If you connect a large electrolytic capacitor from the power supply output that your drill will use, to the power supply's ground (note the correct polarity for an electrolytic, and use one with a voltage rating that's higher than your supply voltage), the cap will already be charged when you turn on the drill. The capacitor's stored charge will help supply any current surge that the drill demands from the supply. If the capacitor was large-enough, and could respond fast-enough, the supply might not see the surge demand at all.
If you have trouble getting it to work, you might want to try using several large-ish capacitors in parallel, instead of just one. In this type of testing, I don't think some inital amount of overkill could be a bad thing (unless you get to the point where the capacitors' demands during charging cause the power supply to shut down, or be damaged). So, if you can scrounge some quite-large capacitors from the power supply of a dead monitor, or computer, or whatever, or otherwise acquire them, you could try using some tens of thousands of microfarads (uF), to start with. If you do get it to work, you can then figure out what minimum capacitance seems to be needed for reliable operation, and add some margin to that to try to make life easier for the power supply.
Also, which PSU output are you using? The 15V? I think that some computer (and some SMPS types, in general) need to have some minimum load on them, all the time, to operate correctly. It MIGHT even be the case that your drill-startup problem is more-related to THAT, rather than to trying to draw too large of a current spike.
I'm not sure that that is probable-enough to even be worth mentioning, but, in case the capacitor _doesn't_ solve your problem, then maybe you would want to also try the following.
You might be able to tell what minimum load is needed from the specs on the label, or look it up on the web, somewhere, for your make and model. At any rate, you might need to put a small power resistor on the 5V supply (or 3.3V if you have that), since that's usually the "master" one, which needs to always have a load, if your supply is one of those that does need that.
I have no idea what amount of current you might need to draw, in that case. It might be a couple percent of the 5V supply's rated current. Maybe try a resistor that would pull about 1 Amp, for a first try, since you could then use a common cement power resistor of about 5 Ohms, rated for say 10 Watts or 15 Watts (since it will dissipate 5 Watts, if it's 5 Ohms and draws 1 Amp with 5V across it, and a 5 Watt one would get very hot, then). If you get two or three of the 15 Watt ones, and it doesn't help with just one of them, you could try two in parallel, or three in parallel, to draw more current. If one works, though, then you could try using more ohms, to get less current, and less heat and waste. So then you could try putting them in series.
The currents and resistance I gave were just a totally-wild guess (hopefully on the overkill side, to raise the probability of a correct diagnosis on the first try). So if some minimum load does turn out to be required, you'd want to find the largest resistance that gave reliable operation, and would probably want to end up with one properly-sized resistor with an appropriate power rating etc, and maybe a heatsink (like the case of the PSU, possibly, if you can insulate the leads well-enough).
Good luck.
- Tom Gootee
**broken link removed**
-