Damian Menscher wrote:
Looking in the manual, I discovered this has a "protection circuit"
that turns off the oscillator if the -2.2kV supply is overloaded.
This is accomplished using a "multivibrator" -- a matched pair of
transistors, one if which is normally off, the other normally on.
The behavior of blanking the screen and flashing after four seconds
is consistent with proper operation of this protection circuit being
activated. The problem is, I don't know what's activating it.

I've put a scan of the circuit schematic online at:
http://www.uiuc.edu/~menscher/647_schematic.jpg
Their color scheme is:
black for the circuit
blue for "normal" readings of circuit in operation
pencil for my notes as I try to understand it

The circuit description in the service manual gives a brief rundown
of the theory of operation, which I'll paraphrase. I'll use labels
from the schematic, since otherwise it's hard to keep everything
straight.

The CRT high voltage is obtained by using an oscillator to drive a
transformer. The transformer steps up the voltage, which then goes
through tube rectifiers to convert back to DC. The anode of the CRT
is increased further using a voltage tripler circuit.

The oscillator is called an "Armstrong oscillator", where the T820
primary feeds back into the base of the Q820 oscillator. It is
controlled by Q804, which acts as a shunt regulator. If Q804 conducts
heavily, the base current of Q820 will decrease, therefore lowering
the output voltage.

The conduction of Q804 is controlled by an error amplifier, Q803,
which compares the -2.2 kV cathode supply to the +100 V supply. Under
normal operation, there is a 110-microamp current through the R802
resistors, and a change in the cathode voltage will change that
current. The Q803 amplifies that change to control Q804.

Finally, there's the Q814A-Q814B multivibrator, which protects Q820
from overload conditions. It monitors the voltage on the base of
Q804, turning off the HV if that voltage drops too low. The trigger
is diodes D815 and D816 turning on. The circuit is designed to cut
the HV for 4 seconds. Then it restores it for 250 ms, and cuts it
again if the overload problem persists.

I've measured the voltage at the junction of the protection circuit,
and it does indeed drop shortly before the protection circuit kicks
in. So unless it's R810 that's failing, I doubt the problem is in
the protection circuit itself.

Tracing things back, the voltage change also occurs in front of the
error amplification transistor. Which means the problem could be
just about anywhere.

A note from a former Tek employee said "This scope has a failure mode
with the HV transformer diaelectric constant becomes lossy at the ore
frequency." On his advice, I replaced the transformer, but that
obviously didn't help.

Something in the region is getting fairly hot -- even the chassis is
hot to the touch. I haven't been able to identify exactly which part
is overheating, though Q820 is an obvious candidate (it's fairly warm,
but I'm not convinced it's overheating). I'm still attempting to
freeze parts of the circuit to identify the fault, but it hasn't led
me anywhere yet. Part of the problem is many components are hidden
from view in the HV cage, so I can't easily freeze them.

Any advice appreciated.

Damian Menscher
--
-=#| Physics Grad Student & SysAdmin @ U Illinois Urbana-Champaign |#=-
-=#| 488 LLP, 1110 W. Green St, Urbana, IL 61801 Ofc217)333-0038 |#=-
-=#| 4602 Beckman, VMIL/MS, Imaging Technology Group217)244-3074 |#=-
-=#| www.uiuc.edu/~menscher/ Fax217)333-9819 |#=-
-=#| The above opinions are not necessarily those of my employers. |#=-