Pain and suffering getting the DW01 and 8205A protection circuit work right

[murabat]

I am making a remote solar powered soil moisture sensor which has a lithium ion battery for backup. I have put in the protection circuit by using DW01 and 8205A chips. My schematic is as under:



B+ and B- is where the battery is connected. VPROT and ground are the output terminals.

My circuit follows the schematic from the DW01 datasheet which is like this:



BATT+ and BATT- here are the output terminals, not the battery connections. Andreas Spiess in the video below switched the places of battery with the output terminals; that of course didn't seem to work either and in the comments he explained that it was a mistake:


I have tried many ways to get my circuit working but to no avail. Unless I short the battery negative to ground the circuit does not turn on.

I found an electronic stackexchange question without any real answer:
https://electronics.stackexchange.com/questions/382811/lithium-ion-battery-protection-ic-not-able-to-drive-mosfet/397019#397019

I found some other schematics which show the battery negative is actually suppose to be connected to ground like this:






So the question in the end is what is right way? Did they forget to put ground in the official datasheet on the battery negative? I really need an answer because I don't have any hair left to pull out.

[Peabody]

I don't know why your circuit doesn't work, but I believe the DW01 data sheet circuit and your matching circuit are correct.  The DW01 limits current by using the MOSFETs to control the flow between system ground and the battery negative terminal.  If you bypass them by connecting the negative terminal directly to ground, you have no protection.

If you are getting no current flow, then at least one of the MOSFETs is not turning on for some reason.  The FS8205 may be bad.  Can you measure the gate voltage - OD and OC?

I'm a bit confused by the 8-pin FS8205 shown in your circuit.  It's a 6-pin part.

[ajb]

I seem to recall that some protection ICs require that a charging source be applied before they startup.  Not sure if that's the case for the one you have, that may only be for the devices with integrated chargers. 

[murabat]

Here is the FS8205a datasheet:
https://datasheet.lcsc.com/szlcsc/Fortune-Semicon-FS8205A_C16052.pdf

It is a TSSOP-8 pin part.

Gate voltage on OD = 2.99V and OC = 1.10V with respect to ground. Battery voltage is 4.09V. If I ground the battery negative momentarily, the circuit starts working. Then the OD and OC voltages are battery voltages.

I seem to recall that some protection ICs require that a charging source be applied before they startup.  Not sure if that's the case for the one you have, that may only be for the devices with integrated chargers.

My circuit does start up when the solar panel is connected and keeps working when the panel is removed. If this is some kind of feature, I didn't find it mentioned anywhere on the datasheet.
I remember when I bought some TP4056 boards with protection from Aliexpress, the description did mention that I need to connect the charger first. Is that the key point here?

[NiHaoMike]

My circuit does start up when the solar panel is connected and keeps working when the panel is removed.

That's "shipping mode", to minimize discharge until the product is plugged in for the first time.

[Peabody]

Speiss may have located the battery incorrectly, but I think his description of the DW01 circuit is correct.  When the battery is powering the circuit load, output OD must be high enough to turn on the related transistor.  That permits conventional current to flow from the low circuit rail back into the negative battery terminal.  If that transistor is not ON, its body diode prevents current from flowing.  The OC transistor can also be ON, but doesn't have to be because current will flow through its body diode even if the transistor is OFF.

I would think that 2.99V is more than enough to turn on the OD MOSFET.  If current still isn't flowing, then the FS8205 may be bad, or you may have OD and OC connected to the wrong gates.  Or I guess it's possible that the diode voltage drop across the OC MOSFET is causing problems for your load circuit.

But I would just say again that you should not connect the negative bettery terminal to circuit ground.  That bypasses all of the protection provided by the DW01.

[murabat]

This is the screenshot from the Speiss video:


Pin 6 and 7 of FS8205 is connected to B-. In my circuit they are connected to ground. I think, that is the problem here. But that is not how schematics show it to be!!

[LukeB]

All I can say is maybe check that the pins from the datasheet match the pins that easyeda is connecting them to. I used those two chips in the past and have had them labeled incorrectly in easyeda causing me to wire them incorrectly and have a faulty circuit.

[murabat]

All I can say is maybe check that the pins from the datasheet match the pins that easyeda is connecting them to. I used those two chips in the past and have had them labeled incorrectly in easyeda causing me to wire them incorrectly and have a faulty circuit.

I just double-tripple checked the schematic symbols and pcb footprints. Pins 2,3 are for the mosfet which connects to battery negative and pins 6, 7 are for the mosfet which connects to ground. OD on DW01 is pin 1 which is connected to pin 4 of FS8205a and OC is pin 3 which is connected to pin5. CS pin connected to ground with a 1K resistor. There is no problem there.


BTW, I found another guy confused by the same problem:
https://electronics.stackexchange.com/questions/363432/integrating-li-ion-charging-protection-on-the-same-pcb-how-to-wire-ground-pro

[murabat]

I found this on baidu search:



Here the 8205a pin connections are opposite. Things cannot get more bizarre.

[Peabody]

Can you confirm from the body diodes that the mosfets in the chip match the orientation in the datasheet?  I know that's extremely unlikely to be the problem, but you seem to have eliminated everything else.

Also, when the battery is connected, and OD is 2.99V but no current is flowing, what voltage do you measure across each drain/source?

[Peabody]

Here is Great Scott's video on his lipo charging, protection and boost circuit:



His final version should be available on EasyEDA.  Looks like his DW01 circuit is like yours.

[murabat]

I think I have found the issue in the end. Look at page 8, red note in the bottom right corner, of this datasheet:
http://www.spectron.us/SM6FIE/Electronics/SparkFunLiIon/DW01-G-DS-10_EN.pdf

That explains the problem and its solution.


Thanks for all the help guys, I really appreciate it.

[Peabody]

It's funny, but I have the later v1.1 of the datasheet, and it doesn't have that Note.  I wonder what this is all about.  Anyway, I hope that fixes it for you.

[ledtester]

Couldn't load the above datasheet, but found the note in another version:

Note: When a battery is connected for the first time, it may not enter the normal condition (dischargeable may not be enabled). In this case, short the CS and VSS pins or connect to a charger to restore to the normal condition.

[lbsfilm]

Hey everyone, and everyone who will read this thread later on, I believe I found the real solution. 

This comes from my own PCB not working (without shorting the battery and out ground to turn on), where as one of the blue amazon versions of it just performed as it should, meaning it would just turn on once the battery was connected every single time.

So I tested for about 2 weeks ( ) and found the issue: The little 0.1uF cap that is parallel to the battery seems to be the problem. 0.1 seems right the boarder between having the issue and not having the issue. I noticed this when I replaced the cap on the blue amazon boards with one of my own and suddenly could replicate the issue there too.

When measuring the two caps I realized that the one from the amazon board is about 90nF where as mine was about 102nF. After replacing it with an 82nf capacitor on my own board it would finally perform as I wanted and turn on when plugging in the battery without any shorting.

With the 100nf capacitor I could also test the following: when I initially connected the battery and started the board by shorting the grounds I could disconnect the battery afterwards and reconnect it and the board would start without shorting grounds. This made testing pretty hard as I had to wait 15 min before the next test. After finding out about the cap being the issue I could just short the cap and then reconnect the battery to reliably reproduce the issue

TLDR The whole problem is the 0.1uF capacitor, that should rather be 0.08uF or something like this

I really hope I could help with this finding and hope I can save some people from pulling out hair =D

Greetings,
Lukas

https://lbsfilm.at

[frogblender]

100nF ... 82nF

Mmmmm... if the circuit is failing with 100nF, but working with 82nF...   I would not call that circuit "robust" - there is probably still something fishy in there.  If you build this circuit, in any production quantities, with the 82nF "fix"...  you're almost guaranteed to get bitten on the rumpus.

The capacitors have some % tolerance, plus the nF will vary with temperature, plus the nF will vary with the voltage across the cap... throw in a bit of aging....  82 vs 100nF is, in technical terms, "right on the hairy edge".  Plus the datasheet recommends 100, but it doesn't work....  Fishy.

Is it just that you need to plug the circuit into the charger *once* (as stated in the broken-link datasheet I never read)?   This would kinda make sense, the DW01 is usually soldered right at the tabs of prismatic cells, and will never again be disconnected.  But in your case, having a removeable cell that requires a quick charger shot in order for DW01 to get its shit together... ya, that'll be annoying.

