SEPIC doesn’t require coupling between the coils, although you can often reduce the space taken up by the coils if they are coupled (which increases their effective inductance) and thereby build a smaller conversion module.
See here:
and scroll down to “Using a Transformer”.
As usual with inexpensive Chinese modules, the simplest way to find out is to buy one and see what it does
My modules were shipped already. It is a good thing that Correos (main Spanish postal operator by far) reached an agreement with Sinotrans Limited a few years ago, I can choose Ali Saver or Ali Standard and the majority of times this means Sinotrans - Correos which comes home quite fast.
I'll be able to take a quick peek at the modules next week. ;-)
So it matches the title well, it is based on an LM2577-ADJ integrated circuit. The board layout is a little bit different, namely at the switcher's leads for obvious reasons. Whether it is buck boost or not it a mistery is to me until I try it with my PSU, but it should.
Underside pictures. Notice the current sense track, right at the voltage trimpot's feet.
I've quick tested one. Fed with 7.24V from my PSU I've been able to adjust the output all the way up to 26.9V, and all the way down to 1.237V according to my Bside/Zotek ZT102 multimeter. Voltage adjustment is sort of logarithmic, took a lot of turns down from 3V. Hence, whoever devised this module knows something. ;-)
The unmarked chip next to the 78L05 regulator must be an LF351 or LM351 amplifier for the current sense thing. Compared to other modules this thing is using a quite low resistance current sense track, like 10mΩ or so.
That one has end current percentage turn lamp setting and two switcher ICs: LM2577S + LM2596S. Besides this, the above board is marked as XW036FR4, whereas the one discussed here is marked as XW036NFR4 (check it in my #28 reply).
3rd trimpot is usually for end of charge (low current) detection adjustment
That last kind of converter theorically has a lower efficiency as it's made with a boost converter followed by a buck converter so you have to account for the two stages losses
The turn signal is meant to indicate at which point the current output is lower than the given ratio, for battery charging duties namely. It is set fixed as 1/10th in the OP module.
I have an interesting question, because as I see this module it looks like it is parallelizing friendly. First of all the diode at the output makes it reverse current inmune. Next question is, is it reasonably possible to use the same control trimpots/potentiometers for two or more modules? The idea is to use common voltage and current trimpots/potentiometers, this way the modules should share load with great balance…
Nah, it’s a joke. Like the funny kind, only different.
In car manuals, they’re called “turn lamps”, not “turn signals”, just like “CHMSL” vs “3rd brake light”. So when the spex for this donk kept referring to a “turn lamp”, well…
That doesn’t usually work, as the converters will have very slightly different feedback voltages.
For example, if you put two in parallel, one might go to standby at 0.60V of feedback and the other might go to standby at 0.61V. What happens then is that one will stop converting before the other, so that the other ends up bearing all the load and either frying or shutting down. At that point, the first one has to bear all the load and either fries or shuts down as well.
You might get away with it if you use trimpots on each module to adjust for the different feedback voltages, in addition to your main control potentiometer, and add load-balancing power resistors in series with your module outputs, but that gets very fiddly, especially if you’re doing it for more than one control.
You can also run into problems with things like thermal effects under load making the feedback voltages drift at different rates, so you end up back where you started with one module trying to bear the entire load again.
Easier if you just buy a more powerful converter module in the first place.
The problem is that when i get it i like your kind of humor but sometimes, even if i understand the words, i miss the cultural reference to understand the pun
@maukka : i didn't know that one TY (i hope that when doing that they allow for enough hystheresis to avoid on/off/on/... cycle)
Well, I understand the reasons to prefer a single, more powerful module versus 2 or more in parallel. Anyway, and in this particular case and partly thanks to the anti-reverse output SEPIC topology, 2 or more modules can then be parallelled by properly tuning the output voltage trimpot in each of 'em to match as close as possible, then swapping their current trimpots with a dual, triple, quad, etc. potentiometer. The only concern I see is that, in my experience with cheap logarithmic dual pots, their gangs aren't finely matched, they're like 5% tolerance. This should be nothing major, though, with regards to load balancing. Or does it?
OK, now you’re into a zone where I haven’t actually tried the arrangement you’re suggesting
If you can get the module voltages to stay in alignment (the hard part), then current control deviations from multi-gang potentiometer tolerances are less likely to be a problem. The trick there will probably be to keep the maximum current setting per module at the module maximum minus the largest deviation that the potentiometer could introduce. That way, you shouldn’t overload any one module.
Although those deviations will prevent the modules sharing the loads equally when current limiting kicks in, they should settle in reasonable states. For example, let’s say you have 2 modules, each nominally bearing 50% of the load, and ±10% deviation from the multi-gang potentiometer. You could then have one module bearing 45% and the other bearing 55% in the worst case, which may eventually shorten the life of the module that runs hotter, but probably won’t cause major issues.
Once again, this is not something I’ve actually tried, so don’t rely too heavily on my thoughts here.
Having said that, you might also be able to get the current limiting to catch any overloads on modules whose voltages become misaligned, but that would add still another tolerance to factor in, which means that you’d have to set the overall current limit lower, costing you a bit more still of your total output capacity.
I’ll be interested to see how you get on with this experiment