Re: OPEN SOURCE: New Boost Drive, 2.4A capable from one cell, tested and working
Guys, I really think that the capacitors have to be before the zxct or the output will be horrible!
I may be wrong but the way I see it you are asking for trouble with it set up like that. Think about it like this. Coming right off the coil and schottky the voltage is going to be very choppy, lots of spikes, not flat at all. This is inherent in the design of a boost converter and will always happen. The capacitors smooth this out to get a good DC output. So the current flowing between the coil/schottky and the capacitors is going to be very erratic, have spikes, etc. Each high spike in voltage will cause a high current to flow into the caps, each low voltage spike causes a low current flow. If the caps have done there job to "suck in" any high spikes from the output and "squeeze out" extra current to the output for the low spikes the the output will be fairly smooth. How smooth is relative to the size of your caps.
The current sense resistor shunts off a small voltage to the zxct which is proportional to the current flowing through it. The zxct scales it to something the converter IC can make sense of and that signal is used to determine if the IC wants to make more or less current flow in the system.
So the question is: Do you want your current sensing point to be in a place where there are naturally many huge spikes and ripple and a data sample could be extremely far from the average OR in a place where the flow is smooth, no spikes, and any sample should adequately represent the average?
When I think it through like this it is obvious that you would need your current sensing resistor to be the last component in the path to the diode so that you get the most accurate representation of the current it is actually receiving. If it was before the caps the IC would be trying to raise and lower current constantly for every spike that comes off the inductor even though the flow after the caps would be totally smooth. It would constantly be overcompensating for the natural spikes and that is how you get oscillations in the output.
This is just my two cents and I think everything should be tested, but I have made plenty of drivers and I always put the caps before the sense resistor for the reasons above.
ARG, how about a closeup of the output ripple after its already started up? This is just as important as the startup IMO. At this current level I would hope for less than 50-75ma high spikes on the ripple.
Oh, also could you do the scope tests for current instead of load voltage? Since current is what we are interested in here.