Ultra Precision Reference LTZ1000

[ltz2000]

The immunity against small changes in power supply voltage of the LTZ1000 basic datasheet circuit is "good enough". But has anyone here actually tried / succeeded to measure the output [ppm] / input [V] relation?

[wiss]

Such big labs usally have a supply of liquid He anyway - some even have there own recycling system. Still it is expensive to operate. Usually on can refill the cryostat with new Helium. So there is no need to warm up the setup so often. Transfering liquid He from one container to another is not simple, but it works with well insulated tubes. 

With liquid He one usually can get very stable temperature and no problem with a power outage.  The cryocooler also needs quite some electricity and maybe service every few years. The cooler may also produce extra noise.

An interesting question would be if one could get away with liquid nitrogen and high temperature superconductors.

I think these might be interesting.

A.M. Klushin, L.E. Amatuni, E. Sodtke, W. Prusseit, S.I. Borovitskii, V.D.
Gelikonova, and K.A. Ai. Development and investigation of shunted ybco
bicrystal josephson junction arrays for voltage standard. In Precision Electromagnetic
Measurements Digest, 1996 Conference on, pages SUPL13–SUPL14,
June 1996.

A.M. Klushin, M. He, and A.S. Katkov. Constant-voltage steps at 0.1 volt
from a series array of high-tc josephson junctions. In Precision Electromagnetic
Measurements Digest, 2008. CPEM 2008. Conference on, pages 104–105,
June 2008.

S.K. Khorshev, A.I. Pashkovskii, S.M. Goryunov, N.V. Rogozhkina, Yu.M.
Gryasnov, E.E. Pestov, and A.M. Klushin. Towards a voltage standard based
on an array of high-temperature josephson junctions. In Precision Electromagnetic
Measurements (CPEM 2014), 2014 Conference on, pages 126–127,
Aug 2014.

[Andreas]

The immunity against small changes in power supply voltage of the LTZ1000 basic datasheet circuit is "good enough". But has anyone here actually tried / succeeded to measure the output [ppm] / input [V] relation?

Hello,

I did a measurement when I built my LTZ1000.
But it is with a input voltage regulator of around 14V.
The power-supply was relative noisy. (Thyristor pre-regulated supply).
So it is not clear if the effects in drop out are due to EMI or heating effects.
I really should repeat the measurement with a better supply.

All in all I have 3uV after 2:1 divider or 6uV @ 7V change over the whole input voltage range of 12-18V

With best regards

Andreas

[plesa]

Assembled 4 boards with LTZ1000. For testing I used 25ppm resistors. but soon will be replaced by PWW.
LTZ1000 has been aged in oven for three weeks at 125°C. After PWW will be placed I will cycle it according to John Pickering.
All 4 units I would like to put into one box (similar to which use e.g wekomm use for their resistance standards).
I'm impressed how stable it is on table with 25ppm resistors, temperature of reference is 12k/1k so about 40-45°C.

Time to change Yageo metal film MF0 resistors [25ppm/K] used for basic testing with quality PWW from Edwin.
12k/1k for temperature settings (20°C above ambient temperature ~ 45°C) and 100k for Q1 and Q2 to minimize current for long term lowest drift.
According to initial measurement all resistor except one were far below 0,1% initial accuracy (average about 0.04%)
So I will make poor man tempco measurement (50°C and 10°C) prior assembling resistor to LTZ board.
1k/12k resistor will be close together on board coupled by copper tape to improve their temperature tracking.

[Andreas]

So I will make poor man tempco measurement (50°C and 10°C) prior assembling resistor to LTZ board.
1k/12k resistor will be close together on board coupled by copper tape to improve their temperature tracking.

Dont forget the room temperature. According to my measurements many of Edwins PWW resistors have a parabolic shape with flat T.C. near room temperature.
I guess that copper tape will not be necessary (at least if you have a metal housing for the whole cirquit). Most of the heat transfer is through the wires.

With best regards

Andreas

[TiN]

I don't have VHD200, but I have pair of Z202 10Ks and VPG's custom hermetic (likely VHP101-based) 10Ks network (3-pins), which I can test, but it would take some time.

[ltz2000]

PCB slots revisited (Keysight 34470A)

https://youtu.be/_QpApuKdcqQ

from 15:30

[alanambrose]

>>> Does anyone have a clue about the TCR

Edwin has mentioned the Vishay LT5400 networks a couple of times - 100 ppm (A grade) / 250 ppm (B grade) matching tolerance, 0.2ppm tracking TC. I'm planning to check these out in a  new design. Unfortunately, duh, I didn't notice (below) that they have a central pad, and I dual-foot-printed them with standard 1206 networks for testing purposes. That's not going to work unless I fiddle something in with some kapton tape, so I may be waiting for another spin from the pcb fairies...however they do sound promising. Clearly there's a question there about the impact of the reflow / soldering heat.



Alan

[TiN]

I have few of those too. I can toss all into metal can and cycle all from 20? to 60? , measuring by scanner in 2002, if that is what you looking for.

[alanambrose]

That would be pretty interesting if you have the time.

A.

[TiN]

What is that, time?
Ok, will do on weekend then.

[Edwin G. Pettis]

To Butterfly,

Basically, the TCR / tracking TCR of a resistor is very stable over time and temperature, however external factors can and do influence the actual results.  In the case of tracking TCR, a major factor is how good the thermal coupling is between resistors, all else aside, the tracking TCR can be as low as under 0.2PPM/°C.  Differences in power loads will affect tracking as well, especially when attempting fractional PPM tracking, thermal coupling does help iron out the difference.  Linear Tech's LT5400 has very good tracking TCR but as the resistors are quite small, soldering temperatures are dangerous.  The 'A' grade which has the better specs may or may not cost more than equivalent PWW resistors and note that the resistors are all the same value on the chip, limiting your options for ratios.  Their absolute TCRs are not that good, only the ratio mode TCR are very low.

In many cases, the tracking TCR as specified is done under the most favorable conditions to achieve best performance specs, under real world conditions, that level of performance is not likely to be matched.  In critical performance, it it often a question of trial and error to find the best possible performance under the required conditions.  Depending on the application's actual parameters, a great set of resistor specifications on paper may not be possible.

Anytime performance is expected to be in the low PPM to sub-PPM area, you must expect that otherwise ignored conditions are going to jump up and bite you in the keester, every little thing you could get away with ignoring at higher PPM levels are now going to be demanding attention.

[alanambrose]

Thanks for that Edwin - do you have any thoughts on how these should be soldered given they're smd parts?

TIA, Alan

[Edwin G. Pettis]

Just the usual warnings, solder as quickly as possible, lowest possible wattage iron that gets the job done, ect.  Even a poor heatsink is better than nothing if possible.

[alanambrose]

Well I was imagining a custom made heatsink but then I noticed this...



... which kind of suggests reflow including the central pad? The central pad is apparently floating - so maybe it's there to reduce thermal variations over the die, rather than noise, and doesn't in fact need to be soldered down?

A.

[Edwin G. Pettis]

Unless there is an actual window with metal in it under the package, then no there is no soldering necessary, for that matter the pad may have little effect on heat sinking but it certainly won't hurt to have it there.  Note the exposed die pad option.....so I don't think there is any exposed pad on the MSOP package to solder to on the standard package.

[Edwin G. Pettis]

After looking at the LT5400 data sheet, there is a bonding pad on the bottom of the package and this does need to be attached to a pad on the PCB for best thermal considerations.

[branadic]

After looking at the LT5400 data sheet, there is a bonding pad on the bottom of the package and this does need to be attached to a pad on the PCB for best thermal considerations.

Really? What kind of datasheet do you have? I can't see anything you mentioned in the datasheet.

However, it's all written within the datasheet:

Where to Connect the Exposed Pad

The exposed pad is not DC connected to any resistor terminal. Its main purpose is to reduce the internal temperature rise when the application calls for large amounts of dissipated power in the resistors. The exposed pad can be tied to any voltage (such as ground) as long as the absolute maximum ratings are observed.
There is capacitive coupling between the resistors and the exposed pad, as specified in the Electrical Characteristics table. To avoid interference, do not tie the exposed pad to noisy signals or noisy grounds.
Connecting the exposed pad to a quiet AC ground is recommended as it acts as an AC shield and reduces the amount of resistor-resistor capacitance.

[Edwin G. Pettis]

I do not understand, the pad on the bottom of the LT5400 is a thermal heat sink to be attached to a pad on the PCB as it is said in the data sheet and in your own statement, where is the problem?

"Its main purpose is to reduce the internal temperature rise when the application calls for large amounts of dissipated power in the resistors."  Just as I said.

Regards.

[branadic]

Sure, for thermal issues but it's not a bonding pad whatsoever And the exposed pad does need to be attached in special cases, not for normal use.

[Edwin G. Pettis]

Technically, but for best thermal performance at any level, the pad should be attached to the pad on the PCB, it helps spread the heat around a bit better than without it, but no you don't have to bond it to the PCB in all cases, that was not what I was saying, in my opinion it is better to bond it for all cases for best performance, even mW differences can make a difference.  I do not see any controversy here, perhaps splitting hairs but that is all.

[Andreas]

Hello,

Santa left some goodies for christmas:

Guess it could be to place a LTZ1000/LTZ1000A on it.
Either with Z201 or PWW resistors.
Alternatively a LT1013A or two LTC2057

With best regards

Andreas

[TiN]

LTZ1047B? Huh
That's a most bizzare wierd layout I saw for LTZ circuit. No voodoo magic, virgin nude cutouts and shields? How come?
Looking forward for some tests 

[Andreas]

Hello,

sorry to dissapoint you:
there is some shield intended to place around the LTZ

http://www.reichelt.de/Teko-Steel-Plate-Enclosures/TEKO-3710/3/index.html?ACTION=3&GROUPID=5202&ARTICLE=34042&SEARCH=teko%203710&OFFSET=16&WKID=0&

And LTZ1047B is only the working title for the layout.
(Revision B of a older version started in 2010 week 47)

With best regards

Andreas

[Kleinstein]

The two OPs at the standard LTZ1000 circuit are not that critical with respect to drift - the transistors allready add some gain.

The transitor for the voltage output might be prone to oscillation. I think the LT1013 is perfectly Ok here - it seems to be stable with the extra gain.  If looking for a different OP, I would consider one with external compensation or an added compensation in the circuit.