Project KX : DIY calibrator / reference source/meter

[free_electron]

It have spots on pixels, but at given price it is nothing to complain.
Even will dim pixels its way better than LCD. I'm trying to save some money on non-performance related parts first.

Already expect EMI issues from HV SMPS on VFD. 

I would throw out the display. Make this thing as stable and low noise as possible.
Remote control it or use static displays.

[TiN]

Forgot to show early layout thermal sim.



Conditions:

Inside teflon box (10mm gap from board edges), no airflow, still air.
+25°C initial air and solids temp.
Copper material for LTZ1000 and 0.06mm thick shapes on PCB
PCB 4-layer material
ABS material for other components
Gravity on axis Z (perpendicular to board plane)
0.1W heat source on square inside LTZ1000 package

Now calculating newer layout, it takes couple hours.

zucca

If you need reference meter, take your money and go buy existing one from recognized vendors. This project is just overgrown hobbyist LED blinker, with unlimited timeframe.
I have initial (very optimistic, i guess) plan and Project for this, but I doubt that i'll have everything mentioned here completely working end of this year.
If you think such "DIY" unit would be cheaper - false again, i guess just BOM will be couple thousands USD. And better not to think about calibration expenses, I don't have access to any cal gear.

dr.diesel

That's not possible for amateur hobbyists, I think. Just too expensive. Even to calibrate 6.5 meter need pretty decent standards, not just couple LTZ's hooked around. Of course if we talking to meet original meter specs
Main purpose of this project is to learn something about precision measurements, and waste excessive time in evenings

free_electron

I'll have this mode, cut power to all unnecessary stuff, that's for sure.
So far still want keep display, as it unit will be big and bulky size anyway, and it will be another excuse for EEVBlog members to say "this crap cost triple and don't have graphing function, like 34461A! Useless!" (pun intended)

[Zucca]

Thanks, I will give my money to some vendor then.

PS: with that temperature simulation, "learn Solidworks" jumped in the top ten entry in my to do list.

[Fabiusp98]

I'm a total noob(actually learning how to use transistors) but I'd want to help. Can I do something?
If this project goes ahead and meet the initial specs it will be a good victory for open source hardware on commercial cal stuff!

[TiN]

I'm a total noob(actually learning how to use transistors) but I'd want to help. Can I do something?
If this project goes ahead and meet the initial specs it will be a good victory for open source hardware on commercial cal stuff!

There will be no victory, as commercial cal stuff is traced against primary standards and have guaranteed specs, and traced to primary values.
I don't want for anybody to have false expectation that calibrator is something that is just stable source, and traceability to primary standards is just bundled bonus .
Safe to say, that most of gear in Tek/Kei/Fluke/Agilent calibration labs have more way more money in calibration and tracing all those years, than actual hardware cost. Best I can hope here, is just build stable unit or two, send it to few members with calibrated 3458A's, record values/references and adjust it to match those 3458A's.

As for help - mentioned before, I'd rather need software/firmware development efforts rather than hardware

[CaptnYellowShirt]

Forgot to show early layout thermal sim.

There it is! I've been waiting for this!

[TiN]

Some more thermal sims in calculation now. I need more CPU cores, currently it's taking 4 hours for one study on 10-core CPU 

Take them with a grain of salt, as these are very coarse conditions, not correlated as close as I wish to actual real thing.
Most critical is that LTZ1000 model in simulation geometry is set a solid copper, and my heater is "die" shaped surface inside
can. Real LTZ have silicon die, not copper, and die attach insulator (on LTZ1000A) with 400°C/W theta.
But my model is just copper solid thing.

Purpose of simulation i run is to study what is thermal gradient around pcb, to see if other components have major temperature
mismatch, rather than doing anything with LTZ zener itself. So I just dump 0.05W of power into copper dummy in LTZ model, and
have target for simulation as reaching +60°C in any point. This is about right what real LTZ uses, if I understood all fellow builders posts correctly.

So as a outcome conclusion - main problems with thermal simulation are getting meaning of all those colorful rainbows, and correlation to realworld device, which require learning of limitations and making careful assumptions of model conditions.

-------
Btw, I have an idea to try for community
I will start ordering and making some parts for build, and have camera to make a video's and post on tube.
But since my speaking english is not as well, as writing, not sure if videos would be much fun with brief comments?
What you think, worth to do, or don't waste time, and focus on design only?
Also might get some side videos to cover toolkits and gear I'm using to test various subjects of involved works.

