Author Topic: LM399 based 10 V reference  (Read 329140 times)

0 Members and 4 Guests are viewing this topic.

Offline Macbeth

  • Super Contributor
  • ***
  • Posts: 2555
  • Country: gb
Re: LM399 based 10 V reference
« Reply #400 on: January 18, 2016, 09:29:16 pm »
You should also consider paying Joe Geller a small royalty fee for each of these circuits you build-- (it *is* patented after all)...  I don't know how much he would want, but if you get in touch with him you can negotiate a number...

Surely royalties would apply to each one he actually sells commercially rather than builds for own use?

Also, what of Lars contribution (if I read his earlier post right)?

I also liked that Geller Labs was so open with design, schematic and BOM´s (including my NTC compensation with both first and second order compensation that Joe could have kept for himself)
 

Offline lukier

  • Supporter
  • ****
  • Posts: 615
  • Country: gb
    • Homepage
Re: LM399 based 10 V reference
« Reply #401 on: January 19, 2016, 11:54:09 pm »
At first, I thought like you did-- if not for commercial use, then no fees, right?  Well, I was wrong.  Evidently, it's OK to experiment [research] with patented ideas, but as soon as you put them into actual use, then you are benefiting from the patented design, and the owner of the patent is entitled to compensation. 

Interesting. I didn't know that. I also always assumed the commercial connection.

So, if I employ a patented idea for my own use (e.g. hobby, some gadget for the home etc) how I am benefiting? It often might not have a financial aspect at all, so how should I compensate the owner?
 

Offline Kean

  • Supporter
  • ****
  • Posts: 926
  • Country: au
  • Embedded systems & IT consultant
    • Kean Electronics
Re: LM399 based 10 V reference
« Reply #402 on: January 20, 2016, 07:01:43 am »
Patent law is never that simple, especially internationally - there are exemptions in many countries for private/non-commercial use.
See http://www.wipo.int/meetings/en/doc_details.jsp?doc_id=256317

In this case the patent is probably in the USA, and the USA don't appear to have this exemption, but many people interested in use of the described methods will be outside the US and so it gets awfully murky.  I guess they could sue you if you ever entered the US.

In any case, I fully support the concept of contacting the creator of any unique idea (patented or otherwise not openly licensed) and offering a small donation as a private user.  In most cases they'll be very happy of the appreciation and make a license exemption of some sort, and if not then you might not want to further associate yourself with such people or their "creations".
 

Offline quarks

  • Frequent Contributor
  • **
  • Posts: 830
  • Country: de
Re: LM399 based 10 V reference
« Reply #403 on: January 20, 2016, 09:01:30 am »
About Patent law, it looks like a very interesting situation, because Joe Geller is the patent owner and he also seems to be a registered patent attorney
see last paragraph in link
http://www.gellerlabs.com/about.html
« Last Edit: January 20, 2016, 09:07:25 am by quarks »
 

Offline Theboel

  • Regular Contributor
  • *
  • Posts: 224
  • Country: id
Re: LM399 based 10 V reference
« Reply #404 on: January 20, 2016, 12:46:14 pm »
I just email Mr Joe ask about his patent and here what he reply,

Hi Anton,

   There were no international patent rights, just the U.S. Patent.  Also, I allowed the U.S. Patent to expire, which means that particular patent is now in the public domain.

 

Offline Macbeth

  • Super Contributor
  • ***
  • Posts: 2555
  • Country: gb
Re: LM399 based 10 V reference
« Reply #405 on: January 20, 2016, 11:02:13 pm »
That's great, because Joe Geller is obviously a gentleman. However, the slimey sharks that patent such bullshit as a flashing cursor, shopping cart, hyperlink and all happen to legally exist in Texas relying on what one would consider an extreme version of US patent law. Really - asking them for "permission" to breath is enough for a lawsuit or at least a demand for 1c for each breath for your entire life.
 

Offline babysitter

  • Frequent Contributor
  • **
  • Posts: 813
  • Country: de
  • pushing silicon at work
Re: LM399 based 10 V reference
« Reply #406 on: January 21, 2016, 04:35:05 pm »
Hi there,

just in case you want to run some reference at lower temperatures - at https://www.mikrocontroller.net/topic/386300#new in a german microcontroller forum a guy sells crystalless OCXO "carcasses" (just the PCBs, temp. programmable by external resistor), believed to be 50-100 deg for 2, 50€ per quartet.

