Old fashioned zener 10V reference

[Conrad Hoffman]

So as not to OT the other device specific threads, here's a very conventional zener reference I built some years ago. It has stayed within a couple ppm and isn't terribly temperature sensitive. The trick is fine tuning the zener current for your particular zener, not just using the data sheet value. If the board were put in a simple oven it would be near immune. The capacitor needs to be tantalum for low leakage, or probably better yet, ceramic. I don't have data logging set  up, nor anything more than a 6.5 digit meter, but looking on the high gain scope, the noise level is similar to my 731A. Feedback values are based more on what I had, rather than perfectly optimized, but the performance suggests they're not far off.

You can also do the same trick with a 731A and similar, fine tuning the reference current for a much lower tempco.

[mmagin]

I like it.  Might build some of these with the new-old-stock 1N829A I've obtained and the 2DW233 I'll be getting from the group order.
Now I really need to build a little temperature-testing chamber.

I kind of wonder, are the better grades of 1N82x just ones which have a zero TC point that falls closer to the datasheet spec? 

[Conrad Hoffman]

Good question and I don't know the answer. It seems like any of them can be trimmed to zero TC. I do the whole board, so if some resistor tempcos get trimmed out in the process, all the better. It's the end result that matters. I posted this to show that very good results can be achieved with inexpensive and unsophisticated methods. The same circuit should be ideally suited to the other reference devices.

[enut11]

Hi Conrad
Thanks for sharing. What type/tolerance resistors did you use and what was the nominal zener voltage?
enut11

[Conrad Hoffman]

It's a 6.4 volt zener. Back when I bought them they were inexpensive, but today they start at about $9. The resistors are ordinary Mepco or similar RN55 1% parts. I probably selected  the feedback resistors for value, to get the trim centered correctly. Also, one changes the current setting resistor to the zener for zero TC, which shifts the voltage. Then the feedback to get back to 10 volts output. It may be necessary to go back and forth a couple times if you want perfection.

[RandallMcRee]

Conrad,
How do you determine where the zero TC operating point is? In particular, do you need an oven? Can it be done quickly?

Thanks in advance,
Randall

[zhtoor]

So as not to OT the other device specific threads, here's a very conventional zener reference I built some years ago. It has stayed within a couple ppm and isn't terribly temperature sensitive. The trick is fine tuning the zener current for your particular zener, not just using the data sheet value. If the board were put in a simple oven it would be near immune. The capacitor needs to be tantalum for low leakage, or probably better yet, ceramic. I don't have data logging set  up, nor anything more than a 6.5 digit meter, but looking on the high gain scope, the noise level is similar to my 731A. Feedback values are based more on what I had, rather than perfectly optimized, but the performance suggests they're not far off.

You can also do the same trick with a 731A and similar, fine tuning the reference current for a much lower tempco.

Hello Sir,

always a pleasure to read your articles on metrology.

1. can you tell me about the long-term drift characteristics of this design?
2. it seems that the while tuning the current through the temperature compensated zener, you would also change the voltage gain,
    how about decoupling the current through the zener and the voltage gain?
3. what would be the performance of a zener connected bjt in series with a diode connected bjt instead of the zener, the bjt can be selected for
    noise performance.

regards.

[Conrad Hoffman]

If one worries about all the possible sources of drift, using 1% metals and an IC socket should doom this circuit from the start. OTOH, my experience is once it settles down for a few weeks, it stays within a couple ppm for several years at a time. FWIW, my 731A uses metal films for the zener current and manganin wound resistors for the dividers (I think). The 731A has it's last official cal more than 10 years ago and in comparison with three other reference units, it hasn't moved more than 1-2 ppm. A lot of the drift specifications you see become much better after things have aged a few years. I'd rather have an ancient reference with a lot of history, than a brand new one with zero history!

My "oven" was nothing more than a cardboard box over the board with a lightbulb shining on it. Warm the board up a dozen degrees, cool it down (all slowly) while monitoring the voltage. Change current until you can cycle it with no significant change. If you temperature controlled the circuit, it probably wouldn't be capable of sub-ppm performance, at least not without better resistors. It's very good, but can't break the laws of physics. Or Ohm. IMO, sub-ppm performance is vastly harder to achieve and much more expensive as you need better meters and comparison standards. that's why we have threads on the LTZ parts! This circuit is a great way for somebody to get their feet wet with a discrete design, and for many it's all they might need.

