Ultra Precision Reference LTZ1000

[3roomlab]

After all I never intended KX modules as finished product

I knew it 
you must be happy so many people are "mysteriously" hooked onto an unfinished design
now they have to throw it away
no wait, take out the LTZ, the LTZ is innocent ! 

[Edwin G. Pettis]

If anyone's objections are baseless, irrelevant and misleading it is your comments, if you think that all research is published, you're very wrong, a great deal of research at semi fabs are private and well guarded from snoopers such as the competition and covered by NDAs as anyone would expect.  What gets published is probably only 10% of the real research and if you think Universities are a prime source, think again.  Your sources are very much limited in scope so don't assume you know everything because it has been 'published'.

Do you really think your puny efforts at measurements could detect such things as micro-crystalline changes over any period of time?  Don't make me laugh, some of these fab companies have tens of millions of dollars of the best equipment you can buy, way out of your league, so don't try to blow smoke up every's asses with your ignorance.  You can't verify anything like that, you have no way of verifying it and in such case you have no grounds for making such nonsensical unsupported statements.

The best way to protect a sensitive voltage reference circuit is to remove the external sources of noise and if necessary (which is most of the time) properly shield the enclosure with the correct materials.  Since you have no way of seeing what those capacitors are really doing to the LTZ you have no right to claim that nothing is happening at the crystalline level as it the case.  You already know that thermal excursions can affect the LTZ, whether up or down in temperature, it definitely produces a measurable change in the output when the LTZ has been kept at one temperature and then changed to a different temperature.....that is evidence of changes to the crystal matrix.  That is why you should set the temperature of the LTZ and then leave it alone.  Thermal shock is one example. physical shock is another, try dropping an LTZ on the floor and see what happens to the voltage, small change again.  There is a whole list of things that can affect the output voltage of the LTZ zener, to deny that current pulses from capacitors can't do that is plain silly.  It is just that some affects can be easier to measure than others, thermal shock probably being the easiest to detect.

Just because you can't measure it, don't claim it doesn't exist because someone else is likely able to measure it with the right equipment.

Since you're associated with an University, why don't you try reaching out to one of the German semi fabs and ask them about it and see just how much they'll tell you on the subject.  Let me know how successful you were.......

Oh, by the way, it is well known that capacitors convert voltage pulses into current pulses or has that bit of theory slipped past?

[ap]

Mr Pettis, while Frank and me and probably many others here agree that preventing the RF disturbance from even entering the board is the correct answer to the problem described rather than applying two brute force 100nF caps accross the 120R/1k resistors, and while we vaguely remember that RF voltages applied to capacitors result in a current from the source and through the capacitor and while we concurr that current spikes through junctions (as well as voltage spikes across junctions) may have an impact on these junctions (up to being distructive), would you shed some light based on your very special knowledge, unavailable to us, how an RF signal induced into the circuitry arround said LTZ pins and shorted at these pins would create a related current through that junction, while an RF voltage applied to the junction and not shorted by a capacitor would not create a related current through the junction, and why it would not rather be the other way round.

[Dr. Frank]

Dear Adrian & Mr. Pettis,
I'd like to continue this dispute later, but meanwhile I wish you a peaceful Happy Easter!
Frank

[Magnificent Bastard]

Dear Adrian & Mr. Pettis,
I'd like to continue this dispute later, but meanwhile I wish you a peaceful Happy Easter!
Frank

I suggest single shot pistols at 25 paces, at dawn...

[splin]

Dear Adrian & Mr. Pettis,
I'd like to continue this dispute later, but meanwhile I wish you a peaceful Happy Easter!
Frank

I suggest single shot pistols at 25 paces, at dawn...

Hmm - might want to consider using a few capacitors to supress the noise a bit. Oh wait.....

[Edwin G. Pettis]

Thanks guys, quite amusing, I think we can all agree to cease the overly hostile rhetoric, it really isn't necessary.

We get it, you would like hard copy backup for tech you don't have access to, we all would, unfortunately there is a lot of tech out there which is highly secretive and covered by very strict NDAs that you really don't want to break.  We are privy to only the barest amount of details necessary for our work that intermingles with their needs, details are not generally part of it and even if they were we couldn't spill them.   They don't even want their own people to know too much, very very few know about the whole picture, everyone is pigeon holed, they know their part of the puzzle and little more.  I would love to know more details myself, it is fascinating but that isn't going to happen.  If we say anything it is all that we can or may know except that what information we have is backed by by solid research we know to be proven.

So please don't get mad at us for the limited way we can say anything, on the other hand, if you would prefer we just say nothing when subjects come up and pretend we don't know anything we can do that too.  We really don't want to get anybody mad.  I will be away from here for a few days so don't expect a response right away.

Kind regards.

[guenthert]

[..]
Datron, as one of only two designers and suppliers of an LTZ based 10V reference (e.g. 4910) definitely has applied several components for hardening the circuit against external EMI. One feature are one or two 100nF parallel to the BE of the temperature sensing transistor, the other is an additional RC filter at the 1013 opamp for the reference amplifier.
[..]

I too noticed quite a few capacitors in the Datron 4910 schematic, but are those truly to suppress external EMI?  Since the 4910 uses a PWM scheme as voltage divider, I suspect there will be a fair share of EMI being produced internally.

Further and unrelated, I wonder whether old schematics are simply not quite sufficient anymore in terms of EMI suppression.  Back in the eighties and before, one was typically concerned with the 50/60Hz from the power grid and perhaps a few to a hundred MHz from nearby radio stations.  The latter was probably safely ignored in DC circuits.  These days a major component of house-hold (and hobbyist-lab) EMI are from SMPS, including those in now all-prevalent LED lamps.  At typically 40kHz to 400kHz, I fear they can't as easily be ignored.  And while one could attempt to rid oneself (and the neighbors ;-} of LED lamps, I don't want to feed my PC from a linear power supply (like the earliest micro computers were -- how heavy is a 600W 110V toroid?  Oh, just 6kg and $60 -- I did expect it to be much worse).

[Kleinstein]

Even those 100 MHz from FM radio or some 800 MHz from a mobile phone can be a problem for a "DC" circuit. It usually would not damage the chips, but the RF signal can be demodulated and cause some extra "noise". Especially the pulsed signal from GSM / Dect phones can cause quite some audible signals in audio equipment. The problem with the really high frequencies is that there the circuit diagram is not enough anymore - here the layout also has quite some influence. Not many capacitors still act like a cap in the GHz range.

The emissions from a SMPS are 2 - fold: one is ripple at the low end, where the OPs may still have some PSRR. The 2 nd point are the much higher frequencies from hard switching and ringing - these can extend to 10-100 MHz range. There are ways to keep the RF out, e.g. with a closed RF box and proper filters (e.g. feed through caps). Ideally the emissions to air from a SMPS should be weaker than the radio signals, not to interfere with radio receptions. It's only the supply that may carry more.

LTZ1000 circuit from the DS may not be the best solution to withstand EMI, but it is well tested and changes are not easy. I would not be too much worried about damaging the reference from EMI with the changed circuit. The changes tend to not change that much here and most of the EMI is more like a small signal added that causes a problem as long as it's present. With ESD the original circuit is also not really good. The possible problem is more with the changes effecting the turn on or turn off case - the loop is kind of tricky and could temporarily cause too much current if not properly starting. This can also be a problem with alternative OPs to the LT1013 - the steady state is easy, but turn on / off can be tricky and is not always well defined in the data-sheets or spice models.

For the zener or transistor I won't expect real crystal damage so easy. A more common problem with zeners (especially not burried ones) are hot electrons causing floating charges inside an oxide, a little like an unintended floating gate FET like in an EPROM. Thermal transients could effect the glue.

[iMo]

Therefore you may see a lot of ferrite beads and small chokes here and there in the today's schematics..

[TheSteve]

RF can cause all kinds of anomalies. I know if I transmit with 100 watts or greater around 7 MHz it will cause my 3458A's to generate a fault that requires a power cycle. Lower/higher power levels and other RF frequencies can cause errors in the readings. In my case the antenna is located on a tower outside, not super close to the references.

[babysitter]

As a former tv & radio service tech working for around 4 years (and practicing my hobby ham radio) close to a big AM medium wave transmitter facility I can approve that it makes sense to keep RF out of ANY ELECTRONICS, even non-receiver-type. Any PN junction can be a rectifier.

[Dr. Frank]

Ok, here's my reply to the statements of Mr.Pettis and his alter ego MisterDiodes.

I'd like to point out, that the eevblog is an 'open' forum. Especially in this thread many capable engineers, physicists, technicians and DIYer have contributed many new ideas, schematics, layouts, publications, calculations, limit estimations, test measurements, everything for free, in a copy-able, and most importantly, verifiable manner. I very much appreciate all these great and profound contributions here, and in this sense also the offer of custom specific resistor sets for a reasonable price, from Mr. Pettis.

In the different physics communities, usually all theoretical and experimental publications are freely accessible, papers maybe at a cost, but the openness is a fundamental necessity of this science. Each new theory or experiment requires critical and open reviews, pros and cons, before these new findings are taken for granted. There's also a strict principle, what you can't measure, replicate, or explain properly, simply does not exist in reality. Example: The infamous Cold Fusion experiment of Fleischmann and Pons.

I'm having the mindset of both of these worlds, therefore I absolutely dislike any entry here, which makes a claim without any verifiable and open facts, as of one of the above mentioned items.
Hiding oneself behind references to NDA, unknown, mysterious and frightening companies and hidden documents, and especially omitting ANY quantitative or qualitative detail about the claim, are no arguments for truth or existence of such, either.


In fact, such an argumentation technique fatally reminds me of all these quacksalvers like Solar Roadways, Solus Graphene Heater, Fontus Self Filling Water Bottle, Batteroo / Batterizer, and Theranos as best example for claiming an at first glance evident idea, but then secretly hiding details about supposed machines and tests, and sooner or later being debunked due to missing basic principles.

Mr. Pettis ratings and wordings about serious and profound contributions here in this thread, like 'being puny', 'ignorance', 'unsupported statements', and alike, are a bad substitute for arguments.

Proposals like 'remove the external sources of noise', or 'preventing the RF disturbance' are simply unrealistic. That's possible maybe in an anechoic RF chamber, or in an application where the circuit is not accessible from outside, like in the 3458A.
But this is not the use case of any of the references like 731, 732, 7000, 4910, and of all the DIY LTZ 1000 based references.

All these are used in lower grade cal labs, but also in more or less noisy environments, like in the ordinary houses of the forum contributors, which are likely polluted by RF spikes originating from coffee machines, dish washer, cell phone, WLAN, lighting and PC SMPSUs, and from the 'dirty' mains network.
Therefore the DIY references have to be shielded, that's very obviously necessary. But instead of repeating again and again that common place, would anybody please give a detailed description and practical demonstration, how to 'properly' immunize the circuit further from the exterior noise.

The  design by Datron / Andreas is already proven to efficiently mitigate this problem.
I also cannot recognize a 'brute force 100nF caps' method. Adrian, please explain in detail, what you mean by this.
It's a common technique in commercial electronics to use a lot of 'EMC capacitors' to mitigate EMC irradiation.
Especially in a DC application like the LTZ1000 circuit, such filters are beneficial, as long the circuits are stable.

The sparse hint from Mr. Pettis that EMC pulses are transformed into current spikes (of what magnitude, please?), which in turn shall damage the 'microscopic matrices', does not convince at all.
Maybe the doped silicon crystal  is meant, but the binding energies are on the order of eV, whereas the energy involved in irradiated emission is very probably many magnitudes lower than that, so any deterioration by this suggested mechanism is extremely unlikely, or the magnitude of measurable impact is extremely low.
Also, here we are talking about over 30 years old big, bipolar chip structures, being much less sensitive to supposed 'excessive' electric fields or currents, compared to modern small scale field effect structures.

Please, check all of your own LTZ 1000 applications critically concerning EMC irradiation, present your results here, and bring up some practical ideas and alternative product designs, how to solve that problem differently.

Frank

[ap]

Re. the brute force capacitor: normally when you work on coupled noise that you were not able to prevent from entering a circuitry you try to fight it with tunes means. E. g. we had to do this against demodulated disturbances comming from mobile phones. Based on research on that topic by a big phone company (which we took as given and that finally helped) the most efficient way was to use a specific capacitance value in a specific form factor. Both had an influence on attenuation (in a nutshell, form factor defined parasitoc inductance; resonance effect). Now not knowing what noise source causes the disturbances here in the LTZ design resulted in using high value, kind of 'catch-all' capacitors. Not that efficent at higher frequencies, but covering a broad frequency spectrum. Thats what I call brute force. Others may call this differently.

The other thing is: The lattice distruction theory using caps is not logic in this application. Every ac voltage across an impedance causes a current. If the voltage is essentially shorted by the cap (as it seems to be as the cap helped), most of the current is now flowing through the capacitor, and the voltage accross it (and accross the junction in parallel) is lower than without the capacitor as the capacitor presents an AC short (thats why it is used). So the junction is seeing a lower AC voltage (ideally close to zero) with capacitor than without, so the lattice should also be less affected by the disturbance. In this way of using the capacitor, the voltage is also transferred in a current (as the claim was), but the current generated through the capacitor is not through the lattice/junction, and any potential current through the junction, caused by the noise voltage applied, is reducted as the voltage is reduced /shorted.

[iMo]

..the most efficient way was to use a specific capacitance value in a specific form factor. Both had an influence on attenuation (in a nutshell, form factor defined parasitoc inductance; resonance effect). ..

A pretty important point. Experts do know well, but general public sometimes does not, that for every small chunk of freq spectra you want to block you may need a special value or a type of capacitor (or a combination of Ls and Cs).

The usual quality foil 100+nF capacitors people use here do not work at SW/VHF/UHF/GSM/Wifi/3G/5G/etc frequencies well, thus you get almost no attenuation.

Even people here are concerned on piezoelectrics of ceramics caps, you can hardly block >10MHz frequencies without using ceramics (smd, or special materials and forms for >1GHz).

And even with leadless smd ceramics, you have to use a parallel combination of several values, usually 3 in 1:100 ratio, to cover a larger chunk of spectra, for example 100pF||10nF||1uF, etc., with the lower value cap placed closer to the EMI source.

[exe]

And even with leadless smd ceramics, you have to use a parallel combination of several values, usually 3 in 1:100 ratio

I always wondered how such "multistage" bypassing is designed. What I learned is that just paralleling caps is asking for trouble if not done properly due to resonance effects. Does this 1:100 ratio help to avoid resonant peaks? How to validate the performance?

[iMo]

When you start to look at the frequency characteristics of various capacitors you may see their impedance creates a specific rather narrow attenuation pattern ("V" shaped impedance vs. freq). The 1:100 ratio (advised by ADI for example) usually covers the typical attenuation BW of the capacitors used in such a combination (ie 3 partially overlapping "V"s). The actual combination has to be planned somehow, as it has no sense to combine a 100pF ceramics with 1uF foil, etc. The best verification of the BW of the blocking effect of a capacitors combination is a measurement, as usual.

For example the smd capacitor's V shaped impedance (https://www.avx.com/docs/techinfo/CeramicCapacitors/parasitc.pdf) below.

While the X7R 1206 10nF will block "best" say from 20-200MHz, the 10nF Wima MKS2 will block "best" from  12-28MHz.

[exe]

The actual combination has to be planned somehow, as it has no sense to combine a 100pF ceramics with 1uF foil, etc. The best verification of the BW of the blocking effect of a capacitors combination is a measurement, as usual.

yeah, that's the thing, it's very easy to make things *much worse* just by paralleling caps because they form. This was a big surprise to me, many people still not aware . I'm attaching a picture from Renesas's AN1325, which I think is completely wrong. Doing so because it's a common pitfall cultivated by many sources. The resulting impedance is not a sum of individual impedances on the plot (even this part is shown wrongly on it).

It's not that easy to find info on the subject, so I just post one link here: https://www.edn.com/electronics-blogs/living-analog/4460969/Bypass-Capacitor-Resonances- . A good starter into the subject.

Anyway, point taken. It seem the only way to verify performance is to do measurements. Too bad because it's not that easy as this needs to be done on the actual board. I wonder if measurements can be performed with just a signal generator, oscilloscope, and a piece of coax cable (+ maybe 50ohm resistors for termination). Since we are talking about many tens of MHz, I expect proper measurements to be challenging. At least with equipment I have . (Sorry for going offtopic)

[iMo]

Ok, the initial topic has been on blocking EMI fields such you not get them into the LTZ. The moral of my post above is to show the 100nF Wima (aka "brute force blocking") will hardly block till 10MHz and that is the point.

[iMo]

And here is an online simulation program for modelling impedance, esr, s-params, and creating the Spice model of the various types of capacitors

http://ksim.kemet.com/

[exe]

the 100nF Wima (aka "brute force blocking") will hardly block till 10MHz and that is the point.

Do you mean it will not block above 10MHz? I'd say it will not work above a few MHz....

[iMo]

FYI - here is a simulation with 3 blocking capacitors at an "rail", where the 50ohm EMI source is loosely coupled via 10pF capacitor to the rail, the rail itself is 1k resistive impedance (ie LTZ pin).

The Spice capacitor's Models were generated by KEMET's online app, the values of the parasitics were taken at the frequencies with cap's lowest impedance.

The subcircuits Names are actual partnumbers.

Not an optimal approach, but you may try with S-parameters as a homework..

[babysitter]

And measuring/reproducing suggested e.g. by Dr. Frank is what we should try to become able to do, instead of fishing in dirty water.

I guess not many people here around have a shielded, anechoic basement, a collection of current clamps and a signal generator able to do sub 1 Hz AM or other stuff which would be preferred to tickle our valuable voltage standards. I would like to suggest figuring out forum standards (IDS - International Dave Standard? ) to disturb voltage references and become able to put some value to the supression of the influence when trying to harden them. More like "Raspberry Pi with rpitx with some capacitors to shield/plus/minus and a metal plate of size x*y below it" (tekbox lisn thrown in here or there) or "travelling piezo lighter" testing instead of "Frankonia tin box with €300.000 R&S stack next to it." If we find some empty lab to confirm findings, the better!

Wouldn't that be a great topic for the Metrology meeting?

Best regards,

Hendrik

[eplpwr]

Good evening (in my TZ)! 

I'm in the process of buildung a number of LTZ boards - just waiting for some Ultrohm resistors ... Also, I want to be set up to measure the boards voltage from power on and onwards - got me a Keithley 7002 w/ 7011-S for that purpose. And some 3458A, DMM7510, 34470A usw ... standard voltnuttery. 

Regarding encapsulation I've got som nice Mu-metal boxes from Russia. Inside these I plan to make an alumin(i)um oven box, home CNC style. Qestion is: should the alu chassis parts 1) touch the componens for best thermal equilibrum , 2) have a small distance filled by isolating material like Sil-Pad to be "softer on the compinents", or 3) have an air gap of - say - 0.5 mm to allow all components their "freedom of movement"?

If physical stress was unimportant, I would go for the direct contact, but the more I read, the more I see components wishing to expand/shrink at their own pace.

Recommendations most welcome!

[Echo88]

Do you have a link to those mu-metal-boxes?
The aluminiumbox doesnt need to be thermally coupled to the LTZ-board as no component generates so much heat, that it needs to be cooled. Also you generally want a good thermal insulation of the LTZ-board to the case and mechanical flexibility of the board to avoid strain-induced offset-voltage.
I suggest following AN159 on shielding box construction, for generally good shielding: https://www.analog.com/media/en/technical-documentation/application-notes/an159fa.pdf Page 7.