The following content was extracted from an email received from Terry Walker with comments about the construction of the K2-YJ. Thank you Terry.
From: Terry Walker
To: Joe Sousa
Date: Sat, 27 Nov 2010 12:13:46 -0800
Thanks for the excellent information on the zener and internal construction of the K2-YJ. That is extremely helpful. I looked up the data sheet for the 1N985, and it shows a rated breakdown voltage of 100V +/- 5% at 1.3 mA for the B version. At lower current the voltage will be less by several volts because of the zener impedance. In your photo, the zener does not seem to have an A or B tolerance mark, so I have a suspicion that Philbrick selected a resistor + zener pair which would give the desired voltage drop (your measured value of 115V). This accounts for the pair being shown as a Philbrick part number on the diagram. This strategy was no doubt much cheaper than purchasing accurate zeners. Philbrick was already known for binning parts for other products, so all they would have to do is measure the zener diode drop at a chosen test current and put it in the right bin, to later be associated with a certain standard resistor value.
I pondered the internal circuits for a while looking for the selected resistor without luck, finally noticing that in the edge-on views, the module is seen to have two internal circuit boards, and further that the boards have parts on both sides. If you were inclined, it would be useful for complete documentation if you could remove the two screws at the ends of the circuit boards and fold one upwards, allowing a photo to show the parts and wiring connections on the back sides of both boards. Looking at the nice photos, I suggest that you remove the corner post labelled 'V1 5751' and unsolder the brown wire at the bottom right to make this easier. The colorful wiring makes it easy to trace out the actual schematic from the photos.
On looking at the module, several interesting things appear:
1) The input pair socket looks like bakelite or micalex (cheaper) vs the output socket is ceramic (more expensive, but heat resistant). This looks like a deliberate choice.
2) Several of the 1% resistors appear to have date codes on them which indicate that the module was made sometime after the 34th week of 1961.
3) Just from appearance (transparency) and the construction requirements, it appears that the wiring is probably all teflon, to avoid problems with insulation damage during soldering.
4) The tube pins and board terminals are all very clean, so the unit was probably washed with flux remover after assembly. This would help control parasitic leakages also.
5) The socket pins are all silver plated, not tin as in commercial construction, for reliability. This shows from looking at the tarnish patterns in the top view.
Wishing you a happy thanksgiving and Christmas season,