And futher with the 82nF "fix"... that cap is (obviously) there for a reason;  if it is skimped upon... it'll probably work just dandy, as you've found.   But the *real* acid test will be when the poop hits the fan - ie. hard short downstream, or reverse plug, or something, and DW01 needs to start driving the (heavily-capacitive) gates of those mosfets... THAT will be the time when that cap is pressed into service.  And if it is insufficient, DW01's supply could collapse to the point where DW01 looses it's marbles.  And then your product is on fire.

Plus, during a hard short (poop-fan-hit referenced above), your lithium battery will try (and succeed) to source enought current to jumpstart a locomotive, and grounds get lifted and VCC's collapse even with solid copper on your PCB... so that cap is crucial in ensuring DW01 has a stable supply.

anyhoo... alot of blather, but please let us know the answer to the bolded question above.

And does anyone have an alternative to DW01 that, after being connected to a cell, doesn't require a charger hit to operate?

[ledtester]

...
When measuring the two caps I realized that the one from the amazon board is about 90nF where as mine was about 102nF. After replacing it with an 82nf capacitor on my own board it would finally perform as I wanted and turn on when plugging in the battery without any shorting.
...

I would test your theory with caps from different suppliers - just to make sure the problem isn't with the specific caps you are using.

[virtapp]

Guys, the FS312F datasheet is reporting this:

When a battery is connected to FS312F-G for the first time, it may not enter the normal condition (dischargeable may not be enabled). In this case, short the CS and VSS pins or connect to a charger to restore to the normal condition

...so, to do so (short momentarly the CS and VSS pins), as recommended from Tony Stewart on https://electronics.stackexchange.com/questions/553788/solar-lipo-charger-circuit-design-and-validation/553794?noredirect=1#comment1444777_553794 maybe a simple capacitor will solve this issue! any comment? 

[JoshShabtai]

Apologies for the necrobump but I think I figured out the real issue here (and solution)

Easiest solution: A small (but not super precise) cap across BAT- and CS. I found that 1-2nF was about the minimum. I went with 6.8nF since that's what I had laying around. Try not to add too much as this can slow down the time it takes to react to an over-current/short condition, but a small cap shouldn't be a major issue.

First, to clearly distinguish between the LiPo negative terminal (pin 6 on the DW01) and the circuit GND (the output of the MOSFETs) I'll refer to the battery terminal as BAT- and the output as "circuit GND"

Problem: The reason for this problem comes from the fact that the DW01 chip was intended to be built into the battery, not the circuit itself. So if the DW01 is connected to the load (e.g. if it's built into a bigger circuit) then what happens is that the MOSFETs start "off". The BAT+ gets connected and brings the whole circuit up to the BAT+ voltage (e.g. 4.2v). Since the MOSFETs are off, even a relatively high impedance path from BAT+ to circuit GND will have the voltage go up.

This means that CS (current sense), which is connected to circuit GND through a 1k resistor, is up at ~4V. But that gets interpreted as a high current passing through the MOSFETs, so the over-current protection kicks in. What we need is for CS to stay low long enough for the DW01 to actually turn on the MOSFETs. Adding a cap across BAT- and CS. Combined with the existing 1k resistor, this slows it down enough for the circuit to start up and get into a stable condition.

Another possible solution would be to have another MOSFET connected to BAT+ (so the DW01 controls both sides of the cell). I imagine a P channel MOSFET with the gate connected to circuit ground might actually work, and this would resolve any impact to overcurrent protection (but for many applications a slightly slower response time might not be a real problem).

Hope this helps somebody - I've been having this issue for a while now.
.
Edit: Actually, I think adding the MOSFET on the VCC side is the best way to go. Just use circuit ground to control the gate, and it won't turn on until after the GND side is connected. Some quick tests on my bench seem to work very reliably. Example circuit:

With the capacitor solution, sometimes I would have to pop the battery in, then out, then in again to get it to work. This seems a bit more reliable.

[Peabody]

I ran across a single-chip protection part recently:

https://datasheet.lcsc.com/lcsc/2006240933_XySemi-XB7608A_C669688.pdf

It brings the mosfets into the chip, so it only has three connections - V+, Bat-, and system ground.  And it needs an external resistor and capacitor.  But the whole implementation is tiny.

Unfortunately, it has the same footnote.  And since CS is not brought out, that solution is not available.  So you'd have to short the two negative pins temporarily, or connect a charger.