If the dealer is helping you to cross-examine your HA-300, I couldn't offer any solid advice beyond what they have been doing. Your dealer is always your best bet for repair and warranty service because they can examine the amplifier in person. If you want to seek a second opinion from the Cayin service department, you can drop an email to
service@cayin.cn
My comment therefore focuses on the non-circuitry issues.
2 sets of Sylvania 6SN7W and 1 set of Melz 1578 lasted a couple of days or a week, but PSVane CV181-TII and LinLai E-6NS7 lasted 1.5 years. This is abnormal IMHO. If all 6SN7 tubes last roughly the same time (say from 1 year to 1.5 years) and they are significantly shorter than the norm, then we probably will suspect the amplifier is the problem because the amplifier is the only common factor. However, since we are talking about 1.5 years vs 1 week, the logical deduction implies the quality of your NOS tubes is questionable.
Did you keep the stock 6SN7 vacuum tubes? they are the Shuguang WE6SN7? Maybe you can use the stock 6SN7 to verify this hypothesis. If the stock tube can last 1 month without showing any problems, then we have a stronger case to suspect your NOS tubes are part of the problem.
The other possibilities, again a matter of logical deduction, suggested that the amplifier has developed a certain problem or has been subject to certain changes (I have noticed the Duelund cap in your photo) in due course, and this led to my second comment. If you were to send your HA-300 back to the Cayin factory, the Duelund cap mod is going to be a problem, Maybe you should restore the HA-300 circuitry to the original condition before you do that. Even with that, if our service engineer has identified a trace of circuitry mod, we might refuse to offer repair service or charge you extra for testing and restoring the amplifier to its original condition before we proceed to repair. Unfortunately, under these circumstances, we cannot take the customer's word for granted. We have no choice but to examine the amplifier in detail to make sure we have documented and restored all the changes.
Last but not least, the PSVance and LinLai are not particularly impressive on the longitude tests we performed in-house. I don't have the original HA-300 testing figures around, but from what I noticed, the Tung-Sol 6SN7GTB that we used in the current HA-300MK2 is more reliable in the longitude tests.