[CaptnYellowShirt]

-------
Btw, I have an idea to try for community
I will start ordering and making some parts for build, and have camera to make a video's and post on tube.
But since my speaking english is not as well, as writing, not sure if videos would be much fun with brief comments?
What you think, worth to do, or don't waste time, and focus on design only?
Also might get some side videos to cover toolkits and gear I'm using to test various subjects of involved works.

I'd love to see what you're doing in that lab of yours, even if its not narrated. Or a picture slide show with sub-text?

 

[dr.diesel]

What you think, worth to do, or don't waste time, and focus on design only?

Your forum posts are quite clear with plenty of picts, what you're currently doing I think is working well.   

[Jebnor]

Interesting Project.  I'm commenting so I can follow this better. Keep up the good work.

[TiN]

Okay, some time had pass, and I did not had any new ideas, so will proceed with current order for voltage reference modules.

Here's BOM list with all parts.



I will hand-assemble 5 pcs and test them first.

[Andreas]

Here's BOM list with all parts.

Not shure about the real positions. (Missing the schematics):

R5 might be too low for a LTZ1000A device.
Usually you would need about 12.5K

R2 should be deleted for LTZ1000A. (only necessary for LTZ1000)

with best regards

Andreas

[TiN]

I posted schematics couple posts ago.

I'll have 12k and 13k resistors ordered too. And 400K is left on PCB for option if one want use non-A ref.

[TiN]

Ordered PCBs for LTZ references.

Going to buy some parts for them, and probably 2 more LTZ1000ACH and 2 LTZ1000CH.

Also got some low tempco resistors today for this project



VPG VCS101 0.02 ohm 1% TC=20ppm shunts for secondary measurements

.

[TiN]

Some parts from Digikey have arrived today (Exact list)
Hope to build first LTZ reference this week, still waiting for PCB.
Got LT1012's too, so will try both LTC2057 and 1012 on same design, to see difference, if any.
1012 is single version of famous LT1013 used widely in LTZ refs.

Will use 25-100ppm resistors, as it's best I have at moment.

[quantumvolt]

I will hand-assemble 5 pcs and test them first.

If you should want to sell one of the assembled boards, I would like to buy one. I never seem to get around to make one myself ...

I have a 34401A and 2x 34970A - all around 10 y.o. and apparently stable in relation to each other and matching within ca. 50 ppm - so I would like to have a LTZ1000 board set to 10.0 Volt elsewhere, and then follow it with my instruments.

That is - if you want to sell one ...

[TiN]

I'll have assembled boards later, but they are all direct 1:1 output, as goal with these modules is
to get most stable unit, with less error sources. And most accurate source of 10V i have here is
EDC MV106 (30ppm spec) calibrated against 90-day cal'd Keithley 2001.

And plan to use Z202 and VHP resistors for VREF modules, so price is relatively high, you can see
couple posts above from BOM. Just to make sure, noone get false expectations of having 60$USD for
LTZ reference or so If still interested, i'll let you know in couple months when I get few modules
aged and tested.

PCBs should arrive tomorrow or day after  O0   

[quantumvolt]

Sure - I am fine with raw appr. 7 volt. I have an HP RefAmp 1820-0001, a SZA263, an LTFLU-1, an LM399 and a 1N829A all running raw output 6.abc... to 7.def... volt without divider/amplifier.

And I expect you to do a profit on the board. Please put me on the list. Thank you.

[TiN]

PCBs arrived

Little leaser:

[quantumvolt]

Looks very beautiful indeed. I like the mix of high-tech patterns and discrete matching colors.

Don't forget I am on the list to buy a board 

[Dr. Frank]

Concerning raw output:

In former times, VOLT relied on Weston cells, i.e. 1.018xxx V.

That's the same situation, you had to transfer this standard voltage to the desired Cardinal Points (... 1.000, 10.000, 100.000 V ..), by means of a transfer standard. e.g. Fluke 720A.

AND those raw voltages were always much more stable than the amplified "round" value.

A pity, that Flukes references do not output and specify the raw reference output.

Anyhow, my planned LTZ1000 ref will have an additional amplifier for 10.000V output, because including that on the same PCB is more stable, i.e. less thermo voltage disturbance, less ground shift problems, less external disturbances.

btw.: The raw output is very susceptible to external HF signals.
Have you included any blocking measures?

