It may be very low with normal input voltage (8.4V), causing large sporadic current spikes as the feedback loop tries to correct its output voltage.
Did you read this my last message with image. (there is now my bad english also so that I can not very clearly tell what I mean. It need also keep in mind possibility that these special high random peaks are not real. Becouse this is possible in theory and also seen in practice it need take one possibility also in this case as long as know what it really is.
If look total noise trace where is sum of noises what travel between different GND points and partially changed differential using probe as used when gonnected to GND. What we see. We see sum of separate sources generated signals what we see as "noise".
Now, if think situation we have only one source there, example generating fast "peak". Lets simplify more. If there is example 10ns constant level peaks say example 10us period and 100mV high. Ok, we see these pulses. We can now use low horizontal speeds and get just nice fat trace even if we turn peak mode on. Just nice band without any peaks what are peaking over this noise band.
But what happend if now add other pulses (peaks) say example other 10ns wide 100mV peaks and timing so that they arrive allways example 50ns different time (syncronized so that always this timing - syncronized signals). We see now double amount of nice pulses dependent of horizontal speed. Now we see two pulses. Then we change horizontal speed to very slow and perhaps also turn peak mode on. Agen we see just nice fat band and level is same as before and no any special peaks.
What happend if we now connect off these signals syncronization and they are nearly as before but some amount different freq. Mostly they are different position and we still see this nice fat band without separate peaks. But...wait a moment.. oops, now I just see one or two short 200mV peaks. If frequencies are very very close and they only overlapp each others more or less we see our "noise band" vary between 100mV and 200mV. If they only randomly or hit each others we see just randomly peaks between 100mV and 200mV level.
Of course in practice this is much much more complex how these separate sources "noises" interfere and this is what we see - sum of these "noise" signals.
There IS this kind of short high peaks and these noise peaks are more or less short but there is many sources and more fun there is also variable frequency SMPS circuits and many of these noise "peaaks" are short "oscillations". SDS7102 is some amount too slow scope for analyzing this noise but with right methods it is well enough for regognize sources and analyze if modification give better or worse result (and least this need method where can detect individual source made "noise")
If do not use right methods for analyzing we are perhaps just hunting "ghosts" (or not, but we do not
know it)
But more fun come if take this same kind of simplified case but, now, there is agen these two signals but one level is 100mV and other level is 50mV. Then we see 100mV noise band with low horizontal speed... but now if we look more deep we find there 50mV peaks and 100mV peaks and if they run with different frequency... there appear third peak 150mV. But now if we know only what we see there we see 50mV, 100mV and some random 150mV peaks.
Lets start find curcuit what produce just these 150mV peaks becouse we think this is more important becouse this is most high and very odd random peaking just as some "failure". We do not find never circuit what produce this highest and random peaks becouse no one do it.
It is "interference." (in this example).
It disappear just if we stop these 100mV or 50mV peaks (or both) and it go more low if we reduce other of these lower peaks. Also it disappear if frequencies go so that they do not appear same time or they appear so that 99.9% of random peaks fall to scope dead time. Perhaps we need collect 1 day continuous data we can hit this peak if it is rare random. No matter if trigger is set if it is rare and we have auto trig on becouse auto trig generate trig mostly before this rare peak occur in time what is ok for trig to it.
So, before can be sure, there need keep eyes and mind open and avoid hunting "ghosts".
I have looked many times this gif animation and there may be "thousend" reason for it, but one is this possible interference (including also scope itself features with dead and displayed time and trigger)
But, also there may be failure in 7.6V SMPS circuit what produce some odd things in some cases and dependent of this circuit input voltage.
So, this circuit is good to look how it works. Battery lowest is near 6V before battery module shut off its output and highest is around 8.5V and it need work normally in this range.
In normally working SDS with primary PSU voltage around 8.4V this -7.6V have near 50% duty and around 60kHz (center freg).