There is no real need to measure temperature of the LTZ1000 itself. The temperature sensor is on the chip and the temperature is very well regulated - down to the mK range.
Temperature is more effecting the resistors and maybe thermal EMF from the contacts LTZ to copper. Another point is the change in heater power that might effect the reference.
So a temperature sensor would be better placed at the case or at the resistors.
Branadic, quick question for you, do the LTZs have >4.2 mm of vertical clearance when in the Hammond case? I'm considering socketing the LTZs if it will work. Also, do you have any plans for how you're using the thermistor? I'm wondering if it's thermally a terrible idea to drill a small hole in the LTZ plastic covers (thanks for the models) and placing the thermistor in direct contact with the case. I have to imagine the thermal transfer would be minimal, but...
There is for sure enough space inside the Hammond case above the LTZ, but do me a favor and don't use a socket for your reference. I do have a minimal gap between board and LTZ to avoid any stress transfer between them.
The temperature sensor be it a thermistor or a platinum sensor can be used to measure temperature dependance of the whole circuit to verify everthings works a treat. And as Kleinstein already stated, it's not the temperature of the LTZ you want to monitor, but all the other components such as the resistors.
-branadic-
Received some of Mr Pettis' resistors during the week so finished stage 2 of a x6 DrFrank's PCB based Ltz Reference Bank
It is located above TiNs KZ PCB reference bank (plus one old 3458 LTZ reference board alone)
(The 'code' next to each output refers to
A or C - LTZ1000 A or C,
1 or 2 - op amp LT1013=1 or 2057=2
Resistors V =vishay. , U= TE's UPW series, P=Pettis Engineering
All have 2 stage linear PSUs
1st common rail LM38xx,
2nd stage each LTZ has its own TPS7A4901 ultra low noise in a separate box
All connected to mains UPS
Pomona 3770 posts for Ref 4-15
My Thanks to EEVBlog VoltNuts esp TiN, DrFrank, Edwin Pettis just to name a few.
Robert
Hello,
Looks like a serious ageing test.
with best regards
Andreas
Thanks, references 1 to 9 and `3458` have been almost always switched on since Aug 2017. Reference 8 up until Sept was doing brilliantly but then dropped about 8uV. But now is pretty consistent again - within 1-2uV
My results also may be effected by the 800km road trip my old 3458 went on, for calibration in September - all in spec - so no adjustment. The results before and after cal don't appear to have a consistent trend.
One I get a few months of data from the new references , I read every few weeks I will publish here.
Robert
Hi!
I bulid my LTZ1000A Reference based on Dr.Frank's PCB,and the LT1763 Supply designed by Cellularmitosis.
The Resistors ratio is 12K4:1K.
Keep the long leads of LTZ1000A and protected from airflow by a foamed plastic block.
The plastic box is temporary, I'll replace it with metal shell later.
I put a layer of aluminium foil on the outside of the plastic box to reduce EMI interference.
I'll do some tests later.
Thanks to Dr.Frank,Andreas,branadic,Cellularmitosis,TiN and others.
@ha110ween
Nice Job. Thank you for sharing, and welcome to the forum.
Some information about 10V 'divider' konfiguration and what type of binding posts you use would be nice, after some days of settle down maybe also some logged data.
I like it.
@ha110ween
Nice Job. Thank you for sharing, and welcome to the forum.
Some information about 10V 'divider' konfiguration and what type of binding posts you use would be nice, after some days of settle down maybe also some logged data.
I like it.
Hi hwj-d,
Thank you
The output of the Zener is 7.08840v,so the resistors are (18+250+10K):25K.
The posts were removed from my old instruments.
I have a Keysight 34465A and a Fluke 2638A Data Acquisition Unit,I'll use that to record drift data.
Would building this into a custom copper box enclosure be better or worse than building into an aluminum box? Would a solid copper box aid at all in reducing outside emf from the circuit? Please and thank you in advance.
Would building this into a custom copper box enclosure be better or worse than building into an aluminum box? Would a solid copper box aid at all in reducing outside emf from the circuit? Please and thank you in advance.
A steel box would be best for EMI, Aluminum and Copper would provide good RFI shielding but not much for EMI.
What about the other boards that have been spread over the whole world?
As many others: waiting for resistors ... I'm in the last batch for Edwins resistors. The other day I got a mail from Texas Components that they had misplaced my special order from half a year ago, and had produced exactly none of the ordered resistors (some strange story about not beeing able to mfg resistors with 0.125" spacing).
Regarding the Vishays, my alternative is to buy Z201 and test them for tempco, which will cause further delays.
Precision takes time.
What about the other boards that have been spread over the whole world?
As many others: waiting for resistors ... I'm in the last batch for Edwins resistors. The other day I got a mail from Texas Components that they had misplaced my special order from half a year ago, and had produced exactly none of the ordered resistors (some strange story about not beeing able to mfg resistors with 0.125" spacing).
Regarding the Vishays, my alternative is to buy Z201 and test them for tempco, which will cause further delays.
Precision takes time.
Not at all happy with Texas Components and custom Vishay delivery times lost orders and email excuses
Here is one unit based on the previously shown board. It uses Rhopoint 8G16D resistors with 1k:12k for temperature setting at the LTZ1000CH (manufactured in 2018) and 4k (2x 2k) + 10k UPW50 resistors plus a 20ohm 64W pot with 22ohm resistor in parallel to one branch of the pot for the 10V boost circuit.
The voltage reference was adjusted to the corrected readings (reads 9,9999945V instead of 10,000000V with respect to the calibration report) on our 3458A (still within 90day specs). It will now take the big trip overseas in its double thermally insulated package to reach its new owner. Hopefully it does a good job there.
-branadic-
The graph is imo very well laid out and easy to read. If possible, I would like to know what program / software was used for the multi-graph template?
The answer is GNU Octave / Matlab.
-branadic-
branadic demanded photos some time ago:
here are my first two references, at this time without insulation/housing. For a first test i populated the first one with a LTZ from a broken HP3458-reference with strange high 7,245V and the second one with a new LTZ bought at Digikey. Both are compared with a KVD and a Null-meter (it is a clone of the one in Conrad Hoffmans calibration lab article) against a Voltagestandard-Reference and each other.
Hi,
I have measured my LTZ1000 References(Dr.Frank 's PCB) with my 34465A.
The data was loged by RPi_LogNut through the LAN port.
The python code of the DMM setup like below:
inst.write("CONF:VOLT:DC 10,MIN")
inst.write("VOLT:IMP:AUTO ON")
inst.write("VOLT:ZERO:AUTO ON")
inst.write("VOLT:NULL:STAT OFF")
inst.write("TRIG:SOUR IMM")
inst.write("TRIG:COUN 1")
inst.write("TRIG:DEL:AUTO OFF")
inst.write("DISP OFF")
The data is not averaged and the temperature(red line) is readed from the DMM internal sensor.
The python code is:
Temperature = inst.ask("SYST:TEMP?")
The drift is caused mainly by the LM399 or the resistor networks of the 34465A,so the T.C. of the 35565A is about 0.5ppm/°C(Non ACAL).
I will measure the T.C. of my three LTZ1000 References later.
Thanks.
What should be a tolerance for non-precise resistors please? I just do not have 1% resistors in all nominals and wonder if I can use resistors with tolerance 5%? I am talking about R9, R10, R11, R6, R8 etc.
The extra resistors at the temperature regulator (e.g. base, 400 K nominal, 1 M and 10 K) are not critical in there value / tolerance. The 1 M and 10 K set the frequency response - so no real need to be more accurate than the capacitor. The 400 K resistors is used to compensate for some residual TC - the best value depends on the temperature, thermal insulation and reference chip. The best value could be considerable away from 400 K (e.g. 270 K - 1 M). Especially if a small resistor is used, it should be stable and thus more like 1% grade.
branadic demanded photos some time ago:
here are my first two references, at this time without insulation/housing. For a first test i populated the first one with a LTZ from a broken HP3458-reference with strange high 7,245V and the second one with a new LTZ bought at Digikey. Both are compared with a KVD and a Null-meter (it is a clone of the one in Conrad Hoffmans calibration lab article) against a Voltagestandard-Reference and each other.
