After some Metrix tests the thing is not outputting anymore if loaded, somewhere is a contact problem.
Gently poking different parts is difficult, fingers are burning, must find a small wooden stick, maybe I go to the woods with a knife.
I think I'll leave the hybrid desoldered for the next guy.
+/-V filter caps (5+5) are also off and nuts and bolt will secure the hybrid.
813 816 ely 100uF/50V
821 822 tan 2.2uF
807 810 pol 1uF
806 809 cer 100nF
818 819 cer 1nF
Pictured two left hybrids are correctly positioned, pin 1 is up left.
Low left of the board is RF out, next relay is its enabler, other relay is for internal terminator, its parallel resistors beside it.
Top left diode is +V reverse polarity protection, resistor next down is 10k to feedback.
Pin 13 from right hybrid is out, so two inputs to the golden one, golden pins to nowhere are +/-15V supply (2+2) and GND (5x), right hybrid pins 1 and 11 are also GND.
R818 between golden input and output is stamped under it on the other side, took a moment to realize, first I thought it's a capacitor, but it's two resistors stacked, around 3k5.
Other green of top left of right hybrid is insulation.
Gently poking the right hybrid almost anywhere with a wooden cocktail pick created severe output disturbances.
Bushing socket under it is not a socket but two separated bars and plenty of space.
I'll leave the reseating to the next guy.
For overloading,
first I thought it's a simple thing since the output was completely negative, now after some poking the output is completely positive.
Selected output is +/-2.5V all the time, so overheating is actually a feed driver trying to meet its deeds.
Slow slope problem seems to be also disappeared, very funny, maybe it's because filter caps are off, earlier mentioned BE is still shorted.
The machine is also sending sync pulses and receives external trigger, and had an intact cal seal, so pretty original, or something.
--[whitespace]
Then something completely different,
how to evolve with this "my prospects to our gold" system.
Now it's quite limited, first the target must be interested, then part that is leaving behind must be valued correctly and then what if the result is not what it was meant to be.
Maybe next level could be something where the whole bad batch is the original pool.
Then possible resource base could be wider, but it probably would also be more time consuming all the way.
For Ice-Tea's situation, actually for many others also, I'm practically behind the sea.
Closer resource would probably have more options.
Then the semimandatory BTW,
old rule of bad subcontractor, what if the new one is even worse.
So even if you are a bad subcontractor, don't give up, you don't know how bad the other guy is.
For face to face situations, stay at least a broom length away, just in case the other guy is actually pretty good.
Other BTW,
my 465 lost its A channel, dang, only one channel is actually quite limiting.
It's also a bit hard to measure if you forget to swap and the other end of the probe is still resting at trigger input.
I think I'll check if I can remember how the 2215A is broken, it's at least sort of a single board thing.
Maybe I'll finally get that real DSO.