DALY BMS Teardown & Failure Review (16S 40A)

methods

1 GW
Joined
Aug 8, 2008
Messages
5,555
Location
Santa Cruz CA
Particular Model
16S 48V 40A LiFePO4 ....... (simple, no uart)

Exact link where I bought 15 of them ..................(not prime, shipped from China)
https://www.amazon.com/gp/product/B0876ZT5M1

DALY_001.png

I have already posted all the positive things I have to say. The BMS are amazing in every way but one... the output fails to short (at least on this batch). Classic Counterfeit Mosfet overheat-to-failure. Once in failure, the device acts like a diode passing any current that comes along, while dropping 500mV and producing lots of H E A T

HOW TO DETECT FAILURE MODE
* If it runs super hot at mid to high current
* If it lets you run past LVC
* If you can measure Diode from Blue to Black wires (anode facing out) while powered off

Now onto the teardown, if you want to know more about what lead up to it, ask.

PXL_20221112_010515193_compress87.jpg

-methods
 
I have two units in Failure Mode that we can open up.
Forgive my slow and inconsistent response, I have to work from Android Phone.

I witnessed a video on Youtube some time back where a dude was able to pop the red cover. Inside he found a potted PCB with the mosfet body molds exposed to heat transfer material. Not the drain side, but the plastic. So -> Lets get to it. Progress will be slow as I have other activities going on (including a dispute with the vendor)

MY ASK:
Take these failed units back, give me something that matches the wiring harnesses I already spent dozens of hours installing on a bunch of packs. :idea:

-methods
 
Here is the unit that we will open first. I have marked it with Quality Inspection 001 so that we don't get them mixed up.

This will be a destructive teardown because the unit does not function and never will again. Our purpose here is to inspect the heat sinking mechanism.

PXL_20221112_152807028_compress16.jpg

-methods
 
I am compressing pictures for upload so they are not mega-huge.

I am going to try side-cuts for starters. The BMS is a monolithic clump of potting with the only exposed aluminum being the anodized red, aluminum, top cover. I will pry that now.

..> In my brain?
The mosfets should be directly heat coupled to this aluminum with an Insulator. This is not my product tho, and this may be thermally conductive potting material. I was VERY SUPRISED when I saw the other teardown exposing funky heat transfer pads. The first thing I thought was:

"Uh oh"
Dubious thermal path... and... a dubious thermal path is exactly aligning with my overheating failures over here. This device should have NO PROBLEM thermally protecting itself. There is NO EXCUSE FOR THIS DEVICE TO OVERHEAT in any condition.

Again
methods says:: "There is no excuse for this device to overheat to failure in any condition, since it can read its own temp, read its own current, and has control over the flow of that current."

I would rightfully expect this product to protect itself thermally... which could result in repetitive shutdowns... but it should never be able to actually fail to an ON state. That would be hazardous, and reason for recall.

-methods
 
Adhesion is pretty good, and I think I'm cutting away more of my finger than the potting material.

PXL_20221112_154240262.MP_compress14.jpg

-methods
 
In the background is a GROWATT SPF 5000 ES. This is what I use for Battery Cycle testing

* Will charge 16S LiFe @ 5KW from 240V wall source
* Will charge 16S LiFe @ 2.5KW from 120V wall source
* Will charge 16S LiFe @ 5KW from arbitrary DC input 150V - 420VDC (solar input, it will MPPT)

* Will discharge 16S LiFe @ 5KW into 240V AC load
* Will discharge 16S LiFe @ 5KW into 120V AC load with an Auto Transformer

You can set
* LVC
* HVC
* Max Charge Current

I control max discharge current via the AC load. ANYWAY -> That is just my test rig. I was using it to apply

* 20A Constant discharge
* 40A Constant discharge
* 60A burst (trying to hit the over-current protection limit) (died after 10 seconds)

-methods
 
As for GroWatt?
I have 3pcs of the 5KW unit. Got all of them from Signature Solar out in Texas. These can be paralleled up to 30KW and so far I have been 100% satisfied. The only negative experience I have had was that my second unit shipped with a battery nut banging around inside the main cavity, its power switch was not plugged in, and it was missing a button. My third unit shipped with the same battery nut banging around inside.

BUT _> They (GroWatt) perform
I have 2 units in continuous duty out in pretty harsh environment. Anyway - They have not let me down. DALY has, so let us get to the bottom of this and get

* A bad batch confirmation, purge them from reality
* A firmware update to protect faster, purge old ones from reality
* A mosfet upgrade, purge old from reality
* Etc

But something IS going to give here, as these units are not meeting specification.

-methods
 
Here we go! First Contact!
These really are very well built.

PXL_20221112_155522908.MP_compress74.jpg

Now it is delaminating very easy. There is no bond between the potting material and the bottom of the PCB board. It's going to peel up like butter.

But first
My finger is leaking, I need to go tape it up.

-methods
 
I was going about it all wrong. Without doubt you open these up from the bottom. They peel open like a banana.

* Run the blade along the edge
* Peel away the potting material

PXL_20221112_160022916.MP_compress41.jpg

Easy ... For someone who does this for a living.

-methods
 
(I have to go do family stuff, will return before the sun falls from the sky)

-methods

P.S. This is freeform thread for any and all to speak truth or ask questions. Feel free to post links, etc.
 
It's methods Jr #2 first birthday today. Methods Jr #1 is now 10 years old. Methods wife #2 is baking cakes etc. . .

PXL_20221112_173520187_compress56.jpg

Nice imprint above... Conformally coating for sure

PXL_20221112_173535898_compress6.jpg

I presume the spots where the potting stuck is where the mosfets that failed reside. It got hot.

PXL_20221112_173546140.MP_compress62.jpg

So this PCB is good up to 17s... Which may bode well for part voltage ratings (or not). You want a really solid margin, like 50% + tons of TVS

PXL_20221112_173556101.MP_compress6.jpg

I always like to see a maker's mark

PXL_20221112_173601047.MP_compress7.jpg

PXL_20221112_173609597_compress44.jpg

PXL_20221112_173613547.MP_compress15.jpg

Enjoy
Once the festivities are over we can look at the other side. It'll be a harder dig but based on how poor the adhesion is, I am sure I can expose the mosfets for what they are

Hell
At this point it's looking like I might be able to replace the mosfets. Wouldn't that be sweet irony.

Methods buys a nice potted BMS, just to have to depot it and replace its mosfets, then potted again with proper heat sinking.

Lol - teasing but not at the same time. This is a safety device and they better figure it out real quick.

-methods
 
Battle wound from fighting DALY BMS to pull down it's pants and show me it's

* MOSFET voltage
* Mosfet RDsON
* Mosfet brand
* MOSFET genuine indicator

PXL_20221112_185349408.MP_compress19.jpg

Then
Since I spent about $300 in parts (+labor!!!!) and it's going to cost me over $3,000 in engineering time to maybe clear this up maybe not clear this up... We will get a snapshot of their solution while in there.

* Output mosfet configuration
* How many in parallel for discharge
* How many in parallel for charge
* How are charge and discharge stack together (common drain etc)

Almost everybody does it the same way, and that is the way of common source, so you can have a simple gate Drive that is referenced to same ground.

Maybe yes maybe no we find out... Maybe if it never failed we never find out.

-methods
 
The extra balance wire contact to support 17S is interesting. There was supposition toward that in the other thread where both 13S and 14S LiIon Daly models were labeled 48V. Nice to see the reality that they use the same boards to cover multiple SKUs, though. Nice investigation!
 
Methods!!! Great to see you back here again!

methods said:
Since I spent about $300 in parts (+labor!!!!) and it's going to cost me over $3,000 in engineering time to maybe clear this up maybe not clear this up...

Yep, that sure sounds like depot (like you said) LOL
methods said:
Methods buys a nice potted BMS, just to have to depot it and replace its mosfets, then potted again with proper heat sinking.
 
Have not read back yet
Remove 4 fasteners from that point, get here

PXL_20221112_230647468_compress47.jpg

From there get here
PXL_20221112_230656893_compress15.jpg

BINGO

-methods
 
First impression
Seems like you might want the thermal pads on the other side... Copper is a excellent thermal conductor, and on the back of the PCB board you basically have a direct shot to the heat sink.

Oh well, I am thermal novice. All I know is that a configuration like this tends to not survive peak overcurrent. The mosfets heat soak and Spike before they can disperse the heat.

So in other words
The path for heat to escape is a treacherous one, and I think we can prove that.

But
First we will look up those mosfets... Then determine if they are legitimate. I will leave it to the reader and not spoil anything.

-methods
 
I'm a software engineer, not EE, but '042N08D' - does that mean 42A current limit, 8V breakdown voltage? BMS label does say 40A discharge limit.

Sure would have been nice to have the part burn out at 60A instead of 42A, though, to have some more leeway.
 
Part numbers can be made up in any sort of way. I wouldn't infer much from the numbers.

I would however make something of the fact that they are absolutely ghost on a Google search. There are people around here who I'm sure can ferret out the details.

Without looking
* It will be over 60 volts
* It will be N channel
* It will have overstated rdsON

A really good mosfet for this application would be at least 80 volts and probably 100 volts. It would have 2 or 4mOhm ON resistance. This one is probably 40 or worse, with some parallel to make up for it.

They are probably like $0.10 each rejected from some process or another, then re-stampes and sold on the Chinese Gray market.

... I can open a new unit and see

* Extract mosfet or 6

* Set power supply to 100mA, crank voltage until it conducts. That's its avalanche voltage***. Can't work anywhere near that

* Bias ON with 15V gate to source, then CC 10 or 20A drain to source. Measure Drain to source voltage and divide by your current to get the on Resistance

-methods
 
Anyone who recognizes any of the stampings please pipe up.

-methods
 
Chiming in to say that I’m grateful for this investigation and your work, methods. DALY brand BMSs are widely used, and it’d be nice to have confidence in the product. I hope your work here leads to accountability and better product reliability from the manufacturer. Maybe that’s a stretch, but one can hope for good outcomes!
 
I'm good at finding stuff but I can't even find a reference to a similar FET, not even on Aliexpress. :/

Is the potting thermally conductive, or insulative?



methods said:
I am compressing pictures for upload so they are not mega-huge.
Unless your bandwidth is limited, Neptronix created a thing for the forum that automagically resizes/etc the uploaded images to help with this so you don't have to do it before uploading. :)

methods said:
Here is the unit that we will open first. I have marked it with Quality Inspection 001 so that we don't get them mixed up.
I'd've used FA (failure analysis) rather than QC (Quality Control), but I appreciate the teardown and analysis. :) (I enjoy many sorts of failure analysis)
 
Daly junk.

Ohh shiny red.
 
UPDATE

Questions: Is the Potting thermally conductive?
Oh got I hope it is. If it is not then we are all in trouble. I will have to think up a test for this, but I can say that the "real stuff" is pretty expensive and this stuff is very heat sensitive. It gets sticky just from the heat gun

TEST RESULTS
In another thread I heated the unit up with a heat gun in an attempt to get it to thermally protect. Here are the very disappointing results.

* BMS External Probe is very responsive
* BMS itself appears to have no thermal limiting !!!!

PXL_20221113_132941400_compress69.jpg

I heated up the unit until the aluminum (tied directly to mosfet plastics via thermal pad) was too hot to touch. I then heated up the potting in the bottom until it was pliable and tacky. I then heated the top again etc. The unit WAS THERMAL SOAKED and it did not (N O T) shut down

Cough
So flipping stupid


To confirm my test I then went and heated up the probe - click - turned off right away. So there you go. I am now over 90% sure that we are heading in the right direction. Now the real quest is what kills the units

* Voltage
* Heat

Heat would come from current and the only thing we can do there is draw less. Voltage would come from fly-back, and we MAY be able to do something about that. We could possibly try clamping the primary output with TVS, but since there are likely a pair of fets stacked vertical... not sure how much it will help

SO
I guess that means we need to try and peel the rest of the potting off to investigate further.

So far what I am seeing is that you get every penny of your $23 worth with this unit. ASSUME it performs no HVC/LVC duty, and only does top balancing. If you really have to use it... I suggest having this unit trigger a secondary current cut circuit.

-methods
 
Starting here
(Now with two puncture wounds to my finger, so we can prove the time domain)

PXL_20221113_135915697_compress7.jpg

Starting here where we left off yesterday. Using nothing but basic tools I bought at Home Depot.

PXL_20221113_140330476_compress27.jpg

Glue is just like anything... The whole is stronger than the single. The way you break a bond is with a thousand tiny cuts.

Deboned all the way around the edges by Rolling... Grab and roll, grab and roll.

PXL_20221113_140446043.MP_compress7.jpg

And... A perfect teapot (I mean depot)
Now we can populate this piece of s*** with real mosfets from America.

C O U G H

-methods
 
Back
Top