EEVblog Electronics Community Forum
EEVblog => EEVblog Specific => Topic started by: EEVblog on December 27, 2015, 03:53:07 am
-
An impromptu play with Dave's Wekomm 10K Transfer Resistance Standard.
What is the strange discrepancy in drift on the graph on the Keithley DMM7510 multimeter?
http://www.wekomm.de/metrologie_e.php (http://www.wekomm.de/metrologie_e.php)
This is Part 1
https://www.youtube.com/watch?v=dIXbjNb82xk (https://www.youtube.com/watch?v=dIXbjNb82xk)
This is Part 2
Part 2 of an impromptu play with Dave's Wekomm 10K Transfer Resistance Standard.
Was the Keithley DMM7510 multimeter the culprit in the discrepancy in the thermal response?
Dave finds out with the help of a thermal chamber.
https://www.youtube.com/watch?v=gEDs0OtOkGE (https://www.youtube.com/watch?v=gEDs0OtOkGE)
-
I find this stuf fascinating also and some times go way down the rabbit whole with it.
I do wonder a few items.
What level is your lab on in the building? Just wondering about shops below you that keep the air/heat on higher that would force air up from other levels into your lab?
For the LCD of the meter, does it turn on and off?
I think this meter has a fan, could the speed of the fan change to compensate for the ambient air change as the fan tries to regulate the temp of the unit?
I could go on, their are so many factors to consider :)
I recently built a voltage reference and saw a difference of 0.00000 to 0.00010 and the bread board, turns out that was the ambient air temp change alone so I logged my lab temp and the data showed a 1.2C change despite the outside temp change of 13.9C. So not bad on lab temp.
Now I'm working on a cheap way to heat the refencse chip at 40C to keep I stable and still run long on battery.
Nice videos, gave me some other ideas. But my bench meter is a old HP 34401A calibrated 2 months ago so no internal fan to worry about I guess. This could be good and bad, presumably more good as long as I allow the 2 hour warm up first.
-
One thing I think people often forget:
- It doesn't have to be perfect, but does have to be good enough to get the job done.
-
Yey, Dave using 7510, finally :)
Those banana wires could be causing some EMFs as well.
-
Looking at the noise some screened leads would not go amiss, and connecting the guard to the ground of the precision resistor would eliminate a lot of the noise from the thermal chamber maintaining its set point.
-
Guard? There is no external guard port on 7510 AFAIK.
-
Guard? There is no external guard port on 7510 AFAIK.
Should have a grounding point, it is a precision unit and shielding does help a lot more than averaging.
-
Guard? There is no external guard port on 7510 AFAIK.
No it doesn't.
-
Thank you for the fine experiment. It shows, how hard live is in the area of high precision and why the effort grows exponentially.
The "big" periodic changes during operation of the thermal chamber could be produced by the waste heat of the peltier cooler more than by the temp changes within the chamber itself.
The temp of the chamber was set to a temp below the room temp. So it has to cool a bit. And by cooling the poor efficiency of the peltier leads to a temp change of the heat exchanger on the back of the chamber.
And the Keithley is pretty close to the back of the chamber and potentially sucks in the waste heat of the thermal chamber.
So, we could have two "amplifications" to the disadvantage of the Keithley readings:
- The low efficienciency of the Peltier cooler amplifies the heat changes on the back of the thermal chamber
- The specified thermal coefficient of the keithley is bigger than the coefficient of the precision resistor. The Keithley potentially sucks in a part of the waste heat produced by the chamber and converts it by its bigger thermal coefficient to a bigger change in reading.
To improve this it could be helpful to set the temp in a chamber to a higer temp than ambient, so that the peltier has not to cool down so much. But maybe that the controller switches on the peltier anyway.
Socond improvement could bring a different local arrangement so that the waste heat of the peltier is not sucked in by the Keithley
-
It does seem the temperature chamber is not that stable for measurements at this range, even when Dave was showing the current temperature inside it was changing quite a bit and he did mention that it varies.
@11:48 he states the temperature chamber was at about 20C +-0.5C while he was monitoring it.
