Here is picture from -7.6V SMPS.
This is "normal" in ok working PSU.
Powered by 230VAc
Connect probe to D12 kathode (same as 34063A pin 2)
Probe GND to GND
Probe 10x, scope 10x
Memory 100k
Horizontal speed 5us/div
Voltage 5V/div DC
Persistence 1s
Trig Auto, level 0mV, rising edge
There can see around 60kHz square wave with some overshoots.
Frequency and duty is jittering some amount as see in image. (it is normal)
This is continuous waveform without markable changes over time in level or frequancy/duty if circuit works normally. There is NOT any low frequency components what may relate to long periodic high peaks.
It is good to note that these "peaks" in some slow horizontal speed images. (with long random period time) they may be false.)
It was not "without reason" when I have told that it NEED analyze with synchronize method. Not so that trigger channel is same as noise "image" channel. Why: Becouse there is lot of noise peak sources. They are not synch with each others. Some of them use timing jitter for share noise components to more wide spectrum.
Of course some times there is two (or more) peak what in some random time point occurs exactly same time we see sum of these two peaks, if more quite high peaks randomly happend same time we can see "super peak". There happend lot of this kind of interference as we look this "total noise".
For this reason we need go to inside this spread spectrum noise and take lock to each separate noise source, one after one to look.
After then we know what they are and what is real and what are two or more source produced sum. In other way, we see chaos and we can not trust what we see. (if look this total noise and then try use itself to analyze it. We can see lot of mix products and we can not know what is real and what is just some kind of interference "peaks" and then loose time if find in circuit what produce this high peak. (if this peak is this kind of "chost")
One method for analyze it is just go inside this chaos and try lock every separate true noise source (we know these typical sources in this construction) and this locking to separate signals can do with this 2 probe method very very easy and after this we know what are these original real noise "peaks" and stop caring possible random peaks what are perhaps only "chosts" (sum of peaks when they randomly meet each others and produce image of random high peak).
I do not try tell or claim it is just this
in this individual case, but it need keep in mind that it is possible - so that do not try find unreal chosts.
Perhaps there is these high peaks what have long and random period time but - keep in mind also this theory what also is possible.
(last spring I use my time more than two hundred lab hours with this case. I have been totally in "forest" before I regognize that nearly only way analyze it is really separate these noises and stop looking this total noise figure - and how to do it. Very very simple. Use two channel, other for lock and other for look signal.)
And more. This noise channel (not synch channel). There travel now high amount of random noises, but some part there is locked to trig point. Now it may be difficult or some small peak cases nearly impossible look this becouse eye need care only this locked part of noisy trace.
After locked (stable trig) this (trig) channel can turn even off if like. (example, if CH 2 is used for trig, after have adjusted trigger as need, CH2 can turn off if do not want momentarily look this channel blinking there. It still can use for trig if it is OFF))
If it is difficult to look locked signal middle of screen when there run randomly lot of high and fast signal over it.
These others are random, only this one what we want look now is not random related to trig time position.
Oh... why we have average mode. It can help in some cases lot of. It can use for filter out these others and now really can see what part is related to source where we get trig.
In this method, trig channel sensitivity need keep so low and probe need go so close to source what we inspect that it do not pick-up any other signals what may disturb reliable trig and give false results.
Here normal D12 kathode (ref GND)