LM399 carrier/aging board

[AlfBaz]

I haven't been able to find a discussion on mounting an LM399 on a carrier board to avoid creating a new aging cycle when soldering them in after burning them in for preselection.
I've recently purchased 10 LM399AH and wish to start the aging process on them. Some of them are intended for projects which may or may not require amplifying to 10V
I was thinking something along the lines of the attached picture, perhaps slightly modified to enable encasing both sides to prevent air drafts

Is this a feasible idea or do the extra connections create thermal emf's which render the idea useless

[David Hess]

Is this a feasible idea or do the extra connections create thermal emf's which render the idea useless

The additional junctions are a potential source for additional thermal EMF errors however the usual steps can be used to mitigate this problem.  Keep opposing junctions close to each other so their temperatures track, etc.  So do not place the signals which matter, for instance the sense signals if used, at opposite corners of the board.

[AlfBaz]

Thanks for that David.
Its  also occurred to me that the small board would be a lot warmer than the pcb its mounted on thereby  increasing thermal emf at the mounting pins.
I may have to rethink this. I'll build a couple of different ones, measure and compare to ones mounted directly

[Tj138waterboy]

I have been wondering where the sockets are from that some people have used for lm399 and LTZ references they favor tube sockets also wire wrapping the leads could be an option for temporary burn in. I'm going to experiment with wire wrapping myself for a permanent fixture it will avoid Heating the legs of the reference and avoid stress from PCB Flex

[Squiddaddy]

Is the LM399 really that sensitive??? They are installed in tons of equipment over the decades.
HP didn't seem to go over the top installing them in their meters.
I'm no volt nut, so maybe I'm missing the point.

[David Hess]

Is the LM399 really that sensitive??? They are installed in tons of equipment over the decades.
HP didn't seem to go over the top installing them in their meters.
I'm no volt nut, so maybe I'm missing the point.

You would be right about a lessor 20ppm/C reference but inadvertent thermocouples can easily produce errors several times larger than the 0.5ppm/C or 3.5 uV/C performance of the LM399.  If the joints like socket connections are maintained at the same temperature which is most easily done through proximity, then most of these errors can be canceled out.

[Kleinstein]

At least when the leads are left relatively long, it is a good question if the LM399 is really that sensitive.  Thermal EMF could cause some extra voltage, though this to a large part compensates due to symmetry. The other point here is whether the LM399 is that sensitive to soldering:
the LM399 runs rather hot and is know not show very little hysteresis. If the leads are long soldering might not even bring the pins at the case to normal operating temperature.

So there might be the option to use a rather normal socket for the burn in phase and than solder the chip directly to the board, with long enough leads and short solder time. For the LM399 I would not expect the classical reset of aging due to soldering.

If using a small daughter-board I would consider having the critical pins closer together (less thermal EMF) and possibly using separate current / voltage sense pins if plugs are used. So more like 6 pins in total. The connections could be soldered copper wire.

[iMo]

I've been experimenting with 4 various LM399 variants. Inserting them into 4pin socket made of 2.54mm female header. While measuring the stuff with my differential uVoltmeter I can see the 399 (with the white plastic cap) is pretty sensitive to temperature changes and when in the socket also to any even pretty small mechanical stress.
Without an 1cm thick foam cap, inclusive a piece of foam around the legs (actually the legs go through the foam) a slight blow from 30cm creates a dozen of uV drift.
When in the socket a gentle touch from a side or knocking at its top creates a random jump of +/-XXuV.
When I carefully touch the top of the original white plastic cover (no foam) with my finger in order to cool it down (such I do not create a mechanical stress) it drifts immediately maybe 30++ uV off.
All the recommendations our volnuts post here in that regard have to be taken seriously.
For burn-in the sockets are ok.