-
Yes, and if the temp inside changes a little bit, the temp at the backside of the peltier changes even more due to the low efficiency of the peltier.
-
I like the small thermal chamber, I wonder where these could be bought?
-
LOL not here. "All need at least 13 3/4 digits multimeter" :scared:
Quote from: pickle9000 on Today at 03:47:54 PM (https://www.eevblog.com/forum/index.php?topic=60494.msg830113#msg830113)One thing I think people often forget:
- It doesn't have to be perfect, but does have to be good enough to get the job done.
-
I like the small thermal chamber, I wonder where these could be bought?
I think he modified a peltier based mini fridge like this one:
http://www.ebay.com/itm/Thermoelectric-Cooler-and-Warmer-Old-Fashioned-look-/221971503797 (http://www.ebay.com/itm/Thermoelectric-Cooler-and-Warmer-Old-Fashioned-look-/221971503797)
He did make a video, if I find it I'll link it here.
Edit: it was a thermal incubator (HERP Nursery II), related video:
https://www.youtube.com/watch?v=AI0Q6-h3EGU (https://www.youtube.com/watch?v=AI0Q6-h3EGU)
-
I like the small thermal chamber, I wonder where these could be bought?
I think he modified a peltier based mini fridge like this one:
http://www.ebay.com/itm/Thermoelectric-Cooler-and-Warmer-Old-Fashioned-look-/221971503797 (http://www.ebay.com/itm/Thermoelectric-Cooler-and-Warmer-Old-Fashioned-look-/221971503797)
He did make a video, if I find it I'll link it here.
Edit: it was a thermal incubator (HERP Nursery II), related video:
Thanks, definitely gonna look in to that :-+
-
I like the small thermal chamber, I wonder where these could be bought?
Designed for Reptile eggs:
http://www.petsathome.com/shop/en/pets/lucky-reptile-herp-nursery-ii-incubator (http://www.petsathome.com/shop/en/pets/lucky-reptile-herp-nursery-ii-incubator)
-
I bought an 'identical' ReptiPro 6000 a couple of years ago. I replaced its on-off-on temperature control with a PWM controller with a tiny thermistor placed in front of the internal recirculating fan intake. It now modulates the temperature to within +-0.05 C.
The PWM temperature controller I bought does require me to swap the output leads to the Peltier thermometric module. but it still cost 2x what the incubator/test chamber cost.
Controller = TE Technology Model TC-48-20. http://tetech.com/product/tc-48-20/ (http://tetech.com/product/tc-48-20/)
-
Hi Dave,
very nice experiment and demonstration of the sensitivity of analog instruments in this ppm region..
Some remarks, though.
These measurements very obviously are extremely sensitive to any stray noise. Therefore, all disturbances you see, probably come from switch mode power supplies or from 50Hz mains.
The 400sec oscillations may be a beat frequency, if the temperature chamber strongly induces 50Hz mains, and if the 7510 does not measure (suppress) the line frequency exactly.
It may also well be an Alias frequency of the 8 sec sampling rate versus the 50Hz noise.
The 7510 (and other Keithley instruments) probably offer too few averaging possibilities. NPLC 100 (n/a for the 7510), with additional statistics averaging would be better.
Speculations about EMFs: The 7510A also has the Offset Compensation feature, which removes any EMFs.
It was not evident in the video, if this feature was switched on.
The cables are from multi contact, I assume? Gold plated contacts?
So the EMFs are probably very low, at least these are constant.
But there's another problem with these cables.
Their isolation is probably PVC, which resistance is too low, on the order of 1E10 .. 1E11 Ohm only.
As you have twisted the red and the black one, that already might cause errors of several ppm.
Better twist both positive and both negative ones and don't let black and red touch each other.
Or better use Teflon cables.
Frank
-
I spent the whole video wanting to peel off that screen protector... drives me nuts!
-
In the #835, you suggested to hook up a battery to the terminals to measure tempco for the meter.
So I'm just letting you know, that batteries have a voltage change for temperature change. The lithium cells I've tested had a few PPM/K, a magnitude higher than the 3458a used, but I've seen the change for FiFePO4 and lead acid. Maybe NiCd has less than noticeable change, but at 7-8 digit, the battery is not a perfect voltage source.
You should also take a battery around 50% charged, and let it settle for some time. And use good big cells, otherwise you will measure the 10 megaohm discharging the batteries and self discharge. Maybe for 6.5 digit.
Ultimately, I think batteries are OK for a few second transfer standard, but not suitable otherwise at this precision.
-
In the #835, you suggested to hook up a battery to the terminals to measure tempco for the meter.
Yes, I realised that was dumb after I uploaded it.
I've done a video on this sort of stuff:
https://www.youtube.com/watch?v=fJSNAyMmNGk (https://www.youtube.com/watch?v=fJSNAyMmNGk)
-
I spent the whole video wanting to peel off that screen protector... drives me nuts!
That's the idea >:D
-
How is you air-con physically delivered?
Maybe the "undershoot" when you start it up is the duct work warm air getting flushed out.
-
IIRC Dave has a split unit in his office, with a separate outdoor unit. No multi split, no central unit and ducting. The fan in those has a very short run from the evaporator to the room, so not likely. The undershoot is probably a temperature differential or a cable thing, or a mains induced artefact with the drop in supply voltage with the unit turning on.
Time for Dave to look out for one of those older ferroresonant power conditioners, which provide both short term dip ride through and very good spike and noise filtering, in addition to the voltage regulation. The higher third harmonic and the increased line reactance are a bonus, helping to reduce the current spikes through the SMPS input diodes for those units without active PFC, and even for active PFC they will reduce radiated noise down the line.
-
Very nice video, Dave!
Great to see the Keithley 7 1/2 digit meter in action.
May be part 3 could be the same setup with the Keysight 34470A 7 1/2 digit meter in comparison.
Interesting also, that the Keithley meter 7510 does not have a guard, like the 2450 SMU.
-
Thanks for the video, I enjoyed it.
-
Interesting videos.
For these high accuracy voltage measurements, I understand that a solid copper conductor may be best, I note DrFranks advice re Teflon/PTFE insulation, some shielding also advised.
Some cables come with silver coating, I don't know what that would do at DC (i.e. not RF) and it may introduce some thermal EMF I wonder?
Most banana plugs are coated brass, good ones seem to be gold, ? true or not, is soldering better or screw clamped? I suspect there may be pros and cons for each way but I suspect this road has been trodden before!
-
The difference between copper gold and silver plated are not big
Junction µV/°C
Copper-Copper <0.3
Copper-Gold 0.5
Copper-Silver 0.5
What I learned from one member here is the fact that PFTE insulation is not compatible with bare copper, that why you can find on market only silver plated PTFE wires.
Surprisingly good is TiN patented solution :)
https://dev.xdevs.com/projects/kx/repository/entry/img/vref_a01/ltz1000_initial.jpg (https://dev.xdevs.com/projects/kx/repository/entry/img/vref_a01/ltz1000_initial.jpg)
-
@plesa, thanks for some real data. I should have looked myself, a quick search revealed this paper, interesting copper-copper oxide >1000uV/C ! A bit of silver plating might be useful, also this paper suggests a low thermal mass to allow more rapid thermal equilibration.
http://www.google.com.au/url?sa=t&rct=j&q=&esrc=s&source=web&cd=3&ved=0ahUKEwjVg7Py8YXKAhUDJKYKHXlND3sQFggoMAI&url=http%3A%2F%2Fwww.minerva.pro%2Fdownloads%2Fproduct%2F172%2Fpapers_articles%2FWatch_Out_for_Those_Thermoelectric_Voltages_Cal_Lab_Journal_reprint_182_MB.pdf&usg=AFQjCNHCMy9LQOmXT8lo4T8OEg3SyEAJ0w&sig2=x7LtiRau4irXT9RrkBXGqg (http://www.google.com.au/url?sa=t&rct=j&q=&esrc=s&source=web&cd=3&ved=0ahUKEwjVg7Py8YXKAhUDJKYKHXlND3sQFggoMAI&url=http%3A%2F%2Fwww.minerva.pro%2Fdownloads%2Fproduct%2F172%2Fpapers_articles%2FWatch_Out_for_Those_Thermoelectric_Voltages_Cal_Lab_Journal_reprint_182_MB.pdf&usg=AFQjCNHCMy9LQOmXT8lo4T8OEg3SyEAJ0w&sig2=x7LtiRau4irXT9RrkBXGqg)
-
I bought an 'identical' ReptiPro 6000 a couple of years ago. I replaced its on-off-on temperature control with a PWM controller with a tiny thermistor placed in front of the internal recirculating fan intake. It now modulates the temperature to within +-0.05 C.