Both those LTZ's are the ACH version and so you shouldn't need the 400k resistors as they are for the CH version AFAIK.
branadic demanded photos some time ago:
here are my first two references, at this time without insulation/housing. For a first test i populated the first one with a LTZ from a broken HP3458-reference with strange high 7,245V and the second one with a new LTZ bought at Digikey. Both are compared with a KVD and a Null-meter (it is a clone of the one in Conrad Hoffmans calibration lab article) against a Voltagestandard-Reference and each other.
Both those LTZ's are the ACH version and so you shouldn't need the 400k resistors as they are for the CH version AFAIK.
Even the 'A' can benefit from TC fine tuning. Also, 400k for the TC resistor is just a place to start and the optimal value can be above or below this. I think I found one of my 'A's needed 680K, while another needed 300K.
See here;
https://www.eevblog.com/forum/metrology/kx-reference/msg1369941/#msg1369941
branadic demanded photos some time ago:
here are my first two references, at this time without insulation/housing. For a first test i populated the first one with a LTZ from a broken HP3458-reference with strange high 7,245V and the second one with a new LTZ bought at Digikey. Both are compared with a KVD and a Null-meter (it is a clone of the one in Conrad Hoffmans calibration lab article) against a Voltagestandard-Reference and each other.
Both those LTZ's are the ACH version and so you shouldn't need the 400k resistors as they are for the CH version AFAIK.
Even the 'A' can benefit from TC fine tuning. Also, 400k for the TC resistor is just a place to start and the optimal value can be above or below this. I think I found one of my 'A's needed 680K, while another needed 300K.
See here; https://www.eevblog.com/forum/metrology/kx-reference/msg1369941/#msg1369941
Interesting.
Pardon my ignorance, but how did you come up with the TC resistor values? Purely empirical?
I have a non-A LTZ1000 and thinking what values I should start from.
Interesting.
Pardon my ignorance, but how did you come up with the TC resistor values? Purely empirical?
I have a non-A LTZ1000 and thinking what values I should start from.
I have published an elaborate description of my setup, and also how I measured and iteratively trimmed the T.C. of the assembly:
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg1336573/#msg1336573See the small table in my 4th thread.
I always started without R9 (= inf.). Any resistor in between 240k and 1M will increase the T.C., so you can only trim reference assembly with an initially negative T.C.
I did not check for the A version, it's initial T.C. , maybe its sign might be different, but I assume that the increase of the T.C. by decreasing value of R9 will be the same.
That bootstrap method I have used the measure the T.C. can probably also be done with a LM399 based DMM.
Frank
branadic demanded photos some time ago:
here are my first two references, at this time without insulation/housing. For a first test i populated the first one with a LTZ from a broken HP3458-reference with strange high 7,245V and the second one with a new LTZ bought at Digikey. Both are compared with a KVD and a Null-meter (it is a clone of the one in Conrad Hoffmans calibration lab article) against a Voltagestandard-Reference and each other.
Both those LTZ's are the ACH version and so you shouldn't need the 400k resistors as they are for the CH version AFAIK.
Even the 'A' can benefit from TC fine tuning. Also, 400k for the TC resistor is just a place to start and the optimal value can be above or below this. I think I found one of my 'A's needed 680K, while another needed 300K.
See here; https://www.eevblog.com/forum/metrology/kx-reference/msg1369941/#msg1369941
Interesting.
Pardon my ignorance, but how did you come up with the TC resistor values? Purely empirical?
I have a non-A LTZ1000 and thinking what values I should start from.
Very much the same way Dr. Frank explains. But basically built an environmental chamber out of an ice chest to chill and warm the reference while monitoring the temperature and voltage deviation with a 7.5 dig DMM. Repeated the tests many times over a few weeks.
Hello
my heater is an old ion. (Bügeleisen)
Do not take it too serious.
Just a fast solution for me. :-)