Stacking voltage references

[Tj138waterboy]

I finally have acquired my 10+ ref102 cp voltages to tinker around with to try a 100v and or low noise paralled 10/1v reference. I have what may be a simply answerable question but am thrown off by data given on how to power each ref.  The uploaded schematic suggest you pretty much have to supply around 1.2v more than ref value x number of stacked refs. I tried powering two references each with its own 12v power supply and just connected output in series which resulted in 20v as expected. Is there inherently anything wrong with using this method to obtain final output voltage rather than the way the attached example shows? I will upload schematic of my setup when i get back to a computer if needed.

[Kleinstein]

Using a separate supply for each reference chip is possible, but tends to be a lot of effort.  With more references in series it may be easier to have the supplies in series and thus use only one higher level suppl.  This is especially simple with a reference chip that can work in shunt mode. However this is not as flexible as totally separate reference circuits.

[RandallMcRee]

Well,
Have you seen this? (two references in one, hackaday article and actual technical document it references)

https://hackaday.com/2016/10/03/stacking-voltage-references-to-high-voltage-extremes/

Not sure what technical foot you are standing on.

[Tj138waterboy]

Those documents are what puzzles me. They mention stacking by connecting Vout from ref1 to Vin of ref2. They also mention there is no technical documents on proper way to do this as it was an experiment. My process will be more like how you would put batteries in series, Vout from ref1 to ground on ref2. It sounds easier to do it this way as I have already created makeshift single trimmable modules for individual testing as per the ref102 datasheet precision trim circuit. I have tested 13 ref modules and can trim all within 1 lsd according to my hp3478a. This is using non precision parts as well. 100ppm 1meg resistor and cheap ebay 3296 multiturn 20k pot.

[Tj138waterboy]

I am also trying to wrap my head around if it would be possible to use a single 12V rail and power each ref through a zener diode. Will this isolate each ref as well or would I also have to isolate ground as well, if so can ground be isolate through a zener as well?

[Vtile]

Hello, you might be interested of this application note from TI. It is different ref-chip though.
http://www.ti.com/lit/an/sbaa203/sbaa203.pdf

[Gyro]

There's another thread about stacking REF102s / AD587s around here somewhere.

EDIT: Ah, here...  https://www.eevblog.com/forum/metrology/stacking-ad587s-for-higher-output-voltage-(20v-30v-etc-)/msg1317592/#msg1317592 (link fixed)

[Tj138waterboy]

Thanks for the link. I have been over that document as well. I think the method I am trying to use is undocumented by the looks of it. I thought maybe someone with better equipment than me may could cobble up 2-3 of these in series to see if there is something that I can't with 5.5 digit resolution, in terms of drift or noise because mine seems rock solid with minimal effort.

[Gyro]

Try relocating your thread to the Metrology section (I think you can do it by editing your OP - either that or hit "Report to Moderator" and ask them to do it). You might have a better chance there.

[David Hess]

Using multiple floating power supplies instead of a single higher voltage power supply can work however it shunts the connection points to the reference chain with common mode capacitance to ground which may or may not matter.  There is also the consideration that at high voltages, the floating power supplies need to be able to hold off the common mode voltage that they see so it is likely easier to use a single power supply.

[Tj138waterboy]

I did wander about about how the common ground would come into play using both methods and also thought about if implementing a virtual ground would be possible or even beneficial.  I have a telecom background so in my mind im seeing common mode voltage as being induced by long leads/traces from AC source. I don't comprehend how any meaningful amount can be introduced if I go with using individual A23 12V battery cells for each reference other than atmospheric and self made zener noise. Again I don't know if this will even play a part with only 5.5digit resolution.

[Tj138waterboy]

So I tried a couple of configurations using diodes as well as virtual ground with no success. Now waiting for 10 a23 battery holders and batteries to just give each ref its own supply. My understanding of diodes and virtual ground led me astray in trying to isolate supply ground from each other but I would like to understand whythis didnt work if anyone cares to explain.

[IanJ]

Hi, I haven't read through this thread fully to see what exactly you are trying to achieve......but how about this DaveCad drg I just knocked up.

Ian.

[Tj138waterboy]

 Yes I have found quite a few application notes similar and can verify as far as the resolution I have available there are no negative side effects of placing ref102cp in series this way. The output seems perfectly linear. There is a roadblock as far as my Idea of using the 12 volt a23 batteries to power each reference individually as they are only around 55 mah. That would probably give less than 12 hours runtime given typical current draw is 1.4ish mV.  Now looking into some form of isolated ldo or dc-dc converter but the ones I found soo far are extremely noisy.

[Tj138waterboy]

I will try this method using single supply this afternoon but my goal was to have isolated ground on each ref and have capabliity to switch vout from series to parallel to essential have a semi precision 10-100v ref or low noise 10v by means of 10 refs in parallel or even possibility paralleling 3 groups of 3 refs for low noise 30v ref. I will post schematic of what I have soo far to see if anyone has better syggestions later.

[ArthurDent]

This scheme seems to be much more complicated than necessary. For decades Fluke and others have been making high voltage standards by using one low voltage stable reference, a voltage divider, and an opamp. much simpler and you can get very high accuracy.

It's just a modification of the way every linear regulated power supply works. 

[IconicPCB]

While a single reference followed by a suitable amplifier may look simpler, the amplifier now needs to be followed by a precision divider and yet another amplifier... how does this compare to stacked reference solution cost wise?

[ArthurDent]

While a single reference followed by a suitable amplifier may look simpler, the amplifier now needs to be followed by a precision divider and yet another amplifier... how does this compare to stacked reference solution cost wise?

Precision dividers aren't that expensive and wouldn't be followed by another amplifier which would degrade the accuracy. The divider that is used to sense and ratio the output has to be right at the output terminals and any voltage or current amplification has to be within the feedback loop or it could be almost meaningless. Check how it is done on the LTZ1000 thread on the board design by Dr.Frank which I modified by adding a transistor, again, within the feedback loop. Although only increasing the 7.08003V reference to 10.00000V, the idea is the same for any design. I used separate sensing from the divider to try to keep the output within a few ppm despite load or input changes.

[Kleinstein]

In many cases one would not need an "extra" amplifier after the divider.  A stacked reference may also need an amplifier to buffer the output, but this one is way less demanding: the stacked reference is low impedance (like a few ohms, depending on the refs used). A higher voltage divider tends to be relatively high impedance (e.g. 20 KOhms) even at the output, especially if only something like 10:1.

There is another use for a stacked reference: one can compare/measure the individual reference and this way do something like a 10 V to 100 V step. Affordable dividers are usually not that long time stable. It takes something like a Hamon type with periodic adjustment / check to get a long time stable / accurate ratio. Even there the accuracy may be limited by self heating effects and leakage currents.

If a suitable meter is available it could be worth to have all the switches needed to choose between series and individual (maybe also averaged) use. For averaging it depends on the references how much of an effort is needed - with low impedance references one could directly use resistors for averaging. It also depends on the effort for the current supply.

Stacking reference is the accepted first principles way (by definition of voltage) to add voltages. A divider would need calibration.

A cost comparison is not that easy, as many reference in series / parallel are also lower noise and could cancel out some drift peaks. So if would be more like comparing something like a stack of 14 LM399 based references to an LTZ1000 based with step up amplifier. The cost could be comparable, but the performance would be different:  Chances are the stack is better at 100 V, while an LTZ1000 is better than the average over 14 LM399 at 7 V. So it depends on the use.

[iMo]

FYI - from REF102 datasheet.

[David Hess]

A stacked reference may also need an amplifier to buffer the output, but this one is way less demanding: the stacked reference is low impedance (like a few ohms, depending on the refs used).

Some references will only source or sink current well making stacking difficult.  But sometimes an alternative to a buffer which provides source and sink capability is to preload the reference output with a current which may only require a resistor.

[Tj138waterboy]

This seems like something anyone can do with standard parts rather easily and I have yet to see anything for high voltage DC with any means of stability or acuracy readily available other than a voltage calibrator. Only an output buffer filtered for the 10 and 100V with the latter being harder to find any decent offerings, seem to be the only pice missing other than a section for an hamon divider which I want to incorporate as well. The datasheet says the ref is insensitive to different loads but most precision builds I have seen preloads the output to have a bit of current draw. Once again you guys are more experienced and have better equipmnet to be the judge. Should I care about Dc-1Mhz noise on this or putting on a buffer at all? Ideally I would like to make it useful to Cal club seeing how most I have observed here worry more about the 1-10 volt accuracy rather than 100v.

[ArthurDent]

I have a lot of power supplies, some very high accuracy, and none of them use stacked references.  I’ve never seen one built using stacked references by any major company. Check any supply or voltage standard built over the last decades by HP, Fluke, etc. and you won’t find any stacked references for several good technical reasons including you don’t always want an output that is a multiple of 10 volts. That would severely limit its usefulness. Here’s a photo of some supplies I have and the EDC, for instance, has 4 decade ranges from 0mv to 300 volts in .0001 volt steps. To get any type of real accuracy you need an error amplifier inside a feedback loop.

[Kleinstein]

The JJA reference is essentially stacking a lot of low voltage references (Josephson elements) to get something like 1 V or 10 V.

Some of the fluke calibrators start with two  6.5 V references in series.  In the old days of 1N82x zeners it was also common to use effectively 2 in series to get some 13 V.

For a normal instrument there is a limited use to have a stack of many references. This is more like a volt nut thing if the resistors for gain are not good enough.  Ti has an app. note for such a large stack actually used for a high voltage source - not as a commercial instrument, but more like a one off solution at a metrological lab.

[Andreas]

  In the old days of 1N82x zeners it was also common to use effectively 2 in series to get some 13 V.

Hello,

The main idea was not getting 13V but to get a wider "zero TC" range by selecting the diodes accordingly.

with best regards

Andreas

[Tj138waterboy]

The JJA reference is essentially stacking a lot of low voltage references (Josephson elements) to get something like 1 V or 10 V.

Some of the fluke calibrators start with two  6.5 V references in series.  In the old days of 1N82x zeners it was also common to use effectively 2 in series to get some 13 V.

For a normal instrument there is a limited use to have a stack of many references. This is more like a volt nut thing if the resistors for gain are not good enough.  Ti has an app. note for such a large stack actually used for a high voltage source - not as a commercial instrument, but more like a one off solution at a metrological lab.

And this is simply a one off project just to experiment and learn, cheap and simple as possible. The endgame which may go against ultra precision is to make everything semi-modular with header pins or slots. Can always add in a divider/gain loop/current source etc. This is just my attempt to see what I can do without lm399/ltz1000 but get very good stability and drift but mostly reduntacy with self testing against each other.  Am I wrong for thinking this is a better homegamer way of getting as good as or better than advertised vref specs by stacking and paralleling as opposed to spending a small fortune for a hv dc power supply or calibrator?

[Kleinstein]

  In the old days of 1N82x zeners it was also common to use effectively 2 in series to get some 13 V.

Hello,

The main idea was not getting 13V but to get a wider "zero TC" range by selecting the diodes accordingly.

with best regards

Andreas

The usual old circuit used a bridge like circuit, using one diode and a resistor to set the operating current of the other diode. This way there is no need for an extra amplified signal for the current, like in the usual circuit used for a single 7 V reference.
Using 2 references does not really help with the near zero TC. The linear part depends on the currents and thus the accuracy of the currents. So it is only the normal square root factor for random errors that applies on averaging.  The second order part is very similar (especially same sign) for all the diodes - so averaging does not help here.
It is also a little easier and less sensitive to the resistors to go from 13 V to 10 V than to go from 6.5 V to 10 V (it's about the same starting from 7 or 14). 
However I agree that this is different from the stacking idea here, using many reference.

Stacking a set of reference is a valid alternative to a special higher voltage calibrator. Using many independent lower voltage supplies can replace the higher voltage supply - though they are often the more expensive way.

For a high voltage reference the amplifier way, one does not necessary need an LM399 or LTZ reference. The difficulty, limitation could be more with the resistors to set the gain factor. However with only 1 reference needed one could afford an LM399 or even an LTZ1000 in stead of many ref102 or similar. While the noise improves in the series configuration (so 4 x ref. 102 would have good noise compared to 1 LM399), many reference do not help much with the TC and not significant with long time drift of the reference itself. A drift advantage comes from avoiding the divider / gain stage.

[Tj138waterboy]

Just noticed another trap after building my 10 test modules where as some have higher base voltage that requires more trim which in turn is causing more drift. Found quote in ti thread stating "first, you need to know that trimming the output voltage of REF102 will change the voltage drift by at least 0.008ppm per mV of trimmed voltage."  The worst case I have found in my 2 batches needed trimmed over 500 mV so I will probably have to bin a few refs or buy more to samples closer to optimal value untrimmed.