Author Topic: MX Reference  (Read 16818 times)

0 Members and 1 Guest are viewing this topic.

Offline lukier

  • Supporter
  • ****
  • Posts: 614
  • Country: gb
    • Homepage
Re: MX Reference
« Reply #75 on: September 10, 2017, 12:54:06 am »
Thanks for your valuable insights and don't get offended, it is actually a good thing that fellow forum members would like to see a scientific evidence to back up the claims, but well bloody NDAs.

I am not a hobbyist, I make my living making sure high performance semiconductor devices perform to spec long after we're all long gone and forgotten.  From manufacturing to packaging to final application.

At this end we've been helping our customers build longevity into devices for over 30 (35?) years, across hundreds of LTZ's for our own custom test gear - and that know how also goes into billions of devices you use every day

In which industry you are working, if you don't mind me asking. I got curious because of these timescales, 30-35 years or exceeding human lifetime - we only do commercial ICs since the 60's, that is around 50 years and while some ICs are still around (741, 555) I doubt they generate much revenue nowadays and have significant market share.

Military tends to be on the conservative side, but even here 30-35 years I'd say it's much. But then military is not high volume.

What is high volume, is for example smartphones, now I think just above billion a year. But these things have very short lifespan, usually the of royalties for some chip/IP in a phone are mostly gone in 3 years.

I cannot think of any IC that is produced in billions and has many decades of lifetime - except jellybean parts probably.
 
The following users thanked this post: TiN, MisterDiodes

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #76 on: September 10, 2017, 01:19:53 am »
Thanks Diesel ,Lukier, etc.!

Regarding noise shielding:  You can measure some circuit noise at some frequency X, some duration Y and then convince yourself that some sheet metal of some thickness Z from some affected part and it's all coming from a line cord of some length L (like the length of the shielded / unsheilded mains cable doesn't have an effect as well as how many other loads on that power pole pig). 

About then you realize that the current loop in your circuit that's resonating at whatever damn freq it wants to, regardless of whatever external field ripple gave it the energy to start ringing. :-DD

It's been our experience that you -test- an enclosure and find what works best, but in general we start with .05" or .063" (or as I said 1~3mm but 2mm would be a good starting point). steel box, cold rolled steel.  That is the minimum, see if that is enough.  Second box as required.  MuMetal additional as required and yes we have had to do that on more than one occasion.  A -combination- of shielding works well in challenging environments.

Again: All wiring going in and out is protected as required - and if that means balanced chokes and output buffers or whatever extremes you need, that's what you need.  Not everyone needs it all, sometimes one box is fine with some simple ferrites.  Other times you don't get so lucky.

Find out by testing in your environment and what your application -actually- needs.  It could be that an occasional glitch has no effect on the end result, or it could be that the application demands perfection 24/7/365.

A very fun project is to build yourself a mag-loop antenna & amp for detecting lightning, and then use that to sniff around your lab and power mains and see if that matches up with your Vref glitches.  The only thing you have to build is a loop of wire and an amp, and it'll resonate at it's own personal freq. when you find a noise source.  Just remeber that's the res. freq of the antenna, not necessary the noise source.   The blitzortung.org site, etc. has some great ideas and circuit ideas -  Something like that basic amp design & filter is a good starting point to build yourself a simple loop antenna EMI sniffer.

Lukier:  Our company builds and maintains semiconductor process equipment <exact description redacted for privacy>.  The owner started in 1965, Silicon Valley, I've been on board since the late 70's. Started as a young smart ass with a degree and an equation for everything - and then finally figured out what I really didn't know by listening to the older staff scientists who knew what they were doing.
« Last Edit: September 10, 2017, 01:49:16 am by MisterDiodes »
 
The following users thanked this post: vindoline, TiN, Echo88, 3roomlab, dr.diesel, 2N3055

Offline hwj-d

  • Frequent Contributor
  • **
  • Posts: 593
  • Country: de
  • The economy of capitalism has NO social mission!
Re: MX Reference
« Reply #77 on: September 10, 2017, 01:26:00 am »
Over n' Out.  Have fun folks!  Sorry if I've offended anyone here with advice &  tips that has kept our equipment running to spec for some 30+ years, including LTZ's.
I'm at the other end, as hobbyist, that's not so vital to me, BUT, that's extremely grounded to me too, what you are talking about, and i understand every word that you mention here (ok, after some learning effect). I think, your input here is one of the best, and you're right! Others too, for sure. I understand, that must be very frustrating, if someone cannot go too deep in details to prove oneself, if others do because, maybe, they're not so legally bounded. But your sometimes opposite meaning is really an eyeopener to me and others, so, please keep doing that.  :-+ :-+

