Grumps wrote:
Hi All
A friend has acquired an old Goldstar OS 7020 dual channel CRT based
'scope, and I offered to check it out to see if it is all working.
So, mostly it is fine except that the timebase wanders randomly,
sometimes speeding up, sometimes slowing down. And it does this to the
same degree in ALL timebase settings except the X-Y mode.
I have the circuits, they come with the manual, and have scoped various
points in the sweep generator circuit. When in X-Y mode the sweep gen is
still active and is stable.
There are at two main timing caps for the timebase depending on speed
selected, but I have ruled these two out as the source of the problem as
they're not used in all settings. There is also a little cap which must
be part of the timing cct, but even if that was duff, why should it have
a huge effect on the varying speed when in a slower setting as its
capacitance would be dwarfed by the main timing cap.

So, I'm a little stuck. It is an easy unit to probe as it uses all
conventional through-hole components and the sweep gen board is on the
top when you remove the cover.

If anyone can give further clues that'd be great. Or if there are more
relevant forums available to post to.
Ta very much.

OK, so how do you make a sweep ? What principle is involved ?

One way to do it, is take a capacitor and run a constant current source
into it. The voltage potential across the cap, gives you a constantly-rising
waveform. Then, when it hits a threshold voltage, you short out the cap
(with a bit of resistance so the current flow level isn't ruinous) and
that falling voltage interval (with beam gated off) is the retrace. In
my sample picture below, the falling edge should be steep, but not
exactly vertical. It takes some time for retrace to happen.

The result is a sawtooth waveform, with a slowly rising front half, and
a quickly falling back half.

/|
/ |
_____/ |_____ (Apply voltage to horizontal deflection)

__
| |
_____| |_____ Beam gate (turn off beam during retrace)

A varying current flow, instead of a constant current flow, on
the leading edge, makes the sweep rate non-linear. Speeding up
or slowing down in proportion to the deviation from ideality.

If you have a schematic, look for the sweep circuit and see
if you can fit some notion of that to the components present.

Unless there is conductive dirt near the capacitor, I'd be
most interested in the health of the tube providing a constant
current.

In the digital era, you could take a binary counter, and run it
into a DAC, to make a ramp. But of course nobody does that,
because the clock rate needed into the binary counter
would be pretty high.

They also made waveform generator chips at one time, which
is just the embodiment of various waveform generation tricks,
into a single chip. The one I got came from TI, but I believe
it was soon canceled.

Paul