As for decoupling TC and gain, you have to change the zener current to find the zero TC point, and that changes the zener voltage. The zener curve isn't a straight vertical line. Thus, you have to change the gain to get back to 10 volts, and that's just the nature of the thing. Adding more complexity than the minimum shown, isn't going to improve things. Certainly though, you can use this circuit with any sort of two terminal reference device.

[Andreas]

I kind of wonder, are the better grades of 1N82x just ones which have a zero TC point that falls closer to the datasheet spec?

Hello,

according to a post on volt-nuts board the zero T.C. current is usually from 4-11 mA for any 1N82x.
The 1N829(A) is only closer to the 7.5 mA test current.
And of course the A-types are selected for lower dynamic resistance. (which should influence noise).

With best regards

Andreas

[Mickle T.]

Here is a simple zener references I built some years ago.

[Kleinstein]

...
according to a post on volt-nuts board the zero T.C. current is usually from 4-11 mA for any 1N82x.
The 1N829(A) is only closer to the 7.5 mA test current.
And of course the A-types are selected for lower dynamic resistance. (which should influence noise).

With best regards

Andreas

A low dynamic resistance is also important to get a low sensitivity to resistor drift. I have not seen very much on the correlation with noise. They sometimes show noise as a factor relative to the thermal noise of the dynamic resistance, but I am not sure this is more than a way of plotting things. Especially with non buried zeners noise is more a question of defects causing 1/f noise and popcorn noise.

[enut11]

@Conrad Hoffman
Quote
"My "oven" was nothing more than a cardboard box over the board with a lightbulb shining on it. Warm the board up a dozen degrees, cool it down (all slowly) while monitoring the voltage. Change current until you can cycle it with no significant change. If you temperature controlled the circuit, it probably wouldn't be capable of sub-ppm performance, at least not without better resistors. It's very good, but can't break the laws of physics. Or Ohm. IMO, sub-ppm performance is vastly harder to achieve and much more expensive as you need better meters and comparison standards. that's why we have threads on the LTZ parts! This circuit is a great way for somebody to get their feet wet with a discrete design, and for many it's all they might need."
Quote

Conrad
Did you measure the temperature under the 'cardboard box'?
enut11

[Gyro]

One thing to remember wth 1N82x zeners (and all axial zeners) is the need for stress-free mounting. This is specified in the IN82x datasheets and also mentioned in the LM399 datasheet* (by way of comparison). Datron used to handle this by putting a large complete loop in each lead before soldering, basically using most of the supplied lead length.


Edit: * Actually, not the datasheet, the App note, NS AN161.

[Vtile]

Isn't that stress relief a thing in every single component - the PCBs included - when someone tries to archieve zero TC. I would be suprised if IC chips wouldn't suffer it at some degree also.

[Should be taken with grain of salt]1n821 I can't recall the datasheet, but on that were a one sentence that said that best voltage for buried zener is 6.5 to 6.8 volts (1N829 it were one of these zero drift voltage ref. IC datasheets and comparison to traditional zeners, if I did read between the lines they basicly say their IC is not better, but easier for production). [/Should be taken with grain of salt]

What I'm wondering atm. is that

A: How good TC one could archieve with the 1n821..1n827 series, by trimming the current.
B: How much of the noise of components are infact coused by the background radiation from ie. radon.

[Kleinstein]

With the correct current one can bring the linear TC to zero. What is left is mainly the second order TC. That is a parabola like dependence. So depending on the temperature range, one could get the TC to below something like 5 ppm/K over a 10 or 20 K range.
Compensating this second order effect is possible but difficult and may introduce extra drift. However having a stable temperature helps a lot against the square law error.

At the usually rathe high current, the noise due to radiation could not be a really big part of the noise. Radon is more a problem to very high impedance circuits with high voltage. 

[Conrad Hoffman]

I did have my darkroom thermometer stuck in a hole in the box, but this was so long ago I don't remember the exact temperatures.

Stress was a  huge issue for long term stability when I did my voltage reference for the MML articles, and I never had much success with the plastic packages in that regard. Three lead mounting of the device to the board was important, as well as floating the board from it's mounting wires. In this case, the fact that I wired it up on Vector board might be a positive. The leads underneath can easily flex for all the parts and don't put much stress on the bodies the way a PCB might. Stress relieving loops for critical parts sound like a good idea in that case.

One thought occurred to me about making up your own TC zeners from two diodes that I realized from working with a power zener attached to some other parts. Almost all the heat transfer is through the leads. If you solder two diodes together in fairly close proximity and maybe put a Styrofoam sleeve on them, they can't help but be at almost the same temperature. Conduction through metal is way more important than any other thermal path.

Mickle- "simple"? Really nice job!

BTW, I've always thought a perfect case for these references would be to put it in an old Eppley or Weston standard cell case, powered from a wall wart plug on the back.

[Conrad Hoffman]

Nice and similar to the circuit above. I used parallel resistors entirely to get the values I needed, not for any special stability or tempco reason. It was easier to get the trim within range by padding. I stock a lot of values, but not every one ever made! I assume your circuit values were chosen for similar reasons?

[mmagin]

Perhaps this deserves its own thread (maybe once I have a chance to try the idea):
I'm wondering about replacing the trimmer in a circuit like this with a AD5270 digital pot (and some kind of little PIC circuit to generate the SPI adjustments from up/down/save buttons or something).  The specs look quite good.

[Kleinstein]

Digital pots are often limited to relatively low voltage like 5 V. So not many types are actually very useful for adjusting the voltage. Usually the rheostat chips are less suitable, as the include the switch in the active path and thus tend to have a higher TC. The true pots used as a voltage divider can make use of good resistor matching. However they still have the bond wires at the ends.

[Conrad Hoffman]

You can do so much with a PIC these days! I'd think more in terms of injecting a voltage to create an offset, such that the full range of the control was only 5 ppm or so. That way even if the control circuit drifts, it doesn't have much effect. Push it into the negative input via a very big resistor or add a couple ohms on the ground leg of the feedback and again, drive it with a big resistor.

[nctnico]

Interesting circuit! 
Just one remark about a low leakage capacitor: a polyester (PET) capacitor will have an even lower leakage than ceramic X7R (which is somewhere in the several nA range at a few Volt).

[Vtile]

So as not to OT the other device specific threads, here's a very conventional zener reference I built some years ago. It has stayed within a couple ppm and isn't terribly temperature sensitive. The trick is fine tuning the zener current for your particular zener, not just using the data sheet value. If the board were put in a simple oven it would be near immune. The capacitor needs to be tantalum for low leakage, or probably better yet, ceramic. I don't have data logging set  up, nor anything more than a 6.5 digit meter, but looking on the high gain scope, the noise level is similar to my 731A. Feedback values are based more on what I had, rather than perfectly optimized, but the performance suggests they're not far off.

You can also do the same trick with a 731A and similar, fine tuning the reference current for a much lower tempco.

I'm not particularly good at op-amp circuits and analysis of them and my math seems to be off **. I wonder what is the purpose of the 8.3k resistor and tantalum on the non-inverting leg. Is it just a regular RC filter on the input?

I noticed somewhat similar circuit on the OP07 datasheet, but there were a pull up resistor from the zener cathode node to the positive supply voltage. I wonder what is the purpose, my educated guess is that it provides a run up voltage for the reference, resulting faster (but worse? *) stabilization of the system.

**The typical current for 1N4568 seems to be only 500uA so that part were actually correct. What seems to be an series RC-filter is still question mark for me.
* Edit. Added a question mark, to indicate that it is also a wild guess.

[Kleinstein]

The 8.3 K and the capacitor are just an RC filter. Zeners tend to be noisy and filtering helps a little - at this position filtering comes at low extra cost.

Depending on the supply for the OP, they might need a kind of extra power to make the circuit start up well. With just a single supply, the OP07 should be OK without an extra resistor. If at all a resistor to the OPs output might help a little.

[guenthert]

Here is a simple zener references I built some years ago.

  I was browsing the 'tubes for voltage references based on the 1N829 temp. compensated Zener as the prize for the LTZ1000 reaches stratospheric regions and couldn't help but noticing that the first one looks like a well executed  re-implementation of the circuit of the Solartron 7081.   

Can you share how it turned out, i.e. have you noise and drift figures?


(those interested in the original circuit google for SolartronReadme.txt, SolartronRef-1.PDF, and SolartronRef-2.PDF.)

[Tj138waterboy]

Have to resurrect old post to ask if one wanted to make a good current source to tempco test a 1n829, which circuit  construction would be more temperature stable? Opamp current source, wildar current source 2 or 3 transistor version or wilson current mirror 3 or 4 transistor version, or would basic single transistor current source be just as good? Will be powered by stable dc power supply hp6111a @ 12v. Will be cycling temperature of only 1n829a zener attached by wirewraped leads inside insulated cookie tin.  Current control circuit will be built on solderless breadboard or project board in room that temperature varies between 50-65degrees.

[Conrad Hoffman]

Just guessing, but I have to believe an opamp current source would beat any simple discrete circuit. Still needs a reference in there somewhere!

[Tj138waterboy]

I was thinking the same but all schematics of older voltage calibrators such as edc mv106, rv622 use bjt current mirrors when opamps were definitely available.

[Kleinstein]

The circuit like in the original post is a very stable current source. The current is set by the resistors (one ratio for the gain setting resistors and the one from the 10 V to the zener) and the voltage reference. A normal transistor based current mirror is not really that accurate.
In the old time it was common to use 2 zeners in a kind of bridge circuit, so that one zener would set the current of the other and the sum of the voltages would be the final output. In case one wants a 13 V reference, this may still be an option.

For the temperature stability one could with most diodes trim the linear TC via the zener current, but the 2nd order TC still remains. If really needed one may combine this with a regulated temperature. With a relatively low TC to start with the temperature regulation does not have to be perfect. I would consider a regulated temperature simpler than the 2nd order TC compensation circuit used in the Solartron 7071/7081.

[guenthert]

Have to resurrect old post to ask if one wanted to make a good current source to tempco test a 1n829, which circuit  construction would be more temperature stable? Opamp current source, wildar current source 2 or 3 transistor version or wilson current mirror 3 or 4 transistor version, or would basic single transistor current source be just as good? Will be powered by stable dc power supply hp6111a @ 12v. Will be cycling temperature of only 1n829a zener attached by wirewraped leads inside insulated cookie tin.  Current control circuit will be built on solderless breadboard or project board in room that temperature varies between 50-65degrees.

     50 years ago, H.A. Cole at Cern designed a voltage reference for wide temperature ranges.  He used William's ring-of-two reference as constant current source to drive a temperature stabilized Zener (1N825 -- I believe that's the one used in HP735A).
http://cds.cern.ch/record/1063330?ln=en

[dietert1]

In 2009 i made a 10 V reference with a AD587 (small Geller board). That IC included a buried zener, a precision resistor network and an opamp. One could connect a capacitor as noise filter. I glued an NTC on top and made an analog PI temperature controller with a TL071 and several hundred uF for the integrator. Few parts, nothing special except the reference itself.
The reference was good enough to check 6.5 voltmeters. Observed drift was some ppm over 10 years, in comparison to a pool of well aged HP 3456As.

Regards, Dieter

[Gyro]

It's a real shame that Joe Geller wasn't able to justify continuing to produce those boards and keeping his Fluke reference in Cal. I was lucky enough to purchase an SVR-T - the one with the thermistor based second order compensation that Lars came up with. He had a really neat micro thermal chamber for sweeping and compensating them. It spends most of its time turned off, so is hopefully still within a few ppm.

I still kick myself for not buying some extra 586 and 587LQs, and other stuff, when Joe was selling off his stock!

[Tj138waterboy]

That is a great read. Especially the findings that varying voltage 20-24v and 24-40 resulted in positive and negative voltage tempco. I believe that I may have came across this circuit in a Keithley schematic before but didn't understand what was happening with the use of 2 other zeners within the current mirror. Just assumed I was to provide a stable voltage for the current but now realize it more than that.

[TimFox]

A trick used in some older precision power supplies, that had a TC Zener in a small oven, was to connect the Zener and three resistors in a bridge circuit, that could be adjusted to cancel out the Zener's resistance by adjusting the bridge for maximum voltage.  I thought I had an example in an old Motorola diode handbook, but I can't locate it.

