Chinese BC108B Transistors (TO-18): non-existent QC in play...

As part of my 'RCT Revamp'. I've been slowly accumulating the bits needed to build two more Remote Switches, to complement the single unit that I had originally built.  But that's for another post, this is about the BC108B transistors needed that I sourced & bought through Aliexpress - and yes, I'm back buying stuff from this lot again!

After getting my PCB's from JLCPCB - so much easier than messing with ferric chloride, like I did originally! - I soon had one (1 of 5) populated & ready to rock.  Except it didn't work!  Checking the 6 BC108B transistors with a multimeter, I discovered that one was completely open - all B-C-E connections, which seemed decidedly odd.  All the others tested fine.  Swapping out the dud with a new one didn't solve the problem.  I then went off on the wrong tack and ended up destroying two UM3750's, the decoder chip used, which apparently are really finicky about the voltage level applied to its input.

Long story short, with my one working (old) unit to compare against, it soon became apparent that there was something amiss with the transistors I was using.  Starting afresh with a new PCB (2 of 5), and taking measurements as I built, I soon realised that the transistors were not 'amplifying'  the superimposed 200Khz signal from the mains.  Removing all four that I had soldered in, and checking their gain with my trusty old meter, I was shocked to discover that their gain (hFE) ranged between 0 to 107, far from the 200 minimum that these transistors are rated at.  The last two unused transistors were also checked, these came in at 84 and 102 respectively (see pics).

This got me to wondering about the 100+ other Chinese transistors that I've got, which are all of the TO-92 variety.  These as it happens, seem to be perfect, each batch of the same transistors measuring respectable hFE's, and within shouting distance of their compatriots.

So it seems that only the TO-18 type BC108B's that I've bought are crap.  And they were not particularly cheap either!  The reason I ended up ordering them from Aliexpress was that they are considerably more expensive when ordered from Ebay sellers.  But what I got is garbage, which is compounded by the fact that I ordered another 20 of them (from a different Aliexpress seller) [after checking, it's actually an Ebay Seller] before I discovered the hFE problem.

It would appear incredibly unlikely that the 'ST' branded BC108B's  that I got are actually manufactured by STMicroelectronics, instead being produced by some knock-off outfit that employs zero Quality Control testing.

I've being playing with transistors for 40+ years and admit that I had never before come across a transistor with zero gain.  Or 32 for that matter.  Out of the 15 transistors I ordered, only 2 came close to the expected 200 gain (213 & 187).  These two can be seen soldered in-situ and that portion of the circuit (now) works as expected.
 
Edit:
I removed all six BC108B transistors from the first PCB (1 of 5) today, to check their gain.  As expected, the highest gain measured was 102, with one coming in at a mere 10 - so, it's understandable that this RCT build wasn't working!  I figured though that the one 102-gain transistor would work ok in what I deemed a 'non-critical' part of the circuit - wrong!  Removing it again and replacing it with a 187-gain TO-92 transistor, had the circuit working perfectly.

Now all I need to do is triple-check the changes I have made to the PCB layout and re-submit them to JLCPCB.  Although the original batch (of 5) work fine, given how cheap they were, I didn't bother adding mounting-holes first time round, as I was sure that there would be other issues that would be discovered during the build-process - there were, but mainly minor things like component placement & hole-sizes, that need altering.  I am also taking the opportunity to add a single component of my own to the circuit, a 5.1v zener diode, to squish a nasty transient spike, on one of the 'SET' pins of the 74LS74, which is generated during switching, and only discovered today.  Even though it spikes to almost 9v, so well above the 5v operating voltage, the original RCT unit's 74LS74, suffering from the same condition, [maybe not... after checking, it spikes, but less than a volt, odd...] is still working after nearly 25 years, so yeah, a non-critical addition.  But on the off-chance that it might cause occasional false-triggering, I may as well fix the problem.  

Edit1:

Update.  Just received my new order of 20 BC108B's.  Testing, first four out of the bag had me questioning my new-found beliefs about Chinese QC - all 4 had an hFE >= 200 - unfortunately, this run of luck was not to continue.  All told, 10 proved to have gains greater than 200, 4 were less than 170, 3 less than 100 and another 3 less than 50.  A crap-shoot basically!  Even the hFE > 200 varied wildly, with gains ranging between 200 and 700, although admittedly, most were around 200.  So I guess it's up to the buyer of cheap tat like this, to first test, then keep the best & bin the rest.

Luckily, I've now got enough BC108B's of the TO-18 variety to populate the two planned RCT PCB's.  All that's preventing me from continuing is needing to reorder PCB's from JLCPCB, receive the (right!) transformers, and... laziness.