Author Topic: Ultra Precision Reference LTZ1000  (Read 1333473 times)

0 Members and 1 Guest are viewing this topic.

Offline d-smes

  • Regular Contributor
  • *
  • Posts: 101
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1750 on: July 06, 2017, 10:25:46 am »
Dr. Frank-  I soldered Cat-6 to LTZ proto and screwed meter end into Gold dual banana and wrapped meter side with rag.  I'll admit, it's a bit quieter but I also severed the connection to six AD588's under test (Al enclosure at back left of my previous post attachment).  See picture attached present meter connection and LTZ readings following temperature quite nicely.  I further plan to 1) put a LM35 in the banana socket of the 34465A connector to read meter ambient temperature and 2) surround LTX prototypes with copper and aluminum plates for supplemental RF and thermal shielding.  Stay tuned for results...

A few things I forgot to mention before:
* Thanks for pointing me to your excellent review of the then new Keysight meters
* I've looked at and agree my 34465A is within spec assuming the "Non ACAL Temp Coeff" applied within +/- 2C (see Note 7 Keysight datasheet)

Please explain why capacitor feed-thrus are bad from emf viewpoint.  I admit to having a RF-noisy shop.  Even turning on the Hakko soldering iron will cause a noticeable shift of measured voltage (uV to 10'2 of uV).  These feed-thrus are directly soldered to PCB inside enclosure and are mounted upon a (hopefully) isothermal aluminum cover plate.  Surely this is better than standard connectors whose pins are held by plastic.  And unlike standard connectors, feed-thrus provide double-digit attenuation above 10 MHz (assumed 50 Ohm measurement).  What do you recommend?  CuTe jacks will have far greater RF leakage into the metal box.

Regarding shielding, left LTZ is powered by analog rectified and LM317-based regulator in its own metal box about a meter away.  It's shielded cable is tied to its box which is connected to PE (Earth) of the AC mains.  That's why the shield wire is pulled aside on the LTZ end (to eliminate ground loop).  The right LTZ proto is powered via black jack from battery with linear PSU keeping battery in float condition.   Right LTZ also has BNC jacks for zener voltage measurement whereby the shields of the coax were actively driven at guard voltage.  Because I didn't have Triax cable to meter, this picked up noise and was abandoned.  Therefore each LTZ has separate floating supply.  Ground of each LTZ has single-point connection to case.  Case is connected to shields of Cat-6 cable back to data logger where shields are tied to the PE screw of the 34970A data logger.  All voltage measurements are differential via twisted-pairs within Cat-6 cable.

Regarding 100nF capacitors, my SPICE analysis shows this compromises phase margin.  See attached where the Green traces depict the Bode Plot amplitude and phase of the Zener side of the regulator.   Note gain peaking of LTZ transistor stage at 550 kHz (red trace) with 100nF B-E capacitor.  I 'll show analysis circuit in another post (getting late).
« Last Edit: July 06, 2017, 10:35:09 am by d-smes »
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1751 on: July 06, 2017, 03:36:45 pm »
The feedthrough caps use ceramics and often a kind of kovar alloy to avoid high thermal mismatch. However these kovar like alloys tend to have a relatively high sebecke coefficient and thus higher thermal EMF. However it depends on the thermal design around the caps. There are 2 ways to avoid large errors: one is thermal coupling to the case on both sides (inside and also outside). The other is having a symmetric design, thus having a similar design for the Ref. GND too.

For the power supply the caps are a good idea.

Depending on the LTZ1000 circuit used, the circuit might not like to much capacitance at the output.
 

Offline d-smes

  • Regular Contributor
  • *
  • Posts: 101
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1752 on: July 06, 2017, 06:03:21 pm »
Thanks, Kleinstein.  Since these are just resin sealed, not hermetically sealed, I thought lead to be just tinned copper wire.  But I can't find this specified on the product page or in their brochures at http://eis.apitech.com/resin-sealed-bolt-in-filters.aspx   How could I tell for sure?
« Last Edit: July 06, 2017, 06:04:55 pm by d-smes »
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1753 on: July 06, 2017, 07:16:13 pm »
One could check for thermal EMF, when creating a thermal gradient around the feed through. So solder blank copper wired to both ends and measure the voltage. Than use something like the fingers or a soldering iron to create thermal gradients at the feed through. If it is copper only, there should be only a low voltage of maybe a few 10 µV. With a different material and some 10 K temperature difference it would be more in the 100-400 µV range.

Even with just copper (or other wires for thermocouples) wires one sometimes checks them, by moving a hot spot along the wire.  With a sensitive meter one could even detect areas that a more heavily bend than other (annealed) wire.
 

Offline chuckb

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1754 on: July 06, 2017, 07:17:10 pm »
A magnet will quickly inform you if the lead material is KOVAR.
 
The following users thanked this post: d-smes

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2377
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1755 on: July 06, 2017, 07:46:31 pm »


Please explain why capacitor feed-thrus are bad from emf viewpoint.  I admit to having a RF-noisy shop.  Even turning on the Hakko soldering iron will cause a noticeable shift of measured voltage (uV to 10'2 of uV).  These feed-thrus are directly soldered to PCB inside enclosure and are mounted upon a (hopefully) isothermal aluminum cover plate.  Surely this is better than standard connectors whose pins are held by plastic.  And unlike standard connectors, feed-thrus provide double-digit attenuation above 10 MHz (assumed 50 Ohm measurement).  What do you recommend?  CuTe jacks will have far greater RF leakage into the metal box.

Regarding shielding, left LTZ is powered by analog rectified and LM317-based regulator in its own metal box about a meter away.  It's shielded cable is tied to its box which is connected to PE (Earth) of the AC mains.  That's why the shield wire is pulled aside on the LTZ end (to eliminate ground loop).  The right LTZ proto is powered via black jack from battery with linear PSU keeping battery in float condition.   Right LTZ also has BNC jacks for zener voltage measurement whereby the shields of the coax were actively driven at guard voltage.  Because I didn't have Triax cable to meter, this picked up noise and was abandoned.  Therefore each LTZ has separate floating supply.  Ground of each LTZ has single-point connection to case.  Case is connected to shields of Cat-6 cable back to data logger where shields are tied to the PE screw of the 34970A data logger.  All voltage measurements are differential via twisted-pairs within Cat-6 cable.

Regarding 100nF capacitors, my SPICE analysis shows this compromises phase margin.  See attached where the Green traces depict the Bode Plot amplitude and phase of the Zener side of the regulator.   Note gain peaking of LTZ transistor stage at 550 kHz (red trace) with 100nF B-E capacitor.  I 'll show analysis circuit in another post (getting late).

Kleinstein already answered this question.
There are simply too many probable dissimilar metals and too many junctions involved.. better keep it simple, so not to loose the overview over the signal path.
Have a look in the LTZ 1000 datasheet, or the National Semiconductor application note for the LM199, both warn explicitly of thermal junctions, and simply to reduce the number of junctions.. that is the biggest problem in your setup.

Frankly speaking, I'm completely confused about your CAT 6 cable in series with the feed through capacitors plus clamped jacks, and so on.
Clamping the wrong metals, or oxydized metals also makes nice unpredictable thermocouples.


If think hat you have put too much focus on the shielding in comparison to e.m.f.s, that's true also for the absolute values of the different disturbance voltages.

I have seen similar shifts as you, then I used a big ferrite on the externally located DC supply cables, then these two 100nF, at least, maybe a simple grounding of the metal case, and now, no shift any more, even a switch mode LED light creates only minute changes on the the order of one ppm, instead of several ten ppm.
Maybe it's no good idea to ground the supply directly to mains earth. (I don't that, neither)

That's all a bit of vodoo, and fiddling around with the shielding (also keep it simple) , until the reference was nearly immune against RF or mains spikes.

Frank
 

Offline d-smes

  • Regular Contributor
  • *
  • Posts: 101
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1756 on: July 08, 2017, 02:45:08 pm »
A magnet will quickly inform you if the lead material is KOVAR.
Thanks for the tip!  For the 1nF capacitor-only feedthroughs I'm using, there is NO magnetic attraction.  And when I scrape the leads, I see the Cu color.  With a LC type, the magnet IS attracted to the body, but not the leads.  I suspect magnet was attracted to ferrite bead inside, but it could also have been the ceramic capacitor.  High-K ceramic capacitors (Type II & III) are ferromagnetic.   They also have piezoelectric and microphonic characteristics which could superimpose noise onto the reference output.   So if you want to use these bolt-in filters to keep RF out of your LTZ, make sure it has a (Type I) low-value NPO or COG capacitor within.
 

Offline d-smes

  • Regular Contributor
  • *
  • Posts: 101
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1757 on: July 08, 2017, 11:23:10 pm »
Regarding 100nF capacitors, my SPICE analysis shows this compromises phase margin.  See attached where the Green traces depict the Bode Plot amplitude and phase of the Zener side of the regulator.   Note gain peaking of LTZ transistor stage at 550 kHz (red trace) with 100nF B-E capacitor.  I 'll show analysis circuit in another post (getting late).
I take this comment back.  If one uses a 22nF capacitor on Pin 5 to Ground instead of the 2nF shown in the Positive Reference Circuit on Page 6 of the data sheet, the current regulator loop crosses over much lower at 25 kHz and results in plenty of gain and phase margin against the effects of the 100nF B-E capacitor.  OK, now that I understand this, I'll try it.
 
