PX Reference

[cellularmitosis]

Current status:

Current board: v2.4.1

OSHPark: https://oshpark.com/shared_projects/hUZKe65r

Github: https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.4.1



Original post:

It lives!

Off to a bit of a rough start, though:

• Tempco appears to be roughly 5uV/C (nearly 1ppm/C).  Yuck .  I haven't populated the 400k resistor yet.
• While probing around the board I accidentally applied 15V to +Vz and fried one of my LTZ's.  Oopsie.   
• I initially soldered in the R4/R5 divider backwards.  Sheesh.   

I decided to socket the first board so that I could test each of the LTZ's.  It looks like this first one might be jumping by a couple of microvolts.

The critical parts were all from the ebay seller hifi-szjxic, so it was nice to confirm that he sells working parts.

• U1: LTZ1000A, date code 1237
• R1: 120R RNC90Z (2ppm/C)
• R2, R3: 69K8 RNC90Y (5ppm/C)
• R4/R5: 12K2/1K VSR144T (2ppm/C)

The other resistors are cheap metal film "1%" which I had on hand.  They might not be 100ppm/C as HP specs for the A9 board.

I made a mistake on the board and left out Andreas' C15, so I had to bodge that in on the underside of the board.

A couple of the caps are ceramic, as I didn't have 10nF and 22nF in film on hand.

When I cleaned the board with some IPA it seems to have left a slight white film.

Overall, very glad to finally be making some real steps here!   

Thanks so much to the members of the forum!  I have learned so much already!

https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2

edit: oops, intially attached the wrong graphs.

[hwj-d]

Hi cellular,
thanks for your very interesting postings and work. 

Which PS do you use? After some burn in my kx, I'm using pure new unsecured 4s 18650 that gives me a 4 day continuously testing time down to 12v before i change them to a fresh charged set. I see a comparatively cleaner VRef as with my DP832.

[cellularmitosis]

Interesting!  Right now I’m just using a cheap linear bench PSU. I had an idea of buying a small 12v lead acid battery to get a day or two of runtime within a shielded enclosure.  Sounds like you are doing the same!

[Andreas]

Tempco appears to be roughly 5uV/C (nearly 1ppm/C).  Yuck .  I haven't populated the 400k resistor yet.[/li][/list]

I would first remove the socket before I populate a resistor.

With best regards

Andreas

[SilverSolder]

I know what you mean about IPA leaving a white film behind.   You can buy "clean" IPA (e.g. from MG Chemicals) that is of a better/cleaner grade.

[TiN]

I don't know what socket is used here, but on my 2 boards with AUGAT sockets (pin contacts are colet type, same as high-quality DIP panels) there is no visible difference in tempco, due to socket.

Compensation resistor can help to improve tempco from 0.1-0.2ish ppm/K to <0.05 ppm/K. If you see 1ppm/K, something is very much off, so need to fix that first.

in the R4/R5 divider backwards.

Jeez, If that's the LTZ which is in board now, that likely caused problem for LTZ too. It takes seconds for A-chip to reach setpoint temperature due to 400 C/W thermal insulation between little die and metal can.

PSU need to have high isolation from mains and avoid ground loops. It's easy to check if you have power issues: connect LTZ output to meter in normal polarity, record reading. Then reverse output connetion to connect LTZ output in reverse polarity. Absolute voltage level should not change. If it changes, you have power problem or parasitic ground loop that steals current.

[cellularmitosis]

Luckily I caught the R4/R5 mistake while I was still assembling the board

[cellularmitosis]

v2.2 of the board.

Changes:

• Smaller capacitor footprints, tighter layout
• Added an output buffer (LTC2057).  Thanks to Andreas for inspiration, see https://www.eevblog.com/forum/metrology/cx-reference/msg1381896/#msg1381896
• Added a jumper to optionally tie ground to the enclosure via brass standoff

OSHPark: https://oshpark.com/shared_projects/etqxI0Ok see follow-up post below.

github: https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.2

edit: attached schematic

[Andreas]

Hello,

there is a "bug" in the schematic:
R31 has to be connected to the other side of R32. (directly to the output connector).
(like R22 in my schematic above).

with best regards

Andreas

[cellularmitosis]

Andreas, thank you so much for taking a look. 

If I could bother you a bit more, could you explain why?  I was a bit puzzled about the placement of that resistor as well as your 10nF capacitor.

[kj7e]

Andreas, thank you so much for taking a look. 

If I could bother you a bit more, could you explain why?  I was a bit puzzled about the placement of that resistor as well as your 10nF capacitor.

Looks like you may have implemented part of Andreas 10v gain in to your unity gain buffer.  R31 and C31 is not needed for unity gain, only needed if you are going to also use ~25K resistor to ground between R31 and pin 2.  This would give you 7 to 10v gain.   It looks like you just want 1:1 here though.  So just tie pin 6 to pin 2.    R32 is just an inrush current limiter for C32.

[Andreas]

If I could bother you a bit more, could you explain why? 

see e.g. here (Figure 17)
http://www.ti.com/lit/ug/tidu032c/tidu032c.pdf

in your case with a 10 Meg Load you will have 2 ppm DC error because the series resistor which isolates the capacitive load is not within the feedback path.

R31 and C31 is not needed for unity gain,
 So just tie pin 6 to pin 2.   
 R32 is just an inrush current limiter for C32.

no
no
and no

please read the application notes from the Op-Amp manufacturers.
"isolating (large) capacitive loads"

with best regards

Andreas

[cellularmitosis]

Thanks Andreas!

kj7e, this circuit is essentially a combination of an input bias compensation technique
 (equal resistors on the inputs of a buffer, see https://www.allaboutcircuits.com/textbook/semiconductors/chpt-8/op-amp-practical-considerations/) and the capacitive load technique Andreas linked to.

I have also read that some op amps need resistors in the buffer configuration for stability reasons (e.g. the AD797).

[cellularmitosis]

Updated v2.2 board with Andreas buffer corrections.  Also, I added a small copper zone around the LTZ pins to encourage constant temperature in this area.

OSHPark: https://oshpark.com/shared_projects/5n56VbPg

github: https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.2

edit: corrected out-of-date bottom-copper.png

[TiN]

I'd not leave those floating capacitors around LTZ .
Also what is the reason of thinny-skinny neck to the "star" points?

[cellularmitosis]

I'd not leave those floating capacitors around LTZ .
Also what is the reason of thinny-skinny neck to the "star" points?

C11 and C12?  Are you concerned they will disrupt the circuit, or are you concerned about their placement (that they will get hot)?

Initially I had the star traces full-width, but when they diverge at small angles, this creates a large area of overlap, and I was concerned that this was not really a single-point ground and also that it would be difficult to solder.  But I don’t have a lot of experience with star grounds, so maybe this is I’ll advised.

Edit: “hot”, not “hit”

[cellularmitosis]

In order to maintain stability, the RC time constant of R_f*C_f must be (at least) twice that of the RC time constant of R_iso*C_load.  This is easy to simulate on LTspice.

Ah, interesting.  10,000R * 0.00000001F is much larger than 22R * 0.0000001F, so I should be stable.

[TiN]

No, I meant those unconnected PCB shapes. And on 2 layer PCB everything solder easily  .

[cellularmitosis]

Getting some results which are a bit better now.

I've put the reference in a Hammond 1590B aluminum case, used some "cotton rounds" below and a cotton ball above the board to protect from air currents (this might be too much insulation, we'll see!), and I've connected the case to the negative terminal of the power supply.

For comparison, I also put an LM399 board in a Hammond case.  The LTZ is a bit less noisy, but not as good as the datasheet spec of 1.2uV p-p.  However, my 34401A uses an LM399 reference, so I may have reached my measurement limit.  Next step will be to populate a second board and compare them back-to-back on the 100mV range.

[Dr. Frank]

For comparison, I also put an LM399 board in a Hammond case. 
The LTZ is a bit less noisy, but not as good as the datasheet spec of 1.2uV p-p.  However, my 34401A uses an LM399 reference, so I may have reached my measurement limit. 

You can hardly judge the LTZ1000 noise by using the 34401A.
Instead, you definitely measure the noise of the LM399 inside the instrument.

[Kleinstein]

When measuring the external reference you get a combination of the noise if the external and meter internal reference. At higher frequencies there might be some filtering, but usually not much. In addition there will be noise from the meter (e.g. ADC) itself. To a certain degree on can subtracts the meters own noise, but this is limited.

For the reference in the case it might be a good idea to have some filtering for the voltage input, something like a cap directly across the terminals and a ferrite ring as a common mode choke.

[Andreas]

(this might be too much insulation, we'll see!)

without a DEWAR? Never!
most of the heat is transported by the pins and the PCB.

with best regards

Andreas

[cellularmitosis]

OK!  Got a second board assembled and have set them up back-to-back.

I'll be out of town next week, so I'll just let it run the whole time (and ssh in remotely to check up on it  ).

Also, Dr. Frank's boards just arrived.  Too bad Rhopoint hasn't gotten back to me about those econistors yet!

[cellularmitosis]

This is by far the quietest graph I've ever recorded

[Andreas]

OK!  Got a second board assembled and have set them up back-to-back.

you should use 2 galvanic isolated power supplies for that.
Otherwise the ground currents (heater) influence the voltage measurement.

With best regards

Andreas

[chuckb]

Just a quick check. Your graph is labeled as 17 uV (and the noise looks way to small). On another post the DVM in the background is displaying 17 mV. Should the graph be mV?
Thanks for all the posts and have a great new year!

[cellularmitosis]

Just a quick check. Your graph is labeled as 17 uV (and the noise looks way to small). On another post the DVM in the background is displaying 17 mV. Should the graph be mV?
Thanks for all the posts and have a great new year!

Yeah, it is 17mV.  If you zoom way in on that image, that’s a comma after the 17, i.e. 17500 uV.

Edit: I just realized it is a bit ambiguous!  If I were in Europe, a comma would have a different meaning.

[cellularmitosis]

When the two references (and the connections) settle down, for the 10-second period (or is that minutes?) between 33.20 and 33.30, you can see the Vpp is about 1uV-- which is phenomenal!  RMS noise is (about) 6X less than that.  Even after the X1.4 boost to 10V, it is still very low.

Now, build some high isolation power supplies for these, and let them "cook" in your lab for a few months (measure them once a week or so)-- and then you will see them settle down even more!

Congratulations!  What does Dave say?  "Winner, winner, chicken dinner!"

-Ken

Thanks!  And now it is time for this hobby to teach me the virtue of patience

[gamalot]

There is an example of using LTC2057 as a buffer in the LTC2508-32's datasheet.

[zhtoor]

well done sir! 

best wishes and regards.

-zia

[cellularmitosis]

40 hours of data, starting shortly after the last graph ended.

Looks like I have some sort of periodic spike I need to figure out.  Also, there is an almost sine-wave like behavior to the voltage, which shows more clearly on the 100-point rolling average.  Typically, I would assume this is from the air conditioning cycling on and off, but in this case I forgot to switch my A/C from cool to heat before leaving, so the temperature is currently not controlled (you can see it falling).  You can also see when I took a shower just before leaving for the airport (in the humidity graph).

[3roomlab]

100-point rolling average.

should it not be repeating average? as in each 100 point data averages to 1 output, and none are re-used in next count? as opposed to next output using 99 old data point and 1 fresh?

[mimmus78]

40 hours of data, starting shortly after the last graph ended.

Looks like I have some sort of periodic spike I need to figure out.  Also, there is an almost sine-wave like behavior to the voltage, which shows more clearly on the 100-point rolling average.  Typically, I would assume this is from the air conditioning cycling on and off, but in this case I forgot to switch my A/C from cool to heat before leaving, so the temperature is currently not controlled (you can see it falling).  You can also see when I took a shower just before leaving for the airport (in the humidity graph).

Looks like you found another little jumper ... fortunately is not inside a 3458a this time. I still have to decide when make transplant of mine beating inside my 3458a.

[Kleinstein]

There is still quite some "noise" or EMI crap present. So it is hard to tell it the one jump seen at about the middle of the curve is a jumpy LTZ or if this is just from the not so good setup.

Connecting the com side of the DMM to one reference adds quite some load to it and might be a path to couple EMI. So at least that one reference should have a buffer.

[mimmus78]

There is still quite some "noise" or EMI crap present. So it is hard to tell it the one jump seen at about the middle of the curve is a jumpy LTZ or if this is just from the not so good setup.
Connecting the com side of the DMM to one reference adds quite some load to it and might be a path to couple EMI. So at least that one reference should have a buffer.

Yes the one big peak in the middle of the chart can be from any source ... but the 1uV periodic look like my jumpy reference inside 3458a. The side by side chart few posts ago look like many of my charts.

I think cellularmitosis need a third one reference ... or better a fluke 732a (noo just kidding) to check what happening.

[Kleinstein]

The more periodic part looks odd - kind of going up and down, more like a triangle and with a spike at the top. I would not be surprised if this is some EMI crap. The low frequency could be something like the line frequency off from the nominal value / sampling rate of the DMM.

[hwj-d]

Last but not least check it with batteries.

[cellularmitosis]

Back in town, and the behavior looks similar over the past 7 days, but with a few even larger spikes.

(currently working on writing a x-y scatter plot script to get a look at the tempco)

[cellularmitosis]

I also just threw the two LTZ references on 3S lipo batteries for an hour.  I no longer see the "spikes", but I now see a couple of bands of 20uV noise.  Hmm...  I will definitely be trying this again.

[cellularmitosis]

The more periodic part looks odd - kind of going up and down, more like a triangle and with a spike at the top. I would not be surprised if this is some EMI crap. The low frequency could be something like the line frequency off from the nominal value / sampling rate of the DMM.

Yeah, I took a closer look at this and it appears the smaller spikes are almost exactly 30 minutes apart.  My current guess as to the cause would be my refrigerator.  The triangle shape is definitely a mystery -- I'll take a longer battery-powered measurement and see if that still shows up.

[hwj-d]

As you will buy fresh 18650's, beware of protected cells. These could be also very noisy, of course.

[cellularmitosis]

Ok, another round with 3S lipo batteries (no regulator, with the negative terminals of the lipo's bridged with a thick wire).  4.5 hours of logging and most of the data is within a 5uV band -- not bad!

I do still see some small spikes on a 30 minute interval, so it appears those must be some sort of EMI (I also realized I forgot to bodge in C15 on the second board -- perhaps that will help?).

I think next I'll have to try altering the temperature of just one of the refs and see if I can get a good look at tempco.

Edit: I waited 10 minutes before starting the logging, to exclude the "settling" from the log.

[cellularmitosis]

I've created a "b" variant of the board which uses two discrete resistors for R4/R5, rather than a divider.  This variant of the board is intended for use with "hifi's" vishay resistors from ebay.

The R4/R5 resistor footprints are adjacent so that they can be thermally bonded.

https://oshpark.com/shared_projects/S1rjgxed

https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.2b

edit: hifi's resistors: http://stores.ebay.com/HIFI-AUDIO-IC/Metal-Foil-Resistors-/_i.html?_fsub=2634142012 and my daily updated indexes of his resistors: https://ssl.pepas.com/hifi/

[hwj-d]

Build two PX-References these days. I used V2.2 boards with ltz1000a and hifi resistors, they are RNC90Y types of 13k, 1k, 2*75k, 120R. For the R4/R5 divider, simply but careful i combined them together with shrinking tube and a little thermal grease between, and bend solder feets together without stress them to much at the place they go into the resistor.

One Reference has ltz1000a 9636 from hifii. As i can see, it is very stable after the first burn-in about 60h, although the can looks a little bit damaged and has no gilded pins, but luckily they are long enough. For this one, i have a Hammond 1590B, but not drilled holes for the Pomona binding-posts in the moment. (edit: in the moment no pictures from this one)

For the other one with new ltz date 1750, i used a TEKO372 steel sheet housing from Reichelt TEKO-372. Normaly this housing, dimension 83 x 50 x 26 mm, is too little for binding posts at two sides and this little PX-PCB with components, but it would be nice if it worked anyway. Without binding posts, it even fits inside the Hammond 1590B for better steel sheet shielding and nevertheless reasonably airtightness, as we can see on one of the pictures.

Because the PX-pcb doesn't really fit inside the TEKO-372, i cut 2 mm at a suitable edge of the pcb, carefully without damaging any traces.
Sorry cellular, now your github url isn't visible anymore...   
But the result is awesome!

Underneath the pcb i put one of this Digi-Key typical silicat-sacs, let's see if that helps a bit, and above one cotton pad before i closed the lid. I haven't had such little, precise, and really low-noise reference before.
More will follow...

Thanks cellular. 

[hwj-d]

Some more Pictures









Edit: brings that pictures to front

[cellularmitosis]

Very cool!  Thank you for the pictures and the report!

Thanks also for providing a link to that case.  Dr Frank has used a similar one, but nothing came up when I googled for “tuner box”.  I like that the steel will give you a bit of magnetic shielding.

Hmm, perhaps v2.3 will have to be a few mm shorter

[hwj-d]

Those chinese binding posts are really crap! I don't want to use them anymore. So, i need an idea, what to use without instead, that uses minimal space inside this box. The inner dimenson left to right was in reality 47mm, the material thickness is .75mm.

[hwj-d]

Another idea is, maybe to make it a little elongated, to have space for four another resistors and maybe one poti, to use the buffer 2057 also to bring the reference to 10V like showing here: Spread sheet aided design of compensation for 7V to 10V step up resistor set?
But now I am getting rude...   

[cellularmitosis]

I've been resisting the temptation to make a 7-10V circuit, because it is so much easier to make a stable 7V circuit, but then I realized that I can only compare it to a 3458A, and not to e.g. a 732B...  Adding 7-10V is probably inevitable at this point 

I have a pile of those banana jacks as well.  They are convenient as a cheap through-board banana option (and the can be made to work for manhattan-style board as well).  They are OK for non-precision stuff.

Edit: also, thanks so much for the link to that tempco compensation post, I hadn’t seen that before!  The idea of winding copper wire around an existing resistor is very clever.

[hwj-d]

That's blog synergies, we all learn from one another. I wish, that would be so everywhere in the world. 

[hwj-d]

This is TEKO - USA

[hwj-d]

Another idea is, to perform the additional resistors of the voltage divider, according to the procedure described, as soldering posts/terminals too. Anyone can then decide for himself, whether someone only wants the buffer, or what values to use there for 10V.

DigiKey ED1087-ND or ED10992-ND, for example

[cellularmitosis]

Drat, I have just realized that I made a mistake on the board which will cause a few degrees of self-heating in the LTC2057.

This op amp has a shutdown pin.  The datasheet says the max voltage at this pin is 5.3V, but I have tied it to the 15V supply (actually, 14.7V after the 1N5819).

The 2057 block diagram shows the SD pin goes through a 10k resistor to a 5.25V zener, so I'll be burning about 15mW total across these components.  If the thermal resistance of an SIOC-8 package is 100 to 200C/W, this will cause 1.5 to 3C of self heating.  If the thermal EMF of copper to solder is 1.5 to 3uV/C, that's a worst-case error of up to 9uV (but will be much smaller than that in practice, as the thermal conductivity of the joint will keep the thermal gradient far below 3C).

I'll post a new version of the board soon.  If you have already purchased a copy of the board and would like to perform a bodge fix, you can lift pin 1 of the 2057 and use a 1meg resistor to connect it either to pin 7 of the 2057, or to the positive pin of C33 (the pin which is closer to the edge of the board).

[mimmus78]

Drat, I have just realized that I made a mistake on the board which will cause a few degrees of self-heating in the LTC2057.

This op amp has a shutdown pin.  The datasheet says the max voltage at this pin is 5.3V, but I have tied it to the 15V supply (actually, 14.7V after the 1N5819).

The 2057 block diagram shows the SD pin goes through a 10k resistor to a 5.25V zener, so I'll be burning about 15mW total across these components.  If the thermal resistance of an SIOC-8 package is 100 to 200C/W, this will cause 1.5 to 3C of self heating.  If the thermal EMF of copper to solder is 1.5 to 3uV/C, that's a worst-case error of up to 9uV (but will be much smaller than that in practice, as the thermal conductivity of the joint will keep the thermal gradient far below 3C).

I'll post a new version of the board soon.  If you have already purchased a copy of the board and would like to perform a bodge fix, you can lift pin 1 of the 2057 and use a 1meg resistor to connect it either to pin 7 of the 2057, or to the positive pin of C33 (the pin which is closer to the edge of the board).

Why you used shutdown pins?

As by datasheet: if the shutdown feature is not required, SD and SDCOM may be left floating. Internal circuitry will automatically keep the amplifier in the ON state.

Inviato dal mio ONEPLUS A5010 utilizzando Tapatalk

[mimmus78]

Drat, I have just realized that I made a mistake on the board which will cause a few degrees of self-heating in the LTC2057.

This op amp has a shutdown pin.  The datasheet says the max voltage at this pin is 5.3V, but I have tied it to the 15V supply (actually, 14.7V after the 1N5819).

The 2057 block diagram shows the SD pin goes through a 10k resistor to a 5.25V zener, so I'll be burning about 15mW total across these components.  If the thermal resistance of an SIOC-8 package is 100 to 200C/W, this will cause 1.5 to 3C of self heating.  If the thermal EMF of copper to solder is 1.5 to 3uV/C, that's a worst-case error of up to 9uV (but will be much smaller than that in practice, as the thermal conductivity of the joint will keep the thermal gradient far below 3C).

I'll post a new version of the board soon.  If you have already purchased a copy of the board and would like to perform a bodge fix, you can lift pin 1 of the 2057 and use a 1meg resistor to connect it either to pin 7 of the 2057, or to the positive pin of C33 (the pin which is closer to the edge of the board).

Why you used shutdown pins?

As by datasheet: if the shutdown feature is not required, SD and SDCOM may be left floating. Internal circuitry will automatically keep the amplifier in the ON state.

Inviato dal mio ONEPLUS A5010 utilizzando Tapatalk

Maybe is better to just cut the two traces near the opamp pins.

Inviato dal mio ONEPLUS A5010 utilizzando Tapatalk

[hwj-d]

Page 22, Applications information, shutdown mode:

Table 1. Shutdown Control Logic
SHUTDOWN PIN CONDITION
SD = Float, SDCOM = Float: AMPLIFIER STATE = ON

So, cut trace to Pin 1 and Pin 8.

 

[cellularmitosis]

Thanks guys, I should have read the datasheet more closely 

So, the work-around for existing board is either to cut the traces to pins 1 and 8 of LTC2057, or lift pins 1 and 8.

[mimmus78]

Thanks guys, I should have read the datasheet more closely 

So, the work-around for existing board is either to cut the traces to pins 1 and 8 of LTC2057, or lift pins 1 and 8.

In my opinion is better cut trace, it may it help to keep the other pins at the same temperature (but this is all to demonstrate).

Inviato dal mio ONEPLUS A5010 utilizzando Tapatalk

[cellularmitosis]

Here's v2.3, with the shutdown pins left floating:

https://oshpark.com/shared_projects/6puqHG9j

https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.3

Also, I've combined the two variants of the board into one (it has footprints for either a divider for R4/R5, or discrete resistors).

[hwj-d]

Hello cellular,

thanks so much for your very quick actions developing this nice little board. The performance ist outstanding but same time by its through-hole construction very uncomplicated to build. Also thanks to Andreas, and all the others, who are involved in the underlying work.

Some thoughts for this todos, to make the board slightly smaller in one dimension to fit within a TEKO case.

These cases are also available in different versions:

• TEKO 371:   54 x 50 x 26 mm
• TEKO 372:   83 x 50 x 26 mm
• TEKO 373: 106 x 50 x 26 mm

(the inner width of these is only 47 mm!)

but also for example:

• TEKO 392:   83 x 68 x 28 mm
• TEKO 393: 122 x 68 x 28 mm

I used the TEKO 372, because to test the effects of better magnetig shielding from his steel sheet, and as a compact alternative to the Hammond 1590B. But for a standalone, maybe TEKO 393 is better, not only to have space for the pomona's, but also to use the pcbs as they are to save your extra work and service.

So, actually, an extra slim pcb is only necessary, if one plan to put a TEKO 371 or 372 without binding posts or banana jacks inside an other housing for additional magnetic shielding.

Thanks for your good work 

[eurofox]

Here's v2.3, with the shutdown pins left floating:

https://oshpark.com/shared_projects/6puqHG9j

https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.3

Also, I've combined the two variants of the board into one (it has footprints for either a divider for R4/R5, or discrete resistors).

Just placed an order for the PCB's and start to collect all components. 

[hwj-d]

Drat, I have just realized that I made a mistake on the board which will cause a few degrees of self-heating in the LTC2057.

This op amp has a shutdown pin.  The datasheet says the max voltage at this pin is 5.3V, but I have tied it to the 15V supply (actually, 14.7V after the 1N5819).

The 2057 block diagram shows the SD pin goes through a 10k resistor to a 5.25V zener, so I'll be burning about 15mW total across these components.  If the thermal resistance of an SIOC-8 package is 100 to 200C/W, this will cause 1.5 to 3C of self heating.  If the thermal EMF of copper to solder is 1.5 to 3uV/C, that's a worst-case error of up to 9uV (but will be much smaller than that in practice, as the thermal conductivity of the joint will keep the thermal gradient far below 3C).

I'll post a new version of the board soon.  If you have already purchased a copy of the board and would like to perform a bodge fix, you can lift pin 1 of the 2057 and use a 1meg resistor to connect it either to pin 7 of the 2057, or to the positive pin of C33 (the pin which is closer to the edge of the board).

Leave one of my finished boards without this modification in continuous operation.
So far absolutely no negative effects are noticeable.

€: Ordered next three boards v2.3   

[cellularmitosis]

I'm on vacation -- time to futz with this board some more! 

v2.4 is up:

https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.4

https://oshpark.com/shared_projects/uh2iEJZp

Changes:

Board dimensions should now be small enough to squeeze into a TEKO 371.16 case.

The "shutdown" pins of the LTC2057 have been exposed via break-out pins (for implementing off-board control, e.g. if you want to disable output until an enclosure oven stabilizes).

The center terminal of the R4 / R5 divider has been exposed via a breakout pin (for off-board manipulation of the temperature set-point, for e.g. Pickering-patent shenanigans).

The chassis-ground jumper has been removed and instead the pad of the mounting hole adjacent to ground has been enlarged (which facilitates soldering a wire from chassis to ground).

Minor routing tweaks.

[hwj-d]

This damn little thing goes to be better and better ... 
Ordered them once more 

Big thanks again, cellular! 

[cellularmitosis]

I’m planning on making a small voltage regulator board which would fit along side it in a Hammond 1590B case.

Andreas has already found a great low noise part — the LT1763, so I’ll go with that as well.

As far as operating voltage, it would be nice to support a variety of battery powered options:

A lead acid battery can be considered discharged around 11.6 volts.

It looks like 10 NiMH cells can be considered discharged somewhere between 11.5 and 11 volts.

A 3S Lipo can be considered discharged at 11.1 volts.

So, if we can operate the 1763 on 11 volts minimum, that would be ideal.

That means we will be operating the LTZ around 10 or 10.5 volts.  Are there any disadvantages to this?  Andreas has shown that the voltage coefficient of the LTZ circuit is slightly negative, but more importantly, is well behaved at least down to 9V, so I think this should be fine: https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg842662/#msg842662

[Andreas]

Hello Jason,

be aware that I do not use the standard cirquit.
I have a FET (BF245C/BF545C) at the output of the LT1013 / LTC2057 which does some kind of level shifting.
So with this modification I need less headroom than the standard cirquit.

I never tested this but you will probably need at least 3 V headroom for the standard cirquit.
(2V for the LT1013 and 1V for the diode). But 10.5-11 V should be fine also for the standard cirquit with not too high temperature setpoint.

I also have a battery monitor for the NiMH cells. It does not only switch off when the minimum voltage (I calculate 1.15V / cell) is reached (then already one of the cells could have reverse polarity) but also when a certain discharge gradient (e.g. 0.6V in 2 minutes) is reached.

with best regards

Andreas

[hwj-d]

With LT1763 firmly installed, we don't need additionally the shottky - voltage drop anymore, right?

[cellularmitosis]

Thanks Andreas!

hwj-d yes I think you are right.

I wonder also if some of the 1N4148 could be replaced with schottkys.  I found the 1N4148 forward drop to be high nearly 0.8V with ~20mA)

[Andreas]

Hello,

With LT1763 firmly installed, we don't need additionally the shottky - voltage drop anymore, right?

Thats right.

but be aware that if you put a variable heat source (depending on battery voltage)
near the LTZ you will have some (bad) thermal influence on PSRR.

I wonder also if some of the 1N4148 could be replaced with schottkys.  I found the 1N4148 forward drop to be high nearly 0.8V with ~20mA)

D2 (5mA) in your cirquit (at the output of the LT1013 current regulator) can be replaced by a LL103A or BAT48 (with larger holes also the 1N5819)
D1 (20mA) in the heater return must not be changed (otherwise you risk to forward bias the substrate diode, which is a no no).

with best regards

Andreas

[TiN]

I can probably share my old design around LT3042/3045 if one is up to little fine-pitch microscope work with 0805 and 0402's

[Echo88]

That would be nice! 

[hwj-d]

TiN, really cool things 

But, what advantages would these have, except the size?

[Andreas]

They are more nice (golden pins) than my LT1763-design with 0805 and SO-8 as smallest parts.

with best regards

Andreas

Edit: the PCB is a single sided design

[hwj-d]

Hello,

With LT1763 firmly installed, we don't need additionally the shottky - voltage drop anymore, right?

Thats right.

but be aware that if you put a variable heat source (depending on battery voltage)
near the LTZ you will have some (bad) thermal influence on PSRR.
...

Question,

is it better to use, maybe 4s 18650 unregulated (shurely observing the discharge level) as to use this variable heatsource regulation down to 11V? In my subjektive practise, I see no negative psrr impact with unregulated 4s over the whole entire discharge voltage range. But maybe I miss something?

[cellularmitosis]

Well, an oven circuit which adheres to the underbelly of the Hammond 1590B case is already in the works for me, so I’m not worried about the variable heat.

However, Andreas did measure a tempco on his LTZ circuit which he did attribute to variable regulator heat, and added another PCB slot to confirm this.  So in his case, both the variable voltage and the variable heat were a concern.

If you’ve measured no voltage coefficient, then you are correct that unregulated cells would be better.  Hmm, I should see if I can measure a voltage coefficient!

Edit: link https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg875274/#msg875274


Edit2: and I have just the tool to perform this test  https://oshpark.com/shared_projects/39B3GV50
 https://github.com/pepaslabs/ProgrammableLM317

[Andreas]

is it better to use, maybe 4s 18650 unregulated (shurely observing the discharge level) as to use this variable heatsource regulation down to 11V? In my subjektive practise, I see no negative psrr impact with unregulated 4s over the whole entire discharge voltage range. But maybe I miss something?

It all depends also on your overall thermal design:

Think of the heater transistor and how he will behave with variable power supply.

In the new 34470a design the heater transistor is non SMD outside the LTZ-cap
(most probably within the air draft of the fan of the instrument).

My personal strategy is:
keep the power dissipation as constant as possible (at constant temperature).
And thermally isolate any variable heat sources.

Your strategy may differ.
At the end only the final result counts.

with best regards

Andreas

[cellularmitosis]

Hello,

I have designed a small LT1763 board.

OSHPark: https://oshpark.com/shared_projects/uVr4yDgz

Github: https://github.com/pepaslabs/lt1763

[cellularmitosis]

The board has provisions for an RLC input filter.  Here's my estimate of performance for a few values, combined with the LT1763's own ripple rejection.

I used this web tool: http://sim.okawa-denshi.jp/en/RLCtool.php

Note that a 10R resistor will add 300mV of dropout when passing 30mA.

[hwj-d]

Hi cellular,

Thanks again, for your nice work.

Seems to me, that 10V isn't that stable as with 12V (- shottky voltage drop). My measurements at the last days are with min 12V, 4s 18650, to prevent discharge limit under 3V per cell.

After your post, I set the voltage to 10V. The buffered voltage is a bit erratic after this. Will see, whether the voltage is still stabilizing the next few hours. Maybe 9.7V is is something at the lower limit.

First half in the picture is 12V with slightly ambient temperature rise of 0.4 °C, second half is 10V (- shottky voltage drop).

I would prefer 12V (€: 11.5V), that's 3k9 for R3, if you allow. 

[cellularmitosis]

Thanks so much, hwj-d!

Perhaps I should have performed some tests before ordering the LT1763 boards

[Andreas]

Here's my estimate of performance for a few values, combined with the LT1763's own ripple rejection.

Hello,

The values above 1 MHz in the bode diagram for the filter dampening are somewhat optimistic.
The self resonant frequency of a 1uF foil capacitor is in the region of 1 MHz so above this frequency the capacitor gets inductive and ripple rejection will be less than estimated.
Also the self resonant frequency of your inductors is somewhere in the 2-5MHz range. Above this frequency the inductor is shorted by the winding capacitance. Reducing further the ripple rejection.

with best regards

Andreas

[cellularmitosis]

Here is an overnight (~8 hour) run to test the supply voltage coefficient of the LTZ circuit.

I measured LTZ #3 vs #4 (back-to-back), powering one from a lipo into an LM7815, and the other a lipo into a programmable LM317, controlled by a laptop.  (see https://github.com/pepaslabs/ProgrammableLM317 ).

I ramped the LM317 from 15V to 9V (up and down) in 0.1V steps, with 2 minutes per step.

It looks like the output swung by about 11 microvolts over a 6V supply change, so supply voltage coefficient is about -0.25ppm/V.

Edit: github link: https://github.com/cellularmitosis/logs/tree/master/20180225-ltz-voltco

Edit 2: Looks like Andreas board has better thermal management https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg875274/#msg875274

Without slot on LTZ#3 the voltage output change was around 5uV / 4.5V (0.15ppm/V) as regression curve.
With slot the change is reduced to 2.9uV / 4.5V (0.09ppm/V).

On LTZ#5 I have 2.1uV/4.5V (0.06 ppm/V) with the slot.

[hwj-d]

Thanks again, to figure that out so precisely.

My observations, that a voltage coefficient is not recognizable for me, referred to directly connect 4s 18650 3400mAh NCR18650B (measured capacity 3200mAh) to the reference. So it tooks about 5 days, to go from 16.5V down to 13V - 12V, (not 9V as a difference), before i took them out to replace them with a freshly loaded set. With PX, I did it for 5 weeks. Before that some weeks with KX. With it i have more trouble with not so constant results, maybe because this is not an LTZ-A type, don't know at the moment.

But with both i didn't see any impact, that I could attribute to the declining voltage. In particular the buffered px shows a little tempco in the expected frame and meanwhile about 4µV voltagedrop, but can't and couldn't see this 11µV difference, that correlates to this 6V supply change.

I didn't made this precise measurements. So i think, the interpretation error lies with me. Maybe some tempdifferences (measured between 21 - 22°C) overlapped, or my PX react differently, don't know. But I believe you, there must be any supply voltage coefficient.

The time brings it to me.

Thanks for your effort.

[TiN]

My observations, that a voltage coefficient is not recognizable for me

Yes, I agree with this one, I was not able to see any correlation to Vout/Supply voltage, given that there are no ground loops in measurement system!
I was trying range 10.2V to 18V, nada, below measurement noise/temperature coefficient on both KX, HP A9 and FX designs.

[Andreas]

Hello,

if not already done:
I would try to do a better thermal isolation around the legs of the LTZ1000
either some micro fiber cloth on both sides of the PCB or some "cotton pads" (the micro fiber version).

with best regards

Andreas

[Andreas]

Yes, I agree with this one, I was not able to see any correlation to Vout/Supply voltage, given that there are no ground loops in measurement system!

I cannot agree with that. I have battery supply on the one side
and kelvin sensed output directly from the zener.
So there is no ground loop in the system.
Despite of that I have a voltage coefficient (individually for every device).

Perhaps in your case the multilayer helps keeping the feets at the same temperature.

with best regards

Andreas

[cellularmitosis]

I can try again tonight by inserting a serial-TTL opto-isolator inbetween the laptop and the programmable LM317.  Currently, this laptop connection is causing the LTZ's to sit on ground, rather than float on batteries.

[TiN]

HP A9 board does not have multilayer design, so PSRR has barely anything to do with thermal equilibrium theory.

[TiN]

I agree, however miracles or not, maybe I'm just lucky to have A9 (STD) that does not show PSU voltage impact (beyond measurement ability of quad 8.5d dmms).
My refs in both 3458s are good too, but they are pimped with 100k for lower temp SV.

I got 6581T, so now I can test its ref module too (which is more "proper" traditional design).

[MisterDiodes]

Digilent:  You should inform the 45ea. 3458a's running in one room that I just walked out of that they have drifty references.  Some old, some new, all of which are well under 8ppm yr drift, and around 18 standard units that are running to HFL spec. in terms of yearly drift.  Oh...those are some of the oldest units too.

1 unit is on the sideline waiting for a new VFD, and one unit went to sick bay for a new power supply cap, and one unit has a squeaky fan.  1 unit had Vref board replaced a year ago but has been fine since.  That's a typical sort of repair list for year or two.

That IS a shining example of making the product work well.  Out of all the equipment in the building, the 3458a's have a very good profit margin - and Vref failures are not too common, but happen now and then.

Again:  None of those units happened by "miracle", and no units need to have Vref pimped, etc.  All running within spec.

[cellularmitosis]

Just to make sure my understanding is correct, if you increase the supply voltage, the biggest difference in circuit behavior is that the 2N3904 burns more power in order to keep the same current going through the heater, is that correct?  This is why Andreas saw an improvement with adding a slot to his PCB?

Perhaps a very small secondary effect is that the LT1013 burns a tiny amount of additional quiescent power?

Edit: Also, I'm not concerned with attempting to correct for the supply voltage coefficient -- I'll soon be including LT1763's with these refs, so the supply voltage won't be varying.

[Andreas]

Hello,

good that you mention it: of course you are right.
on my PCB the LT1763 is the only variable source of heat since it stabilizes all to 14.0 V
On a standard cirquit you have the 2N3904 and the current output of the LT1013 as main variable heat sources.

By the way: in my "heat gun" experiment I got the impression that variable heat sources which are in the "south" of the LTZ (below pin 8 ) on my board have lower influence than the other directions.

https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg861416/#msg861416

with best regards

Andreas

[MisterDiodes]

Cellular:
We get better results using TO-220 (or similar) pak heater switch transistor...  Get that warmer away from the board and into it's own private airspace, then supply voltage changes should have a smaller effect - The whole system will come to a quieter thermal equilibrium.  Otherwise the transistor and LTZ heater can interact. 

You can even play with a heat sink on the TO-220 although usually not needed.

TO-92's tend to drive heat down into the board.  SMD heater switches can cause issues for sure depending on where they are located relative to LTZ.

Everything around an LTZ circuit is treated as a thermal "system", and the quieter that is, the more stable your output.

[cellularmitosis]

Hmm, well after introducing the opto-isolator between the laptop and the programmable LM317, I don't see such an obvious pattern.

I also see a ~70uV offset from the previous run.

Perhaps I did have some sort of ground-loop.

I also changed up the parameters a bit: the ramp was from 18.0V to 9.0V in 0.1V steps every 60 seconds.  The run was ~8 hours (overnight).

I assume the large noise at the very start of these graphs is either me leaving the room, or the display of the laptop shutting off.

csv, etc: https://github.com/cellularmitosis/logs/tree/master/20180225-ltz-voltco/run4-1519715191

[hwj-d]

Much better...

Your PX is better, as you thought 

Moment, 17500 to 17520 -> 20µV?

[cellularmitosis]

Moment, 17500 to 17520 -> 20µV?

yeah, it was a pretty noisy run, especially compared to previous battery-powered runs.

[hwj-d]

In the moment I've shot some fotos from quick manual going through some Voltages with my 832 PS from 18V down to 10V and up again, that are all in the 5µV area. Wait a moment for posting...





That's going from 12V up to 16V down to 10V again up to 12V and then to 18V and back to 12V. All with approximately 5 min settletime and negative coefficient.

[hwj-d]

My second PX shows in about the same characteristic, but is in the 3µV Area. Probably the reason why I haven't noticed the voltage influence as such one.

The voltagedrop, that destroys the statistic, is from setting the voltage to 9V. Minus the schottky drop that is too little. So, i set it to 9,6V as a minimum after that.

[cellularmitosis]

I forgot to mention that these tests were with the reverse-protection diode bypassed (as well as the heater zener).  I did a test run to see when the LTZ circuit dropped out of regulation, and it was something like 8.9V if I recall.

[cellularmitosis]

One more run, while I was at work today.

This time I used a "sleep 120" command before starting the logs, which gave me two minutes to leave the room and for the laptop's display to sleep.  This cleared up the leading noise at the beginning of the charts.

You can see when the aircon kicked on towards the afternoon.

[hwj-d]

Let's see, the noisefloor comes out with this LT1763 board. I'm building a little PS with mains-filtering and 2x200mA Blocktrafo for two LT1763.

[eurofox]

I got my 1st one ready, output 7.0666V on my Brymen BM869S, I just sold my Agilent 34401A and wait for my Keithley 2001 for more accurate data.

Just surprised with low voltages, usually is in the area of 7.1V

Next one will be with resistors from Edwin Pettis.

eurofox

[Dr. Frank]

I got my 1st one ready, output 7.0666V on my Brymen BM869S, I just sold my Agilent 34401A and wait for my Keithley 2001 for more accurate data.

Just surprised with low voltages, usually is in the area of 7.1V

Next one will be with resistors from Edwin Pettis.

eurofox

Beneath sample variation, that depends also on the oven temperature.
Maybe you have used 15k/1k?

Frank

[eurofox]

I got my 1st one ready, output 7.0666V on my Brymen BM869S, I just sold my Agilent 34401A and wait for my Keithley 2001 for more accurate data.

Just surprised with low voltages, usually is in the area of 7.1V

Next one will be with resistors from Edwin Pettis.

eurofox

Beneath sample variation, that depends also on the oven temperature.
Maybe you have used 15k/1k?

Frank

Hello Frank,

I got 12K2955/1K, it should be a little bit higher than with a 13K 

eurofox

[hwj-d]

Hello Dr. Frank,

for what is this temp-resistor between the two primary 110V windings in your block-transformer? Is it necessary to use one, and what type is it?

Thanks

[cellularmitosis]

I got 12K2955/1K, it should be a little bit higher than with a 13K 

No worries, the two boards which I have assembled so far are both using 12k2 /  1k dividers from the ebay hifi seller.  They seem fine in a less-than-72F environment so far.

[Dr. Frank]

Hello Dr. Frank,

for what is this temp-resistor between the two primary 110V windings in your block-transformer? Is it necessary to use one, and what type is it?

Thanks

You're  referring to my Precision Current Source article?
That's been in the EAGLE lib only, it's a principle circuit diagram.
These BLOCK transformers sometimes have a thermal safety fuse (resettable) inside, though.

Frank

[hwj-d]

Hello Dr. Frank,

for what is this temp-resistor between the two primary 110V windings in your block-transformer? Is it necessary to use one, and what type is it?

Thanks

You're  referring to my Precision Current Source article?

Yes. Forgot the thread in the moment.

That's been in the EAGLE lib only, it's a principle circuit diagram.
These BLOCK transformers sometimes have a thermal safety fuse (resettable) inside, though.

Frank

Many thanks. Then I don't need it explicitly.

[Andreas]

I got 12K2955/1K, it should be a little bit higher than with a 13K 

eurofox

Hello,
I think it is the other way round:
The raw TC of the zener is around 50 ppm/K.
with 12K3 against 13K you lower the setpoint by around 14 deg C giving -700 ppm or around -5mV against nominal.

with best regards

Andreas

[hwj-d]

New kids on the block

Build two little gem's with PX V2.4 and TEKO 371 54 x 50 x 26 mm

The pictures says it all:





The resistors i've used:





Thanks to cellular 

[hwj-d]

Some more pictures:







Size comparison:

[hwj-d]

Some measurements: these are also very stable.

[Echo88]

Nice! I particularly like the close thermal coupling of the divider-resistors.

[TiN]

Cute. However hanging binding posts are asking for a trouble  .
Also to apply term "very stable" to LTZ ref, one need use better measurement gear and talk about this after at least few months of ageing  .

[hwj-d]

Hello TiN,

yes, that's right. i know it all. 

Shurely, the cuties are not there to be used with binding posts in this way, but to be used in another housing because of additionally RF shielding. That's an experiment at all.

Also, term "very stable" should mean for this early measuring of course. Maybe I should have expressed myself more cautious, but in the meantime we all know what we have from these ltz's on one or the other kind of this pcb's and costs and effort of them.

[cellularmitosis]

I love it!  Great job hwj-d.  I have some TEKO cases on their way

Have you toyed with the temperature compensation resistor yet (400k)?  I’m waiting on more thermal epoxy to arrive before I can take another shot at making a heated enclosure.

You might try measuring two LTZ’s back-to-back, which will give you more resolution.  Connect the positive output of two refs to the inputs of your meter, then bridge the two negative outputs together.

Edit: Andreas mentioned using isolated supplies for this measurement.

[hwj-d]

I love it!  Great job hwj-d.  I have some TEKO cases on their way

 

But one little caveat are there with my solder posts and TEKO 371. Because they are interference fittings, I can't get out the pcb without destroying two of them (maybe not, i've not tried it yet).

Have you toyed with the temperature compensation resistor yet (400k)?  I’m waiting on more thermal epoxy to arrive before I can take another shot at making a heated enclosure.

No, that's one urgent next point to learn for me. Next step is to build this 2*200mA psu as compact as possible. Then bring it together with temp compensation, output other voltages, maybe multiples, and so on.

What's about a kj7e's 10V buffer pcb from osh? 
(€: no, first, need to play arround with own measuring values, LTC2057, and so on, with respect to kj7e's suggestions.  )

I think, TiN and others are right. The measuring possibilities reach their limit with lm399 based instruments. Now I need a K7510...  unattainable ...

You might try measuring two LTZ’s back-to-back, which will give you more resolution.  Connect the positive output of two refs to the inputs of your meter, then bridge the two negative outputs together.

Edit: Andreas mentioned using isolated supplies for this measurement.

Yes absolutely. A big playground opens up. 

[hwj-d]

After burn in a couple of weeks, i need a calibration of one or some of them.
(as free as possible, except shipping costs of course.  )
Who can I address then, medium term, if possible in Germany or EU?
I know, maybe a difficult question...

Thanks

[Echo88]

You can ship it to me: I have a 3458A, last calibrated about 9 Months ago. Other users in Germany with a 3458A are Dr. Frank, Andreas, e61_phil and High Voltage (if memory serves me right).

[splin]

Some measurements: these are also very stable.

Min:+ 7.124 16,  Max:+ 7.124 18, Span: 0.000 01?

[cellularmitosis]

Oh, I just noticed that 7.12417 is repeated twice on the vertical scale.  What is the vertical scale of that graph?

[hwj-d]

You can ship it to me: I have a 3458A, last calibrated about 9 Months ago. Other users in Germany with a 3458A are Dr. Frank, Andreas, e61_phil and High Voltage (if memory serves me right).

That's very sweet of you! Thanks. 

After a couple of weeks burn in, I would like to use that, after PM.

Big, big thanks again

[hwj-d]

@splin,

there is a small peak, going to Max:+ 7.124 18, that's not going to span, luckily.

@cellular,

yes, that's often criticized indistinctness from many people to Keysight.

But look at the average, that's says 7.124 166 above the middle line, that is 7.124 165 (wrong) roundet to 7.124 17. The whole y-achsis is 10µV, so the lower is 7.124 16 the upper is real 7.124 17. 

[hwj-d]

The other measurement picture seems to be even more opaque. 

Average says 7.153 564V. The y-scale span is again 10µV.

Now the upper line is (wrong, at this point) roundet to 7.153 57 but real 7.153 565.
The middle line is real 7.153 560,
and the lower line is real 7.153 555, but again wrong roundet to 7.153 56

 

[splin]

@splin,

there is a small peak, going to Max:+ 7.124 18, that's not going to span, luckily.

I assume this comes from rounding issues - using 7 1/2 digits internally, but only displaying 6 1/2. One of the reasons that some prefer to use the DMM for datalogging and using a spreadsheet for the statistics calculations so you know exactly what is going on. 

[ap]

After burn in a couple of weeks, i need a calibration of one or some of them.
(as free as possible, except shipping costs of course.  )
Who can I address then, medium term, if possible in Germany or EU?
I know, maybe a difficult question...

Calibrating your LTZ reference after stabilization burn-in against a 3458A does not make much sense. A standard 3458A has a 1 year uncertainty of 8+ ppm, so effectively arround 10ppm. Your LTZ, if properly build and stabilized, should be arround 1 or 2 ppm pa. So if you e.g. do two consecutive CALs, 1 year appart, against calibrated 3458As in their one-year accuracy cycle, to evaluate the stability of your LTZ, you will know nothing. Any drift of your LTZ will be hidden in the 3458As' uncertainties. It deserves a much tighter cal. If you are ready with burn-in feel free to send me a PM.

[TiN]

I still hold KX project calibrations for willing participants, and to support community I can offer one time PX ref calibration (DCV value and tempco as received) too, at shipping cost. Calibration vs unknown history 3458A indeed does not provide uncertainty better than 10ppm, but verified known stable 3458A can do 4ppm without much problem. Both of my 3458A drifted <2ppm from last year, however I'd stress out this is NOT an official or legally traceable calibration, and my meters are not standard stock boxes.

I do my verifications for 10V range in my homelab using bank of 6 x LTZ1000 references and 4 x LTZ1000A-based DMMs versus annually calibrated 732B (unc 0.55ppm). So my safe lab guesstimate is 4ppm/annual, 2 ppm if you hit a lucky day.  . Good verified stable 3458A is great to do DCV-DCV transfers over a day or even week with sub-ppm uncertainty (I have one verified vs 732B).

[cellularmitosis]

Very generous TiN!  Time to assemble my KX board!

[hwj-d]

@ap, @TiN, @Echo88, @cellular and all the others here and over the world,

what can i say now,

here with you, i feel in safe hands. 

After burn in, i will PM you.
Maybe one goes to ap, one to TiN, if i know, whether there are import export customs problems or not.

Thanks so much to all

[eurofox]

Hi PX builders,

I would like to get your input, what do you think about my measures taken on a Keithley 7 1/2 digit, 100 averages:

Max: 07.065917 V
Min: 07.065843 V

For me it does not look very stable 

eurofox

[hwj-d]

Hi PX builders,

I would like to get your input, what do you think about my measures taken on a Keithley 7 1/2 digit, 100 averages:

Max: 07.065917 V
Min: 07.065843 V

For me it does not look very stable 

eurofox

If you switch it on first time, the voltage drops. That's normal. Let it stabilize a couple of hours, days, weeks without turning it off. Don't connect chassis/PSU ground to measurement -Vz. Use a noiseless PSU. Watch the used current, that should be < 30mA (26-27mA).
€: Measure without greater ambient temperature fluctuations for the first time.
€€: Don't use clips, crokodiles etc. with measurement cables. Twist them.

 

[eurofox]

Hi PX builders,

I would like to get your input, what do you think about my measures taken on a Keithley 7 1/2 digit, 100 averages:

Max: 07.065917 V
Min: 07.065843 V

For me it does not look very stable 

eurofox

If you switch it on first time, the voltage drops. That's normal. Let it stabilize a couple of hours, days, weeks without turning it off. Don't connect chassis/PSU ground to measurement -Vz. Use a noiseless PSU. Watch the used current, that should be < 30mA (26-27mA).
€: Measure without greater ambient temperature fluctuations for the first time.
€€: Don't use clips, crokodiles etc. with measurement cables. Twist them.

 

It is on my lab desk not connected to ground, powered by an HP lineair power supply and use less than 30mA.
It is only powered for a few hours, have to build it into a "container" with a dedicated power supply.
I would like to add a opamp to raise the voltage to 10V.

eurofox

[cellularmitosis]

Hi PX builders,

I would like to get your input, what do you think about my measures taken on a Keithley 7 1/2 digit, 100 averages:

Max: 07.065917 V
Min: 07.065843 V

For me it does not look very stable 

eurofox

Hi eurofox,

I have found that my LTZ's have been more sensitive to EMI / waving my hand near the circuit, etc, than my LM399 circuits.

The most quiet results I have obtained so far were:

- With the LTZ board inside of an aluminum Hammond 1590B enclosure

- Measuring the difference between one LTZ and another

- With both LTZ's powered by independent batteries / regulators

I have other boards on-hand (a clone of Dr. Frank's board, and a KX board), but I have not assembled them yet (still characterizing resistors and I'd like to measure the 1/f noise of my LTZ's before soldering them in).  When I get those other boards assembled, I'll be able to compare the EMI / hand-waving sensitivity of the different board designs.

[d-smes]

I have found that my LTZ's have been more sensitive to EMI / waving my hand near the circuit, etc, than my LM399 circuits.

Why don't more folks pour a ground fill in their circuit board layouts and stitch them top to bottom with via holes to improve localized shielding against EMI?

[Dr. Frank]

Why don't folks use more of Andreas RF/EMC blocking capacitors?

Single sided PCB is also fully sufficient, but a Guard / shield enclosure required.

My 'new' circuits / assemblies are completely immune against common hand-waving, SMPSU, or mains disturbances.

Frank

[cellularmitosis]

I have found that my LTZ's have been more sensitive to EMI / waving my hand near the circuit, etc, than my LM399 circuits.

Why don't more folks pour a ground fill in their circuit board layouts and stitch them top to bottom with via holes to improve localized shielding against EMI?

You mean like this?

[hwj-d]

I have found that my LTZ's have been more sensitive to EMI / waving my hand near the circuit, etc, than my LM399 circuits.

Why don't more folks pour a ground fill in their circuit board layouts and stitch them top to bottom with via holes to improve localized shielding against EMI?

That's strange. I've build four PX'es now, and none has this EMI sensitiveness. See attached pictures. Nearby i'm designing/building a little compact 2x15V PSU. That's worstcase  for EMI, normaly. Simultaneously measurement with 10 PLC; 10 µV Y-Achsis.

[hwj-d]

You mean like this?

Does it go inside a TEKO 371

[cellularmitosis]

You mean like this?

Does it go inside a TEKO 371

Nope, a 1590B 

I should put a disclaimer: I haven't actually made nor tried that board yet, and it is also missing one of Andreas' capacitors (his C15, I think), so you'd have to bodge that in.

[hwj-d]

You mean like this?

Does it go inside a TEKO 371

Nope, a 1590B 

I should put a disclaimer: I haven't actually made nor tried that board yet, and it is also missing one of Andreas' capacitors (his C15, I think), so you'd have to bodge that in.

Don't be fooled.
Your PX V2.4 is very good, in my opinion.
These EMI issues must be due to external circumstances. 

[Kleinstein]

I have found that my LTZ's have been more sensitive to EMI / waving my hand near the circuit, etc, than my LM399 circuits.

Why don't more folks pour a ground fill in their circuit board layouts and stitch them top to bottom with via holes to improve localized shielding against EMI?

Ground fill is often a poor solution for precision circuits, as one looses the control over the ground connection. A ground plane can be low impedance, but not low enough when it comes to ppm range.

One might use an extra plane just for shielding, e.g. add an extra shield layer to an otherwise single sided layout.
Close to the LTZ1000 one may not want the extra thermal conduction everywhere, so it would end up with a circle around the LTZ without an shield.  One downside with partial shields is that they tend to concentrate interfering fields near the edge. So a partial shield might be worse than non at all.

[quarks]

After burn in a couple of weeks, i need a calibration of one or some of them.
(as free as possible, except shipping costs of course.  )
Who can I address then, medium term, if possible in Germany or EU?
I know, maybe a difficult question...

Thanks

because I caught a bad cold I am a little late, but feel free to contact me, if you need another meassurement of your LTZ1000A.
BTW very nice build, so thanks for sharing

[hwj-d]

Thank you very much for your offer, will come back to that after burn in a few weeks more. 

Get well. 

[cellularmitosis]

Version 2.4.1

Just a few trivial changes:

I found the TO-99 too difficult to solder, so I made the pads a bit larger.

I also added footprints for an input and output capacitor directly on-board (I had been mounting these caps across the binding posts).

https://oshpark.com/shared_projects/hUZKe65r

https://github.com/pepaslabs/px-ref/tree/master/kicad/releases/v2.4.1

[WillTurner]

@Cellular
  I wasn't quite sure where to put this, but since it's a derivative of your LTZ1000 symbol, I thought you'd most likely find it here.

  Below is a Kicad LTZ1000 schematic symbol broken into heater and zener elements. I'm planning on moving the schematic back to being closer to the reference circuit.

  Here is the heater.



  And here is the rest of the component.


Edit: inserted images inline.

[aronake]

For version 2.5 and 2.6 what connector is intended to be used for the two big areas of the PCB?