The PWM temperature controller I bought does require me to swap the output leads to the Peltier thermometric module. but it still cost 2x what the incubator/test chamber cost.
Controller = TE Technology Model TC-48-20. http://tetech.com/product/tc-48-20/ (http://tetech.com/product/tc-48-20/)
Interesting setup you have there
Can you post us a picture?
-
I bought an 'identical' ReptiPro 6000 a couple of years ago. I replaced its on-off-on temperature control with a PWM controller with a tiny thermistor placed in front of the internal recirculating fan intake. It now modulates the temperature to within +-0.05 C.
The PWM temperature controller I bought does require me to swap the output leads to the Peltier thermometric module. but it still cost 2x what the incubator/test chamber cost.
Controller = TE Technology Model TC-48-20. http://tetech.com/product/tc-48-20/ (http://tetech.com/product/tc-48-20/)
Interesting setup you have there
Can you post us a picture?
Another way can be using LabView and bipolar power supply and some DMM as temperature controller.
Modification of attached .vi is piece of cake.
http://www.tek.com/sites/tek.com/files/media/document/resources/BuildingBenchtopPID_WPaper.pdf (http://www.tek.com/sites/tek.com/files/media/document/resources/BuildingBenchtopPID_WPaper.pdf)
-
I can't seem to find a price on the fancy Wekomm standard, any rough guesses?
-
I can't seem to find a price on the fancy Wekomm standard, any rough guesses?
3750 EUR
http://www.datatec.de/wekomm.html (http://www.datatec.de/wekomm.html)
-
Another way can be using LabView and bipolar power supply and some DMM as temperature controller.
Modification of attached .vi is piece of cake.
http://www.tek.com/sites/tek.com/files/media/document/resources/BuildingBenchtopPID_WPaper.pdf (http://www.tek.com/sites/tek.com/files/media/document/resources/BuildingBenchtopPID_WPaper.pdf)
That is a great paper.
Thanks.
-
3750 EUR
:o
Guess I won't be buying one of those to play with.
-
3750 EUR
:o
Guess I won't be buying one of those to play with.
Have you got $4? http://www.ebay.com/itm/121031115671 (http://www.ebay.com/itm/121031115671)
-
if they plot the wekomm TC curve like in https://www.eevblog.com/forum/projects/t-c-measurements-on-precision-resistors/ (https://www.eevblog.com/forum/projects/t-c-measurements-on-precision-resistors/)
it is suppose to be the "good" well behaved banana curve?
-
Dave,
Looking at your test setup, you might not be looking at all meter drift...Some other areas to look at when you're making measurements in the ppm area - just based on experience and some "gotchas" that we have discovered the hard way sometimes:
1. As others have pointed out, you've got un-shielded cables from meter to ref resistor. I would use a good quality shielded twisted wire paired assembly (as Dr Frank pointed out), and get a meter with a Guard or Shield connection - and make use of it. I use 4ea RG cables with guard amp driven from highest potential driver on your 4-wire setup, or just connect as basic grounded shields. If the DMM doesn't have a Guard point, get one that does - especially if you're going to be doing a lot of PPM-resolution measures.
2. You want another rag or iso thermal shield on those front meter connections. It's easy to get drift of some ppm just on those cheap banana plugs out there flapping in the breeze. Clean Tellurium copper spade connections where you can tighten a nut down to a known torque work well when you're chasing down the last half ppm drift, otherwise gold test clips are good. Good crimped / welded, gas-tight connections are much better than soldered joints at this point. Any crud or tarnish on the connections is a bad thing, and a clean silvered joint is better than a dirty gold or copper joint any day.
3. Turn off all switching power supplies in the vicinity. I've seen many a precision resistor - especially some non-bifilar wirewounds - easily pick up local switching noise. That includes unshielded flouro / LED lamp drivers. We use politically incorrect, electrically quiet incandescent lamps everywhere in the test lab here for illumination for that very reason. Plain old sunlight thru an IR-filtered glass window works too, if available. As long as it doesn't cause bad thermal effects.
4. If this were a '3458a meter, you'd want to run Auto Cal if your lab temp changes more than 1°C, or at least once every 24hrs when you're trying to play in the PPM accuracy pool.
5. You also want to flip your leads to the ref resistor periodically. It is unknown what thermal and diode effects are inside that box, especially when flying leads are present.. We even flip the ref resistor around physically occasionally to check for gravity-force stress effects at different orientations. You'd be surprised at what shows up sometimes - especially on cheap resistors.
6. Keep an eye on lab humidity and baro pressure also. Sometimes we've seen a drift pop up that is only seen during a local heavy storm or exceptionally sunny / rainy weather.
-
Hopefully this is not too late to post.
When I use to calibrate decade boxes many moons ago in 1989, I had to place the standard resistor in a mineral oil bath. This will keep it at a more stable temp in the controlled environment of the lab from the changes of the air around it. Also, using shielded leads and getting rid of power line noise is another factor to consider. Finally, keeping your body and other noisy equipment away from the test will also help with errors. Anyway thats my two cents worth. :-+
-
Re the shielded leads and all the other stuff to get better precision, yes, as explained in the video this was not meant to be a proper controlled test setup, I was just mucking around with no intention of doing a video showcasing the standard, but it turned into one ;D
-
This is my second post on this forum and I'm not an EE but I enjoyed the videos and have a couple of questions.
1) Couldn't a sealed container of water, perhaps a couple of litres, be placed in the thermal chamber to reduce the control authority of the peltier module and smooth the thermal transitions?
2) Would a meter of that caliber have an isolated and thermally controlled analog front end inside the instrument?
-
Curiosity driven, I toredown the same Vishay resistor(VHA518-7) five years ago.
Because it is oil filled, it was very hard to de-solder both for outer ring and for inner Kovar tube, I had to tore the case apart.
As the name suggested, there are seven identical foil elements in the container and they are arranged in parallel probably owning to the fact that medium value around 1k has the best property.Therefore, I bet that that 10k resistor inside Wekomm standard consist of seven 1.42857k elements in series.
-
zlymex, welcome to eevblog! :-+
Every single post you have made so far is 10/10, top drawer!
You purport to be from China, are you active on bbs.38hot.net at all?
-
I thought they put them in parallel to 'RMS average' out the errors ????????
-
zlymex, welcome to eevblog! :-+
Every single post you have made so far is 10/10, top drawer!
You purport to be from China, are you active on bbs.38hot.net at all?
Thanks very much Macbeth. Yes I'm from China, and I'm one of the oldest members of 38hot.net
-
I thought they put them in parallel to 'RMS average' out the errors ????????
Well, in a sense yes. When 7 elements connected together in series or in parallel or mixed, any single element has an error say 100ppm, the combined value will change by only 1/7, that is 14ppm, provided that all 7 element are equally weighted. On the other hand, foil resistors are easy to adjust when manufacture, even a single element can be trimmed to within 0.005% of nominal value.
When more resistors are parallel or series connected, they have more thermal mass and heat dissipation ability thus rated at more power than single element, this is an important factor for standard resistor. Employ foil technology into standard resistors is a relative new thing, there are people(including myself) suspect the performance compare to traditional WW mainly because of its small effective conduction mass. When I DIY standard resistors, I often put a lot of elements together to increase the thermal mass and the conduction mass. And yes, when more elements put together, they make adjustment of deviation easier.