PB 1[4-8][0,5]c? Li-ion Battery

[sutekh]

With all of last week's posting issues resolved, let me try this again...

Not sure if this is most appropriately posted in this forum or over in Hacks, but I think it's more likely to reach the target audience here  As a follow up to my recent Duo Li-Ion battery project, I've been working on a suitable Li-Ion battery replacement for the M5417, M5463, and M5464 NiCd packs used in the PowerBook 140 - 180c.  I'd foolishly surmised that, due to the nicely compatible voltage ranges, this would be a walk in the park compared to the Duo project. When will I learn... The trick here is packing a Lithium-suitable CC/CV charger inside the pack, but only powering it when plugged into A/C power and preventing it from back feeding off of the pack itself the rest of the time. Severely complicating matters is the fact that the charge and discharge ports are the battery's same two + / - contacts.  There are any number of "power path" ICs designed to solve this charge / discharge switcheroo, but they assume the load output and the charge input are separate paths.

This didn't manifest as a problem when designing a pack for the Duo because its charger sends ~20vdc to the battery (which is intended to get pulled down by the NiMh once inserted). The battery-pack-internal Li-Ion 3S CC/CV charger I used requires ~13.5vdc to engage, so when AC is disconnected, it can't back feed off of 3S 18650 cell network that, at most, outputs 12.6vdc. Had I tried the same approach here, the first problem would be a 2S Li-Ion charger needs at least 8.4vdc to function (vs. the PowerBook's 7.5vdc wall adapter output). Once fronted by a boost converter, the second problem would be that, due to the common charge / discharge path, there's no way to prevent the following: Li-Ion cells -> boost converter -> charger -> cells.... in a vicious, parasitic drain cycle that would have the BMS kicking in at low-voltage cutoff in no time.
 

A more sophisticated, current direction sensing approach is needed that works properly in all of the following scenarios:
 

1. Pack installed and providing power (discharge)
   
2. Pack installed and charging from AC (charge)
   
3. Pack removed from unit. No charge or discharge (idle)
   
4. Pack installed and charging, then toggled to discharge (e.g., AC power removed)
   
5. Pack installed and discharging (or idle), then toggled to charge (e.g., AC power connected)
   

Those last 2 may seem obvious or unnecessary, but some of the circuits designs I devised seemingly worked fine for scenarios 1-3, but wouldn't disengage the charger in the case of situation #4 for instance. Anyway, here's what I've come up with:



The sensing / steering is managed by an Analog Devices LT1494 precision OpAmp, and the following is a brief description of what's happening in each of the 5 scenarios outlined above:

1. Current flows out of the pack through D1. The gate of N-channel MOSFET Q1 is pulled low by R2 (10K Ω) and remains disengaged by LT1494 due to current flow from - to + across R1. Charger is off.
   
2. Current flows across R1 from + to - causing LT1494 to pull Q1's gate high, thereby allowing current to flow from Source to Drain. Charger is on.
   
3. Similar to #1. No flow across R1 so LT1494's output & Q1's gate remain low via R2. Charger is off.
   
4. Once current stops flowing from + to - across R1, R2 pulls Q1's gate low. Takes about 1s with a 10K resistor at R2.
   
5. This one was a bit of a puzzle. Even with power applied, D1 prevents current from flowing into the battery (as it should since only the CC/CV charger has any business charging the cells) and with Q1 off, there's no current flow across R1 for the LT1494 to sense and turn Q1 on (Catch 22). Enter R3 at 100K Ω. Once plugged into AC, a nominal current flows from battery + to battery - through R3 and across R1, thereby inducing LT1494 to energize Q1's gate and turn on the charger. Unfortunately, R3 remains when not plugged in and with the pack removed, so there's permanent, small parasitic drain. Left long enough, it'll eventually drain the cells to the point that the BMS intervenes and shuts down so if someone has a better idea, I'd love to hear it! That said, it'll take nearly 5 years to drain with a 100K Ω R3, so perhaps not a material concern
   

Here's a shot of the circuit above on a breadboard doing what it's supposed to do.



And last but not least, here's the finished article with the steering circuit crammed onto a small perf-board between the charger / boost converter sandwich (left) and BMS (right)...


