• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

LTC4412 charger issue

Status
Not open for further replies.

Russbo

New Member
I've put together a PCB based directly on the second application circuit in the LTC4412 datasheet. This seems to work as intended for a power path circuit with an 18.5V (16-21V working range) lithium ion battery as the primary input and a 24V DC PSU as the auxiliary. I chose this application circuit because using a schottky diode for my auxiliary supply would result in a large amount of power dissipation (this circuit is designed to operate up to 20A with FDS6681 MOSFETs).

Here is the page of the datasheet explaining the circuit operation (I have used figure 2 for the circuit, connecting the charger as is done so in figure 3).
http://i.imgur.com/62cdOHY.png

However, when I connect a charger (CC/CV board based on XL4015, from eBay) between the 24V input and the battery in the circuit, the 24V PSU immediately hits current limit and the voltage sags significantly.

If I connect the charger to the circuit with a separate 24V supply powering it, all is fine and dandy, I can charge the battery, switch the source powering the load and so on no problems.

I'm really stumped by this. The only thing I can think of is that the difference in the way the sense pin is configured between the 2nd and 3rd application circuits is what's causing my issue, but I've run a multimeter through the circuit checking the voltages at various points in the circuit against the datasheet in the different states of operation and it seems to check out for the most part.

Could anyone explain to me the behavior I'm experiencing? I can conduct any further tests and give any more information needed to diagnose the problem.

Thanks.
 

ChrisP58

Well-Known Member
Please post a complete schematic of what you're doing. It's really hard to guess from just a handfull of words.
 

Russbo

New Member
As in the image I linked (didn't want to embed the whole thing but it has relevant text in there), this is the exact circuit I put together but with a charger connected between the wall adapter input and the battery.

The MOSFETs are FDS6681z chips and I have a 100uF electrolytic cap as Cout.
 

Russbo

New Member
Forgot to mention, this is the specific variety of board I'm using as the charger:

It's a generic XL4015 based DC DC converter with voltage and current limiting with the addition of C/10 cutoff for lithium charging. I want to make my own circuit similar to these ones to use in my circuit but I want to understand why I'm having the issue I'm having before I go on and do that because if I end up having the same issue it'd be a complete waste of time. I don't have the diagram for this converter unfortunately.


Here is the circuit I put together illustrated with the battery charger like in the datasheet. The primary difference between my circuit with the charger and the one in the datasheet is the use of a P Mosfet versus a Schottky and connecting the sense line to the output rather than to the input. I can't use a Schottky because otherwise I'll dissipate too much power in my circuit.


Here is the circuit illustrated in the datasheet with a charger connected. There is next to no information on its implementation, it just says in the datasheet "here connect it like this" which is somewhat frustrating.
 

ci139

Active Member
i may be an idiot but they connect the P channel enhancement type MOS-fet by it's drain to power source and by it's source to load ?? so - the conduction is done by body diode and the substrate connected to reverse of it's normal . . . though in a conduction mode the D2S voltage difference is not so big -- i haven't yet met the specification of such operation . . .
. . . the datasheet of LTC4412 states that the MOSFET is driven linearly but the datasheets of the mosfets state the continuous body diode forward current is the same as for D2S -- least of which would "compensate" 1 to 3 V drop on a body diode . . . but the entire setup sounds suspicious if not total madness to me -- i mean the chip only does the compensation of the 1 to 3V drop using the undocumented connection of the mosfets ??? and why to drive it linearly by sensing the missing wall adapter (refering to a "Block diagram at pg.6") ???
 

Russbo

New Member
Thanks for the reply ci139, I'm no expert but I think what the aim is in connecting the FET this way is to prevent reverse current flow when the MOSFET is off. You don't want current going back to either of your power sources through the body diode. I could be wrong here, but I think that's the purpose.

I looked into doing this without a chip then realised how whacky things are when your voltages are changing drastically on either side of the MOSFET. This strange configuration might just be a trade off that has to be made in order for the circuit operation to work - which it does seem to when the charger isn't involved. I was whacking 10+A through my circuit in some testing I did and the MOSFETs didn't really get hot and I had little voltage drop at the output.

So far as I'm concerned, if it works at all then I'm happy, but it doesn't when the charger is introduced which is most frustrating.
 

ci139

Active Member
You don't want current going back to either of your power sources through the body diode.
never thought of it - but is painfully true for the battery

. . . if your pin-6 is connected to external supply and the charger puts heavy load (read interference) there might be the bug.1
  • the AUX.SUP is to provide power to LOAD but if charger steals it for BAT (at it's near "empty") the undefined function results
  • the AUX.SUP is to provide power to LOAD but if charger steals it for BAT (at it's near "full" the terminal voltage being some 10% over nominal) the undefined function results (likely the charger has to keep its output high to load the battery and provide power to application LOAD through primary mosfet) -- s.a.n.↓↓
. . . if your pin-6 is connected to application LOAD and the charger puts heavy load (read interference) there might be the bug.1
  • the AUX.SUP is to provide power to LOAD but if charger steals it for BAT (at it's near "full" the terminal voltage being some 10% over nominal) the undefined function results (likely the charger has to keep its output high to load the battery and provide power to application LOAD through primary mosfet) -- s.a.p.↑↑
  • the AUX.SUP is to provide power to LOAD but if charger steals it for BAT (at it's near "empty") the undefined function results (perhaps - see LTC4412 ds. External P-Channel MOSFET Transistor Selection @ pg.8 - both power sources fail to drive your mosfets reliably)
. . . other than ↑ that ↑ -- i've tested mostly unsoldered regulators and for most the output voltage drops off long before their max output current is reached something at 33 to 50 % (rarely higher) of I.out.max no matter the size of the heat sink mounted ??? with the exception for KA7806
-- the point here is to find out the max. LOAD where the setup still functions

shortly put (the forum very impatient logged me off i have to rewrite my edits)
  • get more powerful supply
    • or different supply for charger (full wave rectified stuff is safe to parallel donno about SMPS-s)
    • or don't charge this battery with the charger connected to AUX PWR
  • monitor pins 4,5 on LTC4412
  • e.c. interpret yourself
 
Last edited:

Russbo

New Member
I've used the charger on its own charging the battery and it pulled about 2-2.5A from my bench supply at 24V. A manageable 50-60W.

My aux supply is a 400W 16A 24V unit. I was testing with no external load and the charger was managing to draw 160W from the wall which translates to about 6-7A which it shouldn't get close to. I can hear the charger board whining too and it quite quickly reduces until you can't hear it again which makes me think the chip is going into an overload/shutdown state.

A separate supply for the charger would be a bodge but not very feasilble for the power I need and it wouldn't be an elegant solution.
 

ci139

Active Member
I was testing with no external load and the charger was managing to draw 160W from the wall which translates to about 6-7A which it shouldn't get close to.
what external load

i'm a bit confused here - i think the XL4015 charger drops AUX SUP and does both charging and supplying the external load connected to the source pin on the Primary P-MOS - the LTC4412 likely can't cause the additional 4A to be drawn by XL4015 . . . unless it's due no load present at all (usually the 10% of intended avg. or max. should be present coz many supplies assume the load being present to get to their rated output)

if your 6-7A is likely some X-mode feedback from battery through both FETS _b_a_c_k_ to auxiliary input -- then you should never done that (might blow your equipment before you even notice)

my mistake -- the "min. load" for sure is a "no load" but the voltage at pin 6 of LTC4412 (the same as the source pin on the Primary P-MOS) goes where with nothing connected there

in some other case i'd considered coupling off the AUXsup and charger ... #!¤"¤&#%/ as in fig.
-- if it's not too much of a voltage drop
edit: ha! it'is a correct figure (anyways to confirm)
Random_SMTT_SLF.png -- you must find the correct values of L , C for your need -- i just picked 'em up from the thin air here !!!
___________________

nothing to do on sunday collecting information about the fig.1 at #4 , .rotatedView()
.
--> h t t p : / / y o u t u . b e / <name , replace by following>
  • XxNQ7aZ5q44
  • j6HebRPQdJs
  • kDt-yrz9WJg
  • -ozVzwrOogQ
screw that editor (i don't want the relevant space be occupied by second relevance videos here - so copy paste these links if interested)
 
Last edited:

ci139

Active Member
testing-parallel-PS-s.pngthe leds are visible at daylight starting from some 1mA e.g. (24-3)V/1mA=20kΩ or less
(the shown indicators are speed limited by 200k inputs and pass 10mA through small power (20÷30mA) red LEDs)

edit: "!¤%¤&%#/ -- they drive external source switch with 470kΩ pullup (just noticed) -- so that signal also needs to be amplified prior led & upper P-fet gate OR the Gate Indicators should follow the high-input impedance buffers . . . blah, blah ,blaah . . . . . blah :cool:

edit#2 : it's a P-MOS, it's a P-MOS, it's a P-MOS, it's a P-MOS, it's a P-MOS dummy me :eek::confused:
e.g. a LED indicator update or a very insane one such :meh:
Random_SMTT_SLF-2a.png
edit#3 : fixing misc. . . .
Random_SMTT_SLF-2bc.png

PS! -- i have actually no idea whether you need to perform this test or where to look for the bug o_O
 
Last edited:

ChrisP58

Well-Known Member
Russbo, my first comment on this thread was to ask for a complete schematic.

Since you say that the charger works fine alone, but misbehaves when connected to the LTC4412, I have to suspect that there's some interconnect issue. Please show YOUR drawing of how you have got things connected together. Particularly how grounds are connected. Don't just assume that ground is ground.
 

Russbo

New Member
View attachment 105935
the leds are visible at daylight starting from some 1mA e.g. (24-3)V/1mA=20kΩ or less
(the shown indicators are speed limited by 200k inputs and pass 10mA through small power (20÷30mA) red LEDs)

edit: "!¤%¤&%#/ -- they drive external source switch with 470kΩ pullup (just noticed) -- so that signal also needs to be amplified prior led & upper P-fet gate OR the Gate Indicators should follow the high-input impedance buffers . . . blah, blah ,blaah . . . . . blah :cool:


PS! -- i have actually no idea whether you need to perform this test or where to look for the bug o_O
Your idea of coupling off the AUX supply from the charger is interesting. Only thing is, I can't have a Schottky in the path to the AUX MOSFET - I need to be putting nearly 20A through that MOSFET and I'd not only struggle to find a Schottky suitable but it would simply be too hot to use.

Russbo, my first comment on this thread was to ask for a complete schematic.

Since you say that the charger works fine alone, but misbehaves when connected to the LTC4412, I have to suspect that there's some interconnect issue. Please show YOUR drawing of how you have got things connected together. Particularly how grounds are connected. Don't just assume that ground is ground.
This is the schematic from KiCad. Mostly the same but you can see that all is a common ground.

Do you think I should have a separate ground for the charger with its own plane or similar?

 

ChrisP58

Well-Known Member
I think that the problem lies with the grounding of the charger. Not so much that it needs a separate ground, but that it's input and output grounds need to be separate.

I suspect that it measures charge current with a resistor in the ground path. When you connect it's input and output grounds to the same ground on your PCB, you're shorting that resistor out. When the charger sees no voltage drop across that resistor, it thinks that it's running at zero current, so it cranks up the current. And that is why it's causing your source power supply to go into current limit.
 

Russbo

New Member
Interesting. Well on the PCB the grounds are just connected to a ground plane, is that an issue?

This is my PCB. The charger connections are the second column in from the left, input at the bottom, output at the top. You can see how the grounds are connected to the same plane.

Are they an issue in this configuration? If so, how would I alleviate the issue with the design of my PCB?



Thanks very much for your help
 

ci139

Active Member
VIN and SENSE Pin Bypass Capacitors
Many types of capacitors, ranging from 0.1μF to 10μF and located close to the LTC4412, will provide adequate VIN bypassing if needed. Voltage droop can occur at the load during a supply switchover because some time is required to turn on the MOSFET power switch. Factors that determine the magnitude of the voltage droop include the supply rise and fall times, the MOSFET’s characteristics, the value of COUT and the load current. Droop can be made insignificant by the proper choice of COUT, since the droop is inversely proportional to the capacitance. Bypass capacitance for the load also depends on the application’s dynamic load requirements and typically ranges from 1μF to 47μF. In all cases, the maximum droop is limited to the drain source diode forward drop inside the MOSFET.
Caution must be exercised when using multilayer ceramic capacitors. Because of the self resonance ...


e.g. what i've personally studied -- the bad range for filter cap.-s is (33÷)47µF to 220(÷330)µF -- the causes are misty but it seems to be independent of the I/O power range of the regulator :eek:
 

ChrisP58

Well-Known Member
The problem is that the input and output grounds on the XL4015 charger board are not and must not be connected together. When you tie them together by connecting them both to the same ground plane on your PCB, the battery charge current is no longer flowing trough the current sense resistor on the charger PCB, so the charger can't see it, and therefor, it doesn't limit the charge current.

I'm afraid that you're going to have to find a different charger that uses high side, instead of low side, current measurement.
 
Status
Not open for further replies.

Latest threads

EE World Online Articles

Loading
Top