See:
Dead CV203BU headstage gives crazy temperature readings. The amp appears t o work, but peltier cooling failed.
The problem turned out to be a shorted thermistor. On the DB25 connector, find it at pins 7,18 (resistance reading is below 40ohms, while a good head stage reads ~2K for room temperature.) The peltier cooler is actually OK, just the temp-control feedback has failed.
The thermistor is inside the front end module (gold shieldbox.) It's possi ble to get access by carefully slicing off the cover the box (glued with so me sort of brittle epoxy.) WARNING: EXPOSED SILICON AND DELICATE GOLD LEA DBOND WIRES INSIDE!
A close match for their thermistor is 2.2K NTC NCP18XW222E03RB, digi-key 49
0-12103-1-ND. We used the smt 0603 version.The front-end circuit is three discrete transistor dies on a tiny ceramic/g old PCB, plus one larger thermistor, all connected to the DIP pins with *br ittle* gold-plated wires.
To repair: under the Mantis inspection scope, I removed the gold leadbonds for the dead thermistor, soldered our smt thermistor to the top of the dead one, then used some #35 bare copper to connect it to the original pins. ( Find bare #35 wire inside fine-stranded test probe wire.) A large solder-b lob gives low thermal resistance, to avoid having the new thicker thermisto r leads raise its temperature.
One problem: the clearance between the ceramic plate and the gold cover is very small, and doesn't have room for a second thermistor standing upright atop the dead one. I soldered our 0603 thermistor so it sticks out from th e side of the solder-blob atop the dead thermistor. This gave enough clear ance.
Condensation from ambient humid air is an obvious issue. We don't have a v acuum-oven for baking out and re-activating the dessicant pellet inside the gold shieldbox. We'll try removing the front end shieldbox and giving it a 100C bakeout for a day. Perhaps flush with nitrogen while sealing the go ld cover with epoxy.