Low Parts-Count, Low Heat, 4A, SMD Current Regulation?

[theboom]

My application is a single-cell lithium battery charger. I assume constant-voltage with current-limiting would work (but i may be mistaken).

There will be a group of disconnected cells (not a battery pack). All the cells in the group will be the same chemistry as each other. Need to support both Li-ion and LiFePO.

A single SMPS will supply regulated constant voltage for all the cells. So i think all that's needed is a current limiter on each cell.

The LM338 seems to have low-parts limiter, but apparently not available in SMD package, and (i'm told) not low-heat.

I found this current limiter for the TL431, but i'm told it will disconnect when limit is reached, rather than simply limit current to desired level.

If possible, I like the idea of shunting the excess current to other cells in the group, rather than wasting it as heat.

I'd be fine with a low-parts single-cell charger IC. Analog offers several, such as the LTM8026, but much too expensive for my application.

 

[KlausST]

Hi,

"low heat": if you have a power supply with constant voltage output ..
and a linear current limiting regulator, then the dissipated power is independent of IC, circuit, schematic, mosfet , resisitor....

It`s always: V_drop * I

so if you wnt low heat:
* then you need a low dropout current limiter combined with a reduced power supply output voltage
* or a switch mode current limiter.

Klaus

 

[theboom]

* then you need a low dropout current limiter combined with a reduced power supply output voltage

Since i'm charging multiple cells, at different charge-states, then wouldn't reducing the PS V compromise the charge-rate of some cells?

Still, that's interesting. No matter what, the total charge-time of the entire group will be set by the emptiest cell. Therefor, if we reduce Vsupply to limit current on the emptiest cell, that won't increase charge-time of the entire group. So it doesn't matter if the fuller cells have to wait for the emptier cells to catch up.

However, assume a 2-cell group. Say the low cell is at 3.5V. The high cell is at 4V. If we reduce Vsupply to say 3.6V to prevent overcurrent on the low cell, then won't the high cell discharge and lose charge?

* or a switch mode current limiter.

Can you suggest a low-parts switch mode current limiter?

Parts-count and price on this converter IC is pretty good. 3A max is acceptable. But it's internal overcurrent feature doesn't seem to behave like a continuous current limiter -- it turns off.

https://www.diodes.com/assets/Datasheets/AP62300_AP62301_AP62300T.pdf

Same re this 5A converter: overcurrent feature doesn't seem to act like a continuous limiter.

http://www.aosmd.com/res/data_sheets/AOZ6605PI-1.pdf

This TI TPS5432 chip seems promising re current limit behavior, but parts-count is high.

https://www.ti.com/lit/ds/symlink/tps5432.pdf

The LTM8026 looks OK on parts-count, but way too expensive ($20+)

https://www.analog.com/media/en/technical-documentation/data-sheets/8026fd.pdf

Here are Digikey's in-stock switching converters, sorted by price, with desired power and Vin range. I don't know if any can provide continuous current limiting.

Voltage Regulators - DC DC Switching Regulators | Power Management (PMIC) | Electronic Components Distributor DigiKey

Power Management (PMIC) Voltage Regulators - DC DC Switching Regulators are in stock at DigiKey. Order Now! Integrated Circuits (ICs) ship same day

www.digikey.com

Here's another linear device. I'm thinking that with prudent selection of the mosfet it will run cool. No?

5 Amps Current Regulator with MOSFET – Electronics Projects Circuits

IRF4905 MOSFET with very few elements 5 amp current control, current limiting circuit input voltage line connected on 8 + .40 volts DC current through this circ

320volt.com

i was told the TL431 circuit in my first post will run cool. Parts-count is great. And it looks like a continuous limiter. No?

 

[KlausST]

Hi,

then wouldn't reducing the PS V compromise the charge-rate of some cells?

charge rate is determined by the charging current.
And the charging current is determined by the current limiter.

Since i'm charging multiple cells,

How? in series, in parallel, independent?
Before you talked about "single cell"...

the total charge-time of the entire group will be set by the emptiest cell.

Don´t see why. It depends on the charging method / wiring...

then won't the high cell discharge and lose charge?

Again: depends on wiring..

I can not recommend a current limiter, because the circuit / chrging method / cell count ... and so on is unclear.

If I had to charge batteries, I´d use a battery charger IC. It´s designed exactly for this job. With all the features..
And there are at least hundreds to choose the best for your application.

Klaus

 

[theboom]

charge rate is determined by the charging current.
And the charging current is determined by the current limiter.

