Buyers beware, misleading 34470A reference modules (clones) on eBay.

[TiN]

All know that I have soft spot for ultra-stable DC Voltage references. When I saw some interesting looking module labelled as Semi-finished 34470A Ultra Precision Voltage Reference Board Without LTZ1000ACH. Note that it does not say "Keysight" anywhere except the PCB, and rightfully so. It's not a Keysight product, but a clone. Perhaps sellers main target audience are hobbyists who interested in upgrading and modifying their 34460A/34461A/34465A 6½-digit DMMs for better temperature stability and lower noise? I was thinking "price looks not too bad, maybe should get one or two to test next year?"

Below screenshot from Dave video, featuring real deal Keysight 34470A teardown and showing voltage reference module construction in great detail.

We can see that PCB quality and assembly is impeccable in 34470A with OSP-plated pads and precisely applied solderpaste. While reference circuit design is not very good in 34470A (compared to full potential that LTZ1000A chip is capable of in good hands), it is very good reference totally adequate for 7½-digit DMM. It is essentially replicate somewhat flawed reference circuit design from older legendary HP 3458A, repeating same high oven temperature problem that hurts long-term DCV stability of these instruments. This was deliberate decision to meet +50 °C operating temperature requirements.

Now is photo from the eBay listing for "34470A reference":

By obvious markers it is indeed very clearly different to original design made by Keysight, and does not look same quality even remotely. But it's not the main issue for me, after all soldering quality is fixable. What matters - it even looked pretty close with same looking 15k and 1k R4 and R5 resistors used for LTZ1000A super-zener temperature setpoint (marked with red square and arrow). Ratio of these two resistors is one of the key performance components that define stability of the LTZ1000A 7V output.

Anyhow, I thought it would be interesting to do some measurements and test these little modules in terms of temperature coeffient stability, noise and long-term drift. Seller accepted $30 USD best offer and two modules were on the way. Too good to be true?

Few weeks later boards arrived, but to my great disappointment PCBs I received are NOT what was shown on the photo in eBay listing



Board assembly quality and overall PCB quality is rather poor, but that does not bother me much. What is a much bigger problem - wrong resistors R4 and R5. Both boards that I received have standard SMD1206 1% resistors and not the special better resistors that are SHOWN ON EBAY listing photo. They may look the same to unexperienced eye, but pay attention on the digits font. Only one manufacturer of SMT resistors using special rounded font shown on both EBAY photo from seller listing and also on real genuine Keysight 34470A from EEVBlog's teardown video.

So this is false advertisement and eBay seller is misleading buyers as he mislead me here.



Also second issue - since I bought two modules, they both have this problems with R4 and R5 resistors, but second module also missing capacitors C1 and C2! Why?
I have messaged seller to either modify listing to reflect correct information, or to refund the order. Will see what happens next.

Some more photos. Heater drive transistor is good old 2N3904 in TO-92 package, just like real 34470A ref. No issue here.



And yes, I know that proper LTZ1000A board components just in resistors would cost 5 times the price of this whole module. Have more than 7 years of experience with various LTZ1000-references and have lot of different designs. It was especially interesting to see how bad is "bad" 34470A reference PCB design is versus expensive typical LTZ1000A with VPG foil resistors and fancy PCBs.



At least 34470A module is smallest there is for production-grade module. It is half the size of old 3458A A9 LTZ1000A module and more than 10 times smaller than my latest quad LTZ1000A ultra-low noise xDevs.com QVR module.



Tests and performance numbers are to be done next year.

UPDATE.
Seller promptly responded and will organize return shipping and replacement to two boards with "proper correct" resistors. I was promised a photo of the new boards before shipping. So hopefully this will be resolved to everybody's satisfaction. 

[floobydust]

It's like a science experiment to see how many people stabbed each other in the back. I'm told that once you get your product made in china, the first thing they do is rip off and distribute all drawings and IP.

I saw a documentary going after methlamine and fentanyl producers in china and the video showed knock-off shops in Shenzen, an entire building full of people hand-painting classic art works.

But here it looks like a port from Mentor to Altium, the silk font reveals that look at the "6". I would wonder if the People's Op-amp LM358 snuck in there.
Check the solder bridge on J1 pin 1-3 at R96.

[Dr. Frank]

Hi TiN,

I also have seen this PCB, and from the ebay photos it was quite evident that the components were assembled manually.. leadfree soldering by hand is not easy. 
So chances were high, that the resistors were also not low T.C. ones.

Anyhow I also ordered one, and received it last Monday, see top and bottom shots.

My plan is to fully reverse engineer the PCB, to find out how it's connected to the 34465A/470A, in conjunction with the existing LM399H drive circuit.
There was already a partly deciphered circuit on the forum, but a few tracks were still missing, i.e. these on bottom where a sticker was placed, and a few below the connector and the LT1013.. What I've measures so far, these latter lines make sense. (Kelvin connection  )

Even if this PCB is a clone, they will probably have copied the circuit correctly, and I have the possibility to compare the bottom side against an original KS board.
The quality of the PCB  also does not matter, as Keysight has made the worst design ever, anyhow.. too thin lines, slots, and so on.

TiN, where did you find differences in the layout, already?

My further intention is to replace all 5 precision resistors with small TH metal bulk resistors from AE, the FLCY, which can be squeezed on the existing pads.. and also setting the oven temperature to reasonable 60..65°C (12.5k/1k). T.C. compensation will also be done.

Then I will assemble it into my 34465A so that I can put in the existing LM399H back, if required.     

Let's see, if everything works directly, or if this reference would need a SW enabling for the 34470A.. different analogue interface, different ACAL functionality maybe.

Frank

[floobydust]

I drew this schematic of the board a long time ago, hoping to copy it and make million$... well actually to look at its design.
I could not see underneath the 8-pin header for what pin 2 is about, and if there is a V- rail. I see they can monitor heater output.
Please let me know of any errors/updates and I will update this post.

edit: 2020-12-25  Updated schematic to rev2 based on post #7.

[TiN]

Seller replaced photos in his listing, now showing only same crappy film resistors your and mine boards have. Seems like they dont have proper resistors to sell boards with. Was expected something along these lines to happen ($30 shipped, can't expect $45 resistors on the PCB) but still hope dies last.

I might keep one board as received and replace resistors on second to fancy VPG VSMP1206 15k, 1k and 100 ohm to test. Most likely real 34470A module performance would be somewhere in between. That's low priority project for next year/

[ScoobyDoo]

Hello Folks - The seller stated/states from the early beginning that this board is a replica of the Keysight board 34470A-26503 Rev. 003. I bought a few to use it as test fixture to investigate drift problems with the 34470A unit - as I will not probe & solder on the genuine 26503 board. Big question is if they re-used the CAD source files or gerber data files to manufacture this board (1) - or if they handcrafted the artwork - created new CAD database and generated the gerbers themself (2) - the first (1) is form of theft, the latter (2) is an infringement of copyright under US and EU law but probably not in RPC.
It is little bit similar to situation with HP/KS 82357B GPIB adapter - almost perfect clones are available on market for a fraction of cost ... - and I am sure KS Loveland is reading this ...

Anyhow Merry Christmas ...

ScoobyDoo
Herzlichen Grüßen/Meilleures salutations/Best regards/Un saludo

[Kleinstein]

If one replaces the resistors anyway, why keep it at 15K/1 K and not go for a more moderate 13 K /1 K.
A better grade resistor should be used, but I see no real need to go all the way to foil resistors. Good thin film resistors like the Susumu RG or maybe URG series should be good enough.

I see nothing wrong with a compact LTZ1000 based reference using more moderate grade resistors. However there is no need to do a 1:1 copy of the layout.  The main point is getting the kelvin sensing right. The very thin lines look a little odd as they can add to the resistance. However the resistance of the copper traces is reasonable stable over time and the TC contribution should not be too large and can cancel out to some degree. If tested OK, why not.

Anyway, LTZ1000 reference chips are currently hard to get - so it a more theoretical problem.

[Dr. Frank]

Hi,
here are pictures of an original board. (Thanks to somebody from the forum)
I could not find any difference, even the tracks look identical, apart from a single one, which had a slightly different thickness change.
The date code information varies.

Floobydust, there were really several errors in your schematic, like different numbering of the connector, and especially that pin 5 is current feed for GND and REF- and pin3 is REF- sense. pin 4 is n.c., pin 2 is connected to a capacitor.
I'll post the connector later, including LM399 interface.


Frank   

I had to get the turkey from the poultry farm, so now I added a sketch of the hidden connections, (by continuity measurements), and how the LM399 and LTZ chip connect to the main board.
I designated the white dot as pin 1 of the connector.
pin3 (Ref- sense) and pin5 (Ref- feed) seem to be connected on the 34465A already .. not sure
pin 4 is definitely n.c., pin 2 I could not yet assign meaningful yet to the circuit.

C1, C2, C6 = 150nF
R3, R93 =75k
R4 = 15k, R5 = 1k
R95 = 100

These resistors should be the high precision ones, which I will replace by TH components FLCY

Thanks in advance for helping.

PS: This turkey will also be calibrated, obviously

[floobydust]

[...] Floobydust, there were really several errors in your schematic, like different numbering of the connector, and especially that pin 5 is current feed for GND and REF- and pin3 is REF- sense. pin 4 is n.c., pin 2 is connected to a capacitor.
I'll post the connector later, including LM399 interface.
Frank

Thanks, I've updated the schematic to Rev 2 in Reply #3 based on the new info. Please review. Heading out for turkey now.

The original Keysight board might use Susumu precision SMT resistors for the three parts - 15k 1206, 1k 1206, 100R 0603, based on look (RR1632) and the advice of my psychic. But that series is discontinued 2017 and replaced by RG and typically ±5ppm/°C but Digi-Key has some ±2ppm/°C parts unless in error.
Susumu RG3216L-153-L-T05, 15k ±0.01% ±2ppm/°C  1206 (3216 Metric) AEC-Q200  USD $5.20 ea. in 1-lot seems to be the best you can get in SMT. There are 10,000hr graphs for them. I know WW is king.

