[FT] xDevs.com KX LTZ1000A fairy-tale, or the story of little jumper.

[ap]

Unless you have the 3458A running all the time (and even then) the 3458A is not good enough to verify a LTZ1000. I have seen that many times. That does not contradict 3458A spec, but even if you switch a 3458A of and on shortly thereafter, you see some hysteresis (at least if you have not reduced the LTZ internal temp.) There is no way arround a precise voltage reference...

[Dr. Frank]

Unless you have the 3458A running all the time (and even then) the 3458A is not good enough to verify a LTZ1000. I have seen that many times. That does not contradict 3458A spec, but even if you switch a 3458A of and on shortly thereafter, you see some hysteresis (at least if you have not reduced the LTZ internal temp.) There is no way arround a precise voltage reference...

That's not correct, the way, how you express it..

Here, you compare two equally stable reference against each other.. one thing are the short /mid-term fluctuations, the relevant is the long-term timely drift.
You can't decide with only 2 artefacts, which one drifts...

"A" precise voltage reference (or better : "one" ..reference) is also not sufficient, even if you use a 732B, or similar, which is also equally stable than a LTZ1000 based reference.

Only if you have 3 or more (equally stable) references, you may be able to measure the drift of one of its constituents.

If you have even more references, and more of old, stabilized ones, the higher the probability is to discriminate drifty constituents of the whole group.

The LTZ1000, and also the LTFLU inside the 732Bs in the median tend to drift about -0.8ppm/year, whereas the SZA263 in side the 732A and elder 732B, and also inside the 5440B tend to drift upwards, about +1ppm/yr.
A mix of both types of references would allow better judgment.

Then it's of course necessary to have regular re-calibration of that group.

Frank

[ap]

I dont think I said, and certainly did not mean it differently. Its in the end a question of measurement uncertainty of the gear you use, and its just too low in that case. Not the place going into multiple ref. type discussions here, anyone interested, e.g. Fluke explains it well in Calibration: Phil. in Practice. For Free: NASA cal handbook: Metrology — Calibration and Measurement Processes Guidelines.
I do disagree with your one reference statement though; depending on established history, and time between CALs and so, even DAkkS allows that.

[TiN]

Received today KX LTZ module from VK5RC. At least he was brave enough to take on my cross-check offer. Quick check show it's still working after a trip, so I'll add this guy for monitoring over next multiple weeks and run it thru temperature ramps. I think VK5RC has few weeks of data on it too, so it will be interesting to compare how our test results correlate. According to a label on the box, reference have Edwin's PWW resistors.

[VK5RC]

TiN; Thanks for the great offer! 
enut11 has been leading / encouraging a group of Australian V-nuts, one of whom has a recently purchased 7.5 digit Fluke. A set of mainly voltage references but also some resistors are doing the rounds in Australia.
Just before the KX LTZ reference headed off - I did a quick temperature experiment - my lab heats up moderately on hot days (36C outside) - the temp inside going from 20 to 26C (my 3458 runs about 15-16C above ambient in general), the trace starts at 11AM heats up till 1320 (about 27C ambient)  when I switched on a refrigerative air conditioner (AC) till 1500Hrs - dropping the temp to about 20C. Unfortunately I did not record differences coming from the temp sensor inside the KX LTZ but recall it was only a degree or so. I think most of the temperature change was inside the 3458A - it does have a fan after all!
I use a Prologix (USB_GPIB) and RFScientific software (so far) importing into Excel- I am a Linux newbie. I am admiring the efforts of Tin, plesa and others in RaspPi and various GPIB pursuits!

Below 1 Screenshot of Excel tables
         2 KX LTZ Unit prior to boxing -     Rationale;  I went with a vertical position of Mr Pettis' resistors to get their bodies close to each other thermally - I didn't use any copper tape (risk of short) - also the lead bending was a little more straight forward - the lead lengths are unequal though. The perspex lugs are only holding the board in place - it can move a mm or so but is touching some insulation foam inside its inner metal (diecast) box in an outer plastic box. The linear PSU for the 15V (TPS7A4901) is outside both of these.

Again thanks to TiN and other EEVblog volt-nuts. 
Robert

[TiN]

So that little bodge with 7805 is power supply for temp sensors?
I got home, and hooked ref to a group already, but need modify code a bit as your ref connected directly at 3458B.

Can you post your logs data as well?

That green Teflon cable feels nice, I would mind knowing where to get some.

Sent from my One using Tapatalk

[mimmus78]

Yes I also wanted some of this cable ...

Inviato dal mio Nexus 6P utilizzando Tapatalk

[TiN]

Here's the thing:



First samples:

...
Channel 3: VM1=7.134162383, VM2=7.134162418, AVG = 7.134162400, 0.014 ppm
...
Channel 3: VM1=7.134161875, VM2=7.134161560, AVG = 7.134161717, -0.081 ppm
...
Channel 3: VM1=7.134161595, VM2=7.134161507, AVG = 7.134161551, -0.105 ppm
...
Channel 3: VM1=7.134162050, VM2=7.134162085, AVG = 7.134162068, -0.032 ppm
...

Powered from K2400 (+11V, not 18V). Ambient is around 26°C.

[SeanB]

Green cable is a standard PTFE insulated avionics shielded data cable, used in aircraft generally for either data buses or to connect low voltage sensors like thermocouples to points in the airframe. Makes a nice shielded cable, especially with the right solder sleeve shrinks to get the shield wire connection out.

Also used as Audiophool grade interconnects, where they use the one core for signal and the other for ground, and only connect the shield to ground one end, or use it with XLR balanced connections.

[sergioag]

I have just finished my KX Board  It has been powered like 4 hours now and attached are the pictures of the current measurement on my Keysight 34416A, as well as the board itself.

I plan to have it powered up for about 1000 to 2000 hours so it settles down. Let's see how the voltage varies.

In case you're wondering about the parts:
- The Linear parts were bought from their web store.
- The VPG resistors were ordered from UHP Resistors, which was referred to by Vishay themselves.
- The rest of the parts were bought from Mouser.
- The PCB was made by Seeed Studio. Too bad they can't put silkscreen over bare copper. Green was chosen as it was the cheapest and I had to save a few bucks back then (something I regret - red or black would've looked much better).

I have soldered all the parts using my good old 30W soldering iron - no hot air station for me. The board is completely lead-free. I haven't soldered one of the Maxim ICs (the one on the front side) because I don't think I will be able to successfully solder it because of the surroundings.

All the measurements are being made by my Keysight 34461A, which is about 120 days out of calibration. Hopefully I'll be calibrating it next month. I don't have a more precise meter.

If anyone would like high resolution pictures of the board, please let me know and I'll bring the big camera for that 

[CalMachine]

I wouldn't mind seeing some more pictures!  I'm going to be making 2 boards myself here in the next few weeks. 

Also, something little I noticed..  Your DMM is in auto range.  You want that bad boy on manual range

[sergioag]

Sure, I'll take some good pictures tomorrow. 

[TiN]

Thumbs up for jumping in, we are always happy with newcomers.
Green KX, yikes  No problems with that, now we need to get someone to make blue and white ones. I have red and black, mimmus has purple ones, you has green now. 

Too bad they can't put silkscreen over bare copper.

Most of fabs can do that, but you need to ask specifically. Reason why they don't put silkscreen over bare copper by default - is if you made mistake and put silk over pad, you will not be able to solder that pad.
So they just being extra safe, nothing wrong with that.

no hot air station for me.

This board does not need, and does not want hot air... 

I haven't soldered one of the Maxim ICs (the one on the front side) because I don't think I will be able to successfully solder it because of the surroundings.

I usually solder that one first, and only after put surrounding taller parts. Common rule for PCBA is to put small parts first, then taller ones, and connectors last.

If anyone would like high resolution pictures of the board, please let me know and I'll bring the big camera for that

Sure, show us your gigapixels.

I'd be interested to see which VPG resistors you bought, what their cost was and how long you waited for them. I chose VHP202 70K ones only because there were no 70Kohm in Z202 small package.

Also 34461A can do much better logging, set to manual 10V range, and reset statistics, so you can see variation down to 1uV AFAIK. I don't have 3446xA, otherwise I'd give you exact setup


Good news for VK5RC reference, it's good and stable so far during overnight run, and here's initial chart (bold purple line):

[Andreas]

I wouldn't mind seeing some more pictures!  I'm going to be making 2 boards myself here in the next few weeks. 

Also, something little I noticed..  Your DMM is in auto range.  You want that bad boy on manual range

And I think the more critical is the 10 Meg input impedance instead of being high Ohmic.

with best regards

Andreas

[VK5RC]

Thanks TiN,

You are right in that the little 'bodge board' is a DC PSU for the temp sensor ICs. The reason for the floating - 9V battery is I have some serious earth - loop issues.  The temp sensor outputs are marked - one as REF (close to the LTZ) the other is the sensor close to the op amp.

Thanks for posting the data - I am happy !      I am now pretty sure I know (approx) how far out, my out of cal 3458A is!

I quite like the green cable but it is just a bit stiff - The conductors are quite thick -  I have purchased two lots on eBay from. http://www.ebay.com.au/itm/10-feet-20-AWG-Shielded-Silver-Plated-PTFE-Wire-Green-Twisted-Pair-w-Drain-/232174685320?hash=item360eb08c88:g:YkwAAMXQrhdTQbWh - I would buy from him again.

I will post my raw data later tonight but have to get it off another laptop.

Robert

[VK5RC]

Below are the data combined of two of the longer runs, the first is the run with big changes in temperature in the Lab, and the second is an overnight with relatively stable temperatures. They were actually about 2 weeks apart, The KX LTZ reference was kept mostly powered up in between.
Robert

[TiN]

Well, in that case you are +7.5ppm away from me.
Now we need 3rd person with <2ppm DC accuracy reference to jump in and save us  .

Anyone? Come out of the shadows...

[mimmus78]

Wish I could send mine ...  but mine it's not shippable. After few power on/off 3458a is back to reading kx003 (and others two LTZ1000) as first day (+/- 0.2ppm).

[sergioag]

Here are the pictures I promised. Hopefully you enjoy them

[VK5RC]

I have no doubt my 3458a  is incorrect, if I recall correctly last calibration was in Hong Kong about 7-10yrs ago (by Agilent no less) but another reference is always reassuring. The error of about 1ppm drift per year is not that unusual I believe.
Robert 

[TiN]

Here are the pictures I promised. Hopefully you enjoy them

Those are small pics  .

There are is alarming item though - R3 must be 12.5-15Kohm , not 10Kohm, unless you plan to run your LTZ in freezer all the time.
10Kohm is way too low for temperature setpoint for reference used in ambient.

Also it's better to clean flux residues to prevent gotcha's.

Your Z202 resistors look also very interesting as well, not usual marking font/typeface found on VPG resistors.

[sergioag]

Those are small pics  .

I wanted to upload something larger, but the forum will only accept up to 1000kb per file, for a total of 2000kb... and originals in RAW are ~50MB each 

There are is alarming item though - R3 must be 12.5-15Kohm , not 10Kohm, unless you plan to run your LTZ in freezer all the time.
10Kohm is way too low for temperature setpoint for reference used in ambient.

You mean R5, right? (R3 is 120 Ohm) In that case, I'll order both P/N Y107312K5000T and Y107315K0000T just to be sure. And maybe even Y107313K0000T (as seen in Linear's reference schematic). Taking into account that here the temperature is usually around 18 C on Winter and 25 C on Summer, would you make any recommendation?

Also it's better to clean flux residues to prevent gotcha's.

Yes. I still have work to do regarding cleaning. Even further, when taking the pictures I also found a flux bubble below the LTZ can. I tried to take a picture of it, but it didn't really show.

Your Z202 resistors look also very interesting as well, not usual marking font/typeface found on VPG resistors.

Would this be something to worry about? I bought them from UHP Resistors, which was referred to me by Vishay themselves. Perhaps a newer batch or something like that?

Thanks!

[branadic]

Sorry, but I can't resist. The solder joints are looking very very poor. Most pictures of hobby electronics injure my eyes everyday, so I need to make this statement.

As reliabability of a circuit is within the interconnections of the components in particular the solder joints you want them to be "as perfect as possible". Otherwise thermomechanical stress will lead to micro cracks within the solder joints and furthermore to failure.
Design and fabrication of reliabable circuits and sensors is my everyday job and assembling technology is the major key to meet this requirements. So my request is to improve your soldering skills to guarentee that this expensive components are working fine for a very long time. Even if you think about shipping this reference all over the place it should be your goal that everything works just as it should. Better use non-RoHS conform solder instead of gluing things together with RoHS conform solder, please. You have spent a big buck in those parts, so don't make any compromise in the end by a poor aseembling. Thanks.

branadic

[sergioag]

Yes, I know...  It's not quite good... I'm just transitioning to lead-free solder and thought it would be a good idea to try there - obviously it wasn't. Add that to the fact that I'm just using SMD components for about a year.

Anyway, I've already ordered the VPG resistors for 3 more sets. Lead time has been quoted as "about 17 weeks" so it'll be a while. In any case, those 3 will have regular leaded solder, the one I'm familiar with.

As for this set, I'll try to resolder some parts with leaded solder, but I'm afraid about ruining the foil resistors and the LTZ can. Let's see how it goes. Of course I'll post new pictures after that. They won't be as pretty as the next ones I'll build, but hopefully better than my last pictures.

Sergio

[plesa]

For next assembly keep the wires on foil resistors longer and cool them with tweezers at least.
For your LTZ1000A the  R5 should be >=13k too keep temperature margin.
R5 and R4 creates divider to set temperature, so you can increase R5 or decrease R4.

Also before next reference assembly train soldering on cheaper components Sorry