You suggested reducing the charge voltage. Won't that reduce charge rate, or discharge, on cells who's charge state is higher than the charge voltage?

How? in series, in parallel, independent?
Before you talked about "single cell"

Not parallel or series. It's a group of disconnected cells. There's one SMPS feeding all of them. Seeking a charging solution which minimizes parts-count (and cost) per cell.

Don´t see why.

In my application, we must wait until all cells are fully charged. Therefor, it's ok if the fuller cells wait for the emptier cells to catch up -- that won't affect the overall total charge-time.

I´d use a battery charger IC.

My thought is to minimize per-cell parts/cost by using one SMPS shared between all the cells. The shared SMPS will provide CV. Then we only need current-limiting on each cell. Using an SMPS + current-limiters gives us a lot of flexibility for charging voltage and current. If the TL431 works, we're talking pennies per cell.

I found these three Maxim charger IC's. I believe they're the only Maxim chips which can handle both Li-Ion and LiFePO. But more external parts than we'd like per cell.

 

[KlausST]

You suggested reducing the charge voltage.

You use only one part of a three parts recommendation.
This is my whole statement
1) if you want to reduce heat
2) use low drop current limiter
3) while reducing the power supply voltage

The sentence above this statement (post#4) explains why.
Using a low dropout current limiter does not give a benefit as long as the power supply is not reduced.
Example:
* before: 4V dropout + 7.2V battery --> power supply needs to be 11.2V, power dissipation is 4W per Ampere
* low dropout: 1.5V dropout + 7.2V battery --> power supply needs to be 8.7V, power dissipation is 1.5W per Ampere
(Mind: I did not change the battery voltage)

*****
How can disconnected cells be charged?

--> Draw a sketch. It´s better than a lot of words.

*****

--> use the parametric part search provided by the manufactuer(s)

Klaus

 

[theboom]

You use only one part of a three parts recommendation.
This is my whole statement
1) if you want to reduce heat
2) use low drop current limiter
3) while reducing the power supply voltage

I understood the other parts. Does that affect my question?

Say you have a cell which is 4.2V at full charge. Say it's current charge-state is 4V. What happens if you connect it to a 3.5V supply? (with a low drop current limiter if that matters).

Will it stay at 4V? Discharge to 3.5V?

--> Draw a sketch. It´s better than a lot of words.

Two cells, not connected to each other.

--> use the parametric part search provided by the manufactuer(s)

Unfortunately, i don't know a parametric search that includes "external parts-count" as a parameter.

 

[KlausST]

sketch please

Klaus

 

[cupoftea]

The 3v5 supply will probabaly get its output caps charged up to the batt voltage, and just sit there, in overvoltage shutdown.

 

[KlausST]

Hi,

The 3v5 supply will probabaly get its output caps charged up to the batt voltage, and just sit there, in overvoltage shutdown.

>It totally depends on the circuit and the power supply.

But 3.5V ?
How do you think it can "charge" a 4.2V battery?

Klaus

 

[cupoftea]

Hi,
Here is a cheap constant current smps ...it uses the LT124x, which is pin for pin with the cheap UCC28C4X range.
Here it is driving leds....so you can change it so that it drives battery if you want......if you wish i can change it for you?
LTspice is free download

 

[theboom]

sketch please

Very simplified block diagram:

Alternate design, based on CC/CV at each node:

The 3v5 supply will probabaly get its output caps charged up to the batt voltage, and just sit there, in overvoltage shutdown.

That sounds safe, right? The cell will simply wait until Vsupply is higher than it's current charge-state, and then continue charging. Correct?

Hm, no, if the supply shuts off, then nothing will charge. Question is: if Vsupply doesn't shutdown, what will a cell do who's charge-state is lower than Vsupply?

Keep in mind, in that scenario, the other cell's charge-state is lower than Vsupply, so that other cell will continue to draw current.

But 3.5V ?
How do you think it can "charge" a 4.2V battery?

You said it was necessary to lower Vsupply to benefit from the low-drop limiter. In that scenario, where i'm lowering Vsupply to accommodate the emptier cell, then Vsupply will be lower than the charge-state of the fuller cell. So the fuller cell will stop charging while the emptier cell charges. Once the emptier cell charges up to the level of the fuller cell, then Vsupply can be raised and both cells will continue to charge.

Here is a cheap constant current smps ...it uses the LT124x, which is pin for pin with the cheap UCC28C4X range.
Here it is driving leds....so you can change it so that it drives battery if you want......if you wish i can change it for you?

Many thanks for that! But it looks like about 25 external parts? We'd like to get it down to about 5 parts or fewer per cell.

--- Updated Jun 4, 2022 ---

Hm, no, if the supply shuts off, then nothing will charge. Question is: if Vsupply doesn't shutdown, what will a cell do who's charge-state is lower than Vsupply?

Correction: What will a cell do who's charge-state is higher than Vsupply?

--- Updated Jun 5, 2022 ---

Correction: What will a cell do who's charge-state is higher than Vsupply?

Based on comments i'm seeing here, the behavior of the cell do who's charge-state is higher than Vsupply will depend on the topology of Vsupply. It will either sit at its present voltage without discharging to the level of Vsupply, or it will discharge to the level of Vsupply.

"If your charger is a typical form of power supply charger then it will charge only. If you have a true ideal voltage current source, that can both sink and source current, then you can discharge the battery in a rather rapid fashion limited by the internal resistance of the cell."

--- Updated Jun 5, 2022 ---

If you have a true ideal voltage current source, that can both sink and source current, then you can discharge the battery in a rather rapid fashion limited by the internal resistance of the cell."

"To prevent this in a trickle charge system, a diode is used in reverse to the rechargeable battery to prevent discharging."

If I charge a battery at insufficient voltage, will it discharge?

Sam Hancock's answer: Yes, the voltage will flow through the supplied voltage source like a resister. To prevent this in a trickle charge system, a diode is used in reverse to the rechargeable battery to prevent discharging. However, it would be best to use a chip based circuit for a controlled r...

www.quora.com

 

[KlausST]

Hi,

You said it was necessary to lower Vsupply to benefit from the low-drop limiter.

I gave an example in post #6.... and tried to explain....

**************
About your sketches:
* in both cases I´d say it makes no difference whether you charge one, two or any other number of cells.
As long as each battery has it´s own [limiter] or [CC/CV] circuit.
You may use batteries of different charge stage, age, capacity, size ... it does not matter. Each is charged individually.
**************

"disconnected" ... I see you use this phrase rather different than I do.
For me it means "not connected". But I´d say in your case each battery is connected. They are connected individually.
I did not expect them to be connected in parallel. ..

**************
BTW: did you mention how many charging channels you want?

Maybe consider a microcontroller with
* an ADC input for each battery voltage
* an ADC channel for each battery current
* a PWM output for each battery channel
and a bit of software....

So for each channel:
* 1 MOSEFT
* 1 inductor or power resisitor
* 1 shunt for current measurment
* .. maybe a voltage divider for

Klaus

 

[theboom]

I gave an example in post #6.... and tried to explain....

Thanks for the low-drop explanation! I think i understood. That's potentially very useful for us. Did it seem that i misunderstood?

You may use batteries of different charge stage, age, capacity, size ... it does not matter. Each is charged individually.

But if i use a shared CV and separate current limiters, then mustn't the CV be set to the "full" voltage of the cells? And therefor, mustn't the cells all have the same "full" voltage?

BTW: did you mention how many charging channels you want?

We want a solution that can scale up to any number of nodes.

Maybe consider a microcontroller with
* an ADC input for each battery voltage
* an ADC channel for each battery current
* a PWM output for each battery channel
and a bit of software....

I'm not against a uC based solution, but "dumb" CV/CC circuitry, where each node handles it's own current management locally, seems simpler to implement. A uC solution adds software development and management tasks.

A uC which requires separate I/O pins for each node means we'd need to add another uC as soon as the number of nodes exceeds available uC pins.

I'm not against it, but i'd like to find an analog solution for comparison.

* .. maybe a voltage divider for

for...?

--- Updated Jun 5, 2022 ---

Hi,

Feel free to change title of this thread. Should it be "Charging Parallel Cells"?

--- Updated Jun 5, 2022 ---

Should it be "Charging Parallel Cells"?

or "Charging Parallel Cells with Separate Current Limiting"?

 

[KlausST]

Hi,

Did it seem that i misunderstood?

How an I know?

But if i use a shared CV and separate current limiters, then mustn't the CV be set to the "full" voltage of the cells? And therefor, mustn't the cells all have the same "full" voltage?

yes and yes. Or better say (again: my example of post#6) add the drop out voltage of the current limiter.
(in detail depends on the used circuit / charging method)

True charging IC solutions (CC/CV) have full control about cell voltage and cell current, thus are rather independent of power supply voltage,

We want a solution that can scale up to any number of nodes.

"Any" isn´t a useful number when developing electronics. "Any" could mean 1000000 or more....
Better say "up to 50" (example)
But your power supply also needs to be rated for "any" number times the chaannel power. It makes no sense to use a 3kW power supply when there are just 2 charging nodes now .. but maybe want to add "any" number later.

A uC solution adds software development and management tasks.

True, but your aim was "low part count". And charging a battery does not need much provcessing power, thus a single microcontoller has enough processing power to control 32 channels or more. So one controler for 32 channels reduces the overall part count. ... and is very flexible. Each charging channel may have it´s own CC setup and CV setup (chemistry).
Also may care for not to overload the power supply. (intellingent power sharing)

An ESP32 (or similar) does not need a display or any other user interface parts, since it can communicate with your cellular phone (web page) where it can show additional values like charging state and charged AmpereHours per battery .. and provides input for CC setup und chemistry setup...
So you dont need LEDs, and their current limiting resistors, potentiometers, switches... (reduced part count)

****************
"voltage divider for" adjusting cell voltage measurement range to ADCinput voltage range.

****************

Feel free to change title of this thread. Should it be "Charging Parallel Cells"?

You used the word "connected" in combination with "cells". But as far as I understand you don´t want to charge "parallel connected"batteries. You want to charge batteries individually. Or am I wrong?

********

or "Charging Parallel Cells with Separate Current Limiting"?

I won´t change the title. But I´d rather pharse it: "parallel charging of individual cells .. with seperate current limiting"

*******
In the end I still think that a common power supply combined with individual CC/CV charging (IC) solutions is a good way.

******
Why is "low part count" important? This usually is not a measure of total cost nor of size and effort.
You want aim for "low part count" when the assembling cost is higher than the part price. Is this the case?


Klaus

 

[theboom]

"Any" isn´t a useful number

Fair enough. Up to about 10,000 cells. Yes, i understand the supply has to provide enough current.

32-cell groups is fine.

your aim was "low part count"

Yes, and i want to develop an analog solution to compare.

a single microcontoller has enough processing power to control 32 channels or more.

Which uC? Will that require some sort of hardware multiplexing?

intellingent power sharing

How is that done? Time-division?

it can communicate with your cellular phone (web page) where it can show additional values like charging state and charged AmpereHours per battery

Of course that's very powerful and convenient. And maybe want to add it later. But not important for first version of my application. It's important for the first version to be viable for makers and battery users who might not have computers or smartphones.

"parallel charging of individual cells .. with seperate current limiting"

Ok, that sounds good.

Why is "low part count" important? This usually is not a measure of total cost nor of size and effort.

I disagree. Parts-count certainly correlates to cost, footprint, design-effort, complexity. i think simpler is better, when possible.

As the the system scales up to tens, hundreds, or thousands of cells, the per-node cost matters.

Our first version should be something that can be constructed by hand for up to about 10 cells, and possibly kit form for up to 30. Too many parts = too much effort.

 

[theboom]

I disagree.

Respectfully.

 

[theboom]

If i'm not mistaken, a linear regulator is going to beat a switching regulator on parts-count. So, if we can keep heat down, and get other helpful features, linear seems the way to go for this application.

Here's a linear LDO 3A regulator of interest. Low-parts and low-cost. Claims high-efficiency. I'm guessing that if any linear regulator can work for our application, this one can. I suspect we can't do better for this cheap.

MIC35302​

$1.16 @ 100
This datasheet has a lot of detail, which seems reassuring. Nice and small.

"Ultra-low dropout (600mV over temperature), and low ground current. Current limiting and thermal protection, and reverse current and reverse battery protection, enable pin. Applications include: SMPS post regulator, battery charger (those are our applications). Output current during overload conditions is constant (sounds like what we want). Thermal shutdown disables the device when temperature exceeds the maximum. Transient protection. Output allows voltages in excess of the desired voltage without reverse current flow (i think that means if the cell charge-state is higher than the MIC35302 output, the cell won't discharge, which is good). Thermal design requires the following application-specific parameters:

First, calculate the power dissipation of the regulator from these numbers and the device parameters from this datasheet.

(If we keep Vin =< Vout, then we'll eliminate output current from dissipation. i don't understand what ground current is.)
Where the ground current is approximated by using numbers from the "Electrical Characteristics" or "Typical Characteristics". Then, the heat sink thermal resistance is determined with this formula:

The heat sink may be significantly reduced in applications where the minimum input voltage is known and is large compared with the dropout voltage (it will be in our application). Use a series input resistor to drop excessive voltage and distribute the heat between this resistor and the regulator. The low dropout properties of Micrel Super Reta PNP® regulators allow significant reductions in regulator power dissipation and the associated heat sink without compromising performance. If the output current is too small, leakage currents dominate and the output voltage rises. A 10mA minimum load current is necessary. 5-Pin TO-252-5 (D). "

 

[KlausST]

Hi,

designing electronics is no emotional thing. It has to do with facts given in the datasheet .. and calculations.
Just because it is a 3A low dropout regulator with low external part count does not mean it can do what you want.

So facts are:
It is a voltage regulator. So how exactly do you think it can work as a battery charger? tell us your idea about charging method, maybe with a circuit (draft) and some values.

Important key features of a battery charger are
* precise voltage limit
* current limit
How do you get both?
You can focus on voltage regulation (how it´s meant to be). How do you ensure current limit? Rely on the 3.5A ... 8.5A internal overcurrent protection? Or else?
Or you can focus on current limiting, then how do you get precise voltage limit?

You surely know that LiIon are very sensitive on overvoltage and may explode if you don´t care for it.

And what about input voltage and how do you get rid of the heat?

And yes, there will be no (low) reverse current when battery voltage is higher than charger voltage. But did you also see the low ohmic feedback resistors R1 and R2? .. and the current through them?

****
A word to think about:
I don´t want to discourage you. But basically you don´t want to use a dedicated "charger IC" but want to "mistreat" a voltage regulater as battery charger. While it should be cheap, low part count, and it should do what a charger should do.
Let´s say imagine it´s that easy. Why would the IC manufacturers then design a dedicated charger IC? And why would designers buy and use them? .. when one can simply use a cheap voltage regulator.
No one wants to design a charger with high part count, and no one wants to design an expensive charger.

But if you use a voltage regulator.. then you have to do some "handstands" to make it operate as a charger. It surely is possible. Then add the features for safety ..and other events like input power loss ... reversely connected battery .. and so on... then you need to add additional parts.. which means effort and cost.
And in case the bad event happens and a battery explodes. Maybe somebody gets hurt, maybe a house catches fire. Are you prepared for that? You are the first to be held accountable ... even if it´s a user mistake.

Again: It´s a good thing to have ideas and to follow them. You just have to be aware about the risk and consequences.

Klaus

 

[theboom]

designing electronics is no emotional thing.

So why are you getting so emotional? Let's stick to electronics.

It is a voltage regulator. So how exactly do you think it can work as a battery charger?

i have explained. I will again. The shared SMPS acts as the CV. Then, at the cell level, the current limiter prevents overcurrent. The voltage regulator by itself isn't a charger. The whole system is the charger.

Important key features of a battery charger are
* precise voltage limit
* current limit

I believe you're incorrect. A conventional Lithium charger has two main stages:
* Constant Current (not "limit")
* Constant Voltage (not "limit")

8.5A internal overcurrent protection?

Thanks for alerting about that. That won't do.

Or you can focus on current limiting, then how do you get precise voltage limit?

The desired behavior is constant voltage, not voltage limit. The shared SMPS provides that.

--- Updated Jun 5, 2022 ---

what about input voltage

What about it?

how do you get rid of the heat?

The objective is to minimize heat by selection of components, and add heatsinking if necessary. Isn't that normal in electronics design? According to its datasheet, the MIC35302 is designed to reduce heat. Of course, one would have to run the numbers to see if heat would be adequately reduced, and validate with bench-testing.

low ohmic feedback resistors R1 and R2? .. and the current through them

Thanks for that. Again... necessary to do the math to determine if that's an issue.

"mistreat" a voltage regulater as battery charger

That's a false statement in three ways. First of all, using a regulator as a regulator isn't "mistreating" it. Secondly, the regulator isn't a battery charger. The entire system, including the shared CV SMPS is the charger. Third, this is a system of multiple cell-chargers, not a "battery" charger (if we take "battery" to mean a pack comprised of series or parallel cells).

Why would the IC manufacturers then design a dedicated charger IC? And why would designers buy and use them? .. when one can simply use a cheap voltage regulator.

Again -- the regulator alone isn't the "charger". The whole system is. Also, yes, i have the boldness to think i may devise something different but functional. How dare i!

Then add the features for safety ..and other events like input power loss ... reversely connected battery .. and so on... then you need to add additional parts.. which means effort and cost.

True, we must weigh the tradeoffs! That's why i'm seeking a solution which already contains those protections. The MIC35302 may offer some of these protections without additional parts, effort, or cost. You seem to disregard that.

in case the bad event happens and a battery explodes

No idea what you're talking about. I certainly want to ensure my solution is safe. For example, the MIC35302 contains thermal shutdown. It will help if you let me know why you feel that's inadequate.