Ok, first stage of pump testing.
Here is a scope image of the pump hooked directly to a Fortron 350 Watt supply.
First off, about this and future images:
On an oscilloscope, vertical indicates voltage, and horizontal indicates time, so as you look at the scope trace from left to right you can see what the voltage is as time progresses. The "M 400ns" as the bottom of the screen indicates that the time per division is 400 nanoseconds.
The voltage per division shown in the lower left corner is 1/10 the actual voltage per division, because I am using a 10X scope probe, and the scope doesn't know it. In other words, each vertical division indicates 2 volts per division.
The arrow on the left of the image with the "1" in it indicates zero volts. So by counting divisions up from there, you can see the voltage at the pump input. In general the voltage at the pumps input is about 11.6 Volts.
I'm sorry about the poor contrast. I'll see what I can do to get better images in the future.
Now, about that noise and spike...
Yes, the voltage measured at the input to the pump is actually going down to almost -2V!!! This is only occasional, but it happens often enough that I could easily catch it.
Now, the output capacitors of the PSU will filter that somewhat, so devices that are powered other than through the particular drive cable the pump is connected to won't see such huge spikes on the power. However, looking at the 12V rail on a second drive cable, I saw dips as low as 9.5 Volts.
Next adding a capacitor...