<$30 eBay LM399+LT1001 2.5/5.0/7.5/10V Vref Module Hack- Step-1: PoC

[GEOelectronics]

I'm striving to build a practical, inexpensive 10.000000V DC reference for my Kelvin Varley Divider Calibrator project (ala poor man's Fluke 732A/752B) for my Mini Metrology Lab
My thanks to Conrad Hoffman for the inspiration :  http://conradhoffman.com/mini_metro_lab.html


Several Vref ICs are being evaluated, this weekend's project focuses on a less than $30 USD eBay LM399H based / LT1001 buffered multi Voltage stable reference module which arrived last week. Preliminary testing showed it to be very stable, but also very inaccurate (no surprise with 5% variability of actual output Voltage between samples)
This Asian version's design opted for a (not) precision fixed resistance divider scheme to supply any single output, jumper selected from 2.5, 5.0, 7.5, 10.0. The actual individual "calibration" chart showed all sorts of jumper selected values,but very few zeros present. Quite useless for my 10.00000 + KVD project. (see Fig-1)

9.85773V (/=) 10.00000V

The other V selections held no interest to my project so the jumpers were set for 10V and the plan to add a trimmer to get the accuracy needed began. Basically I want to duplicate the "Portable Calibrator" from the National Semiconductors data sheet for the LM399 series. (see Fig-6)

Pictures of the hack's first crude Proof-of-Concept (PoC) test, data sheets and the module's jumper schedule located at:

http://www.qsl.net/k/k0ff/VOLT_NUTS/LM399H-LT1001_10.00_Vref/

Updates will arrive in that same folder and eventually a project PDF.

Before anything else, all the unneeded resistors and jumpers will be removed and grounding paths cleaned up as required. Once 10.0 V is achieved, all the  other desired Vrefs will be selected by the KVD.

The needed R7 resistance turned out to be 19.182K Ohms (vs. the original fixed 20K), the next step will be to build the trimmer as a low resistance precision 10-turn panel potentiometer ( I like the Spectrol 10T 1K WW and Beckman 15 turn counter for this) with a stable ~18.5K main resistor, and box the whole thing for the next round of experiments and testing.

Have fun.

George Dowell


 

[eurofox]

I order the same and expect it the next coming days.

I got exactly the same idea how to proceed with it .... 

[EmmanuelFaure]

BTW : The "precious" components inside the metal shield appears to be second hand. Date code on the LM399 = 9518, date code on the LT1001 = 9329.

[eurofox]

BTW : The "precious" components inside the metal shield appears to be second hand. Date code on the LM399 = 9518, date code on the LT1001 = 9329.

  this way they are already "aged" 

[GEOelectronics]

BTW : The "precious" components inside the metal shield appears to be second hand. Date code on the LM399 = 9518, date code on the LT1001 = 9329.

Yes and in person it is obvious that the LM399 is dirty looking.

After removing the unwanted parts and redrawing the circuit, I see no reason this couldn't be duplicated quite easily, but frankly this seems to a good enough platform and cost effective for sure, especially considering the simple but effective "can" (the LT1001 spec sheet mentions it should be protected from air currents)


George Dowell

[glarsson]

this way they are already "aged" 

Most likely the aging must be "restarted" after thermal and mechanical chocks (desoldering from old PCB, bumping, cleaning and soldering to new PCB).

[Kleinstein]

The LM399 is getting rather hot in normal operation. So unless really brute force is used the thermal conditions on un-soldering and soldering are not that severe compared to a normal power cycle.

For the LT1001 it is a little different - here a new one is likely better.

A possible problem could however be ESD during recycling.

[cellularmitosis]

This appears to be the origin of these boards:  https://item.taobao.com/item.htm?spm=a230r.1.14.59.6a91279aqTyH3R&id=564137743559&ns=1&abbucket=6#detail

There are a few variants available (TL431, AD584, LM399).

On taoboa, it looks like the LM399 version is about $21.

There's another LM399 board from another taobao seller: https://item.taobao.com/item.htm?spm=a230r.1.14.71.6a91279aqTyH3R&id=538328618275&ns=1&abbucket=6#detail

I've ordered a few of that design, but he seems to either have difficulty with stock, or is a very lazy shipper (I had to harass him to get a second order of a few boards from him).

https://www.eevblog.com/forum/metrology/cheap-ebay-ad584-voltage-references-my-experiences/msg1274955/#msg1274955

[cellularmitosis]

(If there is interest in importing these used / salvaged LM399's from taobao, I'd be happy to do a group buy and distribute the parts).

[Kleinstein]

The circuit around the LM399 looks a little odd and not the highest precision one. So what is the point of buying a board with used parts and a few standard resistors if the circuit has to be bodged somehow and not much more than the LM399 is really useful ?

[EmmanuelFaure]

+1
A brand new LM399 costs ~$13 : https://www.digikey.com/product-detail/en/linear-technology-analog-devices/LM399AH-PBF/LM399AH-PBF-ND/4071473
A precision 5ppm resistor costs ~$1 (Vishay PTF56)
A tinned RFI shield < $2

[GEOelectronics]

Yes that looks correct.
I've added a PDF to the project folder which also shows other eBay boards and parts going into the project:

http://www.qsl.net/k/k0ff/VOLT_NUTS/LM399H-LT1001_10.00_Vref/LM399_Project_05_15_18.pdf

Updates will be added as they occur. I'll try to measure the chip capacitors and generate a schematic today. I see no reason it can't be duplicated in the Home Lab on a much smaller PCB.



George Dowell

[cellularmitosis]

Sorry, I meant importing the used LM399 chips themselves, if people were interested in making their own boards. 

[GEOelectronics]

Sorry, I meant importing the used LM399 chips themselves, if people were interested in making their own boards.

I like that thru-hole board you linked and ordered a few to play with.
My first time ordering from that site. Hoping for good service like eBay.

George Dowell

[GEOelectronics]

This appears to be the origin of these boards:  https://item.taobao.com/item.htm?spm=a230r.1.14.59.6a91279aqTyH3R&id=564137743559&ns=1&abbucket=6#detail

"There are a few variants available (TL431, AD584, LM399)."

These piqued my interest early on as well and my admittedly crude testing indicated they were all reasonably stable (by my newbie standards) but not accurate  to below-milliVolt 10.000V.

So far the best out of the box 10.000V device was the AD2702SD (metal can Mil spec ver.) This is the one I'm mounting up on an Augat Gold-Teflon socket and into a tin can filled with Dow #4 Silicone Compound as a temperature insulator-lag "Castle". http://www.qsl.net/k/k0ff/VOLT_NUTS/AD2702SD/

Another very impressive module but hard to find now is the IET HSVR (High Stability Voltage Regulator- no longer made). So far only the 6.3V versions have surfaced:
http://www.qsl.net/k/k0ff/VOLT_NUTS/IET_HSVR_6.3-FIRST%20_LIGHT/

Physics/Metrology is a hobby for me, has been for a long time, but probably the path I took to get to  Volts and Resistance is far different than most and my Home Lab is not like most here either:
http://www.qsl.net/k/k0ff/Identifying_stable_elements_via_XRF_with_Amateur_Apparatus/Pub2PDF.pdf

George Dowell

[1audio]

Any updates? I need to make a new reference for a vintage JRL Volt-a-vider. It has a 6 digit KV divider an output amp with remote sensing etc. The original seleced Zener reference was pulled and a marginal reference using MC1469 was kludged in which is not up to the task. Modifying one of these may be the easiest way to get a suitable reference to patch in.