Sweet spot between ample brightness and voltage/current consumed appears to be around 2.5v

Would 3.3V work for you? There are plenty of cheap buck 3.3V regulators advertised.

 

Thanks Alec_t. I’m not sure how buck regulators work - is there an overhead between input and output. Will a 3.3 buck regulator operate successfully from an 18650 at 4.1v and how low would it go?

 

I see a failure to communicate here. My intention was to keep the battery packs in each light assembly, so that if the string failed the individual units will still light. The internal battery for each device will provide a stabilizing effect on the voltage, and the external battery will take over the powering as the internal battery voltage drops. That is what I was describing in the post with the series connection details.

 

I see a failure to communicate here…

Not a failure to communicate, MisterBill2, it’s a failure on my part to completely appreciate what you were describing.
I’ll try putting the little batteries back in tomorrow and see what happens.

 

With both the internal battery and the external power sourcethere will be both redundancy and an increase in total charge capacity. Probably it will not be at the maximum possible efficiency, but if the performance is satisfactory then that will not matter much, as the sunshine for charging is free.

 

An 18650 battery cell is 4.2V when fully charged and some protection circuits disconnect it when its voltage is too low at 2.5V.
The battery University says that 3.0V is when a Lithium cell should be disconnected to avoid damaging it and reducing its useful life.

 

Thanks Audioguru. I understand the parameters for charging and discharging Li-ion cells, and when I get to the solar replenishment phase of my project, will be sure to use an appropriate charge controller. In my case, I am using old reclaimed 18650s, and the charger I am using right now (a USB plug-in type), charges them to 4.2v but most immediately drop closer to 4.1 as soon as the charge is over. My arduino based discharge tester stops at 3.0v.
At the moment, as the project is on my workbench, the discharge through the lamps is monitored manually, but when finally in place, it will utilise the charge controller’s discharge protection cutoff to ensure the Li-ion battery isn’t taken too low by the lamps.
It is the voltage disparity between Li-ion cells and the 1.2v Nicads supplied with the lamps that was the reason for raising this post.

 

Cheap solar garden lights stopped using poor quality Ni-CAD battery cells years ago.
Videos show Li-Ion cells from China have a capacity that is much less than their rating because they are full of (rice?) flour and the low capacity of the Ni-Cad and newer Ni-MH cells in newer solar garden lights also might have rice flour in them.
I replace the cheap poor quality battery cells with Name-Brand Ni-MH cells in all my solar garden lights.

 

The last time that we purchased some of those cheap lights they cost as much as the better battery cells. Of course they were on sale at Menards.

 

Videos show Li-Ion cells from China have a capacity that is much less than their rating.

I agree - having done it once, I would never again buy batteries of any type from ebay, Ali express, banggood, wish, or any of the other fronts for cheap Chinese knock offs - especially when the sellers claim ridiculous capacities that are physically impossible.
Note that my projects are being powered by reclaimed 18650 cells from big name laptops, and are all branded batteries from manufacturers such as LG, Samsung, Sony, Panasonic, and Sanyo - so are reliable in terms of their claimed capacity, still relatively efficient for low load usage, and best of all, I already have quite a number of them so for now, they are free.

 

Most of my cheap solar garden lights costed $1CAN at Walmart.
The old ones had a solar plastic-covered panel and LRD that soon got sunburned and a very low quality Ni-Cad battery that rusted away in 1 month.
New ones still costed $1 but last year used a solar panel with a glass cover, no LDR and a modern Ni-MH battery that did not rust but had a very low capacity.
This year I bought some crackle-glass ball solar garden lights at Walmart for $2CAN each and the Ni-MH batteries were very low capacity so I replaced them with Name-Brand cells that cost $4.50CAN each. They are in sunshine most of sunny days and they light all night long.

 

@Audioguru… Unfortunately, my lamps are along a path that is shady, and the batteries never recharge - hence the project intention to provide power and solar recharging from a remote location in my shed.

@MisterBill2 - latest observations on replacing the internal Nicads…
1 - the lamps were noticeably brighter
2 - the lamps reverted to individual switching from both their local on/off switch, and their local solar panel :-|
3 - something was getting warm in each lamp :-(
4 - each pair of lamps started to draw 180mA plus, from the 18650 :-o
Conclusion - something ain’t doing what we were hoping :-/

 

One more - troubling - observation… the units are showing wildly different voltage drops from the 18650, with the first in the string pulling just 1v and the other pulling 2.6v!
I am going to abandon this idea for the moment, and try a different tack…. The solar charge controller has a 5v USB output, so I’ve ordered myself a buck converter, that I will feed with 5v from the USB output, and adjust to 2.5v, then run all the lamps in parallel from the 2.5v - what do you think?

 

The only part that changed was the batteries, and so it appears that some were in a much lower state of charge than others. So the ones with the lower charge drew more current and showed a lower voltage drop. Reasonable results, but not good.
So just running series pairs may be the most effective scheme, I don't think there was any problem with that. No separate batteries because that is when the problem arrived.
Dropping the 5 volts down to 2.5 in anything but a switcher regulator will waste half the power because of dropping half the voltage in the resistance. Not the best use of battery charge.

 

I’m not sure how buck regulators work - is there an overhead between input and output. Will a 3.3 buck regulator operate successfully from an 18650 at 4.1v and how low would it go?

The cheapo 3.3V buck regulators such as this, are usually switch-mode and work with a 5V supply. Some are stated to work with an 18650 supply down to 3.5V.

 

…it appears that some were in a much lower state of charge than others...

I’m not sure I agree with this assessment. Both the Nicads are brand new, fully charged, and when the lamps are working individually from just the Nicads, are drawing very similar current. I have a suspicion that it is the introduction of the series wiring that is causing the issues - perhaps there is some variation between the two lamp modules that is causing the shared power to be unbalanced. The modules aren’t designed to be wired in series after all.
I’ll give the buck converter a go - at least with that and wiring them in parallel, I can be fairly certain that each unit will be getting the same voltage (apart from any drop over the length of the line).
I appreciate your comment about the buck converter not being the most efficient, but as you said in a previous post, the sunshine is free… ;-)

 

The solar lights circuit has a voltage booster that oscillates at a high frequency. If two circuits are connected in series then they each need capacitors to smooth their pulsing current. Try 100uF parallel with a 0.1uF (104) ceramic.

 

Ok - so the buck converter has arrived, and I’ve connected it to a USB plug, and wired up the 6 lamps in parallel. I removed the battery from each lamp, and wired in a 1N4148 general purpose diode into the positive line of each one. The whole thing now works perfectly, with the solar panel on each lamp acting as an individual light level switch to operate the lamp when it gets dark enough, and the whole string of lamps consumes around 100-110mA when all 6 are on. Ironically, the buck converter itself consumes about 65mA, even with nothing attached to it! There is no excessive heat at the buck converter.
I tried without the diodes, and found that the lamps remained on dimly even during the day - which I couldn’t explain, but since the addition of the diode resolved the problem, by effectively isolating each unit from any possible feedback from the others, I’m happy with the solution.