well, this will allow to have multiple ranges and/or multiple NPLC settings shown by one graph on same meter. I don't think comparing 32 graphs for one meter with 32 graphs for another meter would be great idea..

Any suggestions?

Combining multiple sets of data on a graph is the only useful way of doing comparisons - but the overlapping histograms are difficult to read. The only thing I can think of is drawing the different data as a line graph - essentially a line running from the top of each histogram bin. That would probably make it a lot easier to read, but I'm not sure how you would do that with D3.js (I looked at it once, then went with highcharts because it was a lot easier for what I was doing).

With regard to performing the measurements - I'm a bit against sticking some copper wire between the inputs for two reasons. The first is that I don't have any decent wire that is thick enough to do it properly and the second reason is repeatability. I was thinking about getting one of the Fluke shorts (884x-short), but they seem a little expensive (especially as I would like 2 of them).

So I've made a little board (in eagle) that is similar and I wondered what were people's thoughts on it. I'm intending to use Multi-Contact 22.1053 plugs (gold plated copper-zinc plugs that have an M4 screw on them) with aluminium locking nuts (as I can't find any gold plated nuts). I don't think the nuts will cause any problems as the plugs are only making a contact with the front of the board.
 

Offline OldNeurons

  • Regular Contributor
  • *
  • Posts: 78
  • Country: fr
Re: DMM Noise comparison testing project
« Reply #86 on: April 20, 2015, 09:01:21 pm »

Why I want combine multiple data sets on same graph? Well, this will allow to have multiple ranges and/or multiple NPLC settings shown by one graph on same meter. I don't think comparing 32 graphs for one meter with 32 graphs for another meter would be great idea..

Any suggestions?


Yes, there will be too many combinations. You should limit to 2 or 3 data sets per graph, and have 2 to 4 graphs.
You need then, for each graph, a quick and convenient way of selecting what data set you want to display.

That D3.js that I discovered reading this topic looks pretty nice, but I have very limited programming skills, and will not be of great help.

Anyway, for those interested, in the attachement below is an Excel worksheet where I am using the "Offset" fonction ( https://support.microsoft.com/en-us/kb/324991 ) to select in a graph a data set to be displayed. The number of data sets, and number of graphs can be extended.

This is a  very simple way of doing the job. No code, no bug ...
I have been using it numerous times in my job on huge data collections for brainstorming sessions !

Jean
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #87 on: April 20, 2015, 11:52:02 pm »
For shorting PCBs I have few boards I could send out. No banana plugs, only PCB, I did not find copper banana plugs.
Photo attached.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #88 on: April 21, 2015, 12:06:28 am »
I purchased a few of these last year for replacement plugs on one of my Keithley 181 cables

http://www.douglasconnection.com/Furez-TSTWP30NP-Bare-Copper-Banana-Plug-Connectors-Pair-FZTSTWP30NP.htm?categoryId=-1

They are good quality but the red and black covers are made of Aluminum and I used some heat shrink for isolation. I do not know what thread pitch it is and I think it is too fine for my die set.

They do sell these ...
http://www.douglasconnection.com/Furez-TSTW25NP-Bare-Copper-Banana-Plug-Connectors-Pair-FZTSTW25NP.htm?categoryId=-1

You should be able to trim excess copper from the end and solder it in place. I think someone else is selling a similar product in France with the te-b38np part number.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #89 on: April 21, 2015, 02:44:24 am »
I have attached a script that works with the Keithley 2001 and EZGPIB. It will acquire all 20 different Range/NPLC combinations and has been tested with both the Prologix USB and Ethernet adapters.
The original script I posted was rewritten and and is based on the 3458A script. The output was formatted similar to TiN's sample data.
 
I will post the 3458A script tomorrow when I finish the first run and verify the results.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #90 on: April 21, 2015, 05:22:37 pm »
Attached are the scripts for the HP 3458A and Fluke 8846A.

Unless there is some show stopper with these scripts, I will continue to make more of them. I think any issues will be easy to fix and will be specific to individual scripts.

The 3458A ran last night and it went well. The only noted issue is that on the 0.1 and 0.01 NPLC settings, there was a discrepancy with the amount of data compared to different voltage ranges. The 100mV range had x4 the samples as the 1V & 10V ranges. My PC was possibly downloading or installing something even though I disabled Windows Update. A dedicated network connection would be the best scenario.

I ran the two NPLCs again this morning and so far the quantity of data is closer but there is still room for improvement. I will look into separating my test equipment from my home network.
I will upload the data to xDevs for TiN to look at this evening.

The 8846A is based on the 2001 script and has fewer settings to control. The analog and digital filters are off and the NPLC settings have some different allowed values. A quick run was performed and all range/nplc combos were acquired. The connection is through the GPIB port as it was the quickest for me to use and test.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #91 on: April 21, 2015, 06:25:33 pm »
Great, I never used EZGPIB before, but gave it a try, work a threat.
I use NI USB-GPIB-HS, not Prologix. EZGPIB detected it thru NI-VISA, and worked just fine.
I will leave it running overnight and report back.

What is meaning for sequence?

Code: [Select]
Sequence[0]  := '32'; Sequence[1]  := '22'; Sequence[2]  := '31'; Sequence[3]  := '30'; Sequence[4]  := '--'; Sequence[5]  := '--'; Sequence[6]  := '--';
Sequence[7]  := '--'; Sequence[8]  := '--'; Sequence[9]  := '--'; Sequence[10] := '--'; Sequence[11] := '33'; Sequence[12] := '34'; Sequence[13] := '20';
Sequence[14] := '21'; Sequence[15] := '23'; Sequence[16] := '24'; Sequence[17] := '--'; Sequence[18] := '--'; Sequence[19] := '--'; Sequence[20] := '--';
Sequence[21] := '--'; Sequence[22] := '--'; Sequence[23] := '--'; Sequence[24] := '--'; Sequence[25] := '10'; Sequence[26] := '11'; Sequence[27] := '12';
Sequence[28] := '13'; Sequence[29] := '14'; Sequence[30] := '00'; Sequence[31] := '01'; Sequence[32] := '02'; Sequence[33] := '03'; Sequence[34] := '04';

NPLC+Range?
With little help I hope I would be able to modify your .488 script to run on Agilent 34970A, Keithley 2400 and 2002 (this one can run same code as 2001, just have extra NPLC up to 50.
Also I noticed in your code ranges are 1V, 10V etc, but Keithley have 2V,20V etc ranges. Just a minor thing, as it seem to work fine even with that.
« Last Edit: April 21, 2015, 06:29:07 pm by TiN »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #92 on: April 21, 2015, 10:19:08 pm »
TiN,

You are correct on the NPLC + Range. I set it up based on the order you wanted the acquisitions. Some of the code looks odd because I was relearning Pascal as I was writing. I prefer C/C++ but that is not an option, at least not for now.

I started with the 3458A since it had the largest range+nplc combinations. From there, I can subtract unused combinations with the -- . You can also change the values to whatever you want like the 8846A that uses .02 and .2 NPLC. I left the ranges with 1's instead of 2's because they worked, but I should have followed the user manual. I changed them for the 8846A.

Your next three meters are also on my to-do list. Let me know which ones you are writing so that I can work on the others. The 34970A should be close to the 34401A but I have not seen the programming guide on either. That should cover a large number of people with the older meters. I will probably follow those with the 34420A.
 

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 730
  • Country: us
Re: DMM Noise comparison testing project
« Reply #93 on: April 21, 2015, 10:28:22 pm »
The data from the first 3458A has been uploaded to your ftp server. The file was too large to attach. I should be acquiring another set from one with opt 002. The last one still needs repair on the GPIB interface.
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2377
  • Country: de
Re: DMM Noise comparison testing project
« Reply #94 on: April 22, 2015, 10:56:10 am »
Hi,

I also made zero-input noise measurements on the 3458A, full speed with AZ OFF, at NPLC 0.1 and NPLC 1, i.e. 500 and 50 samples/s and NPLC 10, NPLC 100 with AZ ON, each for 1V and 10V so far.

I only collected as many data  as needed to get a stable statistics, i.e. 10000 and 1000 samples for NPLC 0.1 and NPLC 1 (20 sec each), and 100 samples for NPLC 10 and 100 (40 and 400 sec, respectively).

Then I made the usual xls population statistics, deriving the average (= DC Offset) and the standard deviation (= RMS Noise w/o DC offset).

The 10V RMS noise fits quite exactly the HP RMS noise specification, 1V is about 3x worse (instead of specified 2x)

I will measure 100mV also, and will try to program for full speed NPLC 0.01, which requires a different output format and special fast data acquisition.


TiN, besides the distribution diagrams, does your script also calculate RMS noise?

Then I would upload the data files to your server.

Frank
 

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2252
  • Country: ca
Re: DMM Noise comparison testing project
« Reply #95 on: April 22, 2015, 01:29:11 pm »
I have attached a script that works with the Keithley 2001 and EZGPIB. It will acquire all 20 different Range/NPLC combinations and has been tested with both the Prologix USB and Ethernet adapters.
The original script I posted was rewritten and and is based on the 3458A script. The output was formatted similar to TiN's sample data.
 
I will post the 3458A script tomorrow when I finish the first run and verify the results.
I had meant to reply earlier.
I played a little with the earlier version of the script, and it seems fine except for one important change that I would make: Use the FETCH? or DATA? command instead of READ?. The READ command does a full reset/init of the meter before each reading (notice the "---------" display between readings). This is unnecessary, time consuming, and perhaps may lead to less stability in the readings. The DATA? and FETCH? commands will return readings without affecting the instrument in any other way. The DATA? command always immediately returns the most recent reading, and will return the same data more than once if you request data again before the next reading is ready. The FETCH? command differs in that it will return only a fresh reading; if you FETCH again before the next measurement is ready, it will block until one is ready.

So, if you only want to poll infrequently and don't care about missing a few measurements in between, then you can use either DATA or FETCH. If you want every single reading (and why not?), then you can still use either but in different ways. With DATA, you can discard repeats by looking at the reading number (you might need to configure the instrument to include that in the readings). The program will need to deal with the duplicates, but you will never block/hog the IEEE bus. With FETCH, the program is a little simpler, since the readback will only complete when fresh data is available, but it will block the bus until then. This is only important if you have multiple instances of EZGPIB running on multiple meters for example. If the bus is dedicated to this task then it doesn't matter as long as the timeout is set long enough to account for the delay imposed by the meter.
 
The following users thanked this post: smoothVTer

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #96 on: April 22, 2015, 02:22:50 pm »
Dr. Frank
Graphing script does not calculate RMS noise, as well as SD, mean, min, max values and sample window math, but that would be not hard to add after we get all display data format and representation finalized.

macboy

I used READ? as well before. And simply changing READ to FETCH or DATA does not work well, with DATA it does not actually trigger measurement, but just takes reading, while FETCH gives bunch of errors when trying to run on two meters, blocking both.

P.S. Also funny note, running EZGPIB makes my one of my 2001's emit high-pitched buzz. I hear electrons moving in meter's brain, as buzz is different when sampling at different NPLCs :) Did not had it before with LabView.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2252
  • Country: ca
Re: DMM Noise comparison testing project
« Reply #97 on: April 22, 2015, 03:20:30 pm »
Dr. Frank
Graphing script does not calculate RMS noise, as well as SD, mean, min, max values and sample window math, but that would be not hard to add after we get all display data format and representation finalized.

macboy

I used READ? as well before. And simply changing READ to FETCH or DATA does not work well, with DATA it does not actually trigger measurement, but just takes reading, while FETCH gives bunch of errors when trying to run on two meters, blocking both.

P.S. Also funny note, running EZGPIB makes my one of my 2001's emit high-pitched buzz. I hear electrons moving in meter's brain, as buzz is different when sampling at different NPLCs :) Did not had it before with LabView.
The script should issue one "READ" to initiate measurements, then go into the big loop doing "DATA?" repeatedly.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: DMM Noise comparison testing project
« Reply #98 on: April 22, 2015, 05:53:48 pm »
Thanks, will try tomorrow.

Here some data normalized to range.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2377
  • Country: de
Re: DMM Noise comparison testing project
« Reply #99 on: April 22, 2015, 06:51:48 pm »
Not so good, sorry!

Legend is missing (NPLC is the difference????)

And different X-scaling.

Please improve!

Frank
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf