Ultra Precision Reference LTZ1000

[antintedo]

Speaking of 3458A mods, maybe consider not using those Molex connectors and solder wires instead? Apparently they can cause issues: https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg1377590/#msg1377590

[mimmus78]

Once I tested what was effect of moving/removing the A9 board from 3458A and I didn't measured any shift. But this is just my unit and it doesn't prove or disprove anything ... maybe one day I'll test also effect of removing the current compensation resistor.

Anyway I'm not going to solder the board on 3458a ... this seems too much violence to such an instrument.

[dietert1]

On August 23, i showed a possible solution in the LTFLU thread:

https://www.eevblog.com/forum/metrology/the-ltflu-(aka-sza263)-reference-zener-diode-circuit/375/

That is a IP66 (waterproof) case and it takes my LTFLU reference board including a sensirion humidity sensor and a 5 g desiccant bag. The image also shows how to make a thermal clamp for the interface cable, in order to create a thermal "nirwana" inside the case.

Regards, Dieter

[alex-sh]

I need to send my LTZ1000 (non-A version) board to somebody to fine tune it. Can anyone help please? I am fine sending it worldwide, but the UK would be the perfect destination. I do not have equipment to set it up properly.
What's the best way to post LTZ1000? Running in the temperature chamber or …?

[Dr. Frank]

Would you mind telling or showing more details, like PCB,mechanical and thermal design, choice of resistors?
If prerequisites are ok, I offer to do the T.C. trimming, and determination of output voltage(s)

Frank

[mimmus78]

On August 23, i showed a possible solution in the LTFLU thread:

https://www.eevblog.com/forum/metrology/the-ltflu-(aka-sza263)-reference-zener-diode-circuit/375/

That is a IP66 (waterproof) case and it takes my LTFLU reference board including a sensirion humidity sensor and a 5 g desiccant bag. The image also shows how to make a thermal clamp for the interface cable, in order to create a thermal "nirwana" inside the case.

Regards, Dieter

Well consider I have stacked design mainly for this reason ... everything is heat producer is on the top board, and this is connected to bottom board by solid copper wires. The molex connectors are on the bottom PCB where no relevant heat is produced.

[dietert1]

Yes, i guess you want to run the LTZ1000 using its heater.
Anyway, now that i learned how to compensate reference amplfier residual TC using a temperature measurement circuit (like https://www.eevblog.com/forum/metrology/the-ltflu-(aka-sza263)-reference-zener-diode-circuit/msg2578746/#msg2578746), i am wondering how the LTZ1000 would perform at room temperature, i mean in a TEC "oven".

Regards, Dieter

[SilverSolder]

My understanding is that it is very difficult to balance an LTZ1000 for zero TC at room temperature.

Of course, that makes it an interesting intellectual challenge!

[dietert1]

Did you ask a question?
Have a look at the schematic! The same extraction/measurement of chip temperature works for the LTZ1000. It's even better since you have an independent transistor as temperature sensor. Once you have the temperature signal, you can use it to raise or lower the temperature coefficient as you like. This works much better than the usual 400 KOhm resistor, because that signal isn't temperature but heater power.

Regards, Dieter

[Kleinstein]

For compensation of the residual TC, the chip internal transistor is not useful. The internal regulation keeps it at essentially constant temperature.
The residual TC is not so much because of the internal temperature regulation not working well. It is more because of effects on the resistors, thermal EMF and an effect of the heater power in the reference part.

It would need something like an external temperature sensor, or as used in the standard circuit use the heater voltage as an approximation.
An external temperature sensor (e.g. diode or PT1000) may be better than the 400K (R9) resistor from the data-sheet, that uses the heater as an approximation for the external temperature. AT least what I sow here in this thread looks more like a linear TC than a direct correlation with heater current (more like square root of heater power).

[dietert1]

Sorry, this time you missed it completely. This is about the residual TC without using the on chip heater, but using a TEC "oven" at room temperature instead. And don't tell me that an external temperature sensor will work better than an on chip sensor.

Regards, Dieter

[SilverSolder]

Did you ask a question?
Have a look at the schematic! The same extraction/measurement of chip temperature works for the LTZ1000. It's even better since you have an independent transistor as temperature sensor. Once you have the temperature signal, you can use it to raise or lower the temperature coefficient as you like. This works much better than the usual 400 KOhm resistor, because that signal isn't temperature but heater power.

Regards, Dieter

Consider how the LFTLU is made relatively temperature insensitive by a combination of choice of zener current and use of the built in transistor as a compensation device.  That scheme is hard to replicate on the LTZ1000, it seems to me, because its pass transistor isn't connected the same way as LTFLU.

I do like the idea of keeping the LTZ1000 in a controlled temperature environment, using its own heater transistor to measure the temperature.  Just the fact of keeping all the other components and the board itself at a stable temperature has to be an overall win.

[Kleinstein]

The TC to start with at the LTZ1000 is rather high. So it is quite difficult to reduce the TC without using a critical high stability resistor.
There could be chance that this may work at a reduced current (e.g. 0.5 mA). Less current should reduce the TC and the fine tune could than be possible with the transistor current. However this comes at the price of higher noise and I don't know how aging changes at lower current.

Even if still using the internal heater, a reduce TC to start with would reduce the importance of the set point resistors.
Ideally there would be a LTZ1000B with lower TC to start with, e.g. from a reduced voltage zener (e.g. 6.7 V total reference).

[alex-sh]

Would you mind telling or showing more details, like PCB,mechanical and thermal design, choice of resistors?
If prerequisites are ok, I offer to do the T.C. trimming, and determination of output voltage(s)

Frank

This is the board I have - your design   The resistors are Vishay. The two critical resistors are hermetically sealed.
Here is another link - https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg2514555/#msg2514555
Very much appreciate your assistance.

[Dr. Frank]

Would you mind telling or showing more details, like PCB,mechanical and thermal design, choice of resistors?
If prerequisites are ok, I offer to do the T.C. trimming, and determination of output voltage(s)

Frank

This is the board I have - your design   The resistors are Vishay. The two critical resistors are hermetically sealed.

Very much appreciate your assistance.

Well, fine,  , so I'm familiar with that board, at least.
So what is the T.C. of these resistors, the Vishay and the other ones?
Did you determine these T.C.s?

How is the PCB assembled, or is this the current state?

It's not feasible to trim this naked PCB, as the mechanical/thermal assembly also influences the T.C.
Also be aware, that the trimmed 10V output has to be re-trimmed, because T.C. trimming by this nominal 200k resistor changes the LTZ1000 output voltage.

My offer persists, if you would have made a proper assembly.
Frank

[alex-sh]

Would you mind telling or showing more details, like PCB,mechanical and thermal design, choice of resistors?
If prerequisites are ok, I offer to do the T.C. trimming, and determination of output voltage(s)

Frank

This is the board I have - your design   The resistors are Vishay. The two critical resistors are hermetically sealed.

Very much appreciate your assistance.

Well, fine,  , so I'm familiar with that board, at least.
So what is the T.C. of these resistors, the Vishay and the other ones?
Did you determine these T.C.s?

How is the PCB assembled, or is this the current state?

It's not feasible to trim this naked PCB, as the mechanical/thermal assembly also influences the T.C.
Also be aware, that the trimmed 10V output has to be re-trimmed, because T.C. trimming by this nominal 200k resistor changes the LTZ1000 output voltage.

My offer persists, if you would have made a proper assembly.
Frank

Non-critical resistors are 0.01% 5ppm 0.5W Vishay SFERNICE Foil resistors T.C +/-5ppm/C
Critical resistors are 0.01% 0.3W RADIAL Vishay Metal Foil seal resistors (VH102K and VHD200 used as a single resistor) - I think these have T.C. +/- 1ppm/C and 2ppm/C respectively.

The board was assembled by hand. Resistors and LTZ1000 was soldered with an extra care trying not to overheat these components.
I left long legs for LTZ1000 - not sure if it was the right decision. It's been a long debate here about the legs length and TCR.
 
Currently, the board is in the Hammond enclosure (LTZ1000 has been isolated for air drafts) running 1500h+
No, I did not determined the T.C.

Thank you once again.

[Dr. Frank]

Non-critical resistors are 0.01% 5ppm 0.5W Vishay SFERNICE Foil resistors T.C +/-5ppm/C
Critical resistors are 0.01% 0.3W RADIAL Vishay Metal Foil seal resistors (VH102K and VHD200 used as a single resistor) - I think these have T.C. +/- 1ppm/C and 2ppm/C respectively.

The board was assembled by hand. Resistors and LTZ1000 was soldered with an extra care trying not to overheat these components.
I left long legs for LTZ1000 - not sure if it was the right decision. It's been a long debate here about the legs length and TCR.
 
Currently, the board is in the Hammond enclosure (LTZ1000 has been isolated for air drafts) running 1500h+
No, I did not determined the T.C.

Thank you once again.

Well,  I also left the legs long... successful.

Everything sounds good, maybe you have a picture of the completed device, including power supply.

Anyhow, if nobody else in your country (UK?) gives an offer, you may send it to me.. just send me a PM to clarify details.

Frank

[MiDi]

Non-critical resistors are 0.01% 5ppm 0.5W Vishay SFERNICE Foil resistors T.C +/-5ppm/C
Critical resistors are 0.01% 0.3W RADIAL Vishay Metal Foil seal resistors (VH102K and VHD200 used as a single resistor) - I think these have T.C. +/- 1ppm/C and 2ppm/C respectively.

The most critical resistors are the boost resistors for 7->10V.
Their tracking TCR and drift are only attenuated ~3x, whereas all other are >=100x.
The Datasheet seems to be wrong for the relevance of the impact of resistor changes.

Put your money wisely 

[alex-sh]

Non-critical resistors are 0.01% 5ppm 0.5W Vishay SFERNICE Foil resistors T.C +/-5ppm/C
Critical resistors are 0.01% 0.3W RADIAL Vishay Metal Foil seal resistors (VH102K and VHD200 used as a single resistor) - I think these have T.C. +/- 1ppm/C and 2ppm/C respectively.

The board was assembled by hand. Resistors and LTZ1000 was soldered with an extra care trying not to overheat these components.
I left long legs for LTZ1000 - not sure if it was the right decision. It's been a long debate here about the legs length and TCR.
 
Currently, the board is in the Hammond enclosure (LTZ1000 has been isolated for air drafts) running 1500h+
No, I did not determined the T.C.

Thank you once again.

Well,  I also left the legs long... successful.

Everything sounds good, maybe you have a picture of the completed device, including power supply.

Anyhow, if nobody else in your country (UK?) gives an offer, you may send it to me.. just send me a PM to clarify details.

Frank

Thank you
PM sent

[alex-sh]

Does anyone have a spare LTZ1000 PCB? I would favour a board based on a design by Dr Frank and branadic?
Please send me a PM

[notfaded1]

I'd also like advice on source for a PCB either from member here or from online board house that has a design.  I, sadly don't have 4 LTZ1000 but only two older 2001 and 2004 LTZ1000ACH.  Is there anyone that has some PCB's ready that I could purchase?  This seems to be the best route to a nice precision reference vs just buying one (and it's good learning experience for this computer scientist).  You guys have put so much effort into this and hats off  I want in please?  Wow TIN that PCB looks sweet!  I hope you'll post some stats and graphs from it.  After going through hand selecting that many resistors to make your quad... perhaps you've got enough decent resistors left over for a single board set you could part with?

Best Regards,

Bill

[RandallMcRee]

Your wish is our command:
https://github.com/pepaslabs/px-ref

Notice that there is a link in the readme to OSH boards, under Purchase. I have built six and they work well.

BTW your older LTZ1000A are fine, like fine wine, it seems.

addendum: The PX board is the work of Mr. Pepas; as you can see from the github url and the readme therein. I just thought I should publicly acknowledge that this is all his (and others!) great work and we are grateful for it!

[iMo]

There is Dr. Frank's recreation as well
https://github.com/pepaslabs/dr-frank-ltz1000

[Andreas]

Mhm,

do you think that the capacity does not change within 37 days with room temperature?
If you pull the plates of a capacitor apart (e.g. rising temperature) the voltage will increase.

with best regards

Andreas

[Magnificent Bastard]

being distracted by the thread https://www.eevblog.com/forum/metrology/forum-crowd-project-simple-10v-ltz1000-plugnplay-box-(with-1k10k-hermetic)/
i took a peek at the ordering site, and noticed the ROHS and temp range confusion

i just have to ask everyone here

when you bought your LTZ1000 CH/ACH
did you know it was a
1) 0~70C unit? (ROHS complaint)
2) or a -55~+125C unit ? (non ROHS complaint)

digikey/mouser list their as -55~+125C and it is ROHS complaint. there is apparently no such unit on ADI website.

This is a "moot" point.  Since this is an internally heated device, there is an upper and lower bound on the ambient temperature anyway for proper operation.