AD587LQ

[Andreas]

The test started at 40°C, then cooling to 20°C (15 minutes) and heating to 40°C (15 minutes).

Hello,

with a 15 minute ramp: are you shure that the "hysteresis" is real.
Or is it only the temperature difference between temperature sensor and AD587 internal chip temperature?

with best regards

Andreas

[Terry Bites]

octo say there is stock
https://octopart.com/ad587lq-analog+devices-522211

[miklos2]

"with a 15 minute ramp: are you shure that the "hysteresis" is real."

Hello Andreas!
I'm not sure, but I made the measurements with another three AD587LN and the hysteresis was not consistent. The plastic AD587LN I bought on ebay 5USD/pc. They are used but not soldered, I think, worked on socket. Individual mark on the bottom K3, B8, X9, date code 9947, 9947, 9746.
During the temperature sweep measurements I controlled Peltier voltage by hand so I have to learn nice sweep with AD589LQ (3) (60 minutes) and AD587LN (K3) (52 minutes). The CERDIP and plastic PDIP package have different thermal resistance so it would be the same hysteresis the 3 CERDIP and other but same hysteresis the 3 PDIP, and the branches of the diagram would be parallel the full temperature range ( like AD587LN (B8)).
The devices and the arrangement was the same all the 6 AD587 measurements, but the hysteresis was not consistent so there was no measuring error, the hysteresis real.
I attached the voltage-time diagrams of AD587LN-s and LQ-s on the output (LQ1 was not saved).

[Kleinstein]

The curves come back quite well when back to room temperature. So it does not look liks hysteresis, more like some delayed effect, maybe just the internal temperature different from the thermometer. Different cases can have a different delay and heating controlled by hand may mean a not so constant / repeatable speed.

Hysteresis is normally also more expected for larger excursions, like going up to 100 C or so and than back to RT.