Author Topic: LM399 based 10 V reference  (Read 537156 times)

argintviu, Birb, MegaVolt, wry and 4 Guests are viewing this topic.

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 10576
  • Country: gb
Re: LM399 based 10 V reference
« Reply #1100 on: May 25, 2020, 12:25:12 am »
R5, 200K looks like too high a value. the LM399 operating current is 0.5-10 mA, so with a 24V supply, R5 should be between 34K  and 1.73K.

The schematic does not show the connection between Z- and GND.

Look again. R5 is there just to guarantee start-up. R4 supplies the zener's operating current.
Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Offline niner_007

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: us
Re: LM399 based 10 V reference
« Reply #1101 on: May 26, 2020, 01:22:08 am »
R5, 200K looks like too high a value. the LM399 operating current is 0.5-10 mA, so with a 24V supply, R5 should be between 34K  and 1.73K.

The schematic does not show the connection between Z- and GND.
Here you go; and an update on the PCB layout
« Last Edit: May 26, 2020, 01:40:09 am by niner_007 »
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: LM399 based 10 V reference
« Reply #1102 on: May 26, 2020, 02:52:11 am »
Hi niner_007,

Are you aiming to get a 10.0000V reference or around 10V?

The datasheet shows:



and




with your 10K 22K divider you have a nominal gain of 1.4545

So your 10V output could be between

9.81 and 10.6V

Are you going to include a trim circuit?

Here is a picture of Datron 1281 connected to a Fluke 732A with Pomona 5291A low thermal emf patch cords.

They disagree by 4.69ppm  :-DMM



Regards,
Jay_Diddy_B

 
The following users thanked this post: niner_007, KK6IL

Offline niner_007

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: us
Re: LM399 based 10 V reference
« Reply #1103 on: May 26, 2020, 03:11:22 am »
No, I don’t care about the exact value, I’m not aiming for a perfect 10V
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: LM399 based 10 V reference
« Reply #1104 on: May 26, 2020, 03:31:05 am »
Do you have the VPG VHD200 10K / 22K resistor pack?

Jay_Diddy_B
 

Offline niner_007

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: us
Re: LM399 based 10 V reference
« Reply #1105 on: May 26, 2020, 03:46:26 am »
I have not made the order with Vishay yet, I'm planning to this week

The final value not yet decided, probably more like 21.3K and 10K
« Last Edit: May 26, 2020, 03:57:55 am by niner_007 »
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: LM399 based 10 V reference
« Reply #1106 on: May 26, 2020, 03:52:50 am »
Think about LT5400BIMS8E-1#PBF.

You can build a 10K 20K divider with this part. It will give you a gain of 1.5 instead of 1.45

Should be in stock and a lot cheaper.

Regards,
Jay_Diddy_B
 
The following users thanked this post: niner_007

Offline niner_007

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: us
Re: LM399 based 10 V reference
« Reply #1107 on: May 26, 2020, 04:06:03 am »
I have made a design with LT5400 and LM399 before, but I decided against it this time, opting in for through hole parts
 

Offline branadic

  • Super Contributor
  • ***
  • Posts: 2379
  • Country: de
  • Sounds like noise
Re: LM399 based 10 V reference
« Reply #1108 on: May 26, 2020, 05:56:59 am »
Quote
but I decided against it this time, opting in for through hole parts

Go for TDP1603 10k network then or use a NOMCA1603 network the same way and use a SMD adapter board, to make it throughhole. Still much cheaper.
10k:25k gives you the right 0.4 ratio you want and all resistors within the package can be used to form that ratio.

-branadic-
Computers exist to solve problems that we wouldn't have without them. AI exists to answer questions, we wouldn't ask without it.
 

Online The Soulman

  • Frequent Contributor
  • **
  • Posts: 949
  • Country: nl
  • The sky is the limit!
Re: LM399 based 10 V reference
« Reply #1109 on: May 26, 2020, 08:08:33 am »
Quote
but I decided against it this time, opting in for through hole parts

Go for TDP1603 10k network then or use a NOMCA1603 network the same way and use a SMD adapter board, to make it throughhole. Still much cheaper.
10k:25k gives you the right 0.4 ratio you want and all resistors within the package can be used to form that ratio.

-branadic-

Looking at the data sheet these nomca networks may not be good enough?

Quote
TCR: Tracking ± 5 ppm/°C (typical) -55 °C to +125 °C
Stability: RatioΔR ± 0.015 %1000 h at +125 °C
Shelf Life Stability: RatioΔR ± 0.002 %1 year at +25 °C
 

Offline iMo

  • Super Contributor
  • ***
  • Posts: 4675
  • Country: nr
  • It's important to try new things..
Re: LM399 based 10 V reference
« Reply #1110 on: May 26, 2020, 08:48:11 am »
I've been using LT5400 as the 1.5 multiplier. The LT5400 is 0.2ppm/K (AD told me it is rather <1ppm/K) TC of ratio tracking. With +/-5ppm/K with NOMCA you may get similar tracking when lucky (more resistors in the divider may average out the TC of the divider's tracking ratio).
« Last Edit: May 26, 2020, 08:50:45 am by imo »
 

Offline branadic

  • Super Contributor
  • ***
  • Posts: 2379
  • Country: de
  • Sounds like noise
Re: LM399 based 10 V reference
« Reply #1111 on: May 26, 2020, 09:19:29 am »
Quote
Looking at the data sheet these nomca networks may not be good enough?

Please refer to the thread Statistical arrays, they behave much better than the datasheets indicates.

There is a lot of knowledge on this board, that is worth checking.

-branadic-
Computers exist to solve problems that we wouldn't have without them. AI exists to answer questions, we wouldn't ask without it.
 
The following users thanked this post: The Soulman

Offline niner_007

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: us
Re: LM399 based 10 V reference
« Reply #1112 on: May 27, 2020, 03:42:08 pm »
I ordered the PCB with oshpark, I will share the project and gerbers after I tested it.

Just to validate my understanding of the resistors in the LM399 circuit, I understand the ratio setting ones very well. My understanding is that the stability of the 200K resistor is irrelevant, and that the 5K current setting resistor has very little impact. Based on the LM399 resistor thread. Is this correct?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: LM399 based 10 V reference
« Reply #1113 on: May 27, 2020, 03:55:23 pm »
I ordered the PCB with oshpark, I will share the project and gerbers after I tested it.

Just to validate my understanding of the resistors in the LM399 circuit, I understand the ratio setting ones very well. My understanding is that the stability of the 200K resistor is irrelevant, and that the 5K current setting resistor has very little impact. Based on the LM399 resistor thread. Is this correct?
That is correct. The actual zener current is set inside the LM399 and due to the regulation loop inside the output impedance is quite low. So the current setting resistor is less important (some factor of 10, maybe more) than with a zener direct.
 
The following users thanked this post: niner_007

Offline niner_007

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: us
Re: LM399 based 10 V reference
« Reply #1114 on: May 28, 2020, 08:55:30 pm »
Quote
Looking at the data sheet these nomca networks may not be good enough?

Please refer to the thread Statistical arrays, they behave much better than the datasheets indicates.

There is a lot of knowledge on this board, that is worth checking.

-branadic-
Price is of no issue, that's why I was leaning towards VHD200, I'm not planning on selling any of these, they are for me, and me alone
 

Offline KK6IL

  • Regular Contributor
  • *
  • Posts: 55
  • Country: us
Re: LM399 based 10 V reference
« Reply #1115 on: July 08, 2020, 07:00:16 am »

I just bought a 34401 from Craigslist and the seller threw in a free 3468. They both agree with my pair of 3456's that my DIY LM399 standard is outputting around 9.998. Measured 9.9991 on a trio of good but long out of cal meters a year ago. So the time has come for me to build a proper standard using what I've learned on this thread.

Attached is my design for everyone to shoot holes in.  Component selection somewhat driven by contents of my junkbox. 3K 0.01% divider resistors likely from same production batch but I should be able to verify TC tracking by putting 4 at a time as a whetstone bridge and watching null as I vary temp.

I'm considering placing the LM399's upside down on the vectorboard with RTV and wiring all 4 leads with 30 ga wire to avoid mechanical stress, to slot or not slot, etc. Would that likely help or just add more thermocouples?

Main use would be to calibrate my meters, and to drive a KVD, but no reason not to make it as noise free as practical.

Bought a 3455 years ago for $20 if I promised not to return it if it didn't work. Well it didn't, most front panel switches bad as well as unknown problems after that. Saw a thread on the 3455 reference module, so removing that and making it into a calibrator also tempting.

John
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: LM399 based 10 V reference
« Reply #1116 on: July 08, 2020, 07:31:16 am »
The capacitor C3 is wrong - just leave it out, or much smaller (e.g. 10 pF range). If more filtering is wanted, increase C10 instead and maybe add one more RC stage there.

The LH0033 is kind of overkill there.
 

Offline branadic

  • Super Contributor
  • ***
  • Posts: 2379
  • Country: de
  • Sounds like noise
Re: LM399 based 10 V reference
« Reply #1117 on: July 08, 2020, 08:04:04 am »
I'm missing some caps on the +22V rail. I would also add some 100nF foil capacitors across the zener diodes.

-branadic-
Computers exist to solve problems that we wouldn't have without them. AI exists to answer questions, we wouldn't ask without it.
 

Offline KK6IL

  • Regular Contributor
  • *
  • Posts: 55
  • Country: us
Re: LM399 based 10 V reference
« Reply #1118 on: July 08, 2020, 10:39:40 am »

The LH0033 is indeed overkill but I have 1 or 2 dozen of them. Interesting part with their 10^10 input impedance and high speed. Remains to be seen if I can put a 10 Hz low pass filter in the loop without creating a high power oscillator. If it doesn't work, I'll remove it. I can't get a lot of noise reduction from C10 without getting into electrolytics and their leakage and DA. I'm a digital/FPGA designer, not so much an analog guy. Never did understand Bode plots.
 

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 3222
  • Country: de
Re: LM399 based 10 V reference
« Reply #1119 on: July 08, 2020, 07:15:27 pm »

I'm considering placing the LM399's upside down

but no reason not to make it as noise free as practical.

Hello John,

Upside down for a LM399 means: maximum power consumption. (around +10% against normal)
And most likely: higher thermocouple voltage at the soldering joints.

I also hope that you select the LM399 for minimum noise. There is a large stray from device to device.

with best regards

Andreas
 

Offline Inari

  • Contributor
  • Posts: 17
  • Country: ee
Re: LM399 based 10 V reference
« Reply #1120 on: January 10, 2021, 05:58:40 pm »
Hello Mickle T.

What is that "вольтметр постоянного тока" in one of your pictures? I know what it is but seems very interresting device - handmade? Do you share any information about this nice looking instrument?
 

Online dietert1

  • Super Contributor
  • ***
  • Posts: 2022
  • Country: br
    • CADT Homepage
Re: LM399 based 10 V reference
« Reply #1121 on: January 14, 2021, 06:20:15 am »
Just looking through the schematic posted by John/KK6IL, i noticed the OpAmp power supplies pin 4 resp. 5/6 are connected to reference Gnd, while they should be on supply Gnd. The LH0033 datasheet does not contain the term "noise", maybe it doesn't have any..

Concerning the remark of Branadic about "capacitors parallel to the zeners": Those aren't exactly zener diodes, but composite circuits (schematic in datasheet). Don't know if there is experience here about phase margins to observe.

What was the result about the filter loop stability? I think it might be stable since the time constants of the divider with C4 and R13 with C3 are completely different. The advantage of the circuit as drawn is that the filter error vanishes inside the reference control loop. If C10 causes oscillation during power-on one could drive R1..R3 from the output of U4.

Also i'd like to ask for the actual voltage differences between the three references. I think datasheet says like +/- 100 mV or so.

Regards, Dieter
 
The following users thanked this post: KK6IL

Offline iMo

  • Super Contributor
  • ***
  • Posts: 4675
  • Country: nr
  • It's important to try new things..
Re: LM399 based 10 V reference
« Reply #1122 on: January 19, 2021, 11:21:17 pm »
@KK6IL: Afaik we had a discussion on 723's noise and stability. Probably not important here too much, but especially the "new parts" drifts..
 

Online dietert1

  • Super Contributor
  • ***
  • Posts: 2022
  • Country: br
    • CADT Homepage
Re: LM399 based 10 V reference
« Reply #1123 on: January 20, 2021, 12:13:32 pm »
Last week i made a board with 5x LM399 fresh from rscomponents and with averaging. Individual  voltages are within +/- 15 mV from the average. I used 274R resistors for averaging. Then i needed a day or so for making and testing a proper bipolar power supply with guard and the like. Initial heater currents sum up.. Once hot, the whole board runs at 50 to 60 mA at +/- 10 V = 1 W, mainly for the heaters.

After that i started a log using one of our HP3456A. This test is clearly limited by the 10 uV resolution of the HP3456A. Anyway i observed that the average reference voltage started at 7.01531 V, went to 7.01532 V after 3 minutes, to 7.01533 after 10 minutes. 7.01534 V appeared for the first time after 80 minutes. Since then it seems to sit at 7.015336 - with a large majority of samples at 7.01534. So initial drift after startup has been some 3 or 4 ppm. As we know there may be more to come during the first year or so.

Regards, Dieter
« Last Edit: January 20, 2021, 12:42:30 pm by dietert1 »
 
The following users thanked this post: Trader

Offline iMo

  • Super Contributor
  • ***
  • Posts: 4675
  • Country: nr
  • It's important to try new things..
Re: LM399 based 10 V reference
« Reply #1124 on: January 21, 2021, 11:27:45 am »
@dietert1: a pcb or proto?
« Last Edit: January 21, 2021, 11:33:03 am by imo »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf