LM399AH sources

[jonpaul]

Hello: Got the fine Keysight 34465A, was testing my Vrefs.

I Have a LM399 ref board made in Germany, with the LM399 and precision opamp buffer to make 10.000V

the 10.00 output is bad, suspect the LM399AH is damaged.

Lots of LM399AH  on epay, all from China, very wary of knockoffs and clones.

How to get an original genuine Linear Technology LM399AH ? How to spot the Chinese Illegal  clones?

Any better refs available pin compatible or otherwise?

Open to offers if any has spares.

Am in USA as well as France.

Many thanks

Jon

[TiN]

Mouser has some, just buy from them?

[dietert1]

In Germany:
https://www.ebay.de/itm/144082726982?hash=item218bfff046:g:G4cAAOSw7AZgN4RB
We bought from him before, no problem.

Regards, Dieter

[Kleinstein]

From my limited experience the LM399 are not all the same, but the A version is not fundamentally better. Chances are it is just a slightly tighter screened (for TC, but hardly for noise or drift) selection.
If you need a really good one, one would need to do a burn in with a few and than select the best, based on noise and drift - the TC is usually good enough anyway, also for the non A version.

As long as they are available from a known reliable source there is little need to buy cheap and risc getting fakes or the left overs after a slection process.
Reclaimed chips may already have some burn in done, but I would not trust the A very much an a part from China.

[jonpaul]

Rebonjour a tous! Mille mercis,

As new owner of the Keysight 34465A, I am a newcomer to the world of the Volt nuts and references.



0/  Here is the circuit being used, I added a 12V pre-regulator.

1/ Is there a better grade of LM399, than the AH, I  need the metal can 4 pin inside heat shield.

2/ For better accuracy/lower drift what other Vref to consider that can have similar Vout?

Ciao!

Jon

[Kleinstein]

AFAIK there are one 2 grade available: LM399H and LM399AH. There were a few other grades (e.g. LM199 ) before, but AFAIK no langer available.
They come with the plastic cap.

The LM399 heater usually want a relatively high voltage to work best. A 12 V supply for the heater would already be marginal and may effect the TC. The more normal case is some 30 V for the heater. The heat votlage may have a small effect and should be stabilized - though no extremes needed here.

The next step up from am LM399 would be a LTZ1000 based refrence circuit.

If the ouput of interest is 10 V anyway, one could consider other reference chips (5 V or 10 V) too, possibly with a separate build oven to get a low TC.
There are a few in a ceramic case though could come close in the drift.

[branadic]

My suggestion is to go for the bootstrap solution. 15V supply voltage is okay. Even 12V provided by an LT1763 worked reasonable for me.
For the resistors a NOMCA1603 resistor network with trim option is all you need, though you can also go for NOMC1603, TOMC1603 or TDP1603, ... , all more than good enough for an LMx99.

I do have some new LM399A, LM399AH and LM299xx that I could sell to you, if you want.

-branadic-

[jonpaul]

Rebonjour, Kleinstein et autres....

I opened up the REF box,  LM399AH may be OK, Vin 15V Vout 7.03778V stable. on the Keysight 34465A.

the input prereg is LM2940-15 15V LDO. Input to the LM399.

The divider R are fine.

The buffer amp LT1001ACN8 is dead. (8 pin DIP)

Any  good source for LT1001ACN8  (NOT China!) or  recommendations for better part in 8 pin DIP than LT1001ACN8 much appreciated.


Many thanks again!

Jon

[branadic]

LT1001ACN is available at Mouser. LT1006 can be used too.

-branadic-

[Kleinstein]

For the OP an OP07,  OP27, OP177 or OPA177 should also be OK too. The reference is not really high impedance.

The Boostrapped solution for the current mentioned by brandic is defenitely a good idea.

The divider is a bid odd in that the whole current goes through the wiper. This is not the best idea to build a trimmable divider. Better have only a small current at the wiper. A selected fixed value resistor for the coarse trim may allow saving on the trimmer quality.

[dietert1]

The LM399AH source from ebay Germany i linked above is absolutely reliable. Those are original parts and they were found to have tolerances far better than the datasheet. Pins had no traces of having been in sockets for previous burn in. Those statements are plain fantasy. They don't produce popcorn noise when used properly.
As far as i can tell the main difficulty is using precision parts properly. The differences between the parts are often less than you may think.

Regards, Dieter

[mawyatt]

Jon,

We got the LM399AH and LTZ1000 from Mouser awhile back, when the LM399AH recently came back in stock we got another couple LM399AHs.

These were used in a couple custom reference designs shown. They are based on the bootstrap method for both the LM & LTZ references, and powered by 24V and use a 15V regulator to the heater and op-amp. The cases are 3D printed, and the dark blue holds a LTZ1000, LM399AH and another 10V reference (can't remember which one). The lighter blue holds a LM399AH and a 10 volt reference (can't remember which one), and also has a 1V and 0.1V output based upon a 10X type 9 resistor precision voltage divider (3 series and 3 parallel) and CAZ op-amps. We keep these running 24/7 from a cheap eBay 24V SMPS (one LM399AH has ~10,000 hours, the LTZ and another have ~500 hours), and when making a measurement we use a linear supply for the 24V input.

BTW you are going really enjoy the KS34465A, these are the older brothers of the famous 34401As. You can even get a 7 digit result using the MX+B function I believe (no evidence tho) these are the same hardware as the 7 1/2 digit KS34470A with exception to the reference, which is the LM399AH in the 65A and LTZ1000 in the 70A. Maybe someone in the "know" will comment

With your skills you could roll your own 10V reference based upon the bootstrap technique.

Edit: It's seems you could modify the design you've shown to a bootstrap type by moving the 5.6K resistor to the op-amp output and changing it to a 3K. You may need a high value (~200K) startup resistor to the supply rail if the LM399 reference doesn't start.

Edit: Here's some schematics & 3D models for the custom references.

Best,

[dietert1]

A LTZ1000 reference should be in a metal enclosure and it's a shame to build a LTZ1000 reference with loudspeaker binding posts. If you want good results from a LM399 reference, the same applies. There is a LM399 thread with lots of hints. And there has been a nice thread by Andreas about hardening a reference to behave in the presence of RF EMI.

Regards, Dieter

[mawyatt]

The binding post work fine for what we need at the moment, which is mainly to put burn-in hours on the references. Later when time permits a proper enclosure with proper shielding & terminals will be applied, but this must wait until a major research project (128 independent output 15 bit precision AWG with +-100V output capability, later may go to many more outputs) is completed (likely not this year).

Edit: BTW one thought we had when/if we ever get back to this, is painting a 3D printed case inside with a conductive paint with an interior cage for the in/out leads using feedthru capacitors. Has anyone had success using a 3D printed case with conductive paint?

Best,

[jonpaul]

Rebonjour Kleinstein, Dieter, and mawyatt

You are really the tru "Volt Nuts" your response greatly appreciated.

I was EE back to 1960s and had worked at Julie Research Labs (JRL) in NY perhaps 1966..1968. I still have some of their wire wound had calibrated resistors.

But this is my first cut at really precision volt calibration.

Mawyatt very fine designs, will you sell any kits or finished units?

For the Keysight, I discovered the support is the best, the construction is very solid and the unit is in CAL after 1 yr factory CAl.

Several options are "soft" and can be enabled by loading new FW: The 2M point memory and the digitizer.

I can believe the LM399><LTZ1000 reference difference, and can get > 6 1./2 digits with smoothing, averaging, ,etc.

My 10V ref is not completely dead, stupidly I misconnected the power and blew both the LM2904-15 (NNOT -12) as well as the LT1001 opamp.

But the LM399ACH ref is working !

Re binding posts, there are gold plated and also gold banana  receptacles. I have sets of gold plated spade lug jumpers and gold banana plug jumpers.

More to follow, after the Keysight 34465A  is factory serviced and recalibrated.

Bon Journee,

Jon

[bdunham7]

0/  Here is the circuit being used, I added a 12V pre-regulator.

1/ Is there a better grade of LM399, than the AH, I  need the metal can 4 pin inside heat shield.

2/ For better accuracy/lower drift what other Vref to consider that can have similar Vout?

0) That circuit is very basic and will not give ideal results.  You should not run the heater directly from the same source as the reference zener and as others have mentioned, the zener should have its current supplied by a 'bootstrap' configuration whereby the voltage across the zener + resistor is regulated using the output of the LM399 itself as the reference.  The dynamic resistance of the LM399 is low, ~1 ohm, but it isn't zero.  So if you are using a 1mA zener current, a 1% change in that would mean ~10uV change in output voltage.  That's OK for a basic reference, but a needless error in a precision device.

1) The better grade LM399 is the one you select yourself after buying a batch of 10 or 20 and putting them in a test rig, selecting the best ones for tempco, then burning them in for a few thousand hours and selecting the best ones for drift.  I think the only difference with the 'A' version is that it is initially selected for a lower maximum tempco, but I'm not sure the overall yield from a more demanding selection process would be any higher with the 'A'.

2) Search the forum for LTZ1000 and see what these guys have done with that.  It's almost scary.

[mawyatt]

Rebonjour Kleinstein, Dieter, and mawyatt

You are really the tru "Volt Nuts" your response greatly appreciated.

I don't qualify, I'm more of an IC Nut :

I was EE back to 1960s and had worked at Julie Research Labs (JRL) in NY perhaps 1966..1968. I still have some of their wire wound had calibrated resistors.

Recall using some of these WW resistors decades ago, great products form JRL

Mawyatt very fine designs, will you sell any kits or finished units?

Thanks, but they are not for sale. Now we have a proper company "Wyatt Labs LLC", too much liability, and with such little interest outside this group not worth the effort & time to finish up a complete "proper" product

Several options are "soft" and can be enabled by loading new FW: The 2M point memory and the digitizer.

Thanks will give this a try.

I can believe the LM399><LTZ1000 reference difference, and can get > 6 1./2 digits with smoothing, averaging, ,etc.

With aging they should be very stable. Even with the setup shown, the LM399 with ~10,000 hours is very stable. The KS34465A indicates only a few uv SD over a many hours period, and this is in a room with direct AC air blowing over the lab bench (can't be avoided!)

My 10V ref is not completely dead, stupidly I misconnected the power and blew both the LM2904-15 (NNOT -12) as well as the LT1001 opamp.

But the LM399ACH ref is working !

That's good that the LM399 survived, hopefully without any damage! You can probably just drop in a OP-07 replacement for the LT1001.


Best,

[jonpaul]

Again I am stunned by the expertise and generous help received on the forum.

I am off to travel soon and have sent the Keysight 34465A back for service (intermittent dead display) so will pick this up another time..

Cordialement,


Jon

[maxwell3e10]

I noticed that Keysight lists LM399 on their parts store for a very reasonable price. Has anyone tried to order it? It could be even better selected than standard parts.

https://www.keysight.com/my/partDetail/1826-1249

[dietert1]

Edit: BTW one thought we had when/if we ever get back to this, is painting a 3D printed case inside with a conductive paint with an interior cage for the in/out leads using feedthru capacitors. Has anyone had success using a 3D printed case with conductive paint?

A solution not using metal may perform even better, since the "half conductive" shield will absorb energy. As far as i remember people reported using conductive 3D material. It depends on conductivity. There seem to be materials meant for ESD protection, but that's a different story. I found literature where they try to embed conductive paths into standard PLA, so higher conductivity materials also exist. You can use a cellphone to check how your reference performs.
By the way: Pomona 3770 Low Thermal EMF binding posts (gold plated copper) are about €12 a piece and i use them with copper lugs from rs components (433-040, 100x bag for € 6,40+VAT). Don't know if that qualifies for a voltnut..

Regards, Dieter

[Kleinstein]

I noticed that Keysight lists LM399 on their parts store for a very reasonable price. Has anyone tried to order it? It could be even better selected than standard parts.

https://www.keysight.com/my/partDetail/1826-1249

The list of instruments were this part is supposedly used does not include the normal DMMs (e.g. 3446x), where drift is critical. Chances are this may be just the stock part with the custom number. For the DMMs with tight drift specs they need some with a burn in and maybe selction for low drift.

Just some initial burn it would not be so difficult to do at home - drift and noise testing is more tricky.

[maxwell3e10]

It includes E1412A, which is a VXI meter similar in specs to 34461 /34401 with 35ppm 1 year accuracy. I think they don't list 344xx meters because newer meters are no longer considered user-serviceable while 34401 is discontinued.  E1412A was released in 1995 and they post schematics but it is still a current product. 

[bdunham7]

I noticed that Keysight lists LM399 on their parts store for a very reasonable price. Has anyone tried to order it? It could be even better selected than standard parts.

https://www.keysight.com/my/partDetail/1826-1249

What is the price and where do you see it??

[maxwell3e10]

Shows the price of $26.31 if I just click on the link. Says "Not in stock, but orderable" so I am not sure what happens if one orders it.

[mawyatt]

A solution not using metal may perform even better, since the "half conductive" shield will absorb energy. As far as i remember people reported using conductive 3D material. It depends on conductivity. There seem to be materials meant for ESD protection, but that's a different story. I found literature where they try to embed conductive paths into standard PLA, so higher conductivity materials also exist. You can use a cellphone to check how your reference performs.
By the way: Pomona 3770 Low Thermal EMF binding posts (gold plated copper) are about €12 a piece and i use them with copper lugs from rs components (433-040, 100x bag for € 6,40+VAT). Don't know if that qualifies for a voltnut..

Regards, Dieter

I've seen something regarding the conductive 3D printing filaments, but don't know anymore and hoping someone had experience with these or the conductive paint. We have some better quality terminals that are dual types (Red & Black) somewhere, and will look into this and the "shielding" if/when we can get around to this.

Thanks for the information.

Best,