I'm not a feature, I'm a bug! ARC DG3HDA
 

Online barnacle2k

  • Regular Contributor
  • *
  • Posts: 52
  • Country: de
Re: LM399 based 10 V reference
« Reply #407 on: January 23, 2016, 08:39:12 am »
Completely unrelated to previous posts:
i want to know if i killed my LM399:

What happened forgot to connect the heater negative terminal.
In the DS it says:
Quote
The substrate is electrically connected to the negative terminal
of the temperature stabilizer. The voltage that can be applied to either
terminal of the reference is 40V more positive or 0.1V more negative than
the substrate.

so if the heater current was flowing through that (parasitic) diode, the substrate would be elevated by the voltage drop.
Resulting in a negative potential on the negative terminal of the reference of more then the 0.1V allowed in regard to substrate.


Output is half decently stable but is extremely susceptible to input voltage variations and noisy. 
Before i chase other possible causes i'd like to know if i destroyed the LM399 with that mishap.
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2445
  • Country: de
Re: LM399 based 10 V reference
« Reply #408 on: January 23, 2016, 11:32:24 am »
Hello,

how much current does the cirquit take?
(if the heater is off then the cirquit is relative instable).

What is extreme noisy (uVpp)? Which Frequency range did you measure?

Power supply rejection should be below 2 ppm/V above 15 V heater voltage.

With best regards

Andreas
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2604
  • Country: au
Re: LM399 based 10 V reference
« Reply #409 on: January 23, 2016, 12:31:24 pm »
I need a precision trimmed, bipolar reference of +/-7.00V for a precision powersupply I am working on. The reference needs to have low output impedance and low noise.
Here is what I came up with. Filtering is incorporated into the trimmed amplifiers such that the effects of capacitor leakage current is mitigated. The 10uF non-polar will be polypropylene film. No currents are injected into ground, eliminating errors due to IR voltage drops. The -7.00V amplifier provides the bias current for the LM399 zener. The 1N4148 and the 10k to -12V ensure start-up.
     
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2445
  • Country: de
Re: LM399 based 10 V reference
« Reply #410 on: January 23, 2016, 01:10:38 pm »
Hello,

just a question:
how do you ensure that the -12V starts up equally/before the +12V.
(otherwise you could have the oops case above).


I would most probably use 2 schottky diodes to keep Pin 2  near ground level.
And a additional at the output of the negative voltage regulator.

For the 6K8 resistor I would also use at minimum a good metal film (RC55Y with 15ppm/K).

With best regards

Andreas
 
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2604
  • Country: au
Re: LM399 based 10 V reference
« Reply #411 on: January 23, 2016, 01:34:58 pm »
It doesn't matter how the +/-12V supply rails come up (no damage will result) and a run-of-the-mill 50ppm TC for the 6k8 is fine. The DC resistance of the LM399 is 1 ohm or better IIRC.

7V/6800 = 1.03mA.

(1.03/10e6) * 50 (ppm) = 51.5nA per deg. C.

51.5nA * 1 ohm = 51.5nV per deg. C.

That's more than 2 orders of magnitude less error than the ~7uV deg. C (1ppm) specified drift of LM399 itself.
 


 
« Last Edit: January 23, 2016, 01:36:48 pm by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Online barnacle2k

  • Regular Contributor
  • *
  • Posts: 52
  • Country: de
Re: LM399 based 10 V reference
« Reply #412 on: January 23, 2016, 10:15:27 pm »
how much current does the cirquit take?
(if the heater is off then the cirquit is relative instable).

What is extreme noisy (uVpp)? Which Frequency range did you measure?

Power supply rejection should be below 2 ppm/V above 15 V heater voltage.

Thank you for your reply Andreas, i was looking for those members that actually blew up a few LM399. [TiN?, ...]
Circuit takes ~20mA. (including OP)
I isolated the LM399 from rest of the circuit now and tested the PSSR, stabilizer connected to Vin, Rv=2.21k.
10.5Vin - Vz=6.990935V
10.0Vin - Vz=6.990745V
9.50Vin - Vz=6.990546V
9.00Vin - Vz=6.990344V

55ppm voltage difference in a 1Vin change (10.5V to 9.5V).  :--
I think the magic smoke is out of this vintage 1971 lm199.
« Last Edit: January 23, 2016, 10:20:55 pm by barnacle2k »
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 4091
  • Country: us
  • xDevs.com/live - 24/7 lab feed
    • xDevs.com
Re: LM399 based 10 V reference
« Reply #413 on: January 24, 2016, 08:28:09 am »
No, I blew only one LTZ1000 2 years, which is dissected and photographed pages pages away :)
I think one LM399, which had fun in boiling LN2 ain't much useful, but would not call that blown  :-DD

Also since we talking dead stuff, here are graphs from my cryo torture:







CSV-file, heated LM399
CSV-file, unheated LM399
CSV-file, LN2 submerged LM399

Even though some jumps are present (dirty lab environment with lots of gear around randomly running), overall conclusion I already got made for myself.
Datalogger: HPAK 34970A with 34901A card. Temperature reading -  Type T AWG36 thermocouple wire soldered directly to pin 1 of LM399. Current resistor 7.5KOhm (@ambient temperature), supply +15VDC from HPAK linear bench PSU.

Note different scale on ppm's (red on right). Time scale in minutes, sample taken every minute.
« Last Edit: January 24, 2016, 08:30:55 am by TiN »
YouTube | Chat room | Live-cam | Have documentation to share? Upload here! No size limit, firmware dumps, photos.
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2445
  • Country: de
Re: LM399 based 10 V reference
« Reply #414 on: January 24, 2016, 01:46:04 pm »
55ppm voltage difference in a 1Vin change (10.5V to 9.5V).  :--
I think the magic smoke is out of this vintage 1971 lm199.

Hello,

I also fear this.
I have looked up PSRR measurement on my LM399#3.
It is around 0.7ppm/V in the range 8.5 to 15 V.

But the difference is that I use a constant current source
(1V over a 1K resistor with a LT1013 + BF245A and the LM399 as reference).

On the other side 20mA is at the lower limit of a LM399.
My LM399 take usually 25-30mA at room temperature at 9.35V with some additional isolation.

with best regards

Andreas
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6642
  • Country: de
Re: LM399 based 10 V reference
« Reply #415 on: January 24, 2016, 05:14:07 pm »
PSRR comes from 2 or 3 contributions:
1) The temperature regulator part
2) the OP buffering the outout if used
3) the current source to drive the zener part of the LM399 - often this a resisitor from the scaled up voltage.

With modern voltage regulators, the PSRR is not that important any more. Even a 7812 can keep the supply resonable stable - with only a few mV in change 55 ppm/V are not a probem at all.
 

Offline F64098

  • Contributor
  • Posts: 38
  • Country: de
Re: LM399 based 10 V reference
« Reply #416 on: January 24, 2016, 05:47:01 pm »
On the other side 20mA is at the lower limit of a LM399.
My LM399 take usually 25-30mA at room temperature at 9.35V with some additional isolation.
I'm actually testing some LM399H. The older ones (NOS) from 1989 take about 28mA @15V at room temperature, the newer ones (used) from 1998 are running at 35mA.
Strangely, the older ones are increasing their Zener voltage for ~100ppm, when the heater is powered, the newer ones are decreasing for ~85ppm.
Both types are from NS.

With best regards

Frank
« Last Edit: January 24, 2016, 07:14:57 pm by F64098 »
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2445
  • Country: de
Re: LM399 based 10 V reference
« Reply #417 on: January 24, 2016, 07:35:34 pm »
Hello Frank,

You should compare heater power. 30mA @ 9.35V are around 300mW.
At 15V you should expect around 20 mA at room temperature.

is there a correlation of zener voltage to tempco on your references?
see here:
https://www.eevblog.com/forum/metrology/lm399-based-10-v-reference/msg422367/#msg422367

the "sweet spot" seems to be at 6875 mV.
Above I have positive T.C.
Below negative T.C.

With best regards

Andreas
 

Offline F64098

  • Contributor
  • Posts: 38
  • Country: de
Re: LM399 based 10 V reference
« Reply #418 on: January 26, 2016, 11:28:41 am »
is there a correlation of zener voltage to tempco on your references?
Yes, like your's  and i wonder about the much higher deviation from cold (18°C + 2K self heating) to hot of mine.

Quote
the "sweet spot" seems to be at 6875 mV.
Would be nice, to get pre-selected parts from some ebay sellers  ::)

With best regards

Frank
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2445
  • Country: de
Re: LM399 based 10 V reference
« Reply #419 on: January 30, 2016, 08:36:20 am »

Would be nice, to get pre-selected parts from some ebay sellers  ::)


Hello,

keep on dreaming.

In my dream world the manufacturer sends all his production of LM399s to his primary precision instrument manufacturer.
This one selects out the best parts and stamps his personal number (something like 1826-1249) on it.
The rest of production is sent to the other instrument manufacturer.
He selects out the best of the rest and marks them. (with something like SL40057).
(that is my personal explanation that I already found three instruments of this manufacturer which have more than 5 ppm difference when tilting from the left to the right side whereas the other manufacturers instruments have below 5 ppm tilting effect from back to front side (the same orientation regarding the LM399 on this instrument)).

The rest is then sent back to the IC manufacturer and is stamped with LM399.
So how can you tell me that someone will get those LM399 at the sweet spot.

The most probably reason that I have some of them is that I all bought them when NS went out of production. So the manufacturers all selected among the LT parts at this time.

I have put several measurements into one single graph.

With best regards

Andreas

« Last Edit: January 30, 2016, 11:21:35 am by Andreas »
 

Offline F64098

  • Contributor
  • Posts: 38
  • Country: de
Re: LM399 based 10 V reference
« Reply #420 on: January 31, 2016, 11:28:56 pm »
In my dream world the manufacturer sends all his production of LM399s to his primary precision instrument manufacturer.
.....
The rest is then sent back to the IC manufacturer and is stamped with LM399.

That sounds too bad, to be false...

Quote
I have put several measurements into one single graph.

Very nice!
I got another batch from a german seller and their behaviour is much better, than the others.
But the seller increased the price shortly after i bought them, so it's not economic to buy another batch.

Best regards

Frank

 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2445
  • Country: de
Re: LM399 based 10 V reference
« Reply #421 on: February 02, 2016, 07:15:07 am »
In my dream world the manufacturer sends all his production of LM399s to his primary precision instrument manufacturer.
.....
The rest is then sent back to the IC manufacturer and is stamped with LM399.

That sounds too bad, to be false...


Updated diagram.

By the way: when I leave out my references out of the diagram it gets obvious.

That what I would expect from datasheet 6.95V +/-4% for the NS parts is that
68% of the measured devices (gaussian distribution) lies within 6950 +/- 90mV.
(4% = 3 sigma).
If I count in the 2nd diagram there is a large gap...

So either the process is not well controlled (sounds not plausible to me)
or someone did some cherry picking...

With best regards

Andreas
 

Offline F64098

  • Contributor
  • Posts: 38
  • Country: de
Re: LM399 based 10 V reference
« Reply #422 on: February 02, 2016, 12:52:48 pm »
So either the process is not well controlled (sounds not plausible to me)
or someone did some cherry picking...

It seems, we have to look out for parts with the magic datecode...  8)

Frank
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2445
  • Country: de
Re: LM399 based 10 V reference
« Reply #423 on: February 03, 2016, 10:25:03 pm »

It seems, we have to look out for parts with the magic datecode...  8)


Hello,

after reading this thread all sellers will re-stamp their parts accordingly. (and raise the price).  >:D

by the way: today I measured 2 LM399 from LT which I bought from DigiKey recently (DateCode 1351).

results attached and overview picture updated.

The only hope that I have is that perhaps the LM399A are the way to go.

With best regards

Andreas
 

Offline fluxamp

  • Contributor
  • Posts: 7
  • Country: de
Re: LM399 based 10 V reference
« Reply #424 on: February 24, 2016, 09:07:01 pm »
Some more data points for your plot Andreas: I got some LT LM399AH from DigiKey with date code 1502 and did some quick measurements. Room temperature was at 19C, measured with Agilent 34410A at 10nPLC

LM399AH #1:
unheated: 7.077,21V
heated:      7.086,82V
ppm/K:       19.2

LM399AH #2:
unheated: 7.074,50V
heated:      7.085,28V
ppm/K:       21.5

LM399AH #3:
unheated: 7.057,67V
heated:      7.065,89V
ppm/K:       16.4

LM399AH #4:
unheated: 7.060,37V
heated:      7.068,70V
ppm/K:       16.7
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf