MX Reference

[ManateeMafia]

It looks like most of the peaks are during the waking hours though a few are not. It could be the microwave or other issues outside of A/C since mine runs 24/7 nearly year round. My noise generating activities usually include cups of coffee from the Keurig, reheating the coffee that got cold from doing emails, etc... It could also be the refrigerator. The pool pump is just outside the  wall and runs several hours a day. It is not speed controlled so I would expect something around 1200 and 2000. Also, the pump died and was not working during some of the noise.

The Power Designs ps has an internal heater that cycles on and off. It could also be a source but it cycles on/off a 100 or more times a day. I would think there would be more spikes but I wouldn't bet on it. It could be swapped with an HP supply for the next round of testing the tempco.

The board is located inside an aluminum enclosure and the end is sealed with tape where the wires come out. It is temporary and I think the cable needs to be swapped from Cat5 to STP with the shield tied to ground. The end cap also needs to be installed. There are definitely lots of things to tweak with this setup and I want to try modifying the setup before adding parts.

[MisterDiodes]

You can also try some MuMetal or steel foil (or both) around the circuit, depending on the noise source.  Sometimes powering the 3458a thru a 1:1 toroidal transformer or adding additional chokes on the mains wiring helps attenuate power spikes running thru the meter.  Run the LTZ on battery power during tests if possible.  Keep all measure leads twisted, short as practical and shielded - that'll help.   Think "Minimal current loop antenna area", always at low ppm measures of anything.

[ManateeMafia]

I have some batteries charging for the imminent hurricane threat this weekend. I can certainly power the board with a 12V SLA and I will look at getting something on the 3458A mains side.

Could these spikes also be a sign of an issue with the 3458A internal reference? The meter is sealed and under service agreement. I could always test it with a 732B over a period of several hours on battery power.

I am planning on adding a 3D printed cover for the board. Is there any benefit with using a conductive coating on it similar to how Keithley did their 2001/2002 design? MG Super Shield comes to mind but I wouldn't use it unless it is worth the investment.

[cellularmitosis]

Is there any benefit with using a conductive coating on it

I'd say its worth a shot.  Here's an interesting video about using static-shielding bags to block EMI.



Searching for "faraday cage", etc on youtube is a bit of a frustrating experience.  90% of the material is aimed at doomsday preppers trying to save their walkie-talkies in the event of an EMP, and maybe one out of a thousand videos actually has any sort of empirical data (in the above video, he puts a spectrum analyzer inside of the bags!).

[texaspyro]

When I was doing PPM measurements, with a 3458A, I could tell, when the last person left the office down to the minute.

My sub-millidegree res temp logger could easily determine the number of bodies in a somewhat large-ish lab (well, actually body-equivalents... one particularly obese person counted as two people)  from the temperature gradient (BTW,  one body equals around a 100 watt heater).

I also have a double oven OCXO that can tell whenever I go to the kitchen to fetch a cool, refreshing, adult beverage... I can measure that thing to sub parts per trillion.  The temp logger clearly shows person entering room,  opening fridge,  closing fridge, leaving room.

[Kleinstein]

The spikes could be from EMI acting on the meter or the cables as well. If it is just a few rare events, I would not worry that much. Some EMI can be really strong - hard to protect against.

I remember an arc lamp at the university that stopped most computers (different brands) within about 6 meters around, when starting. It was really pain to make is EMC acceptable - they finally used cables inside copper tubing at some places.

The reference might profit from a steel enclosure to keep the magnetic field approximately constant. I Don't think it needs to be Mu metal. Just Mu metal might go into saturation to fast anyway. So it would be more like a second layer only. So far I have not seen how sensitive the LTZ1000 is to magnets. For EMI problems it is likely more the supply leads that bring in noise. Due to eddy currents even a normal metal case can shied AC magnetic fields at higher frequency.

[tszaboo]

So I give them (management) a number, less than 0.3ppm/K. Or something. Buy me a better multimeter, then I will measure it for you.

.
That's the good one, as there are not an awful lot of the voltmeters with tempco better than 0.3ppm/K  .

I think eventually I measured a few. The methodology was:
Select two with <100mV difference in the output. Connect ground, place one in the temperature chamber, measure difference. I know, it is not perfect. Even then I wasnt sure if the 3458A wasnt drifting more than the reference.

