LM399AH#PBF - Low stock worldwide - I need qty=30

[wraper]

The legs are each the same combination of material

Nope, for it to not matter, thermal gradient over such material junctions must be equal across the legs for thermal EMF to cancel out.

     Same combination of materials and same length of legs (as I already wrote), will ensure that the EMF cancel.

Bullshit. There is like 60^oC to drop over the legs (at room ambient temperature). Either you did not read what I wrote or refuse to switch your brain on. You have thermocouple junctions in middle of the legs and each junction is at somewhat different temperature. There is no perfect matching of this welding bullshit.
 

What makes you think that?  Just a few comments above you can see a photo of a new, genuine part which has that very same suffix.

H is prevalent and AH is more expensive.
 

It is refurbished.  No-one guarantees and no one expects that it performs as well as a new product.

Really? If I buy LM399AH, I expect it to perform according to LM399AH specs. It may be not perfect visually but actual operation must be up to spec.

[IconicPCB]

Another well socialised response.

[MadTux]

That's just the truth about the garbage that comes from china.
That guy seems the rare exception, since I think he honestly cares about what he sells:
https://www.ebay.com/itm/1x-LM399H-Precision-Reference-LM399/121069770252?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

60°C at that welded junction can easily make a few mV, a mV on a 7V device is about 150ppm. And since LM399s are frequently used on instruments accurate to about 30-10ppm, it matters, I think.
N-type thermocouple has 1.6mV at 60°C and it uses quite similar metals, NiCrSi-NiSi
One leg soldered to large pad that acts as heat sink might be enough to unbalance the thermocouple pair on both legs to get significant instability.

[FriedLogic]

  It's a bit pointless arguing about the TC and thermocouple effects on the performance without knowing the material used in them or the way that it's finally mounted on the PCB. I tried one for some weeks along with a new LT LM399AH and although the TC for around 25-50°C was a little worse it did not have the popcorn noise that the new one had.

  I cut one open and it was marked LT on the die, but the fake case was marked NS.

  2015 NS parts are amusing

  Reclaimed parts may have suffered a lot, and the 'refurbishment' only adds more. I noticed that someone doing high temperature tests on them got a lot of furry corrosion on the leads too. Others mentioned that the legs came off on some, so long term stability could be another problem.

  Overall, I would much prefer just plain used parts to refurbished ones, but they are usually way too close to the price for new ones. Refurbished ones are a curiosity, but add too much uncertainty for my liking.

[Kleinstein]

Chances are the reclaimed parts are a random mix of 399AH and 399H from different ages with both LT and NS parts. The honest way would be to mark them as something like LM399R or something similar without man manufacturer logo. This could avoid legal problems.

Anyway for a demanding use one would get the parts from a reliable source or at least do testing one all parts. Even the original ones can show higher drift or excessive popcorn noise.

Welding the new leads could be quite some stress too. If the wires are iron and not Kovar like the original there will be some additional thermal EMF. However like with the original, the thermal EMF will cancel out to a large part with a symmetric setup. Iron / steel is usually in between Kovar and copper for the thermal EMF. So the difference is not that large.

For most purpose I would consider the reclaimed parts lower grade than the non A. Even if they were original AH parts, that small difference is probably lost.

[Andreas]

However like with the original, the thermal EMF will cancel out to a large part with a symmetric setup.

How do you guaranty a symmetric setup when one pin is directly connected to the case bottom of the LM399? 
And the others have probably different lengths from the bottom to the welding point. 

 

with best regards

Andreas

[IanJ]

I am not so much interested in how it performs, but rather what did I actually get? Is there an easy way to see diff between AH and non AH if I take the top off the can?

Ian.

[wraper]

I am not so much interested in how it performs, but rather what did I actually get? Is there an easy way to see diff between AH and non AH if I take the top off the can?

Ian.

No, it's exactly the same IC with the same die in the same package. AH is simply tested/binned to tighter spec. But it does not mean that H cannot outperform AH, it's simply that guaranteed specs are tighter.