Ultra Precision Reference LTZ1000

[KT88]

The moment two resistors combine two PWMs the resistor ratio matters again...

[Dr. Frank]

Thought about replacing the voltage divider R4/R5 for the tempreature with a PWM divider. Hoping to get at least some advantages from that.

• Cheaper?
• Adjustable Tempreature
• Higher stability?

What do you think?

I think that such questions should always be started from the (engineering) requirements and from design goals, not backwards from an anticipated solution... as this would be better and professionell engineering practice. Consequently, all the replies to your three questions so far did not give any concrete answer.

So, to answer your questions concretely: I think that a PWM would be a very bad approach in all of your categories.

At first stability: what stability requirements do you mean, like short term (noise), temperature, long term drift, or susceptibility to E.M.C. , pressure, humidity, vibration?

These 2 resistors theoretically contribute only by 1/75 .. 1/100 of their own stability figures, so the LTZ chip itself has paramount influence on mostly all these parameters; that has been measured / demonstrated in this lengthy thread many times. Practically, the residual T.C. of up to 0.3ppm/K, despite e.g. T.C. matching of the resistors, is not yet understood here, and is obviously not relevant, as the overall T.C. can easily be trimmed, down to <0.02ppm/K.
As the oven regulation and in turn the reference voltage is very sensitive to external noise, keyword voltage dips, for a PWM solution you'd need a very elaborate filtering, which would probably yield high cost and in the end worse noise figures than the passive divider solution.

Concerning cost, these Vishay hermetically sealed, oil filled, luxury resistors are not really necessary; much cheaper ones will do the job as well. These only have to fulfill certain minimum technical requirements regarding e.g. T.C. and timely drift.
Please, first make yourself aware of the BOM cost of these cheaper solutions, they are in the ballpark of < 20€ for two b.m.f. resistors, or even much less for other technologies, like e.g. TiN has presented on the discussion about the clone reference of the 34470A.

Concerning oven set point change, that question makes no sense at all to me.
Instead, what would be the purpose/application, or your design goal for that feature?

If you would start from the requirement "timely stability", you would know, that the lower the temperature, the better this parameter.
50°C oven temperature gives about -0.8ppm/year, each 10°C more will double that number. Therefore, simply chose a fixed, lowest possible oven temperature.
Changing the oven temperature will change the reference voltage by about +50ppm/K, which would violate the overall stability requirement as such.

If your feature goal would be to have an easy and cheap(er) possibility to make thermal cycling like in the Pickering patent, well that's also already done in the Fluke 7000 devices, so no need to re-invent the wheel by another complicated solution.

From all these aspects, I can not imagine at all, what would be the benefit of a PWM solution, other than over-engineering an already solved problem  , just like this other voodoo stuff here, as PCB meander cuts, fancy resistors, ceramic and multi layer PCBs and so on.

Frank

[Grandchuck]

One of my LTZ1000 reference circuits failed.  The LTC1052 output was oscillating and the failure occurred when I placed the circuit in a homemade cooler/foam box to measure its temperature performance.  I assume a static charge destroyed the LTC1052.  After replacing that IC, all is back to normal.

The LTC1052 is a CMOS device and perhaps should not be placed in a foam box?  A partial schematic is included ... you might recognize the good work of Dr. Frank.

[guenthert]

It just might be because the LTC1052 is directly connected to the output.  I would have liked a transistor follower in between, also because zero-drift/chopper OpAmps tend to have feeble output drive capability.  Well, next time 

[Dr. Frank]

Hello Grandchuck,
thank you for the flowers 

Anyhow, the LTC1052 was selected because it is one of the few zero offset / chopper amps which also have very low input bias (~ pA) which I originally needed for a precision amplification by a divider with a higher impedance.
As lower impedance solutions (~ 20kΩ) for this divider are available meanwhile, this is not required any more.

All of my 7 references are thermally isolated internally, using styrofoam, but not a single LTC1052 or 2057 has been blown up, especially not by an ESD event which would have originated from the foam.
But it's correct, that the output is not protected at all against external ESD or EOS scenarios.. an additional transistor driver with current limiting would be an advantage .. next time..

I also measured the 0.1 ..10Hz noise directly, and confirmed that the LTC1052 - due to its low bias current - performs very badly in this category, i.e. in total generating 2 times the noise of the raw LTZ1000 output. Briefly, here are the ballpark numbers, at ~7.2V output levels: direct out: ~ 140nV_rms, 2057: ~ 180nV_rms, 1052: ~ 270nV_rms.
These different noise figures are not apparent at all, if you measure the references with e.g. a 3458A at NPLC 100.