[Tj138waterboy]

Finally getting back to test setup after losing my jantx1n829a I found a lot of 10pc 1n825 and have been looking at different setups suggested and wanted to know if there is anyone that can see a problem with the way im looking at setting up a fixture to burn-in all 10 refs at the same time. Given The limitations of 5.5 digit hp3478a, mv106j dc voltage standard calibrator, and fluke 840 nA galvanometer it would seem to make more sense to null the references to calibrator maybe with a mux. The current configuration is feeding 15v into a lm7812 getting output of 11.8v. All 3904 transistors were matched to within 5mV vbe. The goal is to get the recommended 7.5mA to all zeners simplest and most accurate way possible which is why I am currently using a basic multi-stage current mirror I dont know if this has any drawbacks for my specific use purpose but am using a free circuit simulator "Every Circuit" to try to make the Improved Wilson current mirror into multistage and wanted to see if someone thats more knowledgable can tell me if my implementation is correct and or if theres a simpler/better way to make 10 mostly precision tracking current source from opamp better or cheaper. All resistors are 50ppm or better values are exactly as my test setup.

[Andreas]

Hello,

for a simple burn in cirquit I cannot see the advantage of having exactly 7.5 mA.
If I remember right from the discussions on Volt-Nuts the burn in was done at much higher currents (more like 40-50 mA)
(Google for "volt nuts some questions to zeners". I think it was user WarrenS which had a lot of experience)

After burn in the zener was not operated at 7.5 mA but on the "Zero TC" current which is individual for each zener (at a specified environment temperature) and usually in the range of a factor 2 around the 7.5 mA.
So the 7.5 mA is only a test current to select wether the zener is a 1N821 or a 1N829 at 25 deg C.

Essential is also to screen the zeners for "popcorn noise" to sort out the "stinkers".

After my experience the 1N82x sold nowadays have relative high 1/f noise (in relation to the high current).
So I stopped my experiments with 1N829.

with best regards

Andreas

[Tj138waterboy]

Im sorry for not detailing but the circuit is adjustable current mirror so i can adjust from around 6.5mA -  20mA. With the mirrors all tracking each other it should be quicker to find the better zeners per specific current vs testing individually. I will have a look into burning in at higher current thanks for the info.

[guenthert]

The goal is to get the recommended 7.5mA to all zeners simplest and most accurate way possible which is why I am currently using a basic multi-stage current mirror

   As others have already written, the Wilson current mirror isn't expected to deliver exactly the same current nor is the 7.5mA necessarily the right current.  I just can't help, but stating the obvious, that to get the exact same current to all Zeners they ought to be in series (string of Zeners).  That might actually be a reasonable approach to a 100V reverence as it reduces noise by n^1/2.

[Tj138waterboy]

Thanks Andreas, I did come across that exact article you mentioned previously which is why I decided to go with a batch of 1n825s instead on 1 or 2 1n829s. My current set resistor can easily be modified to lower the current to start from 4-5mA. I am just a hobbyist without any schooling on e.e. subjects and just want to build, take apart and learn as I go. My post was more asking if there are any issues noticeable with the construction and simulation of designed multistage improved current mirror due to I have never designed my own circuit only copied others and modified values. Its obvious that all mainstream variations of current mirrors work but there is no data I can find that anyone has built or tested the version I have simulated. Even if using a cascode setup adapted to the design could be useful or if I better match transistors vbe would be worthwhile. I understand this is in the metrology section way above where this probably should be asked but we all have to start somewhere. Thanks for all the knowledge you guys share.

[Vtile]

This was super fun build, I did one years ago. Today I powered it up again after it have been laying on under a pile of trash above the workbench for several years.. After an hour warm up, both my Keithley 197A (LM399) and the 1N823+OP07 reference the value set to 9.9999V which is not bad, it still is "too good" for the best measuring equipment I have the 197A.

The current source I was using were simple.. Forementioned Keithley as current meter in series with some 10 turn potentiometer, powered with Thandar lab supply (or were I driving the PSU against current limiter and using it as current source... can't recal). Then my MeraTester v640 (100meg, 1.5mV FSD) as a null detector against another channel of close enough voltage from Ch2 from thandar supply. Then the heat was cycled first by pinching the dut (1n823), then adjusting the current in a point where no movement of null detector was seen.. After that I did use a 12V lightbulb wrapped to foiltape (to allow only heat) put under and contact to the DUT,  then again adjusting the current to null out the voltage on null detector on this rather violent heatcycle.. This also involved to disassemple the current measuring setup and to use the Keithley 197A to measure the TC voltage after point was found. Repeated this to three 1N823s and did choose the best one (can't recall did I choose the one with widest TC region or something can't recal)..

Then calculated the needed resistance between the 1N823 and 10V output.
Calculated the needed resistance for amplification multiplier resistors.

Sorted out all my handfull of 15ppm 0.1% resistors for TC. This was done again with Thandar as supply and 4 of same value resistors as Weatstone bridge configuration. One as a dut and 3 others as voltage source for null meter. Dut temperature cycled with my fingers and variation detected with zeroed MeraTester v640 in lovest range possible (mostly 1.5mV). I might have used a 10 turn Bourns pot again at one leg of the bridge.. Can't remember all details, it was a mock up one weekend setup, with a few beers.

Calculated series parallel networks to match the needed resistor values and then
constructed the circuit in manhattan style to full copperclad board with attention for mechanical stresses.

As an afterthough I should have put more efford to the amplification multiplier resistor network it is sub optimal part.

[Conrad Hoffman]

Wow, an oldie, but I still have that board. I think an opamp current source would be best but I'm wondering if you can use the DUT for the current source and measure the output, sort of combining the whole thing. No specifics, just armchair pondering. For TC on small parts I use a couple small beakers of mineral oil and just dunk them. Very fast, though you might want a small lab hotplate or immersion heater (or a medium sized low value resistor.) to keep things warm.

[Victorman222]

but I'm wondering if you can use the DUT for the current source and measure the output, sort of combining the whole thing.

Yes, Conrad suggested the same thing to me along with measuring change relative to another reference and it worked well, i was able to characterize some soviet 9V reference diodes to <2 ppm/K over 10 degrees with dt830 multimeter.
I added multiturn potentiometer to control the current of the diode and the output voltage, the diode was on long wires fixed to a light bulb with a rubber band, measurement was done relative to same circuit but with fixed resistors. The graphs in the Microsemi datasheets helped me get my bearings showing the relation of TC to current and zener voltage to current.

Thanks everyone for the great info in this topic!

[Kleinstein]

The trimmers are OK to find the right gain and right amplification. However even good trimmers are usually not very stable over long time. So one should get at least most the resistance from a set of fixed resistors and a trimmer (usually a higher resistance like ~10 K) only for a small part. The is especially for the gain.

[Victorman222]

The trimmers are OK to find the right gain and right amplification. However even good trimmers are usually not very stable over long time. So one should get at least most the resistance from a set of fixed resistors and a trimmer (usually a higher resistance like ~10 K) only for a small part. The is especially for the gain.

Yep, the trimmer circuit is only to find the 0 TC current. Those soviet WW trimmers are something like 500 ppm/C so it's far from a reference. I wrote down the resistance values into an excel sheet and when it came time to using the diode i put the appropriate current setting resistor.

[VE2UM]

excellent idea about the 1N829.

However, you’ll have to replace the two 15k resistors with a 620 ohm device.

1N82x zeners are rated at 7.5 mA bias..

Also, may be a little pullup from the Supply to the opamp output to aid it supply this 7.5 mA 1N829 bias.

[EC8010]

I wasted ten minutes trying to find NS AN161 on the Internet. Then I did what I should have done first and looked in my library, finding it in the National Semiconductor Linear Applications Handbook. Doh!

Although an op-amp CCS is simple and usually best, sometimes you're forced to use discrete component circuits such as the Williams ring-of-two.

Another way to achieve temperature compensation is a string of references. Attachment shows the effect of connecting a string of BZX55C5V6, some in forward, some in reverse. The diodes were immersed in white oil.

[alm]

I wasted ten minutes trying to find NS AN161 on the Internet.

Here is an electronic copy for anyone else reading along: National AN-161: IC Voltage Reference has 1 ppm per Degree Drift.

[Vtile]

This is actually a really good project, not just for hobbyist metrology wannabes (like myself), but to get a stable and relatively (take your best voltage measurement device and call it done) known voltage source for the hobby lab. Output of the reference could be fitted with eg. 25 turn potentiometer which output is then puffered with opamp set as voltage follower (buffer) and/or voltage to current converter. As a single precision potentiometer used as (... eh, well..) potentiometer the tempco shouldn't be that much of an issue since the material is homogenous both sides of divisor. Buffer then makes this arrangement as "ideal potentiometer" as they were called in 70s.

I had originally idea of building three pieces of these and my intention was to make this kind of arrangement to get more stability and controllability for voltage and current for characterise the two or three other zeners and other parts. Only issue is that the first one is too good to begin with and I would need to upgrade to 6.5 digit meter to keep interest. Haha.

[Rax]

"My "oven" was nothing more than a cardboard box over the board with a lightbulb shining on it. Warm the board up a dozen degrees, cool it down (all slowly) while monitoring the voltage.

Actually, great idea. Simple, easy to implement. May steal it?....

Needing to find tempco resistor for an LTZ1000 kit.

[Andreas]

Only issue is that the first one is too good to begin with and I would need to upgrade to 6.5 digit meter to keep interest. Haha.

Hello,

why not measure the difference of 2 references in 100 mV range and optimize further.
(simply use your DMM as null-Voltmeter).

Most 6.5 digit multimeters deliver more resolution than that displayed in default mode.
Just use the math function or the interface to get higher resolution.
So you can still have a lot of fun before you need a higher resolution meter.
https://www.eevblog.com/forum/metrology/getting-one-more-digit-from-a-6-5-digit-meter-without-using-gpib/msg1300529/#msg1300529

with best regards

Andreas

[iMo]

Only issue is that the first one is too good to begin with and I would need to upgrade to 6.5 digit meter to keep interest. Haha.

Hello,
why not measure the difference of 2 references in 100 mV range and optimize further.
(simply use your DMM as null-Voltmeter)..

That IS the best method actually
My first measurements were made with a diy nullmeter (LTC1050 opamp with a 100-1000x gain) wired floating between two references, with a 10turn ww pot to get them nulled somehow. The dmm was a 3.5digit meter, and I saw X uVolts drifting as well 

[Mickle T.]

Simple low noise voltage reference with 4 zeners. TC ~1 ppm/°C, 0.009 uV/V RMS noise (0.1-10 Hz).

Update 1: Fixed a stupid error in the circuit diagram

[EC8010]

I'm guessing that 3C108C is akin to 1N829? An Internet search didn't find them.

[Mickle T.]

This is a typical example of temperature-compensated zener diodes. It is difficult for search engines to find, since the letter “C” is written in Cyrillic.
https://www.optron.ru/catalog/stabilitronyi/2s108s/

[mawyatt]

Watch out as VD1 (12V) is across the 15V power input!!!

Best,

[Vtile]

Nifty design.

[Andreas]

0.009 uV/V RMS noise (0.1-10 Hz).

Hello,

you obviously measure the 6.5V output with relative high impedance ~130 Ohms.
So how is the input impedance of your LNA measurement amplifier?
And did you correct for the "voltage divider"?

with best regards

Andreas

[Mickle T.]

You are absolutely right. The internal resistance of the source is 134 Ohms. At the same time, the minimum input impedance of the amplifier is limited by a differentiating chain resistor of 3.3 kOhm. Therefore, noise amplitude correction was introduced.

[iMo]

In my $5 noise indicator project I reported aprox 2uVpp in 0.1-10Hz measured on a junkbox diy 6.9V zener @3.6mA.
2uV/2/6.6/6.9=0.022uV/V RMS in my case of 4 in parallel. His is 2x lower with special diodes @ 2x higher current, better construction and better LNA, so it fits well..
PS: here..

[Victorman222]

I wanted to measure some diodes for zero TC current and needed to make a reference to do relative measurement. It didn't need to be anything fancy but at the same time i wanted to try to make a hermetic box for it. First try was with some steel that solders nicely, but it didn't turn out hermetic (maybe instead could repurpose something like hermetic paper capacitor case). Then i remembered that i have some hybrid IC cases, quickly drew the most basic schematic, designed and etched PCB. The performance will probably be somewhere between good and bad...

[iMo]

After you test your prototype - for the final version I would cleanup the pcb carefully, also - when you are keen on filing the resistors - try to setup the divider from a larger value and a small value, and then do file the small value only..