The following users thanked this post: Andreas

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1758 on: July 09, 2017, 08:59:42 am »
A large capacitor at the OPs input to GND can compromise the higher frequency noise. Some of the data-sheet circuits use 22 nF, some use 2 nF. Form the higher frequency noise performance I would prefer the smaller capacitor. In the low frequency range the lower transistor noise is effective instead of that of the OP.

A capacitor at the transistor B-E is aimed against RF noise - so even smaller than 100 nF capacitors should have a significant effect - so 1 nF could be enough. In addition to lowering phase reserve the capacitor could also increase the higher frequency noise. Even if not oscillating poor phase reserve can amplify some of the OPs noise. An extra cap in parallel to the zener might help to compensate for some of the negative effects. Usually the better way would be to keep out RF signal at the first place. So have a RF tight case and proper filtering at the supply and some (e.g. ferrite beat) at the output.

In the original circuit also capacitive loading to the output is a possible problem and this problem can add to the effect of the capacitor at the transistor. So an additional cap at the transistor is expected to make the circuit more sensitive to capacitive loading.

 

Offline Vtile

  • Super Contributor
  • ***
  • Posts: 1144
  • Country: fi
  • Ingineer
Re: Ultra Precision Reference LTZ1000
« Reply #1759 on: July 11, 2017, 04:20:05 pm »
It is propably mentioned and maybe even analysed in many times in this enormaus 80-pages thread. How good is LTZ1000s stability without the inbuild heater unit.
 

Online Andreas

  • Super Contributor
  • ***
  • Posts: 3222
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1760 on: July 11, 2017, 07:06:53 pm »
around 50 ppm/deg C (with the standard datasheet cirquit)
with a extra resistor you can get somewhat better.
« Last Edit: July 11, 2017, 07:08:34 pm by Andreas »
 
The following users thanked this post: Vtile

Offline orin

  • Frequent Contributor
  • **
  • Posts: 445
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1761 on: July 12, 2017, 05:07:11 am »
FWIW:

I got the PCB from OSHPark (SvanGool version) and built one up with Edwin Pettis's resistors.

I dialed in my Fluke 343A (which is nowhere near as stable) to match the LTZ output and recorded the difference for 60 hours or so on an Agilent 34461A.  Resulting screenshot is attached.

Enjoy.
 

Online Andreas

  • Super Contributor
  • ***
  • Posts: 3222
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1762 on: July 12, 2017, 05:32:05 am »
Hello,

You will need more references to decide if its the 343 or the 34461A which is drifting ...

with best regards

Andreas
 

Offline orin

  • Frequent Contributor
  • **
  • Posts: 445
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1763 on: July 12, 2017, 06:05:04 am »
Hello,

You will need more references to decide if its the 343 or the 34461A which is drifting ...

with best regards

Andreas


It's about what I'd expect for the 343A.  It was only spec'd to 3ppm/deg C.

I'm more concerned with eliminating the glitches which are likely RFI from WIFI or cellphones.
 

Offline VK5RC

  • Supporter
  • ****
  • Posts: 2672
  • Country: au
Re: Ultra Precision Reference LTZ1000
« Reply #1764 on: July 29, 2017, 11:27:20 am »
Beginnings of an experiment with several LTZs (x6)
- TiNs Kx boards - OshPark vB03 - thanks TiN (and SvanGool for uploading)
- Foam lined metal boxes x2 - metal (grounded) dividers between each reference (hard to see in pictures)
- Each reference has own linear low noise PSU 18v to 15V, TPS7A4901 regulator, linear PSU in metal box hard against reference metal box
- LM7805 supply for temp sensors
- 18V supply from linear LM7818 supply - transformer at some distance ~1m.
- LTZ1000. four As. , two Cs,  one of the C from 1990 from a fellow Aus volt-nut (retired) EEVbloger - Thanks lowimpendance
- Mostly TE UPW50 resistors - one set of Mr Pettis'
- Op Amps  LT1006 x4 ,  LT 2057 x2
- Terminals, Pomona 3770
- cabling twisted solid core copper, ferrite on LTZ output wires (Wurth)
- LTZ mounted low and flush to the board - foam only supports for the boards - see TiNs results with one of my other LTZ references.

Currently they are settling in - the more I look the more I find EMI is a very significant issue.
EMI culprits identified to date
 - LED light transformer (SMPSU - the work of the devil!!)
 - Uniteruptable AC power supply (also SMPSU) supplying the DC power supply.
 - most of my lab other gear, even an HP E3644A - connecting a laptop-GPIB (Prologix) - esp if mains powered.

In the next month or so I hope to investigate this to get some data to guide correction.

I find that all of this gear appears to Lower the readings displayed by the 3458, some as much as 60uV . I have not found any interference that 'raises' the  voltage as displayed.

I would be interested to hear if others have similar findings.

PS Later when in 'serious' use  - boxes will be further thermally insulated.
« Last Edit: July 29, 2017, 11:30:15 am by VK5RC »
Whoah! Watch where that landed we might need it later.
 

Online Andreas

  • Super Contributor
  • ***
  • Posts: 3222
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1765 on: July 29, 2017, 03:44:42 pm »

I find that all of this gear appears to Lower the readings displayed by the 3458, some as much as 60uV . I have not found any interference that 'raises' the  voltage as displayed.

I would be interested to hear if others have similar findings.

Hello,

yes that is typical for EMI.
The reason is that the EMI is rectified on the input/output protection diodes and thus gives a negative offset.

As I knew this already before from my LM399 references I did already some countermeasures on my first design.
(Battery supply to keep mains line EMI out and output filtering).

But obviously not enough so I had to fix this in the 2nd revision: (Caution R19 is needed to avoid oscillating).
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/?action=dlattach;attach=52101

(later improvements are mainly thermal management + different resistors).

In my case the USB cables for my ADCs are the main EMI source.
It also helps somewhat if you place all your gear on a metal plate.

With best regards

Andreas

 
The following users thanked this post: VK5RC, 2N3055

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 457
  • Country: us
Re: Ultra Precision Reference LTZ1000
« Reply #1766 on: July 31, 2017, 01:11:54 am »
VK5RC,
Yes, you found out exactly what happens when measuring low ppm, and I've written about this many times.  No switcher power supply nearby, no motors, no clicking unshielded relays / solenoids, no nearby RF transmitters (WiFi, cell phones, bluetooth etc), no PWM LED lights, try to minimize any AZ chopper op-amps (or add serious bypass and shielding - watch out for chopper freq noise on power rails), even no DMM for some measures - that's why we keep a fleet of analog null meters in the cal room when nulling to 732's + KVD - that will give you less error band than a single 3458a.  That's also why we switch over to 24V battery bus halogen incand. or DC LED lighting when going after best quality measure.  Halogen incandescent emit some IR, but much better color index than LED's of course.   We have several DC halogen lamps with cold (IR trap) filters that get a lot of use, and only use halogen + cold fiber illuminators on the microscopes and cameras (LED's don't have the color rendering quality we need for wafer test & inspection).  The concept here is you try not to generate thermal noise or thermal / photo oscillators with your pure DC lighting.   And you certainly don't shine a PWM-flickering LED lamp at your circuit when trying to get a voltage measure...You find out what parts are photo-sensitive to that PWM freq.

If you're in a clean room you make sure all the ground straps are in place on the Tyvek butterball suit, and test that every time you step out of the airlock.  Of course the table, floor and walls are anti-static, and generally we keep the area right at 60~65% RH (micro-misters overhead supplied with DI water) and generally 20°C in the cal & testing rooms, but it may get warmer on the production floor.  Anti-Stat generator fans of course but not too close to the DUT, and those are shut down just before testing starts.

Don't forget what's on the other side of the walls also.  We keep a clear area of about 15'~ 20' all directions from DUT (incl. above and below), and isolated grounds.  Fiber only for command / control, avoid all USB connections even if isolated.  Typically inner room test area has has steel + copper screen liner and shields most external RF.

It's also handy to have a supply of extra Mu-metal and copper shielding foil on hand for difficult setups.

You can't over-do keeping your test area quiet.  Every PN junction in your precision circuit wants to become a radio  or photo or mechanical stress receiver / rectifier at the worst time.
« Last Edit: July 31, 2017, 01:19:56 am by MisterDiodes »
 
The following users thanked this post: SeanB, cellularmitosis, 2N3055, TUMEMBER

Offline VK5RC

  • Supporter
  • ****
  • Posts: 2672
  • Country: au
Re: Ultra Precision Reference LTZ1000
« Reply #1767 on: July 31, 2017, 12:23:58 pm »
Thanks Andreas and MisterDiodes,
I have had just a little experience with EMI with my Amateur radio work - but they are generally far less sensitive.
I thought I had sorted ground issues - mmmmmm. I am not sure I can come at my 'volt-nuttery' having its own room at home.
Re 'EMI' from light/heat sources ; in preparation for Earth Moon Earth contact (on hold at present) I built a G4DDK Ultra low noise pre-amp for 1.2GHz (~0.3dB NF 37dB gain) and was able to measure the RF (at 1.2GHz) noise differences from 'cold sky' to a 'warm' tree nearby and to the sun and was within 0.5dB of what I should have measured (by calculation)
Thanks again for the advice and direction.
Robert
Whoah! Watch where that landed we might need it later.
 

Offline Do Ma

  • Newbie
  • Posts: 4
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1768 on: August 22, 2017, 09:26:08 am »
Hey all.
I want to built up a LTZ1000 reference with 16 LTZ1000 modules. The single LTZ1000 modules (the pcb design for this I do myself) works fine . But i got a tempco problem when I connect the moduls parallel. ( Connect them to sum point with a 100Ohm Vishay Resistor. I check the noise with 2 modules and I see the noise is going down 1/sqrt(n) (measurement with Fluke8508A 8 1/2digits)
Have somebody experience with this parallel connecting?
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Ultra Precision Reference LTZ1000
« Reply #1769 on: August 22, 2017, 11:10:38 am »
Welcome to forum, that is a very solid first post.  ^-^

I had no problems with 2 modules when I tried, however used higher value resistor (VHD100 10K+10K) and voltage of my modules were close (~40 uV apart).
What exactly issue you have, can share more details? Also blockdiagram of all connections and perhaps few photos would be great illustration.

Did you check for ground loops and power supply connections? Since you can't have direct zener kelvin connection in such setup, unwanted thermocouple junctions could be included as result, visible as "excessive tempco" at your DMM.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Ultra Precision Reference LTZ1000
« Reply #1770 on: August 22, 2017, 01:03:33 pm »
Keithley 2000 is reading temperature sensor? Hard to guess anything from the graph photo, other than something is terribly wrong :). Meter GPIB 3 shows -5.5ppm deviation, and meter GPIB 4 shows almost -10ppm deviation. That tells me grounding/current loops/connection issues.

Zener voltage reading by SMA connectors? SMA is not good for low thermals, as connector package is usually gold plated brass. You want only copper/copper connections with minimal amount of junctions (best is single twisted wire direct from zener to DMM binding posts) to get accurate microvolt-level measurements.

Also I can't see which resistors are used. You want 13K/1K and 120R to be most stable.
What is your single module measured tempco? Should be better than 0.1ppm/K for properly working reference.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline Do Ma

  • Newbie
  • Posts: 4
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1771 on: August 22, 2017, 01:12:12 pm »
Yes I think it is the grounding too but how to realize that with 16 moduls.
The single modul tempco is lower than 0.1 ppm. I got this problem only with more moduls.
keithley 2000 is reading temperatur and the fluke meassure one single module and the sum point (the single modul is much better in single configuration)
I can change the sma connectors in triax connector this is implemented. Also the settings are 13K/1K and 120R
 

Offline The Soulman

  • Frequent Contributor
  • **
  • Posts: 949
  • Country: nl
  • The sky is the limit!
Re: Ultra Precision Reference LTZ1000
« Reply #1772 on: August 22, 2017, 01:38:44 pm »
Zener voltage reading by SMA connectors? SMA is not good for low thermals, as connector package is usually gold plated brass. You want only copper/copper connections with minimal amount of junctions (best is single twisted wire direct from zener to DMM binding posts) to get accurate microvolt-level measurements.

Ok not the best, but realistically what thermal-gradient do you expect there to be on that connector?
And how does the base metal of the connector matter as all connections are made on the same gold surface? When mating the two (assume crimped )connectors it would be copper-gold-gold-copper?

My gut tells me to be more concerned in areas where larger thermal-gradients do exist such as the connection to the kovar legs of the heated reference and to the oxidized binding posts of some warm ancient dmm.  :box:

Not that I'm a expert, just asking to learn something.  :-+
 

Online Andreas

  • Super Contributor
  • ***
  • Posts: 3222
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1773 on: August 22, 2017, 02:01:44 pm »
Here some pictures from the prototype. The setpoint is 13K / 1K

Unfortunately they are not visible. (at least for me)

With best regards

Andreas
 

Online Andreas

  • Super Contributor
  • ***
  • Posts: 3222
  • Country: de
Re: Ultra Precision Reference LTZ1000
« Reply #1774 on: August 22, 2017, 02:13:31 pm »
Hello,

all I get is this:

perhaps it would be better if you upload (smaller sized) pictures here directly in the board.
So they will be also visible after half a year or later.

with best regards

Andreas
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf