DIY Charger - Possible Project

…maybe downscale your pictures:

Thanks for that. I didn’t know the syntax.

I use a 27” display, and sometimes forget to scale.

:+1:

You can make it clickable too, linking to full size:

(opens in new tab)

I made two design choices yesterday. Both are open for discussion. It is easy enough to undo this early in the design.

The LTC4015 can reduce charge current if Vin drops below a hardware set threshold. Typical use would be when Vin also powers a device. If the device starts drawing more power, the charger will throttle back. In this application, it would be used if we program a current that is too big for the power supply.

I decided not to implement under voltage current limiting. I will leave it up to the end user to source a suitable power supply. Leaving it out simplifies the board.

Cell count is set in hardware by three pins. The LTC4015 does not monitor individual cells and cannot balance a series string. I’m leaning towards hard wiring this to one cell only.

The other option would be to tie Cells 2 low with jumper selection on Cells 0 and 1. That would allow 1,2,3…6 all the way to 9 cells in series. 3 and 6 would be for 6 and 12V lead acid. If I do that, I’ll design for 15 or 20V Vin Max and 8A I charge max. The higher current would make this more useful for Pb batteries.

Thoughts?

One more thing about jumpers:

The optimal size of the inductor depends on the charge voltage and current. An inductor sized for 13.8v 8A is non-optimal for charging to 4.2V.

Even if I put in cell and chemistry jumpers I will optimise for 4.2V 4A.

The switching frequency is adjustable between 200kHz and 1MHz. The reference designs switch at 500kHz.

The AM broadcast band starts at 520kHz. 500kHz is the upper range I’m going to consider. Anything higher means I need to worry about interfering.

Here in the UK and EU, we also have AM Long Wave from about 148kHz to 283kHz. Many AM radios use 455kHz IF. I don’t want that either.

That leaves me with 300kHz - 430kHz and 480 - 500. Aiming for 500 could end up with at 525kHz or even a bit higher.

525kHz isn’t a problem in the UK.

*F = 47.65/Rt
*

Where F is in MHz.

Trying some standard 1% resistor values:

93.1k = 512kHz Too high. Worst case will be around 540
95.3k = 500kHz
97.6k = 488kHz
100k = 476kHz :slight_smile:
102k = 467kHz
105k = 453kHz
107k
110k = 433kHz

LOVE THIS STUFF!!! Alot of my ideas never make it much further than design either. I’ve been wanting to learn USB/Charging circuitry for awhile and this is :+1: :+1: :+1: !!!

Thank you so much for sharing!!!

You’re welcome.

C1 is a Nichicon UPA 330µF 25V Radial Lead - 8mm diameter 3.5mm lead spacing. The rest are probably 0805 surface mount.

R3 is 301K 0.1% 25ppm

EQ tied to ground. EQ high turns on Pb Acid equalisation charging. I don’t want that, even for lead batteries.

I’ll keep the cells and chemistry jumpers in the design for now. We will see how things go when I get to layout.

Andrew_Debbie, I believe the people here is more concerned with what this can do and not so much with any particular optimizations. I'd stick a BIG ASS inductor in there and let users enjoy unprecedented charging flexibility and features (like fully adjustable charge voltage).

Others may think different, of course.

Cheers

I plan to size the traces and pads for 8A.

See pages 42 –45 of the data sheet for more about inductor values and PCB layout.

There are some implications for higher power and higher voltages.

I’m trying to keep the board under 100mm x 100mm.

Added a few more parts. The Screw terminals at Vin are rated 15A nominal. Should be fine at 8.

R4 is a 10k NTC thermistor. It will end up mounted off board on Rev 0. On the schematic now mostly for BOM. More decisions later. I’m thinking about sacrificing a $4.99 charger for the battery holders.

R7 and R8 are the current sense resistors. The 4015 sets the current by measuring the voltage between CSP and CSN in 1mv steps up to 32mV. Maximum current is at 32mV. 8mΩ == 4Amps and 4mΩ gives 8A.

I’m going to set this up for 2 resistors in parallel. Put both of them in if you want 8A. Only put in 1 for 4A. — Of course you can use different values… The resistors are going to dissipate about 1/8 W. I’m going to use 2 or 3W resistors so that they don’t heat up too much.

CSN and BATSENS both connect to the battery positive terminal. At 8A or even 4A there could be enough voltage drop across the wire from CSN to the battery to matter. I haven’t checked. I’ll also have to model the voltage drop on the negative side. All of that will happen later on, when I start too look at layout and packaging.

Do we want an input protection diode? I haven’t put one in yet.

I think I will add a 10-20Amp shottkey diode on Vin. Cheap insurance against reverse polarity on the input.

The diode will drop maybe .5 from the input. Probalby make it impossible to charge from a 5V supply. It is easy to replace the diode with a jumper.

I still have to look at the opimal range for Vin. If have have time tomorrow, I’ll start looking at the Inductor and switching transistors.

*FET Selection*

There are people on here that have a lot more experience with small power FETs than I do. I'm happy to have suggestions. There are lot of things to check. The main one is to not over load / over heat the internal gate drivers and the LDO inside the LTC4015. it is possible to use an external LDO to power the gate drivers.

LT's example designs use an On-Semi (Fairchild) FDMC8030 Dual N-Channel MOSFET for the top and bottom switch. The FDMC has good numbers -- rDS on < 14mΩ at 10A , VGS = 4.5V

The FDMC8030 packs two FETs into a tiny 3x3mm MLP package. I like the specs but not the challenge of getting the layout right or being able to solder the part.

Going through the product selector at On-Semi, I found a couple of older Fairchild parts that should work.

FDS 8949 Dual N-Channel At VGS = 4.5V rDS on = 36mΩ 6A.

I'm not sure if this one will work at 8A. Should be ok at 4A but I'd have to do some thermal estimates.

FDS 8447 Single N-Channel At VGS = 4.5V it is 12.3mΩ and 11.4A.

It is single transistor. Trade off of size for better specs and easier layout.

For the first pass, I'm going to use the FDS 8447. We will see...

http://www.onsemi.com/PowerSolutions/product.do?id=FDMC8030

A good primer if you need it:

MOSFET basics Fairchild AN-9010 …

The battery is a dummy to show the complete circuit. There should be a capacitor in parallel with the battery.

I’ve left out the FETs that are supposed to switch between Vin or battery power for the load. SYS is suppose to be the greater of Vin or Vbat. Tying SYS to VIN without a switch could be a problem if you pull the power with a battery installed. I may have to add two FETs. I may do it anyway.

I did first order sizing on the inductor and some of the capacitors. The data sheet has equations and recommendations. I haven’t done any thermal calculations yet.

I’m just catching this thread now…

Inductor Sizing

For those following along, see page 43 of the data sheet -

L = (VBAT •(1– VBAT / VIN (MAX))) / (0.25• f •ICHG(MAX))

In other words, higher switching frequency and higher charging currents make the inductor smaller. A larger different between VBat and Vin makes the inductor larger.

The inductor must not saturate at Imax. So we need an 8A + 60% * 8A inductor at a minimum. I’m probably going to limit mine to 4A but will leave room on the PCB for 8A.

I made an Excel spreadsheet to help size the inductor.

VBat = 4.3 Volts
f = 476kHz
IChg = 8A

Vin = 8Volts —> 2.09 µH - 500kHz is 2 µH

IChg = 4A
Vin = 12V
L = 5.8µH

IChg = 1.5A
Vin = 8V
L = 11µH

10µH should be enough for most of us. I may use a 15 because I don’t care about board size or inductor cost. I’ll have a look at parts later.

I’ve been working on this without posting BUT . . .

I just ordered an ISDT C4 from Banggood.

The C4 can charge at up to 3A per slot and discharge at 1.5A per slot.

I am putting this on hold indefinitely . The C4 does just about everything I was planning.