I would say the original Keysight pc board is sloppy- orphaned extra vias, missing/confusing ref. designators, quirky teardrops, and extra paste on non-populated parts. Usually that stuff gets cleaned up in Design Reviews.

[Dr. Frank]

Hello floobydust,

many thanks for this update! 
The schematics now look much nicer.

It all now makes complete sense.. and I now understand how this  fits into the usual LM399 drive circuit, probably w/o any SW quirk, .. see schematics of the predecessors 34410A/411A (identical A/D converter Multislope IV, and therefore identical Vref circuit).

Frank

[Dr. Frank]

Hi folks,

I have removed these 5 SMD resistors and made a quick temperature test on the 15k00 one, giving about 25ppm/K only.
Both resistors 1k, 15k, were very close to their nominal values, so I assume these might be 0.1%.

I assembled leaded FLCY resistors instead, and an LTZ1000A, so that all components stood less than 10mm above the top side.
That also allowed me to use a copper plated plier during soldering of the LTZ1000A, to avoid strong heating of the chip, and I really don't like to shorten its legs too much.

In the 34465A, I replaced the 4 pin precision DIL socket by 8 pin golden sockets, about 5 mm in height, so that it's possible at any time to plug in the LM399 again.

The connection is made currently with 8 short silver wires of 0.6mm diameter.

A case is yet missing which would give mechanical stability, but also protection from the fan, which blows air directly across the references.
There is 20mm 18.8 mm clearance between PCB and metal cage, so this assembly fits well below.

I made a new stability measurement with the LM399 inside, relative to a silent and stable M7000. .. see also my review in 2016 on both instruments.
The 344465A shows some drift, might be from temperature drift (<1 °C), or from drift after ACAL (for some unexplained reason).
1h noise at a stable time point (7..8h) is about 880nV_rms, which is typical for LM399.

Another contributor of this forum made the same measurement with his 34470A against a 732A, and the better stability and noise is apparent.
A slight temperature change of < 1°C was also present.

The noise looks optically better, but some jumps ruin the data, as it's 1h measurement is 700nV_rms only. On smaller time scales, i.e. 10min., the 470A clearly beats the 465A with 485nv_rms, but that's still twice the noise level of a 3458A.

The third measurement shows the 34465A with the LTZ1000A reference.
It's working, although I assume that the REF- pin 3 is handled differently when the 7.5 digit mode is enabled.

The noise and stability at first sight is disappointing, 782nV_rms, apparently same figure as with LM399.

That could have several reasons.
The most promising is simply the lack of a protection case over the circuit.
I will test that next. Maybe a LM399 also could take advantage of a small enclosure.

Frank

[Andreas]

The noise and stability at first sight is disappointing, 782nV_rms, apparently same figure as with LM399.

Did you check the noise of the LTZ1000A alone (with 1/f amplifier) before mounting into the 34465a?

with best regards

Andreas

[Dr. Frank]

The noise and stability at first sight is disappointing, 782nV_rms, apparently same figure as with LM399.

Did you check the noise of the LTZ1000A alone (with 1/f amplifier) before mounting into the 34465a?

with best regards

Andreas

Hi Andreas,
not yet, one thing after another..

I don't expect an extremely noisy (defective) part, anyhow.

Frank

[notfaded1]

@Frank that's super neat you got it going!  The new 8 PIN socket looks like it almost belongs there.    Can't wait to see the numbers with the case on!

Bill

[Dr. Frank]

All right,

I have applied a thin styrofoam layer underneath the reference. This helps to protect the bottom solder joints of the LTZ1000A (also inside a 34470A).
Then I attached the reference board, and a 18mm deep, 40 x 65mm plastic box over the whole PCB area. The box fits nicely into a rectangular slot around all the components on the main PCB, and also protects this main board area from air draught. The edges of the box had to be trimmed at a few positions to fit over some components, like the ceramic resistor array.

This box now gives mechanical stability to the reference PCB, like the original plastic part.

The 16h stability measurement was started 15min after power on.
It shows an initial 1h warm-up phase and then again a slower negative drift, which finally seems to stabilize in the last 2hours.

The noise now looks like the one of an original 34470A.

The 1hour noise figure between 11 ..12h is 420nV_rms, just like the 470A, but looking a bit better in terms of peak to peak noise.
That's about 2 times worse the noise figure of a 3458A, but over a shorter observation time of 1min that approaches its performance, please compare my review from 2016.

The last screen shot shows my 34465A performing as a 7-digit instrument, variation over 1min as low as 1.2µV / 0.12ppm

Conclusions so far:

1) It's possible to upgrade a 34465A to the performance of a 34470A regarding noise and stability, by simply attaching the LTZ1000A reference board.
2) The reference board needs to be equipped with stable components (at least 5ppm/K resistors), and can be set to a lower oven temperature (60°C for metrology grade performance and environment).
3) The LTZ1000A circuit is used by the main circuit as a virtual zener diode (like the LM399A)
4) Possible special 34470A procedures on this board, like use of the REF- SENSE pin, or the V_heater pin can not be observed, as probably a SW flag for 7.5 digit operation may be enabled.
This might then give rise to these unexplained ACAL problems on 34470As only, which I do not expect in this configuration, for obvious circuitry reasons.
5) The fan air flow is directed over the reference assembly area and the analogue circuitry around. This I regard as a real design flaw.
6) A protecting box around the reference is urgently required.
7) In case of the 34470A, the plastic part is mostly sufficient, although the bottom joints are unprotected.
An additional bottom protection MIGHT improve the noise, i.e. further reduce these irregular bumps in the 1h stability figures.
8 ) In case of the 34465A, this protection box assembly MIGHT also reduce the noise performance of the LM399AH, although these two connections potentially might not be as sensitive as the whole LTZ board. It has to be tested, whether the noise of the 34465A arises directly from the LM399, or from the air flow.
9) The 2% higher reference voltage affects the calibration of all modes, apart from Ohm mode. It has to be tested, if a re-calibration of the DCV modes is sufficient to bring all modes back into spec, or if a full calibration is required.
10) I will do this test, and try to find out, if ACAL will work as perfectly afterwards, as with the original 34465A.
See recent remarks from David Bogdanoff about proper operation of ACAL after a preceding CAL.
11) If my 34465A performs well/better with this LTZ reference, I will keep it inside, replace these silver wires with golden ones, and may send the instrument to KS for a full calibration.. I think it's time anyhow, after 6 years, where it's still inside the 24h specification for many modes and ranges.

Frank     

[branadic]

Thanks for the pre-liminary results, Frank.
As far as I understood the comment by David Bogdanoff

https://www.eevblog.com/forum/metrology/keysight-34470a-calibration-and-acal-problems/msg3358358/#msg3358358

the solution to the ACAL issue is nothing you can solve without the proper equipment

... can be resolved by performing full range calibration on the DC Voltage function ...

So this will be the most interesting part on 34465A to almost everyone, I guess.

-branadic-

[Dr. Frank]

I just have seen, that 1h comparison charts for the different instruments are missing.

So I add graphs with the same vertical scale:

- original 34470A
- performance 34465A with LM399A
- 34465A with LTZ1000 without protection
- 3458A performance

[notfaded1]

@Frank The hat really did make a difference!  Thanks for showing the others on the same scale making it visual.  Super cool stuff. I think you basically proved the upgrade can be beneficial.   Is there a hardware jumper that makes the board know it's a 34465 vs 34470?  I had thought they shared firmware so it must check hardware somewhere to see if it can show the extra digit is that right?

Bill

[Dr. Frank]

@Frank The hat really did make a difference!  Thanks for showing the others on the same scale making it visual.  Super cool stuff. I think you basically proved the upgrade can be beneficial.   Is there a hardware jumper that makes the board know it's a 34465 vs 34470?  I had thought they shared firmware so it must check hardware somewhere to see if it can show the extra digit is that right?

Bill

Thanks, Bill

Yes, firmware is identical, (3.1), as well as mainboard HW, as nobody found any HW difference or switch in the 2015 reviews.
It obviously does NOT check the reference for an LTZ board.
Therefore this will probably be SW enabled, like the MEM and DIG options.

My 34465A still behaves like a 6.5 digit DMM, even after this upgrade.

It displays real 7 digits on the main display and 8 digits for Mean by a simple trick: Engage Math => Scaling with m=1 and b=0.

Frank

[Kleinstein]

It would make sense to have a slight difference in the hardware between the LM399 and LTZ1000 version. The LM399 needs the extra current to drive the reference side. The LTZ1000 circuit can get away without the extra current and this may make a small difference as the connector is in the current path.  So there may be a different resistor value on the board (higher resistance with the 34470 as there may still be input current of the inverting amplifier to compensate).

For sensing the low side of the reference (ground pin like with the LM399 or separate ref-  could be a difference in the software for the ACAL procedure. The normal measurement may use the normal ground also with the 34470.
The software may have a chance to detect the presence of the ref- signal with the LTZ1000 board.

Some of the irregular bumps may be from the resistors. Some Excess noise from the resistors can show up similar to extra 1/f noise in the reference.

[TiN]

Either I'm blind or there is no significant difference between 34465 @ LM399 and 34465 @ LTZ on the plot  ?
I assume first plot is stock 34470 @ LTZ? Chart titles are not very clear, but it does look like stock 34470 have 2 times less noise then 34465 (either with LM399 or LTZ).

So conclusion "It's possible to upgrade a 34465A to the performance of a 34470A regarding noise and stability, by simply attaching the LTZ1000A reference board." does not hold on truth, if we base on above 1 hour results. 

Is there a fan in 34465 and 34470? I remember somebody posted experiment with turning off fan and reducing DCV noise a lot?
I'd solder proper pin-header onto mainboard and plug board like it supposed to be plugged, instead of the strange looking pieces of wire shoved into collet sockets?

