Amanda -
Hadn't thought about using the input voltage as the feedback. Problem with that is, even though it will improve efficiency for the panels, I want to leave the 'reserve' available for other potential uses (ie directly plugging in a cell charger rather than pulling from the battery, etc).
If I use the input as feedback, I'd have to use a dump controller on the battery, which would soak up that reserve coming from the panels. I suppose I could modify a controller to make it disconnect rather than shunt; may have to look into that as a possibility. Right now, I'm regulating via the output, set to 14.4, and when the battery is full, the excess power just doesn't get used, so it's available for other things. Hence the 'pseudo-MPPT'...
Another hack I'm looking at is the chip has built in current limiting, and so I may use that as the limiting factor on the input instead of the voltage. I've bought a few of the 'base units' to play with different configurations and what not, and you guys will be the first to know what I find.
As far as getting it to play nice with a real panel, I think the secret is (as you eluded to) heavy capacitance using multiple small value caps (say, 5-10 for the range I'm using @ 470uF/50V, with bypass caps on each one, say 0.1uF) at the input as close as possible to the MOSFET/Toroid. I've noticed heat issues with caps too if only a couple are used. I think it's all an internal impedance thing. Using multiple smaller caps (with bypass on each one) gives better results.
To trick it out on the bench, use a length of nichrome that will limit the current, making the bench supply mimic the Ip-max (in my case 1.8A) that the panels are rated for, with the bench supply sitting as close as possible to the Vp-max rating (ie 18V). Seemed to work for my 'mini' version rather well, solar output was relatively identical to bench output. Haven't done it with the hi-power version yet. And like you, my two day clear sky window concluded with moderate water coverage, so I won't know how close all this is until after the water blows away.
Also, this system is MUCH more efficient with the panels tied in series rather than using them in their 'native' nominal environment, so the closer you can get to the fire/smoke release valve threshold for the components, the better. In my case, it's 40V. The OTV on my panels is 39 (thanks to having to bypass 2 bad cells in one of the panels).
Which brings me to Ghurd -
As mentioned above, yes the chip is Vmax'd at 40V, but the panels have been partly castrated, so they are just under that limit. The only component currently in the converter is a small electrolytic rated at 35V. Patiently waiting for it to make boom boom.
34063A.pdf (118.51 kB - downloaded 519 times.)
34063A Datasheet (lite version)
The basic brain is yes, a 12V -> 5V cell phone charger (actually GPS, but whatever, output was a USB port), based on the 34063A. It doesn't get any more simplistic than this thing. The datasheet checked out at 40, so I didn't hesitate to throw it on the panels. The only mod I made to the OEM design for initial testing was to replace the two precision feedback divider resistors with a couple of pots, one for coarse, one for fine. Worked well. As stated however, the weather here is not cooperating for the real world test on the beefed up version, so I'm limited to the 'lab'.
I ended up just ditching the LED altogether, input current now is (get this) a whopping 1mA (@33V) when idle. I'm good with that heheh
I may end up putting one back in, just for giggles, but during testing, it isn't really needed anyway. The meters do a more than adequate job of telling me if it's providing output... LOL
My only gripe with the design I'm using so far is that I used the other 'counterfeit' 44N to drive the buck, and that's fine and all - works like a champ, but I had to run it in a drain-follower configuration. Result there of course is, I'm losing a couple volts at the input of the toroid from the gate drop.
Not a big deal at the moment, but I may look into a P-Channel instead and redo the connections on the chip (has open emitter
and collector for output, so I would just tie the emitter to ground and pull the P-Channel's gate to ground in a common-drain configuration. The design for the OEM unit doesn't use an external driver tranny at all, just uses the internal output stage (NPN) to drive the tiny toroid in a emitter-follower config.
This is pretty much the configuration of the current version in the pics above. Primary difference is that the external NPN is replaced with an IRFZ44N. Also, I am currently not utilizing the current sense resistor. Collector (source) in mine goes straight to Vcc. The resistor is still there, but is only 'limiting' the current that the chip draws to drive the MOSFET (almost nothing).
Opera -
Not 100% sure I'm understanding exactly what I'm seeing in the schematic, although it appears you're essentially doing what Amanda was referring to with using the panel voltage as the feedback. Your circuit is based on the same chip I'm using... got a schematic for the whole thing, not just the mods?
Steve