My sub-millidegree res temp logger could easily determine the number of bodies in a somewhat large-ish lab (well, actually body-equivalents... one particularly obese person counted as two people)  from the temperature gradient (BTW,  one body equals around a 100 watt heater).

Interesting. Is it a PT1000 with some 24 bit ADC or is it much more sophisticated than that?

[MisterDiodes]

Some other tid-bits to keep an eye on if you don't monitor already:

1.) http://blitzortung.org/
Use the above map for your location and see if there is a lot of lightning activity within 1000 miles (sometimes even more).  A large bolt will generate a lot of lower freq (some kHz to low MHz) mag field energy that's going to make any susceptible current loop circuit "ring" a little depending on it's orientation - and that ripple in the mag field travels thousand of miles around (and through) the globe.  A steel and/or MuMetal shield (or combo) will help attenuate that stuff.  If you're getting lightning within a few hundred miles might as well give it up for the day if you're trying for quiet measures at low ppm - with lightning that close now you're getting jolts to the local electrostatic E-field which is going to show up everywhere.

You can get MuMetal foil even at Amazon these days, a little thicker like .005" is good.  .01" works well also.  A good -steel- box over your circuit will help and the combination with MuMetal will make a good broadband mag H-field shield.  Aluminum isn't doing too much for mag field shielding but does help with air drafts.

2.) http://www.spaceweather.com/

You're always going to have more or maybe less problems during solar CME's and such (sometimes the CME event deflects cosmic rays, sometimes not) .  Even a random cosmic ray flying through the LTZ crystal substrate will cause a momentary glitch every now and then (usually with a longer recovery tail as the crystal de-stresses again over time) - and not a lot can be done about that.   This is exactly what happens to higher capacity computer memory chips at random times - you'll suddenly see a flipped bit for no reason...but that's the evidence of a cosmic ray on one memory cell.  It will happen more the smaller and more dense the memory cells are, and on analog chips you'll see this effect on die with more dense-packed transistor counts.

[ManateeMafia]

Lightning? https://weather.weatherbug.com/news/Top-30-Cities-in-the-U-S-with-the-Most-Lightning 

I live ~50 miles south of Tampa. Not much to do except move. I have seen the most incredible lightning storms while driving through Cape Coral (100 mi south). There are a few of us forum members living in these areas.

If I wanted to experiment with Mu Metal, where is a good place to buy it? I see several sellers online, but I would guess some sellers are more trustworthy.

[MisterDiodes]

Amazon is easy, but magnetic-shield.com works and a bunch of others.

Yes, summer afternoon are going to tend to show a bit more random noise on your data when there is a storm around - it just depends on how your current loops are going to ring when they get "tickled" with a bit of mag field ripple

[lukier]

You can get MuMetal foil even at Amazon these days, a little thicker like .005" is good.  .01" works well also. 

Is that good enough? I thought that after any mechanical processing (bending etc) mu-metal needs annealing which is done in hydrogen atmosphere. Heating pure hydrogen to 1000 deg C for hours is not something I would like to do at home.

[MisterDiodes]

More noise sources that show up on LTZ measures, especially if you're in a rural area - just as some fun examples literally from "left field":

1) Electric fencers from the ranch down the road. 

2) When the rancher down the road starts arc-welding on his hay trailer... Ask me how I know.  This can even show up when the 3458a is running on a battery-back up inverter.

3) Somebody at the other end of the building operating a spot welder building some test leads.

I can go on all day....but you get the idea.  At low ppm and uV levels you look all around you for the noise sources.  Keep your eyes and ears open.

[MisterDiodes]

You can get MuMetal foil even at Amazon these days, a little thicker like .005" is good.  .01" works well also. 

Is that good enough? I thought that after any mechanical processing (bending etc) mu-metal needs annealing which is done in hydrogen atmosphere. Heating pure hydrogen to 1000 deg C for hours is not something I would like to do at home.

Don't make -sharp- bends in it...You can still see if it has a reduction effect for your situation with somewhat gentle bends.  You might find that a good steel box alone works OK also - you just test and find out.

[ManateeMafia]

custommagneticshielding.com has this product that doesn't require annealing after bending. http://custommagneticshielding.magneticshield.com/viewitems/mumetal-sheet-and-foil/mumetal-foil
They mention that severe forming does not apply. They failed to mention what they considered severe forming. Is it making sharp bends to form a cover? Looks like a CYA catch-all.

I would consider using the adhesive backed foil and apply it inside a printed cover.

[branadic]

My sub-millidegree res temp logger could easily determine the number of bodies in a somewhat large-ish lab (well, actually body-equivalents... one particularly obese person counted as two people)  from the temperature gradient (BTW,  one body equals around a 100 watt heater).

Interesting. Is it a PT1000 with some 24 bit ADC or is it much more sophisticated than that?

This is also of interest for me. I'm sure it's not a PT1000 + 24bit ADC, there must be something more going on. I have a calibrated temperature measurement system based on PT100 + LTC2440 in ratiometric measurement mode (fast digital approach) with accuracy of 0.02°C over a range of -55 ... +150°. The resolution is sadly limited to 0.01°C by the manufacturer.

-branadic-

[texaspyro]

Interesting. Is it a PT1000 with some 24 bit ADC or is it much more sophisticated than that?

It used a precision V/F converter and a synchronized gate frequency counter circuit.  The temp sensor could be anything that can be coaxed to generate a voltage.   In this case a precision micro thermistor (needed for fast temp response) calibrated (data by NIST) with a 24 degree polynomial curve fit.  Absolute accuracy was in the low milli degree range.   Resolution was in the micro degree range.

[MisterDiodes]

custommagneticshielding.com has this product that doesn't require annealing after bending. http://custommagneticshielding.magneticshield.com/viewitems/mumetal-sheet-and-foil/mumetal-foil
They mention that severe forming does not apply. They failed to mention what they considered severe forming. Is it making sharp bends to form a cover? Looks like a CYA catch-all.

I would consider using the adhesive backed foil and apply it inside a printed cover.

..That might work, but really, bending a MuMetal sheet over a dowel isn't going to have a huge detrimental effect.  Usually a "severe" bend is where the inside bend radius is < 2X thickness, or where you take the metal well past its yield point on the outside corner (cracks appear).  Yes then you anneal.  For just playing with a test you get a sheet, put some gentle bends in it or make some corners over a 3/8" or  ~ 1/2" dowel, and see what happens. Keep it simple for first test...You'll know if it will help or not.  A lot depends on your PC board and how long are the traces, how big the current loops are, etc.

Or just try a steel box by itself and see if that helps enough.

Or just ignore the glitches if they don't really interfere with it's final application.  You're in a -very- lightning prone area, you might see the glitches fall off as storm season slows down.

[ManateeMafia]

I will keep you recommendations in mind.

Also, thanks to everyone for their suggestions.

Speaking of CME, I just seen this article. http://www.foxnews.com/travel/2017/09/06/northern-lights-will-be-visible-across-much-continental-u-s.html
Anyone logging data in this part of the country?

[Dr. Frank]

Hello,

to my experience, the initial drift is far too high, should be on the order of < 3 ppm over the whole year, at 50°C oven temperature.
Your measurements are an indication of hysteresis after overheating the LTZ1000.
No wonder, if you did not cool its legs during soldering, and maybe these exterior spikes once made the oven regulator to unlatch, heating it to > 100°C once.

I explicitly recommend Andreas design, especially the additional capacitors in parallel to each base-emitter diodes of the LTZ1000s transistors, that's C11 and C12 in his schematic.
It's definitely not true, that these capacitors affect the stability of the circuit, the opposite is correct.

I have added these two capacitors to my original prototype design from 2009 (it's in principle the datasheet circuit).
That greatly improved the EMI suppression, and the circuit shows no longer such spikes, even when a switch mode P.S.U. is present directly near the box.
Before that, the RF shifted the ovens set point and in turn the reference voltage (reversible change)

Also, I have built Andreas complete circuit once, initial drift was very low, and the EMI susceptibility was much lower than the original circuit... I also tested with switch mode P.S.U. and inductive loads (coffee machine test).


In the 3458A, the LTZ1000A is enclosed in the big analogue guard box, and neither its reference output, nor its supply rail is prone to the exterior, therefore it's not possible, that the 3458A may be affected by external disturbances. Whenever I measure 10V stability of 3458A versus my 5442B, there are absolutely no spikes visible, nor any voltage shifts, even when strong RF disturbance is present.


In stand-alone applications of the LTZ1000 or LTFLU, the disturbance mostly enters via the output connectors, or the P.S.U.
So a shielding / guard box is appropriate, also ferrite rings around the output jacks, like in the 732B. 


I'm just building another 4 new references, that's the successor of my prototype build, including supplements from Andreas design.
It's a single sided PCB, all leaded components, and it's compact, half size of the prototype, i.e. 50 x 80 mm, so four PCBs from one Euro-PCB.
That fits nicely in a small tuner box, which will be assembled inside another case.

T.C. compensation and 7=> 10V step-up amplifier will be trimmed next.

Anyhow, if you want to further improve your circuit, maybe you want to have a look on my solution.

Frank

[Andreas]

I'm just building another 4 new references, that's the successor of my prototype build, including supplements from Andreas design.
It's a single sided PCB, all leaded components, and it's compact, half size of the prototype, i.e. 50 x 80 mm, so four PCBs from one Euro-PCB.
That fits nicely in a small tuner box, which will be assembled inside another case.

T.C. compensation and 7=> 10V step-up amplifier will be trimmed next.

Anyhow, if you want to further improve your circuit, maybe you want to have a look on my solution.

looking forward for schematics and photos...

with best regards

Andreas

[hwj-d]

I'm just building another 4 new references, that's the successor of my prototype build, including supplements from Andreas design.

This awakes a great tension on what is announced now. 

Apologize that i post this here. I'm building a KX-reference too, powered it with simple 4x 18650 in series now about 2 weeks with short interruptions for changing akkus. In my opinion, now it is relatively konstant, but with TC ~1ppm per 1°C wich reflects the overall system with my 1 yr old 34461a. Sometimes, but mostly at the beginning of the overall test period, i have had relatively abrupt voltage rises or falls. But last time getting rare and <0.5 ppm, but noticable, if i touch my dmm for example. Screenshoot shows one caused putting in USB-stick just before. At irregular intervals i have this spikes too. In the near is wlan, router, i can't prevent. 

My  LTZ1000CH sits down on the board, soldering the legs shortly as possible with weller-iron and very little amount of solderpaste taking note of melting point. For measurement im using cat6-twisted with one side shielding to ground at the pcb and alu-housing. At the other side i have this hollow golden spring-suspended bananas, don't know their name.

Let it sit a couple of some more days or weeks, and then i'm lookin for someone in germany or maybe Tin, to kalibrate it. But i would build some more, with A-type LTZ too.

T.C. compensation and 7=> 10V step-up amplifier will be trimmed next.

One question to that, because it is the next logical step to build a useful reference. Most referencepoints of DAC's ADC's need 5V, 4096V or something like this but <=5V. Maybe it's useful to have one of these voltages directly driven from LTZ (second) buffer too, or must it be 10V first? As you can see, this is my first attempt to build such things...

Thanks and best regards

[TiN]

I'm just building another 4 new references, that's the successor of my prototype build, including supplements from Andreas design.

This awakes a great tension on what is announced now. 

I love it. Getting tired to see all these TiN's KXes with noisy choppers. 
Way to go, Dr.Frank. My bank of LTZ's (7 various units) currently cold for 2nd month, but I'll be firing it up when you ready.
Is it some specific project you going to do (number of 4 refs sorta imply that, 3 working standards and one for sanity checks) or you just miss old days building LTZ1000 circuit?
 

I'm going to build few more LTZ soon enough too, need ultrastable 1mA reference..

hwj-d,

One of the possible options to get stable <5V voltage for ADC/DAC is to use LTC1043 to divide 7.2V in half, but that likely to need additional buffer to get driving strength into DAC/ADC.
There is no need to get 10V if you have own DAC/ADC system which can autocalibrate for random voltage.

Hopefully you have cap on top and bottom of LTZ.

[hwj-d]

Hopefully you have cap on top and bottom of LTZ.

Inside the box? Simple i have two of these cotton make-up pads, one under, one over the LTZ.
TiN, you need more than one thanks from me to you ... 

[dr.diesel]

I'm just building another 4 new references, that's the successor of my prototype build, including supplements from Andreas design.
It's a single sided PCB, all leaded components, and it's compact, half size of the prototype, i.e. 50 x 80 mm, so four PCBs from one Euro-PCB.
That fits nicely in a small tuner box, which will be assembled inside another case.

Please share with us!  Possible group PCB purchase?

[texaspyro]

Inside the box? Simple i have two of these cotton make-up pads, one under, one over the LTZ.

I would be concerned with cotton absorbing and holding on to moisture from the air.  Would not ge a good thing if the cotton can contact components / circuitry.  Also make-up pads may be treated with who knows what.