-
@ MisterDiodes: Always a pleasure to read about your experiences, especially since i have the 2508-32-Evalboard sitting in a corner!
Maybe you can have a look at the proposed voltage-reference-filter-design mentioned here and share your opinion: http://electronicdesign.com/energy/filter-trims-ultra-precision-voltage-reference ?
Also: do you know any good ref-filter-design for a LTZ1000 regardless of the stabilizing-time?
Edit: Since the LTZ1000/Ref-Amps are the best voltage references out there apart from the superior JJA and companies like Fluke/Keithley surely like to improve their stuff: Do you know of any successful projects which improve stability/noise of those references or does it just not make sense and the only good way is to handselect and parallel those references like it is done in the Datron 4910?
Ha! That circuit's been floating around a while for sure, in various forms. Nothing to see there, and a poor ad for Maxim - another capacitance multiplier. A great teaching moment in 1/f noise and how hard it is to get rid of.
Just try building a LP filter with cutoff down in the 10's of mHz - watch what happens when you turn it on. You'll be waiting a while... <Grin> You'll still have noise over longer time spans now, and you'll be kicking yourself for blowing that much on those $65 caps, etc.
You don't need the "Teflon standoff's" either if you understand guard circuits - and you sure don't need that extra shunt regulator. That was there just because those goofy Maxims were only 5V parts - typical of the early crop of AZ amps. Noisy buggers also, compared to later offerings. Maxim seemed to always miss the mark there.
Not making noise in the first place is the concept, and paralleling Vref's (up to a point) is probably the better bet. Probably 9 or 10 Vrefs is a practical limit and point of real diminishing returns. 4 Vrefs might be even a better upper limit. Only if the customer is willing to pay though.
If you build this stuff for a living, you'd better make sure there's a practical sense of profit also. If you chase low PPM's for no real reason and no profit - you'll be out of business.
At some point you have to tackle the problem the other way and decide what noise you can live with...and what is "good enough" if you have multiple measuring systems. The 3458a has great linearity and has good resolution for relative measurements over shorter time spans. As good as it is you still have 1hr, 10hr and 24hr uncertainties that are unavoidable - and that's why the most practical approach is to average across multiple systems (Vref and ADC or whole DMM's) to help deal with 1/f noise.
Better yet is to spread your measurements across different devices and techniques - so that why we have 3458a's as well as 732's, KVD's and null meter sets. That way you get a better spread of drift rates and noise, and more likely to fence in an accurate estimate of a DC voltage. What you don't want is everything drifting together with similar 1/f noise characteristics.
Remember: It's always only an estimate measure when you're in low PPM territory. Even a JJ-Array has uncertainty. Most of the time trying to spend an exponential amount of effort trying to shave another half-ppm of uncertainty is a major waste of time AND money. -
LP filter with cutoff down in the 10's of mHz - watch what happens when you turn it on. You'll be waiting a while... <Grin> You'll still have noise over longer time spans now, and you'll be kicking yourself for blowing that much on those $65 caps, etc.
$65? Cheapskate! How about $707?
Quotethat's why the most practical approach is to average across multiple systems (Vref and ADC or whole DMM's) to help deal with 1/f noise.
Hm, you got me an interesting idea for that old DMM noise test experiment. Sample 4 DMMs (2x3458 + 2x2002's) in sync together and average their common result minus the offset, to get total noise figure. That could show if noise of such rig lower than single DMM. -
Again: Learn from the app notes where 1ppm-level values are being tossed around. You'll never see just one DMM being used, it will be usually 3 or 5 units at least.
TiN: Using a $700 cap is unfortunate- let alone a $65 X 2 in that rip-off Maxim LP filter circuit (Burr-Brown was doing that in the 80s, and I think they ripped it off the idea from maybe National, if I remember right). The original idea dates back to mechanical chopper / tube days. If you're having to buy those caps, there is -probably- a better way; or you're not really selling the circuit for a living. The real companies that use wet tant caps won't be paying $700 a pop, guaranteed.
Here's the rule: If you don't like the price of a component, you're not buying enough per year.
Just some other random thoughts here...Even more stuff that has nothing to do with "Story of Little Jumper"
Now class, looking at that Maxim circuit (Which is doubtful it was ever used, it's just an ad for a trade rag) - someone tell me how long it takes to charge that 10uF Teflon cap C1 via 3M at the input to let's say a 10V Vref to within 0.1ppm of Vref (where you'd like it to be to start stabilizing output). Now look at how long it takes to charge that second cap via 1M (actually, what is the net charge across second 10uF Teflon cap?). Like I said, you'll be waiting a while. You'd like to use the circuit in some reasonable amount of time. "Reasonable Time" might be open to interpretation here - but this is something you have to calculate if you're designing a real project. Maybe the device is left on all the time...or maybe the device has to be battery powered, in which case you don't want to wait a long time after power up before you can even measure anything.
Your end application makes all the difference here.
Back to Maxim circuit:
WHOops...Now check the influence of that op-amp input DC and noise current and overall effect of amp switcher spikes, and how does that affect charge time and accuracy? With a 3M resistor is place and the op-amp running, what is the net effect of switcher current spikes when they are not exactly bipolar / cancelling in nature? (That's the other dirty secret of AZ amps ). Will that cap C1 gain or lose more charge over time, and what noise effect does that add to the output?
WHOoops...What happens when you power off or glitch the power rails (maybe another device fails on the power rail)? That 10V stored on the cap C1 has to go somewhere, and that input pin and internal surge diode on the amp is the only real path. Now you find out how fragile that diode in the amp is. You learn to look for this in amp circuits whenever you see a larger cap hanging off an input pin directly, with no limiter resistor. That's a red flag right there nobody really used that circuit in practice - or never designed for reliability.
And on and on and on.
That capacitance multiplier filter isn't a very good approach, and its doubtful it was ever put into real practice beyond a trade mag advertising blurb. For the cost of that filter you could add another paralleled LTZ and resistors at least.
-
The filter circuit with the OPs as shown would have a background noise level of around 45 nV/Sqrt(Hz) (more at low frequency) from the OPs and the additional high noise from the AZ switching (in the 10s of kHz range). So about the noise level one could get from 2 LM329 in the kHz and above range. So if you start with a good reference like the LTZ1000, noise at the output would be higher than at the input side.
The bias current and current noise should not be such a big problem here: the input impedance seen at higher frequencies is relatively low and the bias current is still low for the modern AZ OPs. Even lower noise AZ OPs are often below 200 pA, and this bias is relatively constant over temperature. The offset of A1 is not critical at all (maybe 1 mV DC at the filter cap), and A1 sees a 10 K input impedance. Thus only 1 µV from a 100 pA input bias.
Still the circuit should have a second passive filter stage (before A2) to get rid of much of the higher frequency noise of A1.
One possible nasty part hidden in this circuit is dielectric absorption in the C1 capacitor. The RC time constant is something in the 300 s range, thus something like 1 hour waiting to get it stable - however dielectric absorption can be a really slow process, depending on the capacitor. -
MisterDiodes, little jumper is working happily since, other than jumpy, it's actually stable. And to be fair, those wet slug caps are not for any kind of filter, they are bought for far less than mouser list price for AN124 style low-noise preamp.
-
CORRECTION to my previous post:
The capacitors mentioned in the Electronic Design mag article above - 935C1W10K - are not $65 as I wrote above, but I see TTI has some for $19, generally around $20~$25ea. elsewhere. Somehow I was guesstimating the BOM cost of that LP filter circuit with newer AZ amps, and that number stuck in my head. It will still take a very long time to settle, important if you're building some sort of battery powered application.
Sorry about that. My gift is I know everything there is to know about everything, my bane is I just don't know it all at the same time
-
There are other PP capacitors available at a much lower cost per uF.
-
Thanks MisterDiodes and Kleinstein for dissecting the circuit. Since theres no real cost-effective solution to improve those LTZ1000-references i will stick with temp-stabilizing my reference and be done with it.
-
Sorry to revive this old thread, but I thought I'd have a go at this reference. Boards from OSHPark arrived and look good. I have most of the parts on order, except for the Vishay resistors. Where do people get these from - direct from Vishay?
-
Hello,
in the US: Texascomponents
with best regards
Andreas
-
Thanks Andreas,
I have a question on the resistors. In the xdevs schematic, it shows the resistors as:
R3 120 0.1%
R4 1K 0.1%
R5 12K 0.01%
R6, R7 70K 0.01%
And in the BOM it shows them as
R3 120 0.01%
R4 1K 0.01%
R5 13K 0.01%
R6, R7 70K 0.1%
Which is right? Does it matter? I'm using a LTZ1000CH not A version).
-
13k is correct abd suitable for usual conditions up to 40C ambient.
-
13k is correct abd suitable for usual conditions up to 40C ambient.
Thanks! And the tolerances? 0.01% for all? -
No need for tight tolerances. 2% is well good enough and many could even be 10% if you can find them. The nominally 70 K could be somewhere in the 50-100 K range. For R4 , R5 only the ratio really meters. So 20 K and 1.5 K would also be an option.
-
Which is right? Does it matter? I'm using a LTZ1000CH not A version).
It depends on your environment temperature where the LTZ will work.
For the non-A version common understanding is not to go below 12K for R5 up to 40 deg C environment temperature.
One of my LTZ1000 (#8) needed a trimming with a parallel 22K1 to the 1K resistor in order to get a better T.C. linearity above 33 deg C.
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg2301414/#msg2301414
Generally I would use 0.1% or better resistors for R4+R5 in order to have the ratio tolerance within +/-1 deg C (+/-2mV) to nominal ratio.
(of course the Zener voltage tolerance will add addtionally setpoint tolerance).
with best regards
Andreas
-
Hello,
in the US: Texascomponents
with best regards
Andreas
Is resistorinfo@texascomponents.com the right email for them? I sent an email a couple of weeks ago asking for a quote for resistors and got no reply.
thanks
Keith -
on the webpage they have a different E-Mail see:
http://www.texascomponents.com/store/home.asp
Hmm, not sure where I got the email address I used from! I will give that one a try. -
Hi Keith,
as you are located in the UK, you might have a look on precision wire wound resistors, 'econistors' or BMF from Alpha Electronics, FLCY, which are available on stock from rhopoint electronics in your country.
I have successfully built very stable references with both technologies; the Vishay ones are technical over engineering and budget overkill.
0.1% is fully sufficient for all 5 resistors.
Frank
https://www.rhopointcomponents.com/product/components/resistors/precision-ultra-precision-resistors/general-resistance-8g16-econistor-series/ -
That is good info Dr Frank! I did get a response from Texas Components but the minimum order quantities are 10 per value which would be expensive unless I could get enough interest for reaching the MOQ. These look good but of course they are only 5ppm/C not 0.2ppm/C.
Keith -
the Vishays are 0.2ppm/K typ. only, see datasheet for their max. values, instead.
I also bought these 0.05ppm/K Z-types, just to measure them as having 0.3 ... 1ppm/K.
the econistors might have between 1..5ppm/K, you need to characterise and match them, but the attenuation factors are so high that you can ignore their influence on total T.C. and you need to trim that anyhow.
Most dominant is the LTZ itself.
The bmf types are typically 1ppm/K.
I like the PWW, because they have lowest hysteresis
Frank -
Thanks you Dr Frank, that is very helpful.
Keith -
Just to add I had a quote from Rhopoint here in the UK for the resistors Dr Frank recommended. At between £5 to £8 each in 1 off quantities they are a lot cheaper than Vishays. I think I'll give them a try.
-
I like the PWW, because they have lowest hysteresis
My experience shows that such generalized statement not exactly true for all PWW, many PWW are much larger hysteresis then expensive BMF (VHP10x, VHA, etc).
There are many very different PWW constructions and designs so while some have excellent performance specs and low hysteresis, many others are not so great. And PWW can be more sensitive to handling and physical implementation of design.
I'm not saying BMFs are best there is, but if budget and schedule (9+ months lead time is normal, even way before current chipmageddon times) allows it, they often provide performance without much selection.
-
Have ordered some Vishay from ELcoPC, West Chester last October 2021, package now on the way to me.
VPG VHP202Z Z-Foil Hermetically Sealed Oil Filled Resistor cost between $50 - $85 (not sure why have such huge price range
)
VPG VHD200 BMF Small Package Ult. Track. Herm. Voltage Dividers cost between $37 - $40
All are 0.01%. “EXPRESS LINE” service added $1.2k