 

The battery my 180c had in it when received was an aftermarket pack that just screwed together, which was really quite convenient! I'm using 8 x 800mAh 14500 3.7vdc Li-Ion cells (2 serial banks of 4 cells), which are a perfect lengthwise replacement for the OE NiCds, but leave a bit of space on top (hence the strips of rubber tape to keep things from rattling). If someone made a 16500 (16mm diameter vs. 14mm) cell with a bit more oomph, that would be just the thing! I suppose at 3.2Ah (800mAh *4) though, I shouldn't complain. I've been running the pack as pictured for a couple of weeks now with no issues. Happy to post a BOM if anyone is interested

 

[Fizzbinn]

Very impressive! Definitely interested in your BOM as well as thoughts from others on possible 3D printed cases. It would be very cool if there were a guide, instructions, a kit or a product for those of us that wanted a moDern solution to running these PowerBooks off battery. 

 

[sutekh]

Another week on with a bit more experience using the pack under my belt, I'm not quite satisfied. The voltage drop across the diode and R1 has the notebook complaining about low power. Apple's published figures for "charged voltage" are just wrong. I think they must have meant nominal. I've replaced D1 with an ideal diode, but the relationship between R1 and R3 isn't optimal. If I decrease the resistance of R1, the op-amp doesn't detect the nominal current flowing through R3 unless I decrease its resistance to the point that I'm wasting a material amount of energy at all times.

I'd hoped to keep this mostly analog, but I'm going to try replacing R3 with a Microchip PIC10f200 microcontroller flashed with a few lines of assembly to sleep for 15s, drive a 1Kohm resistive load across R1 for 100ms, and repeat indefinitely. That'll run for years on 3500mAh, and coupled with the ideal diode, should get the charged pack voltage up to around 7vdc. If that works, I'll probably CAD up a little PCB, as it's getting too cramped in there to keep using large, through-hole components...

 

[360alaska]

Maybe you can use a current monitor IC and a diode?

 

[sutekh]

Maybe you can use a current monitor IC and a diode?

Yes! I'm using an LT1494 to that end (see above schematic). Issue is that in order to keep the voltage drop low across the sense resistor, I've reduced it to .01ohm, at which point I have to push a fair amount of current through R3 for the opamp to detect it. As originally conceived, that parasitic drain is continuous.

In what is becoming the most ridiculously overwrought thing I've built in a while, tonight I replaced R3 with an itty bitty microcontroller board running a bit of PIC assembly I cobbled to wait 10s, then ground a 1Kohm resistor for 100ms, on permanent repeat. It means it can take up to 10s to detect the power cord has been plugged in and start charging, but that's NBD.



That little board (gray leads were for programming and then removed) is situated top right in the pic below and the new ideal diode is the little board with the heat sink.



Pack voltage is now up over a volt to 7.2vdc, and all is well and good for a one off, but overall I'm left feeling it's needlessly complex and too hard to build. I'd like to come up with something more repeatable and accessible to others. At a minimum, I may design a single board with SMD components to replace the 3 separate steering, ideal diode, and now microcontroller PCBs, which will also get rid of a bunch of interconnect wiring.

 

[360alaska]

I mean how about something like this:

https://www.digikey.com/en/products/detail/allegro-microsystems/ACS758LCB-100B-PFF-T/2042746

Using a diode so it will only register current in one direction.

 

[sutekh]

Well, I couldn't leave this well enough alone. The first battery I built described above is only 3200mAh, and my 180c is a power thirsty beast. I also wanted another pack so my 180 could get in on the fun. Here's my second effort resplendent in all of its 3D printed housing and 7000mAh glory:







One major difference and simplification from version 1 is the inclusion of a switch to toggle between charge and discharge mode. Perhaps not the most elegant solution, but effective and dead simple, so none of the complex and still not quite perfect steering electronics from v1 are required. The below schematic is functionally accurate aside from the fact that the 2S 8 x 14500 cell network is now a 2S 4 x 18650.



I doctored the original battery's lock tab to include a charge LED window and a small cutout to access the switch:



Here's a post in the 3D-Printed-Objects thread over in hacks with more info on the enclosure:

 

[Trash80toHP_Mini]

Fab! Love the work you're doing in bringing newer battery tech to older PowerBooks.

Suggestion: have a mod move your battery hacking threads to that topic. Hacks more easily discovered or searched out there than when they're buried in general question/answer discussion within model topics. Cross pollination happens more readily in Hacks and Development for hybridization as well. 

 

[sutekh]

Suggestion: have a mod move your battery hacking threads to that topic. Hacks more easily discovered or searched out there than when they're buried in general question/answer discussion within model topics. Cross pollination happens more readily in Hacks and Development for hybridization as well. 

Good call out. I've struggled a bit with whether to post something like this here or there. My impression has been that, especially with something very platform specific like this, it's more likely to reach its target audience posted in that platform's specific forum.  For instance, I put my PowerBook wifi modem replacement thread in hacks, and and it didn't get much attention until I cross-posed about it here.

 

[PotatoFi]

Agree, things don't seem to get much traction in hacks. This is very specific to PowerBooks, so I think it makes sense for it to live here.

 

[pcamen]

I want I want.  @sutekh any chance you'd ever produce these and sell them? 

 

[Fizzbinn]

I want I want.  @sutekh any chance you'd ever produce these and sell them? 

Same!

 

[sutekh]

I want I want.  @sutekh any chance you'd ever produce these and sell them? 

I'd consider it, but I don't think the current design is necessarily production / consumer ready. E.g., forget to flip the switch from discharge to charge after plugging in, bad things could happen. I'm also not sure I'm comfortable accepting the potential liability of selling lithium-based batteries for use in an old, unpredictable platform. My approach thus far has been to expose possibilities and provide as much design info as possible to hopefully inspire others to follow suit if interested.

That said, I am hoping to take a stab at creating a 3rd version with a custom PCB that re-introduces the steering circuitry, obviates the need for a boost converter (by charging the cells independently similar to what @360alaska is doing with his design) and integrates the CC/CV charger. Once done, you'd just need an off-the-shelf BMS, the custom board (which I certainly could sell), to bring-your-own 18650 cells, and an enclosure. I could even, perhaps, package all that together. Enclosure, BMS, charge/steering board, wiring, etc. Just purchase cells of your choosing and solder them in...

Incidentally, @maceffects reached out in my wifi modem thread over in hacks and expressed interest in injection molding replacement battery enclosures. Perhaps a partnership opportunity?

 

[maceffects]

Lithium is an interesting option.  I think if we made the plastic in such away to make the cells user replaceable they could provide their cells and thus greatly reduce the liability.  I can have both injection molded parts and stamped metal contacts made.  What I lack is any practical electrical experience.  It has been my dream to see old PowerBooks become mobile again though!

 

[sutekh]

It has been my dream to see old PowerBooks become mobile again though!

Mine too! Awful video and worse audio, but you get the idea
 

 

[PotatoFi]

Hey @sutekh, could the electronics on the battery pack behave a bit more like the original battery if you did a run of custom PCB's? I did a run of them for a conference badge last spring, and found it to be pretty straightforward. I'm not sure if that would get all of the functionality you want without having several boards packed in there.

 

[sutekh]

Hey @sutekh, could the electronics on the battery pack behave a bit more like the original battery if you did a run of custom PCB's?

Yes, the goal would be a direct OE replacement without any caveats but much higher capacity.

 

[sutekh]

Lithium is an interesting option.  I think if we made the plastic in such away to make the cells user replaceable they could provide their cells and thus greatly reduce the liability.  I can have both injection molded parts and stamped metal contacts made.  What I lack is any practical electrical experience.  It has been my dream to see old PowerBooks become mobile again though!

Thinking a bit more about this, it'd be neat to combine your injection molding and stamped metal contact capabilities and design a battery enclosure that allowed the end-user to drop in cells of their choosing without soldering. It's altogether too easy to short out or over-heat the cells while soldering everything together. I'm imagining an enclosure with the contacts and charge / logic board pre-wired--I supply the boards and wiring, you the enclosure and contacts. The end-user simply inserts their cells (minding the polarity of course) and clips or screws the lid on depending on how its designed.