Frank

[TiN]

Looks very beautiful indeed. I like the mix of high-tech patterns and discrete matching colors.

Don't forget I am on the list to buy a board 

List
Not sure if you still want it, I found very stupid error in design, which need two borged jumpwires 

Assembled one complete board today (as I bought only one LTZ1000A for start) with
available resistors (25-100ppm, best I could get so far).

Also except resistors and zener assembled 4 more boards. Total 3 boards with LTC2057, one with 2xLT1012 and one with 2xLT1097.

.

4-layer FR4 PCB, gold plated, red solder mask.
Idea behind construction - have zener in own shield box with 2 temperature sensors MAX6610 (one glued to top of LTZ, another on bottom).
Total 3 boards with same batch parts will be installed on bigger "carrier" board with dividers/ratios, low-noise LDO for zeners and monitor chip.
Every VREF board will have thermal shield and whole unit will have one more thermal shield.
So total voltage reference system will have three zener boards with 3 temperature sensors on each plus able to generate 1:2, 1:0.5 and prolly couple
more ratios, plus negative VREF.

I still don't see reason why need 10.00000V or 1.00000V outputs for voltage reference, as there will be DAC system later in unit anyway.
And my meters don't ask for 10 or 1V for calibration either, they want 2.0V and 20V.

Let's take a closer look on first PCB

As usual all photos are highly clickable, so enjoy megapixels..



Silk screen on top and bottom around LTZ pins shows areas without copper, so there can be drilled holes if I'll desire to do so later.



Left side (looking on top side) have single opamp and heater control circuit, temperature sensor outputs header.
Right side have input power header (top right) for +15V, zener opamp and precision resistors.



Zener have force and sense outputs on bottom header.
SMD Capacitors are tantalums and polymer film. I decided to eliminate possibility of piezo effects, so did not use ceramics on board. Will have one test module with ceramic X7R's to try later, as 1206 film caps are expensive.



Also there is MAX6610 temperature sensor (U3) near opamp.
MAX6610 glued to LTZ1000 can on top need to have thin wires routed to header J5.



I wanted try different opamps to find which one have best performance with LTZ1000.

LT1012 version (single-amp variant of industry-known LT1013)



LT1097 version



And chopper amp LTC2057 version (this is being primary at this moment, i'll have 2 more units with LTC2057, two with LTZ1000A and one LTZ1000)



Now bad part...

Mistake with this board is that U5 power pins are REVERSED. I had component on schematics mirrored vertically, but got distracted and forgot to
flip power input nets as well. So silly beginner mistake costed one LTC2057 amp.

So had to cut pin4 connection, isolate pin7 from PCB pad and bodge jump wire from a cap.



After that board worked fine, and I got magic 7V output!



Now build wonky box shield around zener.



Then put platinum RTD (1000ohm, Honeywell HEL-705) to measure temperature and get reference idea of stable temppoint.
Later will be replaced MAX6610 after I get temperature logs in various conditions.



And close box. Box sides are soldered to power ground on PCB. There are separate grounds as well for zener and everything else, connected only on single point by resistor R9 (0 ohm on right side).



Had done couple thermal camera photos, but forgot to copy those yet. Will post them later.

Wrapped whole thing in insulation foam (K-flex, commonly used in HVAC industry),
and now left overnight powered with 15V from Keithley 2400, and with my cal'd 2001 storing output.

When cold units just powered on, current draw is 43-42mA, drops to 39.x mA in couple minutes and then slowly decreasing, as heater in LTZ1000 warms up.
Higher temperature is - higher voltage output I get.

[Dr. Frank]

When cold units just powered on, current draw is 43-42mA, drops to 39.x mA in couple minutes and then slowly decreasing, as heater in LTZ1000 warms up.

Higher temperature is - higher voltage output I get.

Depending on your circuitry, the LTZ should heated completely very quickly, and current stable within a couple of minutes only.

I thought, the LTZ reference had a negative TC of around -50ppm/K. Therefore the voltage should decrease with higher temperature.

What temperature(s) you will chose finally for stabilization?

Frank

[TiN]

I think around +55-60°C, as it can be +35°C  here from time to time.

[Vgkid]

That looks really well done, good job.