Ultra Precision Reference LTZ1000

[TiN]

Purchase button still broken, even with new account. But if I use "Enter the Order" and then manually type in LTZ into order list - it finds chips and seems like I can add into ordering system.
Ugh... 

[pitagoras]

Looks like some random page migration strategy is taking place. 

Here: http://www.analog.com/en/products/analog-functions/voltage-references/shunt-voltage-references/ltz1000.html#product-samplebuy
I only see options to buy >=100.
Clicking purchase, though, I get redirected to some http://shopping.na3.netsuite.com/ page where one can add one by one going to "continue shopping".
ACH appears as backorder and CH as available (image attached).
I didn't have Analog account and my linear account got migrated (they sent me an email). When they migrated the account I observed what TiN says: purchase getting redirected to some list of distributors (I don't remember the details, weeks ago).

If they only were prone to hire at least one software guy 

[Insatman]

Attached is a chart showing my present LTZ1000 voltage over time compared to three other references.   It has the worst temperature dependence...yet it should be superior.  The unit is housed in a cast aluminum box lined with foam.  The LTZ1000 is also enclosed in a small hollow Styrofoam block on the PCB.   Schematic is attached.   

[chuckb]

Several things may contribute to this performance.

The OPA177 is not a rail to rail op amp. The data sheet indicates some level of performance at 1V above the negative rail. Based on my previous measurements the input voltage to U3 will be 0.4 to 0.5V. The input voltage of U1 will be around 0.6V. These voltages seem to low for reliable performance from this otherwise fine op amp. A single opamp similar to the LT1013 will work fine here. I don't have a model number for you off hand.

The R5 resistor is listed at 12k. That seems to low of a value to me for a LTZ1000A. It may work with a LTZ1000 non A in a cool environment. You may be running out of heater authority during warm ambient conditions.

The R8 392k Temp Co resistor usually has to be selected based on the particular circuit layout and other magic. Have you had an opportunity to optimize this value?

Hope this helps.

update
P.S. With a 0.6ppm / deg C, the temp co tracking of R11 and R12 in the output amplifier also may be an issue. Monitoring the heater voltage and the raw 7v zener voltage could help isolate the real issue.

[TiN]

OOM heater would render in much higher TC result, but yes, 12K/1K is bad idea for typical conditions.
Using trimmer in critical TC path is big no-no. Output buffer 7V-10V section is way more critical for matched resistors than LTZ ref itself.
Return currents can be also cause issues. Without seeing the board or accurate data result it is not possible to tell better.

selected based on the particular circuit layout and other magic.

[Insatman]

Several things may contribute to this performance.

The OPA177 is not a rail to rail op amp. The data sheet indicates some level of performance at 1V above the negative rail. Based on my previous measurements the input voltage to U3 will be 0.4 to 0.5V. The input voltage of U1 will be around 0.6V. These voltages seem to low for reliable performance from this otherwise fine op amp. A single opamp similar to the LT1013 will work fine here. I don't have a model number for you off hand.

The R5 resistor is listed at 12k. That seems to low of a value to me for a LTZ1000A. It may work with a LTZ1000 non A in a cool environment. You may be running out of heater authority during warm ambient conditions.

The R8 392k Temp Co resistor usually has to be selected based on the particular circuit layout and other magic. Have you had an opportunity to optimize this value?

Hope this helps.

update
P.S. With a 0.6ppm / deg C, the temp co tracking of R11 and R12 in the output amplifier also may be an issue. Monitoring the heater voltage and the raw 7v zener voltage could help isolate the real issue.

I already have plans to replace the OPA177 with the LT1013 for the two op-amps servicing the LTZ1000 directly.  I will use the OPA177 just for the output scaling.   I have a 13K resistor on order to replace the 12K resistor on my next LTZ1000 build.   Is there any guide on what value of resistor value gives what heater temperature?

I recognize that my resistors used on the scaling op-amp are likely a problem, but honestly I have a hard time acquiring decent resistors here in Philippines.  Ebay and Digikey are basically my only sources.   Finding three terminal resistor on Ebay with the correct ratio has been an exercise in frustration.

As for the 392K resistor, this is the closest value to the 400K resistor recommended in the LT datasheet for the LTZ1000.   Any guidance as to how to optimize this resistor is appreciated.

[TiN]

Did you talk to Edwin here on forums to get proper resistors. Using expensive VPG does not give you any measurable benefit in correctly designed and built LTZ ref.

Also you asking questions that were explained and tested many times even in this very thread, leave alone threads about my reference design. I know typing a question is much easier than taking time to read all the pages here, but you will get much clearer picture in the result.

[Insatman]

Did you talk to Edwin here on forums to get proper resistors. Using expensive VPG does not give you any measurable benefit in correctly designed and built LTZ ref.

Also you asking questions that were explained and tested many times even in this very thread, leave alone threads about my reference design. I know typing a question is much easier than taking time to read all the pages here, but you will get much clearer picture in the result.

VPG?

I'll go read more of this very long thread.

[Echo88]

VPG = Vishay Precision Group = Bulk Metal Foil Resistors

[kj7e]

The 400K TC resistor value is just a starting point.  You have to perform your own tests in order to determine the correct value to achieve the 0.05ppm/K or minimum TC drift.  In one case I needed to use 300K resistor and another a 680K.  I was able to do better than 0.05ppm/K with multiple builds, but it took considerable time and testing to get there.

Some of my testing here;
https://www.eevblog.com/forum/metrology/kx-reference/msg1369941/#msg1369941

Also check this thread for information on sources for LTZ1000(a) resistors;
https://www.eevblog.com/forum/metrology/resistor-set-for-ltz1000-positive-standard-7v-circuit/

[MisterDiodes]

Insatman:  As others have noted, an LT1013 will do everything you want it to do for a LTZ amp.  You don't really need expensive resistors around the LTZ, since the LTZ is more forgiving to resistor drift than something like a 7V-10V boost circuit - You just want decent resistors, and the heater ratio pair is most important. PWW are perfectly fine and the preferred choice, and Edwin will give you a deal on an LTZ set, too.

Yes, do some reading, but for sure read page 5 of the LTZ1000 datasheet regarding the difference between the A and Non A version:

The LTZ1000A should be set about 10°C higher than the LTZ1000. This is because normal operating power dissipation in the LTZ1000A causes a temperature rise of about 10°C.

Your heater ratio of 12k over 1k is really pretty low, even if you were using Non-A.  You always want the HEATER circuit to control the die temp, NOT the ZENER.  We always use a range of 13k to 15k over 1k for our needs with LTZ1000A, I know some here use 12.5k / 1k.  The cooler you get on your heater ratio means you have to keep a cooler ambient on your LTZ - and if that heater goes out of regulation (or if the Zener starts controlling die temp) you will have a guaranteed crappy Vref.

Also with the A version you normally don't need the 400k (or whatever value you empirically test).  If you think you must use that, I suggest you get the circuit running within spec FIRST, and then only add that extra resistor as required - and after you learn how the circuit works.  As the circuit ages you might find that resistor isn't helping LTZ1000a TC as much 5 or 10 years from now - and that's why LT recommends against it for A version.  It's your party though, do what you need to for your goal.

[Insatman]

The 400K TC resistor value is just a starting point.  You have to perform your own tests in order to determine the correct value to achieve the 0.05ppm/K or minimum TC drift.  In one case I needed to use 300K resistor and another a 680K.  I was able to do better than 0.05ppm/K with multiple builds, but it took considerable time and testing to get there.

Some of my testing here;
https://www.eevblog.com/forum/metrology/kx-reference/msg1369941/#msg1369941

Also check this thread for information on sources for LTZ1000(a) resistors;
https://www.eevblog.com/forum/metrology/resistor-set-for-ltz1000-positive-standard-7v-circuit/

thankyou so much for the references.  Saved me hours of trying to get through over 1000 posts on this thread

[Insatman]

Insatman:  As others have noted, an LT1013 will do everything you want it to do for a LTZ amp.  You don't really need expensive resistors around the LTZ, since the LTZ is more forgiving to resistor drift than something like a 7V-10V boost circuit - You just want decent resistors, and the heater ratio pair is most important. PWW are perfectly fine and the preferred choice, and Edwin will give you a deal on an LTZ set, too.

Yes, do some reading, but for sure read page 5 of the LTZ1000 datasheet regarding the difference between the A and Non A version:

The LTZ1000A should be set about 10°C higher than the LTZ1000. This is because normal operating power dissipation in the LTZ1000A causes a temperature rise of about 10°C.

Your heater ratio of 12k over 1k is really pretty low, even if you were using Non-A.  You always want the HEATER circuit to control the die temp, NOT the ZENER.  We always use a range of 13k to 15k over 1k for our needs with LTZ1000A, I know some here use 12.5k / 1k.  The cooler you get on your heater ratio means you have to keep a cooler ambient on your LTZ - and if that heater goes out of regulation (or if the Zener starts controlling die temp) you will have a guaranteed crappy Vref.

Also with the A version you normally don't need the 400k (or whatever value you empirically test).  If you think you must use that, I suggest you get the circuit running within spec FIRST, and then only add that extra resistor as required - and after you learn how the circuit works.  As the circuit ages you might find that resistor isn't helping LTZ1000a TC as much 5 or 10 years from now - and that's why LT recommends against it for A version.  It's your party though, do what you need to for your goal.

Thanks for the guidance on the ratio resistors.  Originally I used 12K because it was available and I could not find a 13K at that time.  I have a 13K resistor on order for weeks now...still waiting.  I suspected that cheaper resistors would work in some places and I do realize that the scaling stage is the most important.   I think my current problem is the op-amp used for the LTZ and have a plan to switch to the LT1013 for that service.   I will be building two boards up when all the parts and PCBs arrive.

[kj7e]

I had really good results using a Vishay VHD200 resistor divider network in my 7 to 10v buffer, but it took 6 months to get it and it was not cheap 

More info on the buffer here;
https://www.eevblog.com/forum/metrology/ltz1000-10v-buffer/

I found the temp set divider resistors to be somewhat more critical than the 120 ohm resistor, so I would still use the lowest TC parts you can get here.  In mine, I used 13K/1K, as others have mentioned beware of using anything lower than 13:1 to retain thermal stability.  Also, send Edwin Pettis a note and get a complete set of his PWW parts:  https://www.eevblog.com/forum/profile/?u=96921  His prices are extremely reasonable compared to Vishay and you wont have to wait forever.

[cellularmitosis]

Texas Components seems to have a 3 to 4 week turn-around time, but their VHD200's are definitely pricey (about $48 if I recall).

[kj7e]

Texas Components seems to have a 3 to 4 week turn-around time, but their VHD200's are definitely pricey (about $48 if I recall).

Not if you order 0.1ppm TCR with 0.001 match, I went overboard and ordered the highest spec they would sell me for the 25K2/10K.  That little bugger was more than twice that and they sure took their time to make it.  Unfortunately I had calculated that value for another LTZ1000A chip which I ended up not using due to some strange behavior I saw, so I had to make some adjustments in the trimming section.  The standard 13K/1K VHD200 with 0.01 match I used for the LTZ voltage divider was about  $43.

[Kleinstein]

The compensating "400 K" resistor is only working well over a small range, as the compensation is nonlinear, especially with a low set temperature and thus low heater power.  For this reason the suitable value also depends on the thermal setup and the set temperature /  heater power. The lower the set temperature the higher the value for the resistor. Also the thermal setup will have an influence- so inside a case the best value will be different from the bare board.

I would really question spending much more on the resistors than on the LTZ reference. If one really needs more, a 2nd layer of temperature regulation or a 2 nd LTZ1000 might be more economic than using resistors that don't make a difference. The good point about the LTZ1000 is that it is not that sensitive to the resistors.

[Insatman]

I think I killed an LTZ1000ACH.  I have no idea how/why it died.   It was working for months and recently i posted data here about it's temperature coefficient and got a lot of advise on improvement.  In an attempt test two of the suggestions:
1)  I un-powered the board.
2)  Changed R5 from 12K to 13.7K (i just got a Zfoil 13.7K on Ebay recently).
3)  Changed U1/U3 from OPA177FP to ADA4522 op-amps (in DIP adapters).
4)  Upon power up the output went to positive rail or about 14V.

I did a visual check and removed U4 to check the output of the LTZ directly...it was about 13.8V
I tried replacing the ADA4522 opamps with LT1006 units but same result.

Before I throw away a $50 reference I thought I'd post a schematic with various voltages measured and see if I'm just being stupid.
The heater does not appear to be getting any power.   Maybe the Zener is open circuit?

[TiN]

Which 4522 you used? Maybe you got confused the single and dual package versions?

[Insatman]

Which 4522 you used? Maybe you got confused the single and dual package versions?

4522-1  single package.

[martinr33]

The heater looks OK. If it was broken, I would expect to see a voltage differential across it. The problem is, U1 is only pushing 20mV, insufficient to turn on Q1.

Next step, check the voltage on the inverting input of U1. If it is not the same as the voltage on the non-infringement input, there is a problem with the op amp circuit.

[doktor pyta]

Replace the LTZ1000 with two NPNs and Zener. Then You will make sure if Your LTZ is dead.

[ArthurDent]

Hard to guess without actually having the board in front of us but a possibility could be a problem with the new op amp/adapter wiring. Can you set up a test board to see if the op amps in the adapters do work as op amps.

[Galaxyrise]

I find that pin6 voltage of 19mV to be inconsistent with the other measurements:
  * With the ref output at 13.9V, if the voltage divider is working it should be around 945mV 
  * The pin 8->7 voltage of 0.652 suggests the transistor is conducting, (though not as much as I'd expect with a base voltage of 0.9V). 
  * The heater control opamp is at its minumum output, suggesting its IN- is greater than IN+

The readings on the refamp side would suggest a dead zener to me, I can't otherwise justify that voltage drop across the zener.

Something I did after damaging a couple LTZ modules was to build up a fake LTZ using discrete transistors and a diode soldered on to some CAT5 that I could plug/solder in place of a real LTZ.  If the circuit was working with the "fake", then I'd put in a real LTZ.  Before putting in a new LTZ, maybe put in a fake one to verify the circuit.

[Kleinstein]

The LTZ circuit is kind of sensitive to a wrong output:
too low (e.g. to much load) and the heater will go too high
too high (e.g. oscillating, transients during turn on) and the transistor below the zener can see excessive base current that destroys it. Also the zener might break from over-current, though less likely.

AZ OP might take some time to get working when turned on. One more reason I would be suspicious about using an AZ OP. In addition there is absolutely no need for a low drift OP. Drift of the OP is attenuated by a factor of about 200. So anything better than 50 µV/K would be OK.

The kind of dummy LTZ is a good idea to test a new circuit.