Here is the clearest path to GND. Does it look sufficient to you?
For the BATTERY GND, it seems even wider.
Hard to tell, that's why I don't rely on pours for ground.
What about every other power and gound?
I’m not an expert at all, but from my point of view the return paths look good: the GND plane is present almost everywhere on the bottom layer (B.Cu). I also read in the LT review that the biggest concern is thermal management rather than RF coupling, since the switching frequency is relatively low at 1 MHz
That's not right, should be LOW when charging and go HIGH when done.
According to the datasheet the CHRG LED should blink when the battery is removed.
That's not right, should be LOW when charging and go HIGH when done.
CHRG is low so high side connected to Vin and the led goes ON, It seems CHARGING IS ABOUT C/100 (in my case about 100mA) but in the Solar side i measure 25mA (setting efficiency = 1 best case) i means about 70mA in the output.
But the resistor is R100 1% = 0.1ohm.
And It blinks when I remove the battery!
So the ground must be connected or it would not blink but CHRG is backwards.
Switching regulators can be very sensitive to layout. I would look at the eval board design files and try to make your layout similar.
managed to track down the original Sunny Buddy Eagle layout files from the SparkFun Sunny Buddy product page. I find them much more readable than the official evaluation board PCB, so I would like to use them as the main reference. Before I respin my board, could you help me compare my KiCad design against the Sunny Buddy layout? I am particularly concerned about the maximum charge current limit and the maximum cell charge voltage settings — I want to make sure I have not missed anything there.
I can package up my KiCad files if that would make the comparison easier.
I don't understand. This statement indicates that the battery has been charged to "a slightly lower voltage" than 4.2V:
Charging current tapers and CHRG goes low when the battery reaches ~4.04–4.08 V
In any case, it is completely unrealistic to think about charging a pair of 18650 cells from a wimpy solar panel. Use one that can provide several hundred mA or more at 9V or higher.
I can tell from your layout that you are quite new at it. I would just layout your board as close as possible to the SunnyBuddy layout.
The maximum charge current and voltage are just set by resistor values as explained in the datasheet.
I m using a 20W panel 17V.
LT3652 charges at 25mA even with DC input
Then it should be obvious that your design, or build, is wrong. Post the complete schematic rather than useless pictures of the PCB layout.
I just looked at the SunnyBuddy layout, the bottom copper pour isn't even connected to anything but I assume that the board works.
I thought It miss a vias to Gnd.
The schematic is attached to my first post. Why do you say it's completely wrong?
Well that's the problem with using designs you find on the web, you never know if they are correct or not. Even many of the Arduino schematics contain errors.
All I can say is follow the layout directions in the datasheet.
I thought I had already followed the datasheet recommendations quite carefully, and I believe I read it thoroughly. Below is my personal checklist, which I tried to respect as much as possible first in the schematic and then in the PCB layout.
1.Thermal pad / DFN-12
I soldered the exposed pad (Pin 13) to the PCB ground plane and used it as the main thermal path
I believe this point is respected
2.VIN decoupling
I placed a proper low ESR input capacitor as close as possible to VIN and tried to keep the input ripple under 0.1 V
This is the point I am least sure about, because it is not easy to measure directly, even though with the bench supply and the datasheet limits the ripple should be well below the required delta V
3. BAT output decoupling
I placed a 10 uF ceramic capacitor from BAT to GND, and I kept the sense resistor between SENSE and BAT short and close to the IC
This point is respected
4.BOOST capacitor
I placed a 1 uF or larger capacitor between BOOST and SW, very close to the pins
This point is respected
5.Feedback node
I kept VFB as short and clean as possible, tried to minimize parasitic capacitance, and used the proper divider values
I think this point is likely respected as well
Assembly and cleanliness
I cleaned the flux residue carefully after assembly and used a low leakage / no clean process
I believe this was also respected
In short, I am confident about the thermal pad, BAT decoupling, and BOOST capacitor placement, and I think the feedback node is probably okay too
My only real doubt is the VIN decoupling
In general D5 should not be a schottky diode because they have to much leakage, however the ZLLs500 used in the sunnybuddy has a very low leakage of just 10uA
The working voltage of the SMF15CA is only 15V so your input voltage must be less than that otherwise it will start to draw current. I would check it or remove it.
I’m sorry, but the PCB shown in the previous image is this one, the second version.
The core is the same as the one I reported in my first message, and the issues are the same too: low Icharge and a low maximum LiPo voltage, around 4 V, while Vrf is 3.3 V when fully charged.
In this second version, I added a Burst Mode M1 option, which is a variant described in the LT3652 datasheet, along with a few other secondary features.
The SMF15CA was replaced with a larger 22 V part as well as I upddated the power source (solar from a 12V solar pannel to a 18V 20W).
I chose the SS14 because its reverse leakage current is about uA , according to the datasheet ("reverse leakage current is small, and it can play a rectifying role in the circuit") . What do you think about this choice?
Please let me know which file format I should upload to make the PCB clearly visible, without any references to my local KiCad libraries.
In that case it's D4, the CUS10S530 that is in question.
along with a few other secondary features
You can't have other things connected to the battery connection that may draw current, and that 1000uf cap looks like a dead short when you turn on power.
Why you Say Is CUS? From my naive point of views Is very similari to the Sparkfun One, isn't It?
You can't have other things connected to the battery connection that may draw current, and that 1000uf cap looks like a dead short when you turn on power
The capacitor 1000uF was never connected. And all the other features (LDO) was not present in the First design, which has the same issues.
And, THX so much for your time!