Continue to Site

Welcome to

Welcome to our site! is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Micro generator lipo charger

Not open for further replies.

dr pepper

Advanced Member level 1
Mar 15, 2010
Reaction score
Trophy points
Activity points
Using an arduino, a 6v bicycle bottle dynamo and a 2S lipo battery I want to build a mini battery backed wind generator.
I have a couple of questions on charging the batts, the dynamo (magneto) acts like a current source, so if I was to rectify its o/p and feed that to a buck boost switching reg like a Lm2576 configured to draw 500ma from the supply using a sense resistor, and its o/p goes to charging the batteries via a Schottky (the reg's feedback would be setup so its o/p voltage wouldnt go over 8.4v the safe max for the cells).
As I understand it a balance charger connects a resistor across a cell that is close to 4.2v, this sounds inefficient and wouldnt work so well if the battery was loaded as well as charging. Is there a better way I can balance charge a 2S lipo, or should I make the thing 2P and boost the o/p voltage with another smps chip?

If you are still in the "parts selection" phase I recommend
looking at batterm management chips, which are made
for this. If you're lucky, might find the converter and cell
management in one (this is a hot segment now with a
steady pace of new product introductions).

Now that said, the bicycle (friction style?) alternator may
be a bit of a let-down - back in my youth those were feeble,
only would (barely) light a 3V incandescent flashlight bulb.
I'd recommend setting one up and emulating max wind
speed to see if you need buck-boost, or just boost.

Another point, if these are LiFePO4 cells they may be
capable of tolerating overcharging and let the cells balance
themselves. You'd have to determine that specific to the
cell and mfr.

Thanks for that.
After reading your post I thought of power bank chips, and after 1/2 hour of searching I found a module on ebay that is a battery management system and supports 1 amp of charging current, and takes a 5 to 8v charging power input, and it has adaptable charging on the o/p but doesnt have usb c protocol to switch voltages its just a pot and gives 5 to 24v continuously variable output at 8 watts, and its cable of simultaneous charge discharge, might be just the trick.
As for the dynamo quality busch ones will light a 10w led to eye spot leaving brightness just twisting the shaft with your fingers, I remember the cheap ones from the 80s where you had to do 40 to see anything.

I have the modules, they have 2 Ic's, one is a non descript switching reg, its just marked 'UMB' presumably its made by Umb microelectronics.
And the other is a Tp4056 lipo charge controller, from its datasheet I can change the charging current.
The board has a Usb c socket with the power pin non committed its brought out to a pad.


some issues (for me)
* why do you say the dynamo acts like a current source. I´ve always seen it as a voltage source with series impedance.

and wouldnt work so well if the battery was loaded as well as charging.
* it´s just the question where you connect the load. Connecting the load at the charger side of the current sensing shunt makes the "charger" to see the true battery charging current. Connecting the load at the battery side tells the "wrong value" to the charging circuit: (true I_charge = I_Batt = I_shunt - I_Load)
(also there are other ways)

* efficiency: is a sum of several things. A bicycle dynamo may not be very efficient. A linear charging regulator may not be very efficient. A non MPP type charger may be inefficient. A continously working resisitve charge balancer may be inefficient.
--> thus use a switch mode MPPT charger and a switch mode charge balancer with enable.


You might be right. Some basic testing the generator seems to make 500ma through a load at around 3v, and then the voltage rises with rpm while the current keeps roughly 500ma, so although power increases until the core saturates the current increases less than voltage. Reducing the load makes the voltage jump up.

My question over the battery being loaded as well as charging was regarding the balancing, connecting a resistor across a battery while powering a load wastes a good amount of energy from the battery, this is now solved as I'm going to use one larger capacity cell.

One idea I had before going single cell was a switching constant current charger that had a transformer with 2 secondaries, one for each cell, however that seemed like a lot of complexity.


then the voltage rises with rpm while the current keeps roughly 500ma
If the voltage rises, but the current not ... then you simply dud not use an ohmic load.
With any ohmic load there is a strict relationship between voltage and current.

If you need help you need provide clear and complete informations. ... like your test setup...

Balancing is to adjust for cell mismatch. If there is no mismatch you need no balancing current.
You say "waste a good amount" ... we don't know your balancing current, nor do we know what you consider "a good amount". So it's information we can not validate.
... and I already told you how to "not waste" energy.
Also it's not clear how "constant current" can solve the balancing problem .. balancing needs different current ... best on charging as well as on discharging.


The bike generator "load line" at various wheel RPM or wheel
torque (harder to know) would be useful to you in setting
expectations and component selection. The no-load / load
dump voltage tells your VIN rating.

I wonder whether newer, LED-purposed bike generators
might contain their own regulation, as LEDs can be wimpy
and you get to replace the lamp assembly, not the lamp.
The old timey friction wheel ones were just bare and the
incandescent bulb was useless riding slow, short-lived if
riding fast.

Think of all the factors required to optimize maximum power point tracking (MPPT) and load regulation (CC, CV) before anything else and define or find all the transfer functions.

- starting torque*RPM to speed controlled voltage to motor impedance, to pulsed inductor to low ESR Capacitor to PWM and PFM modulated current with voltage, current and RPM feedback.

Otherwise, you may end up with a roof ornament. If you succeed, you will have mastered dynamic impedance matching and control.

I worked with a small team that developed a wind power floating weather station (750W) and it used both Savonius and Darrieus turbine blades Circa 1975-76.

Last edited:

I'm not here to argue Mr Klaus.

Well it works kinda, last night here we had moderate wind, around 15 to 20 mph and had the setup running.
I fiddled with the charging current on the module by replacing the current set resistor with a pot, the magneto made too much voltage at first, however it seems to work well at 800ma charge current, any less the thing over volts any more it doesnt get above low revs.
Start torque I guess is low as the regulator doesnt charge the battery till around 4.5v or more.

It would be very involved to get all the specs, I'd need to measure the performance dynamics of the magneto and the wind turbine, maybe I'll do that with the prototype, I have a esp32 with datalogging hardware and software.

Its a hawt turbine, and its not on a rotating stand, in location the wind mainly blows Sw.
Yes I'm going to read up some more on Mppt, for now at least I have power to my remote system with enough spare for a couple of lights.


I'm not here to argue Mr Klaus.
It's a technical discussion.
And I was raised not to sugarcoat any problems.

So when I say "current needs to follow voltage" it's just a physical fact for ohmic loads. It's not meant to be offensive.
It's on you what to do with this information. You may ignore it, you may check your meters, you may check your test setup, maybe you find out that you did not use an ohmic load at all.

Maybe my expectation that you used a "predictable" ohmic load for testing is wrong. Then you may clarify my mistake.
No attack at all.


Not open for further replies.

Part and Inventory Search

Welcome to