Binning LM399s

[rodpp]

The pictures anexed using the forum tool was a mess. I uploaded the pictures using Imgur and updated the posts. Sorry for the incovenience.

[rodpp]

Today I received some LM399H, and to my surprise they came without the thermal shield:



I already contacted the vendor to return it.

Is there any reason to not return it, even if the seller offer a discount?

[dietert1]

Difficult to answer, since you didn't tell the numbers.
If i had one of those i would try and make a thermal insulator with a 3D printer. It would look more like those covers used with LTZ1000 references, something that gets fixed to the board. The original LM399 cover that touches the metal case and is free for minor movements, is a good solution yet one may still improve it. What if the so-called popcorn events were simply movements of that plastic hat?
When soldering those references to a large board i would build in some bends into the wires, to take away board bending movements from the reference case by elastic deformation of the wires. I mean something like an s-curve. Bending the wires needs care to avoid stress on the hermetic seals of the metal case.

Regards, Dieter

[rodpp]

The price will be around US$2.5.

I could buy ones with the thermal shield for around US$6.

So the naked ones are less than half the price.

Unfortunately I don't have a 3D printer.

[rodpp]

I found a 3D printing service here that is not expensive. I'll keep the "naked" LM399s and order the thermal shield from that 3D print service.

[Kleinstein]

The original caps are a special high temperature plastics. For a printed cap one may consider a slightly different cap, that does not touch the metal can.

[dkim5nu]

Where can you get these for $6 a piece???

[wraper]

The original caps are a special high temperature plastics. For a printed cap one may consider a slightly different cap, that does not touch the metal can.

Should get ABS as minimum, PLA most likely will become soft as it has 60^oC glass transition temp.

[rodpp]

The original caps are a special high temperature plastics. For a printed cap one may consider a slightly different cap, that does not touch the metal can.

Yes, or a heat resistant material. I suppose that it supporting around 90ºC is ok.

Where can you get these for $6 a piece???

In the brazilian eBay, NOS components. In eBay there are some too:
https://www.ebay.com/itm/LM399H-PBF-Fixed-Shunt-Voltage-Reference-6-95V-5-4-Pin-TO-46-TU/133128046030?hash=item1eff0cc5ce:g:2sYAAOSw~YldP4Qa

The original caps are a special high temperature plastics. For a printed cap one may consider a slightly different cap, that does not touch the metal can.

Should get ABS as minimum, PLA most likely will become soft as it has 60^oC glass transition temp.

The print house suggested ABS, Petg, or PLA HT (High Temperature).

[David Hess]

Interesting graphic, looking at the graphic scale (total of almost 2 years) the CH6 popcorn noise seems somehow sporadic.

Popcorn noise can be sporadic on a scale of hours to days.  It took me a month to track it down in a Tektronix made integrated circuit.

Since it is related to processing, I am surprised that any modern analog IC would suffer from it.

The popcorn noise is usually attributed to single quantum states (e.g. electron occupation at some defects or impurities) that have a relatively large impact because they sit a some sensitive point. One would not note most of the defects with only a weak effect, but they can still contribute to 1/f noise.

While impurities can be reduced, the other prime candidate are defects at the oxide interface, that may not be avoidable. With smaller structures there may be less defects, but more of these can have a large effect.

My favored model is that 1/f noise is the accumulation of different levels of popcorn noise and generated by the same mechanisms.

[2N3055]

Interesting graphic, looking at the graphic scale (total of almost 2 years) the CH6 popcorn noise seems somehow sporadic.

Popcorn noise can be sporadic on a scale of hours to days.  It took me a month to track it down in a Tektronix made integrated circuit.

Since it is related to processing, I am surprised that any modern analog IC would suffer from it.

The popcorn noise is usually attributed to single quantum states (e.g. electron occupation at some defects or impurities) that have a relatively large impact because they sit a some sensitive point. One would not note most of the defects with only a weak effect, but they can still contribute to 1/f noise.

While impurities can be reduced, the other prime candidate are defects at the oxide interface, that may not be avoidable. With smaller structures there may be less defects, but more of these can have a large effect.

My favored model is that 1/f noise is the accumulation of different levels of popcorn noise and generated by the same mechanisms.

I also think that is most probable explanation...

[Dr. Frank]

Today I received some LM399H, and to my surprise they came without the thermal shield:

I already contacted the vendor to return it.

Is there any reason to not return it, even if the seller offer a discount?

Very strange that the bare can IC is sold, I've never seen that before. The plastic cap  is made of Polysulfone, VALOX^TM of General Electric, because the chip runs at 90°C, which lets most other plastics get brittle after short time. Therefore, 3D parts from random plastics will not work.

My only very old stock part looks like this. I assume from this old logo, that it is also from National semiconductor, bought from a regular distributor in between 1992 - 1995, don't remember any more the exact date.
I opened the plastic cap (broke a small edge off), and the metal can is completely blank, no imprint on the top, nothing on the side.

Maybe yours are fakes or rejected ones for both reasons, coming with no plastic cap (which is difficult to fake) and with an imprint.

Frank

PS: The old logo was changed in the era of CEO Gilbert F. Amelio, who took over in May 1991.

[Kleinstein]

It may be a good idea to check a few before ordering more plastic caps, just in case they are fake.

Chances are the chips are reclaimed / un-soldered parts. In this process the cover may get damaged / removed.  I have seen pictures of old chips without the case, that did work OK.

[wraper]

Today I received some LM399H, and to my surprise they came without the thermal shield:

Where did you get them? There is a chance they are salvaged ICs pulled out of their cases, with new legs welded on and marking made right on the metal. And as they stayed in the case previously, they would look new. Not to say you should check them as they may be duds to begin with. I don't see  any mention of part number without a heat shield in the datasheet.

[rodpp]

Thanks @Dr. Frank, @Kleinstein and @wraper.

After your comments I checked carefully and the chips seems to be salvaged ones.

I didn't open the bag, because there was a chance that I could return it.

The legs are silver, not gold. It is not very clear, but it seems to be welded.

So I assume that these chips are salvages.

Even considering that aging in references is desirable, and the price is 1/3 of my other options, I'll return it.

[guenthert]

Today I received some LM399H, and to my surprise they came without the thermal shield:
[..]
Is there any reason to not return it, even if the seller offer a discount?

    You could use them in a temperature controlled enclosure (oven), w/o the internal heater, like a 329.  Given that you then can chose a considerably lower temperature (say 45°C, instead of 90°C), long term stability might actually improve.  Of course you don't want to pay more than for a 329 then ...
    I'd buy a few.

[Kleinstein]

When using multiple units together, it may not be so bad to not have the cover. A common cover for some 2 or 4 units may be OK. The original cover that has contact to the case, but can still move is also not ideal. So an external non touching cover may even have some advantages and may not need to be as temperature stable if there is not too much external isolation.

Without the heater it would compare to the LM129, so it still has the hermetic case, which is a big advantage over the plastic of the LM329.

New welded legs could be a problem, as there may be additional thermal EMF that may depends on the board temperature.

[rodpp]

Today I returned the LM399 without the thermal shield.

Here are the data from the last three days:



Next step is to perform the suggested modifications in the above posts.

[dietert1]

When you start working on your board, please make two separate ground networks, one for the reference parts of all LM399s and a second one for the heaters. This will improve your setup.
If the heater current of a LM399 changes by 10 mA for example and your common ground network has 0.1 Ohm from that LM399 to the reference point that you use for voltage measurements, you get a voltage error of 1 uV. Such errors may very well add up to 1 ppm or so. Better you have two separate ground networks. Current in the reference ground network will be small and near constant. The heaters don't care for uV or mV voltage differrences on their ground network.

Regards, Dieter

[Svgeesus]

Or better yet, if you have a bipolar supply available: run the heater circuit across the positive and negative supplies and keep it away from ground entirely. There is an example in that datasheet.

Also amplify the output voltage and use that to generate the zener current in a feedback loop, so that it is as stable as possible (there are plenty of examples of that, too).

[rodpp]

Yes, after reading your post I checked the datasheet and I could use that symmetric source circuit. The power supply I'm using is symmetrical.

Regarding the amplifier with feedback to generate the zener current, I'll do that in the future. This experiment is only to select the betters LM399.

Early today I modified the grounds of the board as sugested by @dietert1 and added capacitors at the zener outputs. Let's see how it will behave before changing to the symmetrical source, if necessary.