Also no need to enclose whole ref in cover, just an LTZ would be enough In original 34470A bottom pins of LTZ area are also covered by plastic cap.

after 6 years, where it's still inside the 24h specification for many modes and ranges.

Sorry, that is hard to believe

[floobydust]

[...] Also no need to enclose whole ref in cover, just an LTZ would be enough In original 34470A bottom pins of LTZ area are also covered by plastic cap.

I haven't seen a bottom cover on the 34470A ref. in pics    The 3458A has both top and bottom cover.

I think you need both, you don't want cold feet on a hot reference. That just steals heat from it, conducted through the pins to the bottom side pads. Could log Vheater to see what heater output is doing.

Yes there's a wailing, howling cooling fan, in line with the ref. subassembly. As lab A/C cycles on and off, so does the air temperature inside the instrument, and fan speed moves around too. So that adds another variable for the reference's heat loss.

[dietert1]

If you are asking whether one zener reference is better than the other, these multimeters are a good test bed in the sense of convenience and "what can you get with minimum effort, buying and modding a cheap ebay offer". Dr. Frank just demonstrated once more how easy it is to get a LTZ1000 working to a reasonable level of performance, nothing more. He also showed clearly that the solution he found can't compete with another one he has in his lab.

The question whether one ovenized zener will make a better metrology reference than another one is beyond this. Thermal EMF will add in once you are looking at 1 uV and below. And about the oven circuits: Somebody may add a thermal cover to a LM399 reference and it will exhibit less noise than a LTZ1000 exposed to a fan air stream.

Regards, Dieter

[Dr. Frank]

Either I'm blind or there is no significant difference between 34465 @ LM399 and 34465 @ LTZ on the plot  ?
I assume first plot is stock 34470 @ LTZ? Chart titles are not very clear, but it does look like stock 34470 have 2 times less noise then 34465 (either with LM399 or LTZ).

So conclusion "It's possible to upgrade a 34465A to the performance of a 34470A regarding noise and stability, by simply attaching the LTZ1000A reference board." does not hold on truth, if we base on above 1 hour results. 

Is there a fan in 34465 and 34470? I remember somebody posted experiment with turning off fan and reducing DCV noise a lot?
I'd solder proper pin-header onto mainboard and plug board like it supposed to be plugged, instead of the strange looking pieces of wire shoved into collet sockets?

Also no need to enclose whole ref in cover, just an LTZ would be enough In original 34470A bottom pins of LTZ area are also covered by plastic cap.

after 6 years, where it's still inside the 24h specification for many modes and ranges.

Sorry, that is hard to believe

Good morning Illya,
maybe you need a strong coffee first 
Please read the textual descriptions I have given, then it's clear what's been measured.

The 34465A with LTZ AND top cover is displayed one post higher, and it's really on par with the original 470A, i.e. about 400nV rms noise.

465A with LM399 and 465A with LTZ without top cover are both noisy, both about 800nV rms noise. That's in the next lower post, where I also show the 470A and 3458A.

You're right, ScoobyDoo just sent me a picture of top and bottom cover, which I've never seen before.
That means, that the original reference inside the 470A is fully covered, but you also really need complete shields over the whole PCB area ... which the 3458A is missing.


Next experiment will be to put in the LM399 again and try out noise with top cover.

Btw.: I've already re-calibrated the 34465A to the new reference value in DCV, and ACAL works like a charm, each DCV range inside +/- 2ppm, or so, as before with the LM 399A inside.
I'll post results later.

The 34465A currently reads the M7000 you've calibrated, 9.999 873 2V at -1°C compared to 9.999 877 6V what you have measured recently.
T.C. w/o ACAL is determined to be +0.3ppm/K


Concerning the 24h specification: That's easy to believe, just read the specification. It's so barn door wide, i.e. 20..30 ppm in DCV, even for the 470A, that this will be easily met.
All DCV ranges were inside +/- 4ppm during the last 5 years.

Add: With LM399 back in and DCV calibrated, I just checked, that most Ohm ranges are well within 24h spec, a few others inside 90 days. spec.
ACV all <100ppm, low DCI ranges  1µA to 100mA are 50ppm off at most. Unit was last fully calibrated in 2015.

Frank

[View[+]Finder]

(I just now read Frank's excellent post) The pricing for the replacement part on Keysight, versus ordering from China (for ) and buying all the parts, looked too good to pass up without considering.

I placed an order with Keysight for the board (USD441 with tax on my CC) and if their shipping is as I have had in the past, the board should arrive in a week or so.

I don't expect that it will come with its housing--that would be too nice--so perhaps someone will 3D print the enclosure parts? I'll update once the part arrives.

If anyone is interested, send me a PM.

Cheers,
Donal

[View[+]Finder]

In the 34465A, I replaced the 4 pin precision DIL socket by 8 pin golden sockets, about 5 mm in height, so that it's possible at any time to plug in the LM399 again.

What is the spec (or part number) for the 8 pin golden socket? 

Thanks,
Donal

[TiN]

Good morning Illya,
maybe you need a strong coffee first 

Prefer herbal tea, thanks

Ok, so I was looking at "Stab_1h_LTZ_without_hat.jpg" and "34465A_1h_stability.jpg" which is LM399, which looked about same. Didn't see previous post at first, my bad.  .

You're right, ScoobyDoo just sent me a picture of top and bottom cover, which I've never seen before.

Dave's EEVBlog 34470A teardown video shows reference assembly in great details, including covers and everything. That's my only source of knowledge about these meters, as I don't have any 3446*A or 3447*A.

The 34465A currently reads the M7000 you've calibrated, 9.999 873 2V at -1°C compared to 9.999 877 6V what you have measured recently.
T.C. w/o ACAL is determined to be +0.3ppm/K

What is M7000? That is 7000 cell from 7004T unit? What M means? You guys all supersecretive about all these 7000 versions and info adding slight amount of confusion

Concerning the 24h specification: That's easy to believe, just read the specification. It's so barn door wide, i.e. 20..30 ppm in DCV, even for the 470A, that this will be easily met.
All DCV ranges were inside +/- 4ppm during the last 5 years.

Just checked spec, its not that wide for 24 hours, +/-13ppm for 34465A and +/-10ppm for 34470A. Pretty decent for just 6.5d/7.5d meters.

[Dr. Frank]

The 34465A currently reads the M7000 you've calibrated, 9.999 873 2V at -1°C compared to 9.999 877 6V what you have measured recently.
T.C. w/o ACAL is determined to be +0.3ppm/K

What is M7000? That is 7000 cell from 7004T unit? What M means? You guys all supersecretive about all these 7000 versions and info adding slight amount of confusion

Concerning the 24h specification: That's easy to believe, just read the specification. It's so barn door wide, i.e. 20..30 ppm in DCV, even for the 470A, that this will be easily met.
All DCV ranges were inside +/- 4ppm during the last 5 years.

Just checked spec, its not that wide for 24 hours, +/-13ppm for 34465A and +/-10ppm for 34470A. Pretty decent for just 6.5d/7.5d meters.

Hello Illya,
sorry for the confusion with the designators, this 'M7000' was somehow taken over since the Metrology Meeting 2019 where somebody brought a reference which was labeled 'MODEL 7000'.

There were also version from Metron, Wavetek (like the 4th cell) and Fluke, so mine would be better called an F7000, as it has several FLUKE cal stickers .. it is the third reference in the 7004T rack, which you kindly have measured to 0.3ppm uncertainty, thanks again for that!

The specification for the 10V DC range might be relatively tight, but the other DCV ranges and OHM are about 25..35ppm/24hr only, but all these ranges do perform much better than that, over years.
I'm under the impression, that the 34465A and 470A as successors of the 34401A / 410A / 411A simply have gotten the same specification for sort of historic reasons, instead of updating to the real, better data.. and the predecessors might already have performed better than specification.

However, it's not a good sign, that the 34470A was not specified tighter, like a 7.5 digit instrument. But that's no wonder either, as both instruments share an identical main board

Frank

[Kleinstein]

The ADC of the 34401 is quite different from the others. As much as I have read the ADCs in the 34410 to 34470 are still quite similar. There may be some changes in the parts used (one of the OP27 in the 34410 would be a prime candidate for an update for lower noise), but the architecture seems to be the same. So the linearity and stability are probably very similar, though it may only take minor changes to effect the INL likely more toward the better from experience and improved test tools.

The improved the reference for in the 34470 would change that much, mainly reference noise and maybe a little on the reference drift. A LTZ1000 with a 15K-1K temperature set point may not be that much lower drift than a burnt in LM399.
The ACAL  steps should not be effected much from reference noise - this is more a question of the linearity.

[Dr. Frank]

Having calibrated the DMM with the LTZ inside in all DCV ranges, I cooled it down from 23°C to 15.8°C.
The reading was -1.9ppm low, which gives an overall TC, w/o ACAL, of about +0.3pp/K.
At cold using ACAL, the reading was +3.7ppm, so I estimate the TC of the resistor network to +0.7ppm/K, and of the LTZ circuit to +0.5ppm/K, if a perfect ACAL compensation is assumed.

At normal r.t., I made many ACALs, one after the other, the readings in all DCV ranges  then were always within +/- 1ppm to nominal again, no sign of getting that mysterious offset phenomenon.
If you look at all the stability graphs over several hours, you will always see a -1.. -2ppm decrease of reading in the first 8h, or so.
I first assigned that to a warm -up effect, but that's obviously a drift right after each ACAL, instead.
This behavior was not observed when the instrument was new, and I don't understand why this is happening.

As ACV, ACI and most DCI ranges were off by 2% even after DCV calibration, I re -assembled the LM399, put the box over it, and made another DCV calibration.
Again, many ACALs always brought the DMM back to nominal within 1ppm, w/o any shift.   

I cooled it down to 8.1°C, reading w/o ACAL was -14.5ppm, which gives a total of +1ppm/°C.
ACAL brought the reading to - 6ppm of nominal.
That gives an estimate for the resistor network of +0.6ppm/K, and -0.4ppm/K for the LM399.

I made another 12h stability run. Noise was evidently less than the unprotected LM399, about 650nV_rms, but it showed some popcorn noise, therefore having about 1.5 times the noise of the LTZ reference.
Therefore the fan, which is blowing to the outside, creates some noise on both references due to air draught, confirming TiNs hint about the blocked fan.

I will later measure the LTZs T.C. directly and if its calculated T.C. is confirmed, I will probably trim it to zero, assemble it into the 34465A and have it completely calibrated.
Myself I'm missing several stimuli, or a 5720A / 5725A to do that on my own.

Frank

[Andreas]

and of the LTZ circuit to +0.5ppm/K,

 and -0.4ppm/K for the LM399.

Hmm,

what do you mean when you say LTZ cirquit and LM399? is it including the 7 to 10V transfer?
Otherwise the values look rather high to me (near room temperature).

For the LTZ + the 5 Resistors I would expect below 0.2-0.3 ppm/K and for a
LM399 I have typical 2-4 ppm deviation over 10-40 deg C.
Question: which output voltage do you measure at the LM399 (is it far away from the sweet spot of 6875 mV ?)

with best regards

Andreas

[dietert1]

I think every engineer should be able to understand the difference between Dr. Franks LM399 and "the LM399", which is an abstract entity: a multitude of voltage reference parts made after a certain receipt. Dr. Franks part is one instance and he may have caused popcorn noise when unsoldering it, maybe without using his special copper plated tool.
If Dr. Frank wanted to compare "the LTZ1000" and "the LM399", he would have to compare at least several of those parts. In this sense his statements above are a bit subjective. I'd guess using the LTZ1000 in a mezzanine  and plugging the LM399 with its hot pins into that IC socket already causes confusion.

Regards, Dieter

[Tj138waterboy]

 Don't mean to hijack the thread but have also found another semi populated board on ebay  https://www.ebay.com/itm/1-Semi-finished-Board-Ultra-Precision-Voltage-Reference-LTZ1000ACH-LTZ1000CH-/333584715972?_trksid=p2349624.m46890.l49292
I was wondering if this board was designed by any members here as well if anyone sees any issues as pics show its populated with 5ppm .5% resistors.

[Dr. Frank]

and of the LTZ circuit to +0.5ppm/K,

 and -0.4ppm/K for the LM399.

Hmm,

what do you mean when you say LTZ cirquit and LM399? is it including the 7 to 10V transfer?
Otherwise the values look rather high to me (near room temperature).

For the LTZ + the 5 Resistors I would expect below 0.2-0.3 ppm/K and for a
LM399 I have typical 2-4 ppm deviation over 10-40 deg C.
Question: which output voltage do you measure at the LM399 (is it far away from the sweet spot of 6875 mV ?)

with best regards

Andreas

Hello Andreas,

this specific LM399 has a voltage of about 7.020V, and this LTZ circuit has an exactly 2% higher voltage of 7.161V.

I only can estimate the T.C.s from the different temperatures and the ACAL procedure, by neglecting the 7=> 10V transfer non linearity.. but as these shifts are in the 10ppm region, I would assume that that is far less than the 0.5..1ppm non linearity error of the ADC. The ACAL procedure is good for about 1ppm in the 10V range, so you could really have greater deviations of T.C.s..

Yes, the calculated LTZ T.C. seems to be really high, I also have never seen more than 0.3ppm/K, so please regard this as an upper limit/estimation.
At the moment I have no chance to give more accurate results.
I promise to determine it's real T.C. within the next few weeks. 

For the LM399 I have absolutely no alternative experience, but these 0.4ppm/K seem to be in accordance with yours.

Frank

[Dr. Frank]

I think every engineer should be able to understand the difference between Dr. Franks LM399 and "the LM399", which is an abstract entity: a multitude of voltage reference parts made after a certain receipt. Dr. Franks part is one instance and he may have caused popcorn noise when unsoldering it, maybe without using his special copper plated tool.
If Dr. Frank wanted to compare "the LTZ1000" and "the LM399", he would have to compare at least several of those parts. In this sense his statements above are a bit subjective. I'd guess using the LTZ1000 in a mezzanine  and plugging the LM399 with its hot pins into that IC socket already causes confusion.

Regards, Dieter

dietert1,

I don't think it's good practice that you again completely misinterpret my findings. Are you doing that intentionally? 

At first I have NEVER soldered on this LM399, and never made a statement that I needed to unsolder this device, because inside the original 34465A, it's ALWAYS plugged with its short cut legs into a precision socket. I have clearly documented this original socket, and the alternative, better socket I used, and I really wonder why you make such false statements, instead.

I also have NEVER made a claim that my findings are sort of a general ones, I simply reported what I found on my specific unit, and on my specific LTZ1000 parts which I have used.

Therefore this - sort of - popcorn noise I see on my LM399 is clearly specific to my unit, and I have again nowhere claimed any generality about this.

dietert, please publish your own findings on your own specific 34465A / 470A, that would help everybody else much more, instead of such grumbling.

Frank

[Dr. Frank]

The noise and stability at first sight is disappointing, 782nV_rms, apparently same figure as with LM399.

Did you check the noise of the LTZ1000A alone (with 1/f amplifier) before mounting into the 34465a?

with best regards

Andreas

'Aber isch 'abe gar keine 1/f Verstärker..'

Hello Andreas,
unfortunately, I don't have this low noise amplifier you're always using. I have to live with with the limitations of my equipment, but most important, to keep everything in a comparable manner.
Therefore I will measure the noise of this PCB compared to my 3458A, or differentially to another 7.161 V source, when I make the T.C. measurements.

Frank

[Kleinstein]

Seeing some popcorn noise with the LM399 is more the normal case. Selected ones may have need of the popcorn noise, but hardly none. With more other noise it may not be so visible in the initial curve, but it likely was still there and it looks like this too.

For the LTZ1000 reference it is not clear which pins are used for the negative side. It may very well be that the "wrong" (the one with the resistors and zener current) pin of the connector is used. As there is no other hardware change chances are high the normal ADC operation uses that wrong pin, possibly also in an original 34470. ACAL may use the separate negative side contact.
Depending on the way ACAL is done the contribution from using the "wrong" negative side may be seen as part of the reference TC or as part of the ADC's (resistors)  TC.

[maat]

Seeing some popcorn noise with the LM399 is more the normal case.

Indeed. From the last genuine batch (ordered from Mouser) , out of 25, I had to toss 10 due to popcorn noise.

[eplpwr]

I placed an order with Keysight for the board (USD441 with tax on my CC) and if their shipping is as I have had in the past, the board should arrive in a week or so.

I don't expect that it will come with its housing--that would be too nice--so perhaps someone will 3D print the enclosure parts? I'll update once the part arrives.

If you ordered 34470-66303 you get the board without the two housing parts. I ordered an original board 2 years ago, and it didn't come with any shields. When the clone boards appeared, I bought one to disassemble it and figure out the connections, like those under the 8-way connector. My (clone) board arrived days before christmas and has been sitting on a shelf, while I was away. Then, Dr Frank did most of the things I had planned to do, with stunning speed.

Regarding the shields, a modified version with room for better resistors would be most welcome. Maybe the bottom shield also would need a modified version, if thru-hole components are used for an improved board.

[Dr. Frank]

For the LTZ1000 reference it is not clear which pins are used for the negative side. It may very well be that the "wrong" (the one with the resistors and zener current) pin of the connector is used. As there is no other hardware change chances are high the normal ADC operation uses that wrong pin, possibly also in an original 34470. ACAL may use the separate negative side contact.
Depending on the way ACAL is done the contribution from using the "wrong" negative side may be seen as part of the reference TC or as part of the ADC's (resistors)  TC.

Exactly, using the correct pin for REF- is possible only if the 470A uses a special SW switch to take this into account.
If you simply plug the LTZ into the 465A, the instrument has probably no other chance than to use the LM399 REF- pin.
As the LTZ board features a GND current compensation via the 4k22 resistor to V-, I assume that this difference / error on the REF- pin can be neglected, because virtually no current is flowing out of /into this pin. To complete this picture, you might take the 34410A schematic of its reference assembly, page 7, and draw the connection to this LTZ circuit..simply replacing the LM399 zener.
Then you'll recognize, that the 2.7mA supply current nicely complements the 4mA which is required from V+, anyhow.

Hint: 34410/411A share the same Multislope IV converter with the 3446xA/470As, therefore schematics should be identical or very similar.

Frank

[View[+]Finder]

Hello Frank.
Firstly, thank you for the time you spent researching and documenting the possibility of making a '65 into '70 whether by way of China or in my case, by purchase from Keysight. The board should be in my hands next week, so before I do something stupid, I hope you and other contributors will take a minute to review my update plan.

My Plan:
1) From the photos posted it appears the the '65 board is populated with the through holes for the 8-pin connection, so that replacing 2x2 with 2x4 is sufficient modification.
2) Also, Frank's photo shows a new board installed, from which I infer that no changes to circuits with bodge-wire will be needed.
3) If I have read the comments correctly, the firmware (I updated to current version in December) will work with the new board.
4) A calibration may be required to account for differences in reference voltage, however this can be done without shipping to Keysight.(assume that skill, standards, etc. are available)

If anyone has a warning, suggestion or other tip, it will be appreciated. (like a source for the sockets, otherwise I'll find on Mouser)

Thanks in advance,
Donal

[Dr. Frank]

Hello Donal,
you need to desolder these 4pin female connectors and replace them with 2x4 male ones. That's pretty easy because the holes are big, but a bit flimsy to desolder these 4 lead free solder joints. I needed 3 attempts by feeding additional solder.
I used 8 pin female ones, due to using the LM399 again, but for the original board that's not optimal. see picture of the required pin connectors, on the left, these are gold plated standard ones
Just plug the KS board on the pins, no bodge required, also no SW change. Don't forget a proper air shield.
The original circuit heats up to 95°C,unfortunately,  therefore think about paralleling a 100k resistor or less to decrease the oven temperature.
This avoids melting the plastic, and gives much better timely stability. The interior doesn't heat up as the 3458A, so that's easy going.
Calibration for DCV, ACI, DCI and ACV is definitely necessary, lucky if you have all signals available

[TiN]

Melting plastic? LTZ1000ACH barely gets warm thanks to it's blessed bubble foam die attach. 

I decided not to return the module, it's just not worth it and will drop my both reference modules into tecbox to run some tempco slopes. I'll keep one original (after cleaning crap with IPA) and one will modify with Susumu 5ppm/K resistors.

[Dr. Frank]

Melting plastic? LTZ1000ACH barely gets warm thanks to it's blessed bubble foam die attach. 

I decided not to return the module, it's just not worth it and will drop my both reference modules into tecbox to run some tempco slopes. I'll keep one original (after cleaning crap with IPA) and one will modify with Susumu 5ppm/K resistors.

Hi Illya, you're probably correct.
Anyhow HP and also National Semiconductor used a high temperature plastic for the LTZ and LM 399 at 95°C. polysulfone (up to 150°C, glass temperature 190°C) is specified in the LM 399 data sheet or white paper, afaik.
The interior might heat up higher , even if the surface is luke warm only.. and polystyrol or other plastics will disintegrate over longer times, that's the reason why they are using these special plastics.
I've seen many XTAL ovens and similar, where the styrofoam was simply toast after 20..30 years.

Good to hear that you're also experimenting on this PCB, Susumu 5ppm/K sounds good.
Maybe that's also a recommendation for others who bought this PCB, or for pimping the original KS one,

regards Frank

[View[+]Finder]

You need to desolder these 4pin female connectors and replace them with 2x4 male ones. That's pretty easy because the holes are big, but a bit flimsy to desolder these 4 lead free solder joints. I needed 3 attempts by feeding additional solder.
I used 8 pin female ones, due to using the LM399 again, but for the original board that's not optimal. see picture of the required pin connectors, on the left, these are gold plated standard ones

Hello Frank,
I missed seeing the gender switch, thanks for calling it to my attention. Turns out I have some gold plated round pins left over from building a test fixture for some 3458 ref boards. Once the KS board arrives and I open up the '65 I'll have a better idea of what space is available. Might it be possible to trim four pins to fit the existing female sockets with the other four soldered to the board? We'll see . . .

Any reason for round to be better than square?

Thanks again,
Donal

[Fidel]

Here are some ~50mp images of the 34470a rev3 reference board if this provides some help for anyone, including a close up of underneath the female header.

I can also confirm this successfully works in our 34465a,  but I replaced R4/R5/R91 with some nice VPG smd foil parts 

[View[+]Finder]

The reference board I ordered from Keysight arrived about three hours ago and I could not resist the urge to put it to work. As Frank reported, unsoldering "lead-free" is a task that requires patience, for me at least an hour, and a good solder-sucker. I fabricated some gold pins (see photos attached) and fitted them in the new board to act as a fixture to aid alignment for soldering. Soldering and cleanup went as usual. A smoke test-before reassembly went without incident.

I removed the built in fan (temporary) and did not install any protection other than replacing the metal shields that surround the reference. I did an ACAL and hooked up a very reliable voltage source (KY 230) just to see how VDC compared. The '65 had been calibrated by Keysight a couple of years back. Now it is about 130mV too low over range 1 VDC to 100VDC consistently. Should be no problem getting it calibrated.

Many thanks to forum for help with this project. Once the calibration is completed, my plan is to do some analysis of measurements and post the results.

Cheers,
Donal

[View[+]Finder]

The new ref board was installed yesterday and ran overnight connected to a 10VDC reference source. Calibration started about 10AM and finished around 12:30PM. Note: I did not run the full calibration and I did not run it in the sequence suggested. The ADC cal and zero cal were done first, followed by all the low-voltage VDC cal steps. The I stopped to check progress. DC Volts looked good and I ran the 100VDC cal, no problems. Next came the ohms cal followed by DC current cal.

At first I was discouraged by the "dissing" of the Fluke 5720A in favor of the "more accurate" Keysight 33500B series arb function generators to generate the "Frequency Accuracy Calibration" of a 10KHz sine wave at "9 to 10 VRMS." Then I used my little SDG1032 (synced to Rh time) at 7 VRMS and the cal was accepted. Then the low freq cal and finished off with a shot of 500 VDC to close out the HVDC cal.

Left for another day are all the AC flatness curves, AC current and DC amps on the 10A scale.

The work-in-progress is an overnight face-off between my 3458A and the 34465+ both connected to the same 10VDC reference. I'm capturing internal temperature and volts as well as the temperature in the lab at various locations.

More to come . . .

[TiN]

Looking at the photos, I'm concerned of using massive brass pogo pins for precision DC voltage interconnect.
Worth to check tempco on DCV as it could be affected. Also adjusting meter without at least few weeks of warmup for new reference is somewhat wasted effort, other than to check that adjustments still work.

[View[+]Finder]

Hello Illya,
The pins are similar, if not identical, to the ones used in the 3458a and are not as massive as they appear in the photo. I have some a size smaller, however they seemed a bit loose fit in the ref board. Replacement is certainly possible, what would you suggest as a way of measuring the effect of temperature changes beyond correlating voltage measurement and internal temperature over an extended period of time?

Yes, the sole function of the meter for at least the next week will be to verify whether the new board yields better results against the specifications and I rely on the experience of users such as you and others on the forum for advice.

I’ll post the overnight run later today.

Many thanks,
Donal

The new update is an overnight comparison of voltage from a calibrated 10VDC reference connected to both a Keithley 6510 (KY6510) and a Keysight 34465A (KS34465). The test ran from roughly 21:00 on 20210106 to 10:00 on 2021017 over which time the temperature ranged from 24C down to 21.8C and up to 22.9C this morning. Samples were taken at NPLC=10 on the KS34465 and NPLC=15 on the KY6510.

[TheSteve]

Are those pins actually pogo pins or do they just look like it?

[View[+]Finder]

Are those pins actually pogo pins or do they just look like it?

Pogo? No go. The pins are designed to be soldered to wire in a mil spec type cannon connector. There is a shoulder on the pin that, in this implementation, acts as a stop for the ref board. I soldered some cut off pins in place of the wire and fitted these through the eight holes in the MB. Once soldered in place, the pin ends were trimmed leaving the ref board capable of being removed if necessary.

Some explanation of photos posted above:
1) Before, so I would know which way the old ref went
2) Closeup of sacrificial pin being soldered
3) The bottom side of the ref board, the pins go into the holes on the MB
4) The pins sticking out of the underside of the main board, ready for solder
5) The new ref board in place--see the gold pins?
6) The mounting pins with the board removed--see the stops?

Yes, they look massive--not so much so when trying to handle them with fumble-fingers.

[TheSteve]

OK, in the close-up of the bottom of the ref board they look like POGO pins that have had solder added(at the shoulder) so they don't compress anymore.

[View[+]Finder]

OK, in the close-up of the bottom of the ref board they look like POGO pins that have had solder added(at the shoulder) so they don't compress anymore.

Exactly. Minus the spring and the detent. No, just gold pins to give me something to solder in place. Thinking back, I could have used wire, but then what about dissimilar metals and you know someone would call me out on that . . .

[View[+]Finder]

KS34465 with '70 ref board. Latest results from overnight testing . . .


Another test today, 2 hours of data

UPDATE: invalid data, plots removed

[Dr. Frank]

Please indicate, how you have set up this measurement:
-What is the 10V reference?
- Have you connected 34465A and Keithley 6510 in parallel to the 10V reference? (picture of setup?)
- What is the NPLC setting on both?
- how are the data points connected ? (please show a drawn line between them)

There seem to be some sort of oscillation from one value to another.. there's something fishy, either with your 10V ref (not stable) or your setup (ground loop or interference from one DMM to the other)
Frank

[View[+]Finder]

Please indicate, how you have set up this measurement:
-What is the 10V reference?
- Have you connected 34465A and Keithley 6510 in parallel to the 10V reference? (picture of setup?)
- What is the NPLC setting on both?
- how are the data points connected ? (please show a drawn line between them)

There seem to be some sort of oscillation from one value to another.. there's something fishy, either with your 10V ref (not stable) or your setup (ground loop or interference from one DMM to the other)
Frank

At TIN's suggestion I double-checked everything. It looked like there was a problem with the reference (under-volt supply, my fault), however I have just now isolated the source of the divergent data to a mishandling of plotting in the spreadsheet program. I'm working with over 50,000 observation for each data series and using a "line plot with markers" as a starting point with the lines being removed later for clarity. To investigate, I switched the plot type to "2D Scatter" and the divergence disappeared.

That said, the 34456 is showing more drift than in earlier testing and much more than the 6510 on the same reference. The reported temp for the 34465 is 30.7C, the lab is 24.2C, the thermistor in free air 5cm above the case is 32C. I also have an old-style glass lab thermometer stuck inside the 34465 case through the fan opening: it reads 43C with the Hg bulb in the area of the new '70 ref. Temperatures have been stable today.

Graphs and photo of setup attached.

Thanks for your help,
Donal

[TiN]

Oh, a photo of a setup I was so waiting for
So many wrong things on the photo, that hard to pick one to start with.
Probably first - toss the blessed awesome14 "reference" into a bin where it belongs. It was a laughing stock some years back. Don't get me wrong, LT1021 is a nice chip and nothing wrong with it, but it's not adequate to characterize 34465A's performance, even with LM399. I would suggest to build simple LM399A reference (there are many designs available in open hardware) and use that to test 34465A mods. Or if you want something already proven and finished, get one of Ian's PDVS2mini units, which can provide you with adjustable low-noise source up to 10VDC.

Then your wiring need work. Ditch any adapters and bananas where possible. Both DMMs should connect to reference source using own cable, not daisy chain from one meter to another. While usually it works fine at 6.5-digit level, its just bad habit. For most sensitive cases (e.g. unbuffered zener reference) even fancy 8.5 digit meters with AZ may cause issues when used in parallel.

Put reference in some isolated box, that will help to dampen thermal variations a bit. Also since you have another DMM6500 - collect ambient temperature and plot graphs together so it's possible to observe presence/absence of possible correlations.

Your data shows 22 ppms change over 70000 points (hint for future plots - use time scale, not arbitrary samples), which hints that something is very very wrong.

[View[+]Finder]

Oh, a photo of a setup I was so waiting for
 get one of Ian's PDVS2mini units, which can provide you with adjustable low-noise source up to 10VDC.

Then your wiring need work. Ditch any adapters and bananas where possible. Both DMMs should connect to reference source using own cable, not daisy chain from one meter to another. While usually it works fine at 6.5-digit level, its just bad habit. For most sensitive cases (e.g. unbuffered zener reference) even fancy 8.5 digit meters with AZ may cause issues when used in parallel.

Put reference in some isolated box, that will help to dampen thermal variations a bit. Also since you have another DMM6500 - collect ambient temperature and plot graphs together so it's possible to observe presence/absence of possible correlations.

Your data shows 22 ppms change over 70000 points (hint for future plots - use time scale, not arbitrary samples), which hints that something is very very wrong.

Hello Illya,
Thanks for your good advice and also your opinion on bananas (when I get time, I have copper fork terminals and 4-conductor twisted to build proper cables), however, let's not let "perfect" defeat "good enough" just yet. I've been an Ian Johnston fan for years and have his first model as well as his improved version. The mini is a cheaper version of the latter, right? I switched to the Calibratory ones for convenience, I'll try Ian's  . . .

Right now my objective is to determine whether the 34465 with the '70 reference has an error that can be remediated by a proper plastic shield as Dr. Frank suggested or your suggestion that the connection pins might have "tempco" issues. Clearly the 34465 data is out of line with that from the 6510 by an about that might be traceable to "bad wiring"
so it will be easy to test that.

As for the quality of the references, I use the "good, bad, ugly" rule and try to eliminate the "bad" if I can. If you have a suggestion or specification for a reporting format, time scale, etc. that would be helpful. I come from a world of standard deviations, skewness, kurtosis  and R-squared measures to characterize variance in data. PPM seem to hide some easily explained risk in an aggregated statistic.

Thanks again,
Donal

[Andreas]

I come from a world of standard deviations, skewness, kurtosis  and R-squared measures to characterize variance in data. PPM seem to hide some easily explained risk in an aggregated statistic.

Hello Donal,

I think that you will find the Allan Variation as useful tool to describe stability.
There is several Software around which calculates the stability diagrams like
"Plotter" from Ulrich Bangert or "Stable".

Plotter is available e.g. here:
https://www.bartelsos.de/messtechnik/df6jb

with best regards

Andreas

[View[+]Finder]

I come from a world of standard deviations, skewness, kurtosis  and R-squared measures to characterize variance in data. PPM seem to hide some easily explained risk in an aggregated statistic.

Plotter is available e.g. here:
https://www.bartelsos.de/messtechnik/df6jb

That program is exactly what I have hoped to find! Most commercial offerings these days are either bloated (Excel), limited (Numbers from Apple) or slow (Libre Office) and their graphics look like their customer in still in short pants. A quick read of the Plotter documentation got me on the path. I'll download and test ASAP.

Herzlichen dank,
Donal

Sad note: the creator of the Plotter, DF6JB, went silent key in 2014 at age 59. Great program, thanks from KF6OXI

[View[+]Finder]

Here's a run from yesterday afternoon pitting the 34465A with the "Silver" reference (not a favorite of TIN) against the 6510 measuring Ian Johnston's PDVS2. Today, I'm switching the references for another run.

Ian's reference is looking pretty good, I know how diligent he has been in his effort to create a reliable voltage reference for a reasonable price.

[View[+]Finder]

The latest for my KS34465 with the '70 reference from Keysight.
Following Dr. Frank's example, I fabricated an enclosure for the reference board (top and bottom). After a day turned and connected to a 10VDC reference (HP3245A), I created a python script for testing. One hundred readings for taken about 30 seconds apart. Each observation comprised an autocal (*CAL?) command, 20 seconds 'sleep', a query of internal temperature, a query of DC voltage (10V range, auto zero ON, NPLC=10), a record of readings and finally a 5 second sleep. A total of 100 observations over about 50 minutes between 8:40 and 9:20 this morning.

Attached are a plot of measured voltage vs time and a plot of temperature vs time. The plots show a lack of correlation of voltage with time and only a 0.3C range in internal temperature. In short, noise. My next test will involve changes to cables and reduction of EMF sources to see what effect that has on noise.

Some help would be appreciated.
What is the source of the internal temperature measurement in the 34465? What about when the '70 ref board is installed?
I have a thermistor 5cm above the 34465 case, should this be incorporated in testing or is the internal measure sufficient?
I know from TIN and others that LED lights are an EMF source and twisted Kelvin with fork spade copper connectors are preferred, anything else?

Thanks in advance,
Donal

[Kleinstein]

A cycle with ACAL will add some extra noise, as this involves extra convertsion to measure the scale factor. I don't know how long ACAL actually takes, but one may have to extect soemthing like 0.1 ppm "Noise" for an ACAL call, as this would be still better than the nominal resolution.

To check the reference dirft, one should have a run without ACAL calls an with a stable temperature. If at all one would call ACAL in longer intervalls, like every 2-6 hours. Ideally one would than have enough data before and after to calculate back the jump from ACAL.
Chances are one would want to full resolution / minimal noise. So there should be more like contineous readings for some 100 PLC (maybe extra averaging of shorter readings).

The HP3245 is sopposed to use a LM399 ref. So don't expect anything very low noise. It may still be a selcted (e.g. through out the really bad ones) one. This ref would still be not really good enough to fully judge the reference drift or noise.

[maxwell3e10]

Regarding enclosure to prevent air currents, I had some good luck just using tape over the analog section of the meter
https://www.eevblog.com/forum/metrology/turning-fan-off-to-reduce-dmm-voltage-noise/msg2652360/#msg2652360

[View[+]Finder]

A cycle with ACAL will add some extra noise, as this involves extra convertsion to measure the scale factor. I don't know how long ACAL actually takes, but one may have to extect soemthing like 0.1 ppm "Noise" for an ACAL call, as this would be still better than the nominal resolution.

Autocal on my 34465 takes 10 seconds and another 5 seconds for the "OK" message to clear. The Python script sleeps for 25 seconds, however I can't tell whether this includes the autocal time. In any case, the DMM is quiet for at least 10 seconds before a measurement is taken.

At your suggestion, I will try interspersing autocal every 10 measurements and see if that makes a difference.

[Kleinstein]

ACAL after only 10 measurements (still depends on the integration time) would be still very fast and likely a little to short to really see the steps.

ACAL likely does quite a few things (different ranges, maybe even some AC) so there would no be not so much time for the gain calibration for the 10 V range.
One may want a little waiting ACAl for settling - even though there seem to be no such problem as with te 34470. The 5 seconds waiting for the Ok to clear may aready be the settling time for this.

To get most data and thus least noise I would skip the sleep part, but take more data. If looking for possible popcorn noise of the references, it may help to no average to much and accept a longer file. 10 PLC data are probably a good compromise to look for noise from a LM399 or simialr ref. . One can allway avearge later, but there is no way back to finer time resolution.

[View[+]Finder]

Just to confirm, ACAL is the program that I run on my HP 3458a as suggested before starting a set of measurements. It runs for quite a while and tweaks everything to the internal temperature. The 34465 has an AUTOCAL command that takes only seconds to complete. I don’t know exactly what it does, however I assume it is intended to compensate for temperature changes as well.

Last evening I started another run with autocal every 100 readings. It is at about 8,000 of 10,000 observations now and should complete in a few hours. Based on the data so far, there has been a decline in temperature (nighttime normal) and a rise in voltage with the change in voltage 90% explained by temperature.

Post results later . . .

[View[+]Finder]

The data capture (7,766 observations) started at 19:30 last evening and is described in an earlier post. Two graphs are attached. The 12.26.02 shows the internal temperature of the 34465 (X) against measured voltage (Y). The temperature decreased from 30.5C to 25.0C (approx) over the night--the (X) values are in the same order as time.

The second graph,12.24.35, shows voltage (Y) over time (X) as represented by the observation number. The sawtooth pattern is an artifact of the operation of autocal every 100 observations. In effect, the steady decrease in temperature resulted in a recalibration and a concomitant increase in the reported voltage measurement.

I'll leave it to those who know about metrology to help us understand what's going on.

[Kleinstein]

The first curve shows a relatively linear temperature dependence. It looks like a TC of some -0.6 ppm/K . This is little high for just a LTZ reference, but not that extreme. Part of the problem could be how the neg. side of the ref is sensed - not sure if the extra signal is detected. Chances the same pin is used for the normal ADC converstions and only Autocal may look at the different pin.

The curve over time shows quite some effect of the ACAL call.   Here it does not look like ACAL correctig temperature drift, but more like some settling after the ACAL call - so a little like the ACAL problem of the 34470 !

[View[+]Finder]

Just finished reading all of the '70 ACAL error posts--some really good work was done by forum members on this issue! Dr. Frank's post points out the need for a full range DCV calibration to get the ratios corrected. Looks like that is my next task.

Thanks for helping,
Donal

[Dr. Frank]

Just finished reading all of the '70 ACAL error posts--some really good work was done by forum members on this issue! Dr. Frank's post points out the need for a full range DCV calibration to get the ratios corrected. Looks like that is my next task.

Thanks for helping,
Donal

Exactly. It was also my main goal, to search for the root cause of this ACAL failure by inserting the LTZ reference into the 465A.

Your continuous ACAL engagement during your measurements also does not seem to provoke any shift errors.

For targeted stability analysis, i.e. over temperature or time, you should better do appropriate test runs w/o ACAL.

I also estimated a too high T.C. of about 0.5ppm/°C for the LTZ board, this will be directly tested on the  board, outside the 465A.
Maybe you could also do the same with your original reference board.

Frank

[View[+]Finder]

Just a general FYI

When extracting data from Keithley 6510 and 6500, I use the web interface over ethernet and the measurements are 13 digits for voltage. With the Keysight 34465, I use a USB stick and get 9 digits for voltage--the same when using SCPI in a Python script. By comparison, my 3458 provides 10 digits in a Python script.

Questions:
What happened to the data dump software for the KS344xx DMM's? I used it a lot in years past and now can't find it. (ditto, my glasses)
Has any one had a look at the hex code for the firmware to see what differences might be in place for the several meters?
How many digits of actual information are in the SYST:TEMP reading on the 34465? To me it looks like everything past ##.# is suspect.
Is it worth the trouble of putting a thermistor (the little silver cylinder, not a chip) in the 34465 to get a reading?
If the 34465 main board is the same as the one in the 34470, how do we know that the voltage ref connections are ok for the '70 ref mod in the 34465?

Whew . .

[TiN]

eBay 34470A module temperature coefficient benchmarks

We often hear from gray bearded metrologists and analog designers’ importance of using stable and good components in analog circuits, especially voltage references. Hundreds of hours were spent in debates about voodoo PCB slots, swastika-shaped LTZ fanouts, vacuum-insulated thermal chambers, cadmium-enriched solders and quadruple shielded PTFE wiring. But often such discussions lack of any quantitative performance data to actually show benefits of all that effort. Here we address this important issue in terms of thermal stability measurements of voltage reference module shown above.

Keep in mind, reference module tested here is NOT manufactured by Keysight Technologies. This is aftermarket clone copy of original design, with subpar assembly quality and unknown source of the parts. That said, I do not have 34470A original module to compare 1:1, but based on my experience with LTZ circuits I do expect similar results.

Equipment used to perform experiment is ubiquitous characterized Keysight 3458A 8½-digit voltmeter, DIY temperature chamber with 40W Peltier heater/cooler, Keithley 2510 TEC controller, calibrated Fluke 1529 Chub-E4 with Omega RTDCAP-100A-2-P098-050-T-40 sensor and our own xDevs.com teckit Python application to control all equipment and process data into DSV-file for further analysis.

Temperature in chamber was swept gradually from +18 °C to +50 °C with speed around 0.03 °C per minute with static step at 50% of sweep (+34 °C) for 1 hours and 1 hours for peak temperature. Ambient room temperature and humidity stability was maintained at +23 °C ±1°C and 15% RH ±2% respectively.

Reference was powered with Keithley 2230-30-1 linear power supply with supply voltages +12 and -12 VDC. Current consumption from module was ~26-29 mA on positive rail and about 3mA on negative rail, which is typical and expected for LTZ1000A datasheet circuit. Module was assembled with LTZ1000A chip manufactured in 2018 week 38 with legs cut flush to the PCB surface. None of this voodoo LTZ standing on long kovar legs rubbish . Resulting plot reveals output voltage variation with change of temperature inside the airbath box.



Reference performance was quite bad +0.114 ppm/°C (coefficients &alpha; = 0.107 ppm/°C and &beta; = -0.0003 ppm/°C^2) with onboard resistors as is shipped from eBay, as I already expected. This is about 3 times worse than good LTZ reference should have.

Good LTZ1000A design capable to provide thermal stability better than 0.05 ppm/°C and this is what I have used as a target during testing and tweaking of all my xDevs.com KX, xDevs.com FX 10V and xDevs.com QVR low-noise DC Voltage references. With some effort stability better than 0.03 ppm/°C can be achieved from typical datasheet circuit, without slots, cadmium solder or trimming LTZ1000A kovar legs trimming to micrometer precision. After all, LTZ1000A is power in/power out device and require just careful system design.

Typical 1% SMT chip resistors as used by seller have temperature coefficient from 100 to 250 ppm/°C, not the 2-5 ppm/°C expected by LTZ1000A reference. Now idea is to replace cheap resistors with something much better and repeat same temperature stability test to narrow down if the excessive tempco caused by resistors or something else like poor PCB component placement, SMT-type LT1013 or routing of the copper traces around.

I bought some reasonably-priced Susumu SMT RG and URG series resistors in 1206 size, specified to have tempco better than 2 ppm/°C.

Digikey part number	Susumu part number	Resistor description	Price
408-1788-1-ND	URG3216L-101-L-T05	RES SMD 100 OHM 0.01% 1/4W 1206	$8.38 USD
408-1645-1-ND	RG3216L-102-L-T05	RES SMD 1K OHM 0.01% 1/8W 1206	$5.20 USD
408-1648-1-ND	RG3216L-153-L-T05	RES SMD 15K OHM 0.01% 1/8W 1206	$5.20 USD

Of course, there are even fancier metal foil VPG resistors such as VFCP Y1630 (over $11 USD/pcs) or FRSM Y4023 (over $18 USD/pcs) but in standard LTZ1000A circuit I doubt we will see much difference in DC Voltage output temperature stability compared to more affordable Susumu 2ppm/°C parts. Long-term stability is another aspect, but its completely out of scope of this article. In the end of the day Keysight 34470A is a benchtop general-purpose DMM and it is not suitable or supposed to act as a laboratory DC Voltage reference equipment.



Pay attention to twice smaller scale on vertical ppm axis. Benchmarks show significant 257% improvement after simple resistors swap. Now reference performance matches expected LTZ1000A circuit capability, with box tempco just +0.046 ppm/°C and slope parameters &alpha; = 0.0431 ppm/°C and &beta; = -0.0005 ppm/°C^2

Same data presented in RAW timescale format also presented below, together with some resistor tempco runs. I often run multiple experiments and measurements at same time and combined plots showing multiple unrelated values, which my friends and people can find often confusing. Our DUT +7V reference measurement presented in olive color chart line on this example plot. This is a data from a temperature run with Susumu resistors and presented to show the timing of the experiment. Temperature used for tempco calculations is shown in maroon color (from Fluke 1529 thermometer).



P.S.To provide justice for the seller, I have contacted seller about misleading resistors discrepancy and after discussing performance drawbacks of the unit's resistors as in eBay listing (with photos at time of purchase, seller replaced photos with cheap resistors already) seller issued full refund on my two modules. Buyers beware, as with everything that comes from secondary market!

[View[+]Finder]

Attached are plots of supplemental testing of the autocal sawtooth in my 34465 with a Keysight '70 reference. The top plot was captured between 9:00 and 10:30, the middle one between 11:00 and 12:30 and the lower one between 12:30 and 14:00. As suggested, the last one was done without intermittent autocal. Autocal was run at the start of the sequence with a sleep of 25 seconds prior to measurements. The voltage source was the PDVS2 on battery power from 16AHr lithium 4S.

[View[+]Finder]

"Temperature in chamber was swept gradually from +18 °C to +50 °C with speed around 0.03 °C per minute with static step at 50% of sweep (+34 °C) for 1 hours and 1 hours for peak temperature. Ambient room temperature and humidity stability was maintained at +23 °C ±1°C and 15% RH ±2% respectively. "

Hello Illya,

You show temperature at alpha=zero at 237.565C. That's hot! Is this a result of the quadratic regression?

Why did you use quadratic on data that look amenable to linear regression with a pickup of one degree of freedom in the regression?

By definition, one degree in Celsius is equivalent to one Kelvin. In other fields of science, temperature-related comparisons are based in Kelvins; why not here? Does it matter?

If anyone would like a raw data file for any of of my temperature/voltage measurements related to the Keysight reference, I would be happy to post here. Reciprocity would be nice.

[TiN]

Using Celsius scale is just convinience in regular room-temperature metrology, so we don't have to say that temperature was changed from 292.15 K to 323.15 K, while room temperature kept at 296.15 K?

As for math, I used same approach to calculate alpha and beta coefficients just as it is common and well understood in resistance metrology.

This particular LTZ1000A reference does look fairly linear but it is not given and many other (LTZ-based) references are not as linear. Quadratic law fits bigger picture better. Sure, +237.6 C "zero" point does not have significant value here, more of an artifact number from my automated math script. 

If you want to share RAW data, it is always good to include more details about actual experiment and methodology. I was interested in tempco of this 34470 clone and impact of different resistors on tempco results. Might try some other resistors in future as well, but for now I got my answers here. Full disclosure : don't have any 344* LCD-based instruments in my lab to contribute anything else or make tail or horns out of ACAL tests posted above and alike, sorry.

 To me adding extra delays and pauses or running ACAL so often (which is not ACAL like on 3458A) sounds bit strange. Typical experiment run is using DMM for logging data for some time, and then if temperature or condition changes a lot, perform corrective ACAL. Usually always try to collect at least day or more of data when its about stability or noise parameters. But without more clear idea what you testing or expect to see in results it is hard to suggest anything specific.

One important thing forgot to mention - measurements of the 34470 module output were made with wires (both power and LTZ output) directly soldered to the module. Original connector with female sockets was desoldered. Voltage output wire was twisted pair PTFE-insulated copper with shielded connected to GUARD port of 3458A. Cable on the DVM end are terminated to bare copper spade lugs. LTZ1000A chip was covered with some packing foam on both sides with help of kapton tape  . Typical stuff when dealing with LTZ1000A-designs.

[Kleinstein]

For the test of the reference in the 34465 it makes sense to have both a run without any autocal calls and a rund with a relatively frequent (though ideally not as often as shown - need to have at least enough time to fully settle) autocal calls. One wants a low TC / low drift in both cases. The case without ACAL is limited by the ADC gain but still needed as normally one would take a curve without frequent autocall calls. The case with autocal would be relevant for the long run. The difference would tell on how much drift is from the ADC gain plus maybe taking the reference low side from the "wrong" pin.

The slightly higher than normal TC may be due to not so good resistors, but also from the way the negative side is sensed.  With the separate pin Dr Frank may be able to a test where the extra ref- signal is not connected - so one would be sure the DMM would use the normal ground like with the LM399. If there is no difference, chances are the meter does not use the extra ref.- pin.

[View[+]Finder]

The case with autocal would be relevant for the long run. The difference would tell on how much drift is from the ADC gain plus maybe taking the reference low side from the "wrong" pin.

The slightly higher than normal TC may be due to not so good resistors, but also from the way the negative side is sensed.  With the separate pin Dr Frank may be able to a test where the extra ref- signal is not connected - so one would be sure the DMM would use the normal ground like with the LM399. If there is no difference, chances are the meter does not use the extra ref.- pin.

The schematic posted in this forum shows the -REF LTZ connected to ground on the ref board and J1 P5 connected to ground, presumably on the mainboard. I agree, the reference is probably operating as designed, albeit in a different venue.

I have a longer (24 Hour) test underway and will post here when completed. Preliminary results show PDVS2 voltage stable in a less than 10 micro-volt range over a 12 hour period when measured internal temperature decreased from 29.5C to 26.5C.

In the meantime, I could use some advice concerning my newly acquired Keithley 182 Sensitive Voltmeter. How sensitive? Well, really too sensitive and embarrassed that I have posted the attached photo of its input jack stripped down to show all. Noted as made of unobtainium, is it possible that the pins could be available somewhere? If no one offers a source, I can make them on my lathe. Which leads to my second question: what is the material? I presume some sort of copper alloy? I don't want to waste time making something unsuitable for the next phase: low voltage testing on the 34465 '70 Keysight ref.

Illya, if you are done LoL at my pure copper short, how can I make a proper one?

[TiN]

Would recommend radical method - before you spend hundreds $-$-$ ditch the obsolete unobtanium connector on 182 and replace it with something modern like LEMO connector used in 2182/34420A (still few $-$-$, but at least you are more futureproof). But I'm lost on what you planning to do with nanovoltmeter and LTZ reference? 
Connector is primary reason why I nearly never use my 182.

My PDVS2mini (prototype 5, if I remember right) has very good tempco and p-p noise under 5uV.

[View[+]Finder]

Would recommend radical method - before you spend hundreds $-$-$ ditch the obsolete unobtanium connector on 182 and replace it with something modern like LEMO connector used in 2182/34420A (still few $-$-$, but at least you are more futureproof). But I'm lost on what you planning to do with nanovoltmeter and LTZ reference? 
Connector is primary reason why I nearly never use my 182.

Good advice! I watched your video on the battery replacement, which led to having a look inside my 182-M--missing its battery--now it has one. Thanks. I also have a 181--that how I came by the connector--so I'll look for the LEMO connector for the 182.

Nanovoltmeter and LTZ ref? I might be "lost" as well. I was thinking about doing some millivolt verification on the PDVS2 (presently connected to the 34465) using the 182, however looking back, it was a dumb idea. Using an unknown 182 anywhere in the process of testing an unknown implementation of the Keysight ref in the 34465 just add to confusion.

Thanks for the assist.

[MiDi]

Noted as made of unobtainium, is it possible that the pins could be available somewhere? If no one offers a source, I can make them on my lathe. Which leads to my second question: what is the material? I presume some sort of copper alloy?

K-181/182 Connector:
Amphenol 97-3106A-16-11P & 97-3057-1008-1 only awailable with silver plated pins.
For nickel finish add (689) to order number.

[View[+]Finder]

Preliminary results--overnight 21:00 to 09:00 on Sunday. PDVS2 set at 10VDC into KS34465 w/ '70 Keysight reference.

[Kleinstein]

The curve looks good for a LM399 base reference. It does not tel very much about the meter or LTZ based refernce - the LM399 alone could explain all the noise. The temperature coefficient looks good, but the temperature range is to small to definitely see something.

[View[+]Finder]

The curve looks good for a LM399 base reference. It does not tel very much about the meter or LTZ based refernce - the LM399 alone could explain all the noise. The temperature coefficient looks good, but the temperature range is to small to definitely see something.

Good point about the source of noise.

My reasons for buying a reference from Keysight rather than the China knock off stemmed primarily from having a reliable baseline comparison installed in a 34465 as I knew there would be interest on this forum. And also curiosity, sloth (didn't want to DYI), lust (Keysight delivered in less than a week), maybe getting better performance from the meter. So my focus was on limiting the effect of temperature on the voltage measurements while determine whether the replacement of the reference was beneficial.

TIN showed an excellent test across a broad range of temperatures for both the China ref and the PDVS2 mini. Is it safe to assume that the "best case" for performance of a 34465 with a '70 reference from any vendor is the performance of a Keysight 34470? From your comment, the voltage source should be two levels above the meter--9.5 digits for a 7.5 digit meter.

I would love to have a 9.5 digit voltage source . . . I have an option 2 board for a 3458A wired up in a metal box that has been run for a few months. I haven't verified performance on it, however I'm sure others on the forum have tried this as a reference. Time for some research . . ..

[Kleinstein]

No need to have the external reference much better than the one to test. The same grade reference would be about the minium - higher grade is better, but not that easy at that level.
An extrernal source of lower quality make the life very difficult and limits the result.

With the upgraded reference I would expect the performance of the 34470 at best - though by luck the DIY thermal cover may be better than the original. There may still be some other higher grade or better tested in the '470 - not just the reference.
I have a slight suspicion that there may be some mathematical correction active in the '470, that is not used with the '465 and this may be the part that sometimes / after some drift in other parts cause the ACAL problem.

The resolution / noise of the 34470 also limits on how easy a test of the refernce can be made. There are parts in the ADC circuit that act like reference noise (e.g the first OP where the ref signal goes to).

[View[+]Finder]

Attached are plots showing "Serial Correlation of Voltage Readings from an LTZ1000 Reference (HP3458 A9)" as measured by an HP3458 and for comparison, a KS34465 with a genuine Keysight 34470 reference. The testing comprised about 14 hours of data collection. Discussion and setup photos here:

https://www.eevblog.com/forum/metrology/making-do-with-what-you-got-an-hp-3458-opt002-reference-with-no-heater/msg3430502/#msg3430502

From what I can gather from the time spent installing testing, adding the official '70 was worth the effort. (mostly for the fun of it; if someone were paying me, buying a 34470 would be the better choice.)

Oh, the plots on top of the three PPM plots are of the regression residuals. ("Nothing to see here folks, move along . . .")

[branadic]

I had a play with our 34465A at work and measured some LTZ1000 on it to see, if this unit suffers from popcorn noise and indeed it does, about 2.5 ... 3 ppm. So I followed Frank's advice and covered the area around the reference with foam to prevent any air current. The unit is currently warming up and then we will see, if that improved anything.

-branadic-

[branadic]

By the looks of it the foam slightly improved noise by preventing air current, but other than that this LM399 is pretty volatile and suffers from strong popcorn, worth to throw it away.

-branadic-

[Kleinstein]

The amplitude of the jumps looks pretty high. The expected size would be some 6 or 8 µV and not 20-25 µV. It could also be something different from the normal popcorn noise (e.g. contact problem, or maybe a cell phone).

So this may indeed be a reason to look for an exchange - maybe even one of the new ADR1399 as a slight upgrade if the current provided to the reference is high enough.

[dietert1]

In other discussion it was proposed that LM399 popcorn noise may be caused by minor movements of the plastic hat and/or by RF instability caused by directly wiring a resonator made of traces and a capacitor parallel to its output. The two states would then be oscillator/non-oscillator.
Another difference between LM399 and LTZ1000 with its external temperature control is that one is a P controller, while the other one is a PI controller. I don't think the integrated temperature control of the LM399 has an integrator as a proper capacitor doesn't fit the scheme.

Regards, Dieter

[branadic]

It's neither a contact problem (checked that when I installed the foam, by removing the reference from its socket) nor a cell phone. But I agree, this jumps are rather high, even for the average LM399.
A moving cap can't be the case here, the legs are pretty short and the reference originally sits in this "precision sockets" so that the cap touches the socket and can't move. Also with the foam the cap is prevented from moving.
RF instabilities should have been reported by other customers before I guess, if that was the problem.

ADR1399 would be an option, as Frank reported lower noise with LTZ board installed, but that would require a full adjustment and at least the change of one additional resistor on the board.

In summary the unit is nice, but the GUI needs more work, there are still some issues such as auto scale that doesn't fully work as expected.

-branadic-

[Kleinstein]

The ADR1399 does not need that much more zener current than the LM399. It depends on how much current is actually used and in some cases the current can be enough also for the AD1399.  It would thus only be the extra RC (or tantalum cap) connected in parallel to the reference - either free floating or a tiny PCB.

The meter would not be able to tell the difference from LM399 or ADR1399 - so I don't think it would need extra adjustments. One would still need to account for the different absolute voltage, but that would also apply to a normal LM399 replacement.

[randymatt444]

That also allowed me to use a copper plated plier during soldering of the LTZ1000A, to avoid strong heating of the chip, ...

I know this is an old thread, but what would be the harm in using a good socket for the LTZ1000?   

[Kleinstein]

The extra socket would introduce possible extra thermal EMF from thermal gradients. To a large part this would be constant, bit still avoidable. For the current setting path the contact resistance would add.
Worst case an open contact could cause the ref circuit to apply high than normal current or temperature.
Usually the LTZ1000 is used in conbination with the resistors as a set and not real need to remove it. Except for a separate burn in or test circuit for selecting chips I see no use for a socket.
As a normal reference it is more a disadvantage.

The chip is also not that sensitive to heat, as it is normal to run it at some 60-90 C. So unless one uses excessive long soldering on all pins I would not expect an effect of the soldering heat. This is especially with the full length pins and not mounted flush to the board.