LM399 based 10 V reference

[Mickle T.]

Hello Macfly,

S5-61 is one of the old russian hermetic foil resistors, equivalent to the Vishay. I don't have full specs.

Mickle T.

[bingo600]

Hi guyzz

I have bought two of Branadic's LM399 boards. 
Schematic is the same as in the LM399-DS : Buffered Reference With Single Supply - but with a LT1001 in Cerdip.
https://www.eevblog.com/forum/projects/lm399-based-10-v-reference/msg219527/#msg219527

And have also gotten the "econistors" from Rhopoint (Outchhh $$)
I couldn't get the 7.5K for the heater , but have some 6.49K 5-ppm resistors.

Can i use one of those for the heater ?

I was thinking of using a 7815 for the psu , a led to get the 7815 in regulation , a diode for reverse polarity protection , and feeding it 18v. Maybe from my HP-E3610A.
Is a LM317 a better choice (I have the 7815 in the drawer).
 

I'm an analog noob , so go easy on me

TIA
Bingo

[Andreas]

I couldn't get the 7.5K for the heater , but have some 6.49K 5-ppm resistors.

Can i use one of those for the heater ?

Hello,

the heater of the LM399 does not need a 7.5K resistor.
I think you speak of the zener pull up resistor.
(the blue resistor on branadics photo).

In this case a 6.5K will do its job.
You will have slightly more than 1mA at the zener.
-> ageing drift increases slightly.

For the 15V supply you should chose one with low tempco and low ageing drift
to minimize changes on output voltage.

With best regards

Andreas

[Mickle T.]

7815/7915 works very well. I didn't find any measurable impact on the reference output.

[Mickle T.]

VRE310J have a large TC (even typical 0.6 ppm/c is too large). Low-frequency noise - no comments, see picture 
Resistrors - selected russian metal-foil S5-61, 0.1 ppm/C.

[Vgkid]

That is an impressive increase in stability.

[Mickle T.]

Side note: 0.1 ppm/C is only after selection. All resistors in my reference have a 5 ppm/C class (max tempco). Typical is 10 times better. By selection is possible to achieve 0.1 ppm/C (20-50 C range).

[bingo600]

In this case a 6.5K will do its job.
You will have slightly more than 1mA at the zener.
-> ageing drift increases slightly.
Andreas

Thanx Andreas & Mickle

I'll use 6.49k for the Zener pullup , a 7815 and some Red Wima's

/Bingo

[Andreas]

Hello,

the 10 Ohms resistor in series to the heater of the LM399 in some designs did not get out of my mind.
I decided to make some measurements on PSRR (power supply rejection ratio)
with different resistors in series to the heater.

I choosed  0 R, 10 R, 22 R and 47R.

From my LM399 ageing box I decided to choose LM399#13 (see April 9.th) it is the one which has the largest TC between heater off and heater on. Additionally I will do the measurement with LM399#14 which is the one with the lowest TC from heater off/on measurement.

Measurement conditions:
Zener in parralel with 100nF is supplied from battery with LT1763 stabilisation to 14.08V via a 6K8 1% metal film resistor.
Kelvin sensing of zener voltage with 2:1 precision divider and measured with 24bit ADC (LTC2400).
Heater is supplied from programmable power supply and fed via the chosen resistor to the heater.
47uF + 100nF additional in parallel to the heater (avoid oscillations due to series resistor).
Start was with 12V increments 0.25V to 30V. Then back to 9V and finally increasing to 12V again.

X-axis is heater voltage
left Y-axis is 2:1 divided zener voltage in mV (red dotted line)
right Y-axis is local PSRR gradient in ppm/V (green dots)
(But due to measurement noise the PSRR gradient is nearly not usable)

Here the results for LM399#13:

Amazingly this device cannot read the datasheet 
The heater should work downto 9V if used without series resistor.
But below 9.1V there is a large change in zener output voltage.
Of course with additional series resistor the minimum working voltage has to be increased.
I decided to "zoom" and cut off the "odd" measurement values.

Evaluation between 12 and 30V

  0R: 34uV   (17uV * 2)
10R: 37uV   ignoring the "event" due to bad contact
22R: 38uV
47R: 37uV

So there is no enhancement in PSRR for the heater on the zener output with a series resistor.
Best seems to be when using no series resistor.
Since PSRR decreases on higher voltages the heater voltage should be choosen as high as possible.

with best regards

Andreas

Edit: continued on:
https://www.eevblog.com/forum/projects/lm399-based-10-v-reference/msg443181/#msg443181

[bingo600]

I did the first LM299 ref.
Thanx for letting me buy the boards Branadic 


See schematic in post below , this one has an error

Had to put a 200ohm trimpot in series with the 9k feedback , now i can hit 10.00000v  - using approx 9k140 ohm

I have added a RC , and a feedback cap to Branadic's version , my EE Guru said it was a good idea.

/Bingo

[Andreas]

my EE Guru said it was a good idea.

The standard cirquit would tap R3 on the other side of R5.
So voltage loss due to R5 is canceled out.

Edit: see http://www.ti.com/lit/ug/tidu032a/tidu032a.pdf figure 17

Usually R5 can be choosen around 100R or even somewhat below
without loosing stability with large capacitive loads. (C2 necessary).

With best regards

Andreas

[bingo600]

my EE Guru said it was a good idea.

The standard cirquit would tap R3 on the other side of R5.
So voltage loss due to R5 is canceled out.

Usually R5 can be choosen around 100R or even somewhat below
without loosing stability with large capacitive loads. (C2 necessary).

With best regards

Andreas

You're absolutely correct Andreas.

My EE has just corrected me with exact same info.
I have updated the schematic.

Edit: See newer schematic below

Thanx to all you electronics (volt) Gurus.

/Bingo

[Hideki]

If you remove all the small green junctions where a single wire simply connects to a component, it would make me very happy

Absolutely keep them for three-way junctions. If you are drawing this using "Wire" in Eagle... don't! Use "Net" instead, and the junctions will for the most part automatically show up when they are needed.

[bingo600]

If you remove all the small green junctions where a single wire simply connects to a component, it would make me very happy

Absolutely keep them for three-way junctions. If you are drawing this using "Wire" in Eagle... don't! Use "Net" instead, and the junctions will for the most part automatically show up when they are needed.

Thanx for the NET tip , will use that for new schematics

My EE suggested to rearrange the trimpot , should create better tempco i think.



/Bingo

[Mickle T.]

My version of the quick-and-easy LM399 voltage reference 

[Andreas]

Hello Mickle,

just one question.
From where can I get those tiny dewar cylinders.
Do you have a source with reasonable pricing?

With best regards

Andreas

[macfly]

Side note: 0.1 ppm/C is only after selection. All resistors in my reference have a 5 ppm/C class (max tempco). Typical is 10 times better. By selection is possible to achieve 0.1 ppm/C (20-50 C range).

Hello Mickle,

would you let the cat out of the bag, where we can buy S5-61 foil resistors?

Regards,

Macfly

[Andreas]

Hello Mickle,

would you let the cat out of the bag, where we can buy S5-61 foil resistors?

Hello,

You could also order some VHP100/VHP103 or from the HZ series resistors from vishay and select for TC.
Most probably the S5-61 resistors are no longer manufactured. Otherwise you could google after them.

with best regards

Andreas

[bingo600]

My version of the quick-and-easy LM399 voltage reference 

Mickle ... Easy 
Is that transistor arrangement for controlling the Zener from the 10v , meaning super stable ?
Why the transistors , to avoid loading of the Vref Out ?

Edit: Beautifull finish 
        Is there an "outside screw" that connects to the trimpot on the 10ppm adjust ?

/Bingo

Ps: I just modifyed my LM299 to implement the changes that are in the last schematic.

Here is a graph from the night (before the R3/C5 mod , and the trimpot mod)
Blue (K2015) is the LM299 , strange that the LM299 (Board and/or trimpot) seems to have a negative tempco.

Well see what the next day's graphs reveals

/Bingo

[Andreas]

Mickle ... Easy
Is that transistor arrangement for controlling the Zener from the 10v , meaning super stable ?
Why the transistors , to avoid loading of the Vref Out ?

Hello,

the transistors are just a standard cirquit to give a short circuit proof output with at least 10-15mA current.
In case you want to attach a thermal rms converter for precise AC calibrations.

with best regards

Andreas

[Mickle T.]

From where can I get those tiny dewar cylinders.
Do you have a source with reasonable pricing?

Hello Andreas,

These tiny dewars was taken from "Giatsint" TCXO (~10$), but I don't know is it currently available or not.

Mickle T.

[Andreas]

Hello Mickle,

thanks for the information.

When I look for dewar I only get to pages with LN2.
And the lowest cost dewar is around one HP34401A.
$10 sounds reasonable.

With best regards

Andreas

[Andreas]

Blue (K2015) is the LM299 , strange that the LM299 (Board and/or trimpot) seems to have a negative tempco.

Interesting would be also the raw value of the voltage.

Which type of trimmer do you use?
AccuTrim series of vishay?

Otherwise you should use the "standard precision" trimming scheme shown e.g. in the AD587 datasheet (figure 3+4).
extra 10-100k trimmer from the output to ground and the tap via a "precision resistor" (RT in datasheet) to the center tap of the 9+20K divider.

with best regards

Andreas

[bingo600]

Andreas

Else i think you should look at Thermo Cups

http://www.aliexpress.com/item/Chinese-tea-cup-280ml-261-280-elegant-tea-art-double-layer-glass-water-cup/1375175051.html?s=p

The above one seems to big (280 ml) , but ....

http://www.aliexpress.com/item/free-shipping-by-CPAM-10-pcs-lot-Bombs-Away-Shot-Glasses-Missile-Shot-Glss-44ml-Double/671805147.html

http://www.aliexpress.com/item/Free-Shipping-Set-of-6-hand-blown-double-walled-glass-Espresso-Coffee-mug-80ml-teacup-Thermo/1760091959.html

http://www.aliexpress.com/item/4pcs-1-7-fl-oz-50ml-Heat-Resisting-Glass-Double-Wall-Cocktail-Wine-Tea-Cup-Mug/1545114611.html


Not perfect , but maybe usable

/Bingo

[bingo600]

Blue (K2015) is the LM299 , strange that the LM299 (Board and/or trimpot) seems to have a negative tempco.

Interesting would be also the raw value of the voltage.


Which type of trimmer do you use?
AccuTrim series of vishay?

Otherwise you should use the "standard precision" trimming scheme shown e.g. in the AD587 datasheet (figure 3+4).
extra 10-100k trimmer from the output to ground and the tap via a "precision resistor" (RT in datasheet) to the center tap of the 9+20K divider.

with best regards

Andreas

raw value

Is that the LM299 Vout ?

The trimmer has CERMET written on it (but from ebay .. so ??) , i have 100 & 50 ohm also but only 0805 1% smd's in "General values" (csd-electronics.de).
I might measure the current value on the 200R , and try to replace it with a resistor & a 50ohm trimmer.

I don't have any of the stable trimmers.

Will have a look on the 587 DS (thanx)


/Bingo