Dewalt DCB118 18/20V battery charger

[GraemeG]

This Dewalt DCB118 charger showed a solid red light (normal) but would not detect or charge a battery.

When I opened it up there were some pretty obvious burn marks on the base of the cover from underneath the circuit board. There was no sign of any loose wire or metal inside the case but it looks very much like something had shorted between R11 and C39. Note that this version of the DCB118 is designed for 230-240V AC and the rectified 240Vrms input, about 340Vdc, is stored on C1, a 180uF 400V electrolytic capacitor. The end of R11 is connected to the +340 volt DC rail. There was no significant damage to R11 but C39 was partly burned away and measured as a resistance of about 5 ohms!

Another burn mark on the case corresponds to Zener diode ZD2 which had exploded and partly vaporised. Looking more closely I found several other scorched or burned components. In all:
C39 arced and virtually short circuit.
ZD2 exploded and open circuit.
R17 very burned and one pad was completely gone from the board.
R42 and R44 burned and open circuit.
Q103, Q104 and Q212 all open circuit. The plastic case of Q103 had blown open.
R31 burned and measuring around 130Kohms, should be 51.1Kohms.
IC31 physically looked OK but various pins were shorted to ground.

Checking around and testing a few other components also found:
A fine copper trace leading from C11 to J5 was melted and open circuit.
Q1 11N80C3 power MOSFET shorted between Drain and Source.

Common sense says I should consider this one a lost cause but I like a challenge!

[GraemeG]

On a side note, While searching for information about this charger I found several mentions of the toroidal inductor L101 coming loose from the board and fracturing one of the solder joints. This seems to be a common fault with this model and I found the same problem with mine.

The silicone compound used to stabilise and glue it to the board had come unstuck and there was a clear fracture in the solder all around one lead. This looked bad enough to stop the unit from working but has nothing to do with the shorted components as far as I can tell. If this is your only problem then resoldering and some fresh silicone or glue will have it fixed in a few minutes.

You can see in the photo that the silicone was never stuck very well and only in a small area.

[GraemeG]

Luckily I was able to find a schematic for a 110V version of this charger at https://www.acidmods.com/RDC/DeWalt/DCB118.pdf This file seems to be an orphan at that site as I can't find any other details about the DCB118 or why the circuit diagram is stored there. I have reproduced the file here to help make sure it isn't lost but I can take no credit for creating this diagram and I am very grateful to have it. Note that the diagram is for the 110 volt version and there are some differences for the 230 volt model that I have.

There is also a possible error with the part number for 2 components, Q212 and Q213. The circuit diagram lists both as PNP bipolar transistor MMBT2907A. But looking at the circuit surrounding both Q212 and Q213 there is nothing to limit the Base-Emitter current for either device and I suspect they are actually a P-channel FET BSS84AK. These 2 SMD devices are labelled WVS in my charger and it took some tracking down but I believe this refers to the Nexperia manufactured BSS84AK where the code is described as %VS where % indicates the manufacturing site. Another clue is that the N-channel FET 2N7002K is used for several devices across the board and it is the complementary device to the BSS84AK. The 2N2007K SMD devices are labelled LNW in my charger and this corresponds to the datasheet where the label is listed as LN% and % is again the manufacturing site. Q213 is apparently undamaged on my board and doesn't show the typical "2 diodes" of a BJT when I test with my multimeter but it does appear to show the single S-D diode of the FET.

Component differences that I noticed in the 230 volt model:
F1 main fuse is 5amp.
R1, R2, R3 all 200K and are connected after FL3.
FL1 is replaced by two 5 watt wire wound resisters. One 0.82ohms on the neutral rail and the other 0.91ohms on the live rail.
C1 180uF 400V electrolytic.
Q1 SPA11N80C3 power MOSFET in a fully isolated plastic TO-220 package.
R10, R12, R13 are 2W resistors of 2.4ohms, 3.3ohms, 2.4ohms respectively.

EDIT: Do NOT use this schematic if you are working on one of these chargers as there are a number of errors in it. Use the corrected schematic attached about 6 posts below. No guarantees it is totally fixed but it's closer and also covers the 230V model.

[GraemeG]

The end of C39 that I suspect was shorted to the 340volt supply is the VCC power rail indicated on the circuit diagram. The VCC rail seems to supply the drive circuitry for the power MOSFET Q1 and the regulator IC31 so it's no surprise that most of the damaged components are in this area.

I had my doubts about IC32 which is the switch mode regulator that supplies VCC and the 5 volt rail for the rest of the circuitry. I scavenged a 100nF SMD capacitor from an old scrap board to replace C39, used a 1/4W through-hole 10K resistor for R17 and a 1/2W 27V through-hole Zener for ZD2. The through-hole components were fairly easy to mount on the bottom of the board and solved the problem of the burned track under R17. With the other dead components removed I added a temporary 1K resistor from VCC to ground as a load for the IC32 regulator, crossed my fingers, and powered it up.

I was a bit surprised that nothing smoked and VCC was solid at about 12.3V and the 5V supply from the 3 pin regulator IC135 was exactly 5.00V!

So IC32 has survived and now I am waiting on deliveries from China. They should arrive in a couple of weeks and then we will find out if I have missed anything.

Here's another shot of ZD2 and the melted open circuit copper trace leading to J5.

[GraemeG]

While I am patiently waiting for the components to arrive I have noticed a few other differences with this 230 volt version. Or maybe it's just a different generation of the circuit from that in the diagram I found. As you can see in the photo above, this one is a DCB118-XE Type 1. I suspect there are other versions.

Other component differences:
R122, R123 and C110 are labelled but not installed.
R103 and R119 are 82 ohm 2W resistors.

[GraemeG]

It's a work in progress... In the mean time I have noticed a few more problems with the circuit schematic.

I suspect Q214 is actually a P-channel MOSFET. It's labelled P1N6G which seems to be a Vishay Si2343 30V 5.9A device used to drive the cooling fan.

The next problem had me confused for a week! The entire BAT+(Sw), BAT+(SwA) and BAT+(SwB) network is actually connected to the 5V rail. So the BAT+ rail exists but everything labelled as BAT+(Sw?) should be labelled 5V. The missing link is under Q215 and it took a lot of track tracing to find it. I am working on a modified schematic and will upload it in the next few days.

So far I have replaced all the bad components (that I know of) and found a problem where the optocoupler IC34 doesn't turn on quite enough to fully turn on Q212. Instead of the full 12 volt VCC supply getting to IC31, I'm only seeing about 7 volts. I don't fully understand why but this may be due to a higher gate threshold voltage on my replacement Q212. I'm using generic Chinese components so it wouldn't surprise me if they have different specs. I plan to test this by swapping the replacement transistor Q212 with Q213 or Q215 which are both original components.

I'm also waiting for a high voltage differential probe to arrive so I can better check out the hot side of the circuit. Even when I supply the full 12 volts to IC31 it doesn't start the main switch mode circuit and I'm not sure why...

[GraemeG]

This is turning into a saga but it keeps me entertained 

I was right to question my cheap replacement components. I swapped the original mosfet from Q215 with the replacement at Q212 and now the Vcc power supply fed to IC31 is about 11.4 volts and high enough to run the IC. But Q212 is still not fully turning on and tracking back I still find 10.3 volts across the optoisolator IC34. With Vcc at 12.2 volts we are only getting about -1.9 volts on the gate of Q212.



So what I don't understand is why the designers put ZD216, a 3.6 volt (nominal) zener, in series with the LED of the optoisolator. With the circuit powered up I see 3.1v across ZD216, 1.05v across the LED and that leaves just 0.85v across R261. The outputs of IC132 are open collector and I have confirmed they are in a high state. So with less than 1mA through the LED it's no surprise the output of the isolator is barely turned on! It works with an original MOSFET at Q212 but only just.

Weirdly an almost identical circuit driving IC36 includes ZD218 but this time it is a 3.3 volt zener and, as far as I have tested so far, this difference is enough to more completely turn on the isolator.



So what were the designers thinking? Why use two different value zeners? Why include them at all? I can't see any purpose for either of these zener diodes. If any of the comparator outputs are driven low (I've measure them around 0.2 volts in the low state) it will easily turn off the optoisolator LED even if the zeners were replaced with a link. All I can think of is the zeners raise the threshold needed to enable the isolators so they make some allowance for noise on the comparator outputs.

At this point I think I will replace both zeners with 2.7 or 3.0 volt devices. Or maybe just a standard forward biased diode giving about a 0.6 volt drop. Or maybe simply a link and do away with the diodes altogether.

I'd be happy to hear if anyone can explain this design choice!

[GraemeG]

Success!

But somehow in all the testing I totally missed that the current sense resistors R10, R12 and R13 (two 2.4 ohm and one 3.3 ohm 2 watt in parallel) at the source of the power MOSFET are all open circuit. I could swear that I checked them but they LOOK perfect so maybe I assumed they were ok because they are not even a bit brown. Assumptions will get you...

Anyway I have the new semiconductors installed and cobbled together some 1 watt carbon film resistors from our local Jaycar store and the charger is up and running. Now I have to wait some more for the correct resistors to arrive from RS. The reason the originals died without burning is that these current sense resistors also act as fuses in the event of a catastrophic failure. They are designed to die without smoking, burning or starting a fire. The replacements I have ordered are Vishay PR02 units and they have "Defined fusing characteristics" and "Meets active and passive flammability requirements".

I'm still running a few tests to make sure everything is working correctly, particularly the dual rate charging and the safety circuits that check for the battery ID, temperature, etc. There is an interesting write up by Reddit user Tool_Scientist here https://www.reddit.com/r/Dewalt/comments/1ershh9/dewalt_battery_schematic_temp_and_low_voltage/ that provides good information about the inner workings of Dewalt batteries. In particular the ID line controls the allowable charge current for the various sized batteries.

The ID pin on the battery is simply a resistor connected to the battery negative rail. According to Tool_Scientist the allowable charge rates for various ID resistances are:

<110 ohms - Bad, no charge
110-379 ohms - 4.2 amps
380-890 ohms - 7.8 amps
891-928 ohms - 5 amps
929-8250 ohms - 4.1 amps
8.26-9.06 Kohms - no error but no charge
>9.08 Kohms - Bad, no charge

This matches fairly well to the small sample I have seen. I only have one battery, an XR Flexvolt 6Ah N632052 with an ID resistance of 608 ohms, and the DCB118 charger. Looking at the circuit of the charger I can see it will lock out any battery with an ID of less than 111 ohms or greater than 9000 ohms and it will halve the charge rate if ID is less than 374 ohms or greater than 1083 ohms.

I haven't measured the actual charge current yet but I have seen it halve the charge rate as the battery approaches full charge. There is no discrete circuitry to do this, it is managed inside the microcontroller and I'll try to note the voltage it switches the next time I run a charge.

[GraemeG]

The charger is finally repaired with the Vishay non-flammable resistors and the temporary through-hole components replaced by the correct SMDs.

While waiting for the various deliveries I poked around quite a lot and run a lot of testing. I found several errors in the original schematic but my thanks once again to whoever put in the time to create the diagram. This repair could not have been done without it. The corrected schematic is attached below and the main issues were a few devices shown as BJTs when they are actually FETs. I have also updated the diagram with components in the 230V version and then added a bunch of notes describing how I think the circuit works.

Curiously this charger only runs at about 4.5 amps even though the DCB118 is supposed to be an 8A charger. The components that set the charger current were not damaged by the fault so it seems 4.5A is "normal" for my unit. I was tempted to adjust bits to give it a boost but I wonder if Dewalt found heat or reliability issues when running at 8A? A bit cheeky if they dropped the charge rate and didn't tell their customers! Or maybe my unit is abnormal in some way. There is a second or 2 soon after power on where it pushes about 8.5A into the battery but then the regulation kicks in and it drops to about 4.5A, so the circuit is well capable of 8. A mystery to me.

I think the final count was 14 dead components and a melted track. I also changed zeners ZD216 and ZD218 to 2.4V so that optoisolators IC34 and IC36 will be turned on more effectively. In particular ZD216 must not be reduced any lower than 2.4V or the MCU will not be able to use Q225 to turn off the charge when it needs to eg. battery full, over temp, fan failure, etc.

[GraemeG]

The full schematic is too big to print on one A4 (or Letter) sheet. It really needs to be A2 size to be readable so I have chopped it into 4 equal sized pieces. If these are all printed in the same way on A4 or Letter pages then a bit of physical cut and paste should let you reassemble them into one sheet. You will probably need to use the "Shrink to Fit" option when you print them.

I also collected a bunch of scope captures and voltages so if you are working on one of these and run into problems or need more detail, let me know.

[voltsandjolts]

Nice work and congrats on the successful repair!
Could you pls provide the schematic as a PDF, one sheet, A2 size or whatever?

[GraemeG]

I had to tweak the compression and other pdf settings to get this down to a reasonable size so the previous png file is probably slightly higher quality but I think this is quite readable.

This was one of those repairs that was completely uneconomical, particularly with the hours spent. But once you start you have to see it through!

[voltsandjolts]

There is something amiss with your method of generating the PDF, because it is just a raster image embedded in a PDF. The correct method would produce a PDF with vector graphics, such that zooming in never pixelates.

What EDA package are you using?

Edit:
Ahh, my bad. I now understand that you're making edits on top of the DCB118 schematic PDF which you found, using a raster graphics.
BTW, Inkscape imports that original DB118 PDF perfectly, allowing vector edits and output.
Thanks again!

[GraemeG]

This is the problem! Sorry I didn't make that clear before but the only circuit diagram I could find for this device was already a raster file embedded in a pdf. It is posted back in the third post from the beginning of this thread.

I don't have any EDA software and I wasn't about to learn one and recreate the entire circuit. So I took the easy option and used Photoshop (I am very familiar with it) and directly edited the raster image. I know it's not ideal. Even then I had some troubles extracting the full resolution raster file. I'm not sure if there is something wrong with the original file but I spent some time trying to get the full resolution image.

I gather there are some free or cheap EDA packages. My needs are pretty basic. Is there one you would recommend if I want to go down that path for a future project?

[voltsandjolts]

Ah, yes, understand now. Was editing my post above when you replied.

Any EDA package is difficult to 'dabble with' in a short space of time. They all have quite expansive feature sets, have their own quirks and take quite some time to get efficient at using them. If you want that challenge I would recommend KiCAD (Free software) or DipTrace (Free to use version has limitations).

Alternatively, if only doing simple schematics (not PCBs) you could consider Inkscape, which is a general purpose vector graphics program that is well worth learning the basics of.
https://www.eevblog.com/forum/eda/drawing-schematics-using-inkscape/

[GraemeG]

Yes it was the learning curve that totally put me off. Although I have just been wondering how much AI could have done for me    I now know about ProtoFlow but probably just as well that I didn't spend a week playing with software!

Thanks for the Inkscape tip, I'll keep that in mind too.

[voltsandjolts]

Although I have just been wondering how much AI could have done for me 

Nothing, AI is currently terrible at schematics AFAICT

[GraemeG]

Recently @RDC discovered I have been reaping the benefits of his hard work creating the original circuit diagram for the DCB118 charger! He has now incorporated the corrections discovered during this repair and created a separate 230V version of the schematic. Then to add icing to the cake he has reworked the layout to make it even more readable!

Many thanks for your hard work RDC and also for allowing me to post your diagram here. Greatly appreciated.