Just as an aside, Thomson ( RCA, GE, Proscan , and many other rebrands I'm sure ) had a similar problems with a line of chassis #CTC175-187. It was all due to poor soldering around the micro / eprom and the shield cage. A whole myriad of weird problems arose. If the problem was left go for a while it only got worse, to the point where the EEPROM could be damaged. Then the set could do it self expensive damage or be completely inoperable, since the EEPROM held values for proper operation of power supplies, scan rates, etc.
If caught when it first happened, simply resoldering the bad joints ( Thomson actually issued a special solder and flux kit ) was all that was needed. If it progressed, you needed to replace the EEPROM with one that was preprogramed to a level that would allow the set to turn on, then you had to recalibrate everything, full factory alignment. The fun was that you only got one shot at it, hoping the correct values got written to the EEPROM, or you would be back at square one, replacing the EEPROM again.
Lesson? - get it fixed before it gets worse, and possibly more expensive. Just turning the set on, even if it appears to work sometimes, could be a bad idea.