Anyhow, in a new design the LTC1052 will be replaced by an ADA4522-1, and for some of the existing references by means of an adaptor PCB... I will report the outcome of this improvement (?)

Frank

[Grandchuck]

Hello Dr. Frank,

I have 3 iterations of your design and one of them (the one that failed) would occasionally latch up ... the output would drop to some low voltage.  This only happened when cables/connections were being changed/moved.  Power cycling always restored normal operation.  Some kind of clue there?

Please do share additional details as you can.  As always, it will be appreciated.

Grandchuck

[branadic]

Grandchuck

I had a similar issue on one out of nine references based on the redesign of Frank's schematic I made a few years back. After replacing the LTC1052 similar things never happened again. However, I added a 120R resistor in the feedback loop of the reference amplifier and in series with the diode to prevent any issues during startup, in case that would cause any trouble.

-branadic-

BTW: There was a component listed on Arrow, ADR1000AHZ, but for some reason they removed their listing. However, it can still be found in the cache https://webcache.googleusercontent.com/search?q=cache:snoBuWs4HP0J:https://www.arrow.com/en/products/adr1000ahz/analog-devices+&cd=2&hl=de&ct=clnk&gl=de

[cemelec]

I've been following this topic for a while; but is there anywhere today (May 2021) that you can buy a LTZ1000 at a sensible price? Searching comes up with ridiculous pricing....
This part is of sufficiently high value that I don't trust Ebay or any Chinese supplier, if they have it,  - its well worth faking.

Here in the UK the distributors I trust don't have it.
Is it still in production? According to AD website yes, so why so hard to get one?

Another part I'm interested in, the AD5791 20 bit DAC, has the same problems

Charles

[Grandchuck]

I have an order in for an LTZ1000CH @ $58 and the supplier has flagged it to be shipped 6/29/2021

[Dr. Frank]

Hello Dr. Frank,

I have 3 iterations of your design and one of them (the one that failed) would occasionally latch up ... the output would drop to some low voltage.  This only happened when cables/connections were being changed/moved.  Power cycling always restored normal operation.  Some kind of clue there?

Please do share additional details as you can.  As always, it will be appreciated.

Grandchuck

Grandchuck,

Well, I can't guess, which 'iteration' you have used, as my schematic has been used by several other people.
Myself, I only published one schematic and one single sided layout, which has been been working well so far.

Also, you did not tell, whether the 1052 amplifier, or the reference amplifier itself showed a latch-up behavior.

So sorry, but I can't give any hints on this sparse information.
Any further hint is welcome.

Frank

@ cemelec:
The LTZ1000 may be under allocation also. It was rumored to have a lead time of 25 weeks, or so.

[Grandchuck]

The three references differ in the resistors.  They all use the PC board based on your design and schematic.  One uses Edwin Pettis devices (wire wounds) in all the important locations, one uses some of those and some low ppm Vishay foil types, and one uses some odds and ends purchased from various ebay stores.  All three compare favorably with each other and with Tin's FX reference, but not THAT good   I honestly don't feel that my 3458A is as rock solid as some discussed here by the serious volt nuts.  I have seen the term "golden" used to refer to some 3458As.  Mine is perhaps silver.

The LTC1052 is the device that latches up.  It only happens on occasion.

I will probably try your replacement for the LTC1052; if and when you release the information.

[Dr. Frank]

Many, many Thanks to Andreas! 
He donated me one of his x10000 noise amplifiers, so I also was able to make absolute noise measurements on my 10 LTZ1000 based references.
For each reference, I only have taken one 10sec long sample, and let the scope calculate the Peak and RMS noise, see example screen shots.

Prototypes Ref_1 and Ref_2 are based mainly on the original datasheet schematic, plus two blocking capacitors in parallel to the 120 and 1k resistors.
LTZ #5 is fully assembled on a PCB from Andreas, lacking those 2..3 ESD / angst capacitors, and it is battery powered.
LTZ # 1 - 4, 6, 7 are based on my design, i.e. also lacking these ESD capacitors.
LTZ # 1-4 feature a 10V amplification.
On the LTZ #3 I repeated the measurements with an AD4522, instead of the LTC1052.
LTZ #2 was battery powered only for this experiment; running on a PSU, I estimate about 10% higher noise figures.
The FLUKE 7000 is also operated on battery power.

Nearly all measurements were done inside a cooking pot, apart from the box for Ref_1 and Ref_2, which is too big.

I also added the 1h noise figures measured with my 3458A at 100 NPLC, i.e. that's a convolution of the noise of the DUT and the 3458A internal LTZ1000A.
That measurement indicates, that the noise of the references does not play a dominant role when using them as a calibration source at this aperture.

I will update this table, as soon as I change more and more buffers from 1052 to 4522.

 
Frank

Updated the table, now 4 modified references. Measurement on a 9V battery gives about 41nV rms for the amplifier

[Andreas]

For each reference, I only have taken one 10sec long sample

Hello Frank,

since noise is very statistical I usually do at least 15 samples 10 sec long and calculate the average + std dev of uVpp and uV RMS.

When looking for popcorn noise I even use 100 sec long samples.

with best regards

Andreas

[The Bootloader]

Hi there,

I would like to build myself a LTZ1000A reference.

I am a bit lost among all the versions available.
If I quote the signature of our fellow inmate cellularmitosis, there are :
KX FX MX CX PX Frank A9 QX

It seems like KX is the original one, Frank is a popular one, and PX is a compact one.
The other ones are either specialized (made to fit within a specific instrument) or a bit unpopular (so I might have difficulties finding support if needed)

I was leaning toward PX, but it seems like TiN offers KX calibration which is a big deal to me, so I might do this one instead.

Any opinion ? It seems like a jungle, I want to enter this world, and would like guidance as to what is my best bet so I won't be frustrated to have built the "wrong one".

I already have a set of resistors from this group buy (70Kx2, 1K, 120, 12.5K). Originally intended for Dr Frank board, but I think it can be used for any board as long as they fit physically.

Thanks

[leighcorrigall]

Hi KX-ref owners,

I am about to solder the critical resistors and LTZ1000ACH onto TiN's reference board. As this is my first build it would be great if someone could perform a 'sanity check' on the component positions. See attached. I have ensured there are no shorts and I tried my best to orient the components correctly based on reference pictures. The board was cleaned multiple times with a new toothbrush and fresh isopropyl alcohol.

Bismuth solder paste was used for all non-critical components thus far. The total soldering time was much less than 280 seconds. Approximately much less than 4 seconds per lead. Priority was given to the ease of mounting and as suggested by the original xDevs post.

Even though they are greatly excessive , I used the following critical resistors that were purchased from AVNET:
1x Y4078120R000T9L (120 Ω)
2x Y607270K0000T9L (70 kΩ)
1x Y5076V0169TT9L (1 kΩ / 13 kΩ)

This reference board will be used in combination with an output buffer stage (based on the FX-ref board) to obtain approximately 10 V.

Thanks for your help!

[TiN]

Looks alright. 70k VHP resistors are indeed pure wasted gold

[MegaVolt]

In the fresh LTSpice I found models for LTZ1000 and LTZ1000a from AD.

[iMo]

The diode model has got a tempco, it seems..

[magic]

Most zener models in LTspice have TC.
The question is: is it accurate over current and temperature?

I tried to reproduce the s-shaped TC of compensated zener references with some built-in diode models and failed.

[iMo]

And finally the complete Vref source while changing ambient temp from 20C to 60C.
LTZ1000: You may see the heater current changes with ambient (from 27mA to 0mA) and the Vref's TC is -7.8uV/C within 20C to 55C ambient temperature range.
LTZ1000A: You may see the heater current changes with ambient (from 17mA to 0mA) and the Vref's TC is 0uV/C within 20C to 55C ambient temperature range.

PS: the previous sim of the TC of the LTZ1000A "zener diode only" showed +2.7mV/C.

[iMo]

Most zener models in LTspice have TC.
The question is: is it accurate over current and temperature?

I tried to reproduce the s-shaped TC of compensated zener references with some built-in diode models and failed.

The LTspice zener models usually lack a few parameters which are needed for creating S shape with some compensation.
Look at the first line in my above model of the "iMo6V9" zener. The numbers are rather guesses..

[branadic]

The model is missing the resistors shown by John Pickering at Metrology Meeting 2021 too.

-branadic-

[iMo]

With highest numeric resolution the TC of the LTZ1000A Vref shows -3nV/C (20-55C amb).
A small (ie 8ohm) resistor in anode does not change TC much, the output voltage increases by a few mV.
There is also an ADI's "testing jig" in the model - showing the response of the Vref on the heater power (switching it on/off). See below.
I think this is not the last revision of the model..

[chris_11]

To the LT1000 LTSpice model. You can set the start temperature with a .temp statement and get an idea of the conditions in that case.

To the LTC1052 "Latched" reference. Educated guess: The LTC1052 gets phase reversal with both inputs at GND and you get a stable situation with the circuit not starting up. If you prebias one input enough in the right direction it should start.

A CMOS IC latch up is usually a destructive event with the CMOS IC turning into a switched on Thyristor over its supply rails with no real current limit.

@branadic.  You can add the resistors external and see what changes.

[aronake]

How do people get hold of the fancy sealed metal film resistors?

They are pretty much out of stock in all places I have seen.

Put an order and wait?