I purchased an inexpensive two channel thermocouple thermometer on eBay. It was very low cost (A$24 inc post) so there was risk that it was low quality and since cost of returns is so high, that it was wasted money.
But it was junk…
In fact, it did turn out to have a major problem.
The contact springs of the thermocouple sockets were very weak and with a couple of plug insertions, were so loose that contact was unreliable.
The instrument was truly junk, and consigned to the rubbish bin. The seller offered $4 as inducement to rescind the negative feedback, but since the fault makes the instrument unusable, that was inadequate and declined.
A possible fix
On a rainy afternoon, the thing sitting in the bin was a challenge, possibly some entertainment for an hour or two.
The fundamental problem of each contact spring was that there was almost no material at the root of the spring, and they bent at that point. In fact, there was so little material that they would quickly fatigue and break at that point.
So, a solution might be to bind the sides of the contact spring together higher up the contact so as to relieve stress on the root and limit flexing at that point, and to solder the root to give it a little less chance of fatiguing.
Above is the implemented solution. A piece of 0.5mm copper wire wave tied around the springs about one third of the way from the root to still leave plenty of spring above the tie to flex, and clear of the plastic housing. The copper tie was soldered to the spring on both sides. Additionally, a small dob of solder was applied to the root at one side of each spring to reinforce the root. The work was cleaned in solvent to remove flux residue and the instrument reassembled paying attention to location of the uninsulated thermistor leads used for cold junction compensation.
Above, the repaired instrument which seems to retain good contact pressure after repeated insertion of a range of probes with different types of plugs.
The rest of the thing was not bad quality, it is Chip On Board (COB) construction with few components and box standard tactile switches which can be easily cleaned or replaced. The thermocouple sockets were the Achilles heel of the thing and time will tell whether the repair endures.
Accuracy was checked against a UT713 calibrator and was within spec of both instruments. The cold junction sensor is mounted in air near the contact springs and so there is some latency in cold junction compensation tracking.