Thanks, thanks to all
The german goverment has fired his own nation and all his citizens
 
The following users thanked this post: MisterDiodes

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #78 on: September 10, 2017, 02:02:12 am »
Thanks. I try to give as much detail as I can but I really have to stay out of trouble with customers, and I have to not mention the company name.  The end result here is sometimes it looks like a wishy-washy answer pulled out of my butt I know - and sometimes it's better if I just stay quiet.  I have to respect our customer's respect for keeping their R & D private; so sometimes I just have to observe here, and toss out a tidbit only if it's common helpful knowledge.



 
The following users thanked this post: hwj-d

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #79 on: September 10, 2017, 04:57:58 am »
Dr. Frank:

I was looking at the photo of your setup, and I wasn't sure where your analog signal ground is referenced to Earth.  You probably already did this but just in case -   In your basement lab, do you have good access to a buried metal water pipe leading outside that you can use as a ground point? - or some way to setup a separate dedicated lab signal ground?  If you're using the ground from the power mains, that can be a source of noise issues.

We have a separate dedicated lab signal ground (we like the lightning rod downlead for low impedance / good bandwidth) that travels outside (and shares no close routing with any power mains wiring) to its very own dedicated ground rod.  The idea is that your lab test area has it's own dedicated clean analog lab signal earth ground point with no power mains wiring bundled in close proximity. 

At least that's what we do here to help prevent some of those ground-coupled power line noise spikes sneaking into a test.



 
The following users thanked this post: CalMachine

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6206
  • Country: de
Re: MX Reference
« Reply #80 on: September 10, 2017, 08:48:14 am »

At least the simulations show that adding capacitance at the transistors (base to emitter) reduces the phase margin. Together with capacitive loading this could go all the way to oscillation.


no.

you have overlooked that loop stability is maintained by a additional 10K (in series to negative OP-Amp input)  + 100nF across the Op-Amp
in Franks and my cirquit.

Opposite to your assumption the simulation shows more overshoot with the original AN86 cirquit than with Franks design.
Of course the original AN86 cirquit will oscillate with a load capacitance of more than some nF.

with best regards

Andreas

The modified circuit with the additional 10 K and local feedback around the OP is a different thing. Here these modifications are effective in improving stability even with capacitive load. But even in this circuit the capacitor at the transistor is reducing the phase margin, though only slightly. Without the extra 10 K and capacitor in direct feedback, adding a capacitor to the transistor brings it close to the edge - thus not a good idea.

In my opinion the modified circuit looks like a good idea - still not convinced of the capacitor at the transistor.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4045
  • Country: tw
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Re: MX Reference
« Reply #81 on: September 10, 2017, 10:07:53 am »
Thanks. I try to give as much detail as I can but I really have to stay out of trouble with customers, and I have to not mention the company name.  The end result here is sometimes it looks like a wishy-washy answer pulled out of my butt I know - and sometimes it's better if I just stay quiet.  I have to respect our customer's respect for keeping their R & D private; so sometimes I just have to observe here, and toss out a tidbit only if it's common helpful knowledge.

We indeed value all the feedback. I honestly didn't even expect to see any feedback from people who actually build and use ultrastable references like LTZ circuit for real applications, due to NDA's, R&D NRE agreements, legals, etc. So you have my respect on trying to share what's possible.

Hobbyist which built 1-5-10 references for fun rarely even have ability or dedication to measure the results over decade, maintaining constant environment (temperature, RF, EMI, noise, etc) over all that time. It's just amount of man-hours required to get a proof. Same reason why HPAK/Keithley/Fluke can provide those tight specs on their instruments and charge $-$-$-$-$ for it. Surely, some big customers can get even special versions of 732, with even tighter spec if they need so, for attached $-$-$-$-$-$ :).

I'm bit lucky as my daily professional field is much far from precision analog design, that I'm not bound by NDA's on these topics so far. (yet?).  :phew:

Quote
I understand a hobbyist has budget restraints, but that means you get creative and learn.
Or move the restraints, but that does little good actually, leading to equipment hoarding issues with zero knowledge learned.  :-DD
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 
The following users thanked this post: MisterDiodes

Offline ManateeMafia

  • Frequent Contributor
  • **
  • Posts: 719
  • Country: us
Re: MX Reference
« Reply #82 on: September 10, 2017, 12:50:36 pm »
Or move the restraints, but that does little good actually, leading to equipment hoarding issues with zero knowledge learned.  :-DD

I am making a storm shelter out of my gear. I hope I don't have to find out if any of it floats.
 

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #83 on: September 10, 2017, 01:43:33 pm »
Or move the restraints, but that does little good actually, leading to equipment hoarding issues with zero knowledge learned.  :-DD

I am making a storm shelter out of my gear. I hope I don't have to find out if any of it floats.

Good wishes to you Manatee...That's a bit of a breeze blowing your way.  We'll send you good thoughts and hope you're on a little higher ground.
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2385
  • Country: de
Re: MX Reference
« Reply #84 on: September 10, 2017, 01:57:42 pm »
Hello MisterDiodes,

don´t take it personally.

Have you ever heard of an NDA?

Of course, but that´s not the way a public forum is working.
And usually on important things the NDA is running out as soon as the patend is claimed.
If you say throwing capacitors is a "no no" you should be able to explain that.

I get the impression that with your statements some of the members are insecured somewhat.
And I can not estimate if it is really worth that.

with best regards

Andreas
 

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #85 on: September 10, 2017, 02:01:25 pm »

Or move the restraints, but that does little good actually, leading to equipment hoarding issues with zero knowledge learned.  :-DD


Ha!  Good one...  At least you have some really good boxes to study, and then you try to build you're own version you come to that moment: "Oh - That's why they did it that way!"  THEN you appreciate the details that went into the design of very high quality equipment.

So you're not hoarding, you just bought some very good educational class time - and a much better value than a lot of university courses.  Sometimes you don't realize right away when you first pop the lid for a peek inside.

It's true:  If you're into low ppm / high stability / high reliability, it's not for the faint of heart when it comes to spending time & money on equipment for what amounts to something less exciting than watching paint dry - if it works correctly.

"Nothing happened this decade...and I'm thrilled!! Yay!!"   Volt-"nuts" indeed.  :-DD

It's all fun.
 

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #86 on: September 10, 2017, 02:15:29 pm »
Hello MisterDiodes,

don´t take it personally.

Have you ever heard of an NDA?

Of course, but that´s not the way a public forum is working.
And usually on important things the NDA is running out as soon as the patend is claimed.
If you say throwing capacitors is a "no no" you should be able to explain that.

I get the impression that with your statements some of the members are insecured somewhat.
And I can not estimate if it is really worth that.

with best regards

Andreas

If I gave you the complete explanation that would cross over into current R & D onto die aging / stability / reliability.  So take that little gem of info or not, it's entirely up to you how you want your LTZ to age or how you want it to drift long term.

What I do invite you to do is to think about the effect of converting what you see as a small voltage noise spike into a surprisingly large current noise spike flowing across a zener...  That noise spike is the symptom of a bigger problem.

You're treating the LTZ zener like it's a digital IC or an op-amp power rail.  There are times you don't ever want to heavily bypass a PN analog junction, and this is one of them.

I'll leave it at that.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4045
  • Country: tw
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Re: MX Reference
« Reply #87 on: September 10, 2017, 02:51:36 pm »
The way NDA works is that you get resources and honor to acquired knowledge to be kept private between the involved parties.
Also the way public forum works is anything published is up to poster discretion. Be it accurate or not, complete or not, up to the poster, and nobody else. It's also better be safe than sorry, and stay out of sensitive subjects, no matter how much we would love to see the data.

Surely MD will be happy to offer you full details and certificates and test data results, if you become his customer and pay all that research and time, and likely you would wish that data not to be freely distributed for anyone else, undermining your product advantage. :)

I'd suggest we leave this OT at this, and move on  :-DMM.
After all any of our data or results posted (including yours, with all due respect) are yet to be proven to be the only "hard rock solid truth". I don't see issues with my AZ modules without extra caps, nor I see any improvement with the caps. But that only means it works for my specific case, in my specific environment, and with my equipment, and those few references for hobby voltnutting purposes only.

« Last Edit: September 10, 2017, 02:53:13 pm by TiN »
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline Echo88

  • Frequent Contributor
  • **
  • Posts: 567
  • Country: de
Re: MX Reference
« Reply #88 on: September 10, 2017, 03:21:13 pm »
Back to the topic: Does anyone know if our diligent chinese pals on the http://bbs.38hot.net-forum have already built working power-supply/battery-pcbs which resemble those working in the 732B? This way i wouldnt have to layout one myself.
 

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #89 on: September 10, 2017, 05:47:45 pm »
The way NDA works is that you get resources and honor to acquired knowledge to be kept private between the involved parties.
Also the way public forum works is anything published is up to poster discretion. Be it accurate or not, complete or not, up to the poster, and nobody else. It's also better be safe than sorry, and stay out of sensitive subjects, no matter how much we would love to see the data.

Surely MD will be happy to offer you full details and certificates and test data results, if you become his customer and pay all that research and time, and likely you would wish that data not to be freely distributed for anyone else, undermining your product advantage. :)

I'd suggest we leave this OT at this, and move on  :-DMM.
After all any of our data or results posted (including yours, with all due respect) are yet to be proven to be the only "hard rock solid truth". I don't see issues with my AZ modules without extra caps, nor I see any improvement with the caps. But that only means it works for my specific case, in my specific environment, and with my equipment, and those few references for hobby voltnutting purposes only.

Exactly.

Back to LTZs, hard rock data style:  When you build your own circuit, take a look at at the datasheet circuit and HP3458a implementation.  Tell me what you -don't- see added, and tell me why.  Sometimes its what you -don't- see is the lesson.  Pay attention to that also.

When you understand that an AZ amp does absolutely -nothing- for an LTZ circuit except increase current noise in the zener (Vofs in the LT1013 has no effect on output), then it'll dawn on you what's happening with added direct-connected bypass caps energy storage (especially see C5 and C9 on Dr. Franks circuit, which are covering up the much bigger problem).  You've taken the same problem with the AZ amp current noise and magnified it to an even more glorious scale.  The voltage glitch is the LTZ crying out for you to help, but strapping a pillow over its face is not the correct answer.

Again: the Dr. Frank / Andreas circuit will work fine - at least for a while, considering what you're doing to the zener.   If you did this for a real company:  If you shipped that circuit out in production you'd certainly see some customer returns after a while, and some would be fine.  And then you'd be wondering why.

I have given you every clue you need right here.  The rest of discovery I leave to you.  The datasheet circuit will be all you'll need if you're building LTZ -and- shielding correctly, and that is my 85+ pages of EEVBlog summary.  That's what has been proven for decades over some hundreds of thousands / millions of devices and centuries of man-hour development work if you are after the hard facts.

That's how we do it here anyway.  It is a very good circuit that works fine.  A much much more interesting project is the conversion from 7.2VDC down to another voltage in a ppm-stable way.  If you ned an ultra-stable 10V, stick with a 732...you'll be done much faster.

 
The following users thanked this post: dr.diesel, Edwin G. Pettis, hwj-d, 2N3055

Offline NANDBlog

  • Super Contributor
  • ***
  • Posts: 4428
  • Country: nl
Re: MX Reference
« Reply #90 on: September 10, 2017, 05:56:00 pm »
The way NDA works is that you get resources and honor to acquired knowledge to be kept private between the involved parties.
Also the way public forum works is anything published is up to poster discretion. Be it accurate or not, complete or not, up to the poster, and nobody else. It's also better be safe than sorry, and stay out of sensitive subjects, no matter how much we would love to see the data.

Surely MD will be happy to offer you full details and certificates and test data results, if you become his customer and pay all that research and time, and likely you would wish that data not to be freely distributed for anyone else, undermining your product advantage. :)

I'd suggest we leave this OT at this, and move on  :-DMM.
After all any of our data or results posted (including yours, with all due respect) are yet to be proven to be the only "hard rock solid truth". I don't see issues with my AZ modules without extra caps, nor I see any improvement with the caps. But that only means it works for my specific case, in my specific environment, and with my equipment, and those few references for hobby voltnutting purposes only.
Yes thank you. I also made a design for the LTZ1000, few hundred was built, tested, calibrated and shipped.  And I had a lot of data about it, a lot that I produced myself. Under NDA. And though my NDA is not applicable anymore, I dont feel like I should be going around and spreading the info that I have. I could from memory, re-draw the schematic, aproximate the PCB, and open source it, or even start selling it.

But it is bad karma. I was payed to do it, and I dont own the IP. It was well over a year ago, and my NDA is not even valid anymore. And the product wasnt even a voltage reference, it was a very expensive test equipment, with a lot of other parts, in fact this was a only a tiny tiny part of it.

So that is why you would only hear tales from us, who worked with the LTZ1000 at work. Bad karma.
 
The following users thanked this post: MisterDiodes

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #91 on: September 10, 2017, 06:07:37 pm »
A lot of people don't realize that you sign an NDA just to step into the back room of someplace like Broadcom.  Or Keysight. Or wherever.  It has nothing to do with patents, and nothing to do with a timeframe.

Many NDA's have no expiration date - so you sign it, and take the knowledge to your grave. 

 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 9842
  • Country: gb
    • Having fun doing more, with less
Re: MX Reference
« Reply #92 on: September 10, 2017, 06:18:39 pm »
While in HPLabs, we were specifically instructed never to sign NDAs. Why? Because, when push comes to shove, NDAs didn't have much value but they did open the doors for the other party to sue HP at a later date.

No doubt NDAs were occasionally signed, but to do so would have been a heavyweight business decision.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline MisterDiodes

  • Frequent Contributor
  • **
  • Posts: 454
  • Country: us
Re: MX Reference
« Reply #93 on: September 10, 2017, 06:22:55 pm »
BY the way - since this is apparently the LTZ100 thread now - and somebody might have already posted this bit but to be formal:

The LTZ1000 datasheet Page 6 that is currently online right now - that cap shown on Non-inverting input of '1013 is .022uF...NOT .002 uF as shown.

You will see on every other datatsheet for LTZ that cap as .02uF, and even on the first page of current sheet it is correct.

That error crept in 11/2015 when they made some pinout corrections to the negative circuit reference. This datasheet will get corrected at some point, but everyone who knows the LTZ knows what cap value for that position.

They do not really want .002uF in there as that leads to instability, but the circuit will work with that value.  Just use the correct .022uF value.

Confirmed with LT apps engineering in a phone call 3/16 and a subsequent email, for hard-core-facts record-keeping.
 
The following users thanked this post: lowimpedance, plesa, TiN, dr.diesel, lukier, hwj-d, 2N3055

Offline lukier

  • Supporter
  • ****
  • Posts: 614
  • Country: gb
    • Homepage
Re: MX Reference
« Reply #94 on: September 11, 2017, 11:23:08 am »
The LTZ1000 datasheet Page 6 that is currently online right now - that cap shown on Non-inverting input of '1013 is .022uF...NOT .002 uF as shown.

Thanks! When I was ordering bits and bobs for my references from Digikey I, of course, ordered 2.2nF. Now I need to get 22nF. Luckily, I'm slow with my DIY projects so I didn't even design the boards yet.

Side question to fellow LTZ enthusiasts. How much value there is in ovenizing (i.e. not only insulating but putting a heater as well) the reference board? In Fluke references it is obvious as LTFLU doesn't have a heater, but LTZ1000 does and the surrounding resistors have something like 0.01 factor impact on the output value. I'm planning to have 13k/1k resistor ratio, therefore AFAIR 55 deg C so the oven temperature would need to be be below that to allow LTZ heater operation and that would get close to room temperatures in the summer.

I'm asking because originally I was planning to design LTZ + LT1013 + other page 6 bits mini-module, very similar to shape and form factor to the 3458A reference, but inside a case filled with Spacetherm insulator. No oven. Then on another mini-module I was planning to have 10V step up and VHP202Z 10k (I want this modules to be 10V and 10kOhm standards) - this mini module would be similarly insulated, but also ovenized to 55 deg C, this could age the resistors slightly faster but would provide better stability. Overnizing only the second mini-module would bring the power consumption down which is important as I want this to be powered by 4x18650 cells.

I've ordered VHD200 10k/25k dividers from eBay for the 10V step up and I'm thinking if I should provide any means of adjustment at all. This divider in the non-inverting configuration would bring LTZ voltage 6.9-7.5V to 9.66-10.5V, close enough for any equipment that would need ~10V and adding extra parallel/series resistors/trimmers could be detrimental, I will be happy with just this divider and whatever value it settles onto and let it age. But on the other hand adjustment capability would allow nulling to another 10V standard, instead of just measuring my standard with e.g. 3458A which is less accurate. Any opinions?
 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 1697
  • Country: de
Re: MX Reference
« Reply #95 on: September 11, 2017, 12:13:55 pm »
Side question to fellow LTZ enthusiasts. How much value there is in ovenizing (i.e. not only insulating but putting a heater as well) the reference board? In Fluke references it is obvious as LTFLU doesn't have a heater, but LTZ1000 does and the surrounding resistors have something like 0.01 factor impact on the output value. I'm planning to have 13k/1k resistor ratio, therefore AFAIR 55 deg C so the oven temperature would need to be be below that to allow LTZ heater operation and that would get close to room temperatures in the summer.

I'm asking because originally I was planning to design LTZ + LT1013 + other page 6 bits mini-module, very similar to shape and form factor to the 3458A reference, but inside a case filled with Spacetherm insulator. No oven. Then on another mini-module I was planning to have 10V step up and VHP202Z 10k (I want this modules to be 10V and 10kOhm standards) - this mini module would be similarly insulated, but also ovenized to 55 deg C, this could age the resistors slightly faster but would provide better stability. Overnizing only the second mini-module would bring the power consumption down which is important as I want this to be powered by 4x18650 cells.

I've ordered VHD200 10k/25k dividers from eBay for the 10V step up and I'm thinking if I should provide any means of adjustment at all. This divider in the non-inverting configuration would bring LTZ voltage 6.9-7.5V to 9.66-10.5V, close enough for any equipment that would need ~10V and adding extra parallel/series resistors/trimmers could be detrimental, I will be happy with just this divider and whatever value it settles onto and let it age. But on the other hand adjustment capability would allow nulling to another 10V standard, instead of just measuring my standard with e.g. 3458A which is less accurate. Any opinions?

It makes no sense to ovenize the LTZ1000 board.
That would only lead to problems with the LTZ1000 own oven stability.
And as you already mentioned, the influence from the 5 precision resistors is only 1/75 , down to 1/1000.
Typically, you'll end up with about +/-0.05ppm/K overall.
And it's much easier, to use compensation methods to bring the overall T.C. to around 0.01ppm/K.

If you put the step up resistors in an isolated oven, that would make sense, as this is the main feature of the 732A/B circuits, to have a low T.C. for the 10V output also.
Obviously, you're planning to build the very same solution/form factor, which I have published here, a few days ago.
For different LTZ voltages, there are several 'magic' ratio values to trim to near 10.000V, which are easily obtainable in PWW or BMF, like 4k over 10k, and 5k6 over 15k.
I decided to provide a trimming to exactly 10.000V, for usage in a precision Bridge configuration, where it's more convenient to have a 'round' value.
The trimming does not affect the stability of this amplifier circuit, and you can also omit that and just live with odd values.

If you don't design in a trimming, you'll have no choice, afterwards.

Frank
« Last Edit: September 11, 2017, 12:25:22 pm by Dr. Frank »
 
The following users thanked this post: lukier

Offline NANDBlog

  • Super Contributor
  • ***
  • Posts: 4428
  • Country: nl
Re: MX Reference
« Reply #96 on: September 11, 2017, 12:55:01 pm »
Well, you can still order a custom resistor from Vishay Foil. In fact, if I would be volt nutting at home, I would do that. The S series has some 0.2ppm tracking, and the two resistors in one package. It will be for sure expensive, about an LTZ1000 brand new. They trim it to 6 digit. Just build the reference, burn it in, measure the value, and custom order the precision divider to get the 1-2ppm accurate  10V reference. If you really really want, trim that with a million turns pot.
 

Offline lukier

  • Supporter
  • ****
  • Posts: 614
  • Country: gb
    • Homepage
Re: MX Reference
« Reply #97 on: September 11, 2017, 01:00:41 pm »
It makes no sense to ovenize the LTZ1000 board.
That would only lead to problems with the LTZ1000 own oven stability.
And as you already mentioned, the influence from the 5 precision resistors is only 1/75 , down to 1/1000.
Typically, you'll end up with about +/-0.05ppm/K overall.
And it's much easier, to use compensation methods to bring the overall T.C. to around 0.01ppm/K.

Thanks! That's what I thought, but I've seen some of the builds here with the oven so it made me wonder.

If you put the step up resistors in an isolated oven, that would make sense, as this is the main feature of the 732A/B circuits, to have a low T.C. for the 10V output also.
Obviously, you're planning to build the very same solution/form factor, which I have published here, a few days ago.

Similar, but I wanted separate 3458A-like ref module, so I could use that elsewhere (I want to build a variable resistance standard, based on Valhalla 2724A, like Mickle T did) and separate step-up/10k standard in the oven.

I have parts to build 4 (and 1 extra ref-module), so the final version will be a DIY Fluke 734A. I need to find a suitable case that would fit a toroidal transformer for charging and PCB slots to slide the modules in.

Initially, I was thinking about something sophisticated, with an MCU in the chassis for control and monitoring over isolated I2C, rerouting the reference voltages through the backplane for the averaging output or nulling, step up done by an ovenized MCU per-module doing the PWM like in 5440B DAC and doing other things like reading temperature sensors and switching a latched relay to switch the voltage output between raw zener and 10V but this all gets too complicated, noisy and I think I'll take MisterDiodes low noise advice and do instead:
* chassis with just some toroidal transformer + linear regulators to provide voltage for the LiPo charger, this voltage will go through a 2 pole switch per module, isolating the module from the PSU and turning off the on-board charger (so I don't need to slide the module out),
* each module will be a base board sitting on the plastic PCB guides, like EuroCards, on the base board there will be:
* 4x18650 - wouldn't last for many days, but small + some 4S charger IC (TBD) + low noise LDO to make 12V (might not be enough, if 15V is really needed then I need to rethink this section),
* a 3458A-like ref module, insulated with SpaceTherm and maybe instead of 3D printing a cover I'll get some 2mm steel sheet and spend some time with my sheet metal bender,
* another insulated and ovenized module with 10k/25k step-up (similar to your schematic)  and 10k resistance standard, need to add some protection circuitry as well (TBD, short circuit etc).
* cat6 cable coming from the oven to the front Taobao copper binding posts (on order) on the front with 7V, 10V, 10k force + sense + probably some cheap binding posts for guard and earth and LEDs to indicate oven temp OK, charging ON and LiPo low voltage warning.

I decided to provide a trimming to exactly 10.000V, for usage in a precision Bridge configuration, where it's more convenient to have a 'round' value.
The trimming does not affect the stability of this amplifier circuit, and you can also omit that and just live with odd values.

Maybe I'll provide footprints for series/parallel resistors/trimmer but leave them unpopulated. BTW What about using digital potentiometers, like MAX5436? They seem to have better TC than typical trimmers. As my step-up will be in the oven then the access to the trimmer will be difficult.

Well, you can still order a custom resistor from Vishay Foil. In fact, if I would be volt nutting at home, I would do that. The S series has some 0.2ppm tracking, and the two resistors in one package. It will be for sure expensive, about an LTZ1000 brand new. They trim it to 6 digit. Just build the reference, burn it in, measure the value, and custom order the precision divider to get the 1-2ppm accurate  10V reference. If you really really want, trim that with a million turns pot.

I've already ordered the page 6 circuit resistors (S102K and similar), VHP202Z 10k standard and VHD200 10k/25k from ebay hifi-szjxic. Not the best source, probably not the best batches of Vishays but good enough.

The reason is that well over a year ago I've created myself an account on VPG website, filled the Get quote form for 5 sets of page 6 resistors + 4 pcs of 10k oil filled and send the quote and guess what - nothing, zero, no reply. Very nice.
 

Offline NANDBlog

  • Super Contributor
  • ***
  • Posts: 4428
  • Country: nl
Re: MX Reference
« Reply #98 on: September 11, 2017, 01:09:01 pm »
You should definitely talk to sales manager of your region from Vishay foil. On phone. It is not a huge company, and works differently than Vishay. Because they might not have something on stock at all since years, or you might break their production schedule completely by ordering a few thousand of their resistor, and they would deliver it in tiny, 5x5 trays.

No that definitely did not happen to me at all.
 

Offline lukier

  • Supporter
  • ****
  • Posts: 614
  • Country: gb
    • Homepage
Re: MX Reference
« Reply #99 on: September 11, 2017, 01:14:42 pm »
You should definitely talk to sales manager of your region from Vishay foil. On phone.

Hate talking to sales people and also I would be wasting their time with my order for a couple of resistors while just before they were negotiating $1m supply contract, so I thought their website would be a better option.

That is also what it says on their website "How to Order" -> "Get a Quote", it must be there for some reason.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf