Hacking my Ping's 48v bms to accept 15s

[swbluto]

I recently emailed the castle creations guys over the issue of my "48 volt battery" being 57ish volts off the charger and being used with their phoenix HV 85 amp 50 volt controller. I received this reply:

Max input is a strict 50.4v unfortunately. You'd have to get the pack to cut off at this voltage in order for the pack to work. Anything over this and your headed down a nasty path...could be immediate, could be 10 flights from now, but it will happen. I wouldn't do it. There are FETs on the market that will handle the higher voltages, but they're not quite up to par yet (and cost a small fortune). We've been testing with our SHV (90v) ESC...very hard to get a controller that handles over 50v to be totally consistent...why ours is not out yet. Basically, I wouldn't do it, and I work here.

Joe Ford
Product Specialist
Castle Creations




Andy wrote:
> Hello, I noticed that your HV phoenix controller was able to handle 50 volts which seemed like it would cover my nominally 48 volt lifepo4 battery. However, I realized that it's "hot off the charger" voltage is usually around 57 volts but its "working voltage" is usually around 47-48 volts or lower at 20 amps. Would this be a problem? If so, what component would be the limiting factor and would you happen to have a component number so I can check the specs? If it's the FETs for example, would I be able to manually upgrade them if needed?
>
> If it's an "Oh no, don't even go there" type of problem, I could possibly build a circuit to lower the charging voltage(the charger is not designed to be adjustable.) to lower it to 53.5 volts "hot off the charger". Would that still be a problem? The BMS kind of makes removing a few cells essentially unworkable.

So, it seems like it's a forced issue to drop a parallel cell group to drop the voltage by 3 volts or so and then I can effectively limit the battery's "hot off the charger voltage" to 50ish volts in conjunction with an external battery turn off circuit. But, the BMS is the issue as it's designed to work with 16s. It's a Ping 48V 1.0 battery with a 16s configuration and I have to repair a cell group in the middle of the pack. Is there a way to simply cut out the cell group and have the BMS function "correctly"(I.e., w/o tripping the LVC detector circuitry or frying something). Does anyone have a schematic that is or is pretty likely to be similar to Signalab's BMS? I'm going to try to hack this thing, but I'd like all the information I can get to prevent myself from ruining the BMS.

Or I might end up using an amplifed zener for the discharge. Uggggg... so much soldering to do. There are single component darlington BJT transistors that can handle >20 amps and >150 watts, right? I need me some of those, otherwise I'm soldering together of upwards of 10 NPN transistors(5 darlington pairs, since my high-amp ones don't have a high enough minimum gain for the zener diode at the base), which sounds like it'd be a PITA.

Hmmm... now I'm thinking and I imagine I could use the "darlington" transistor for when the current is at a lower current/higher voltage, and then switch it to a regular power mosfet when the running voltage is low enough. That'd minimize the amount of soldering and component count, while also maximizing energy efficiency during discharge. I'll have to code the microcontroller for this though(And I wonder if it would "switch fast enough" so the controller wouldn't be fried?), and I worry about other safety concerns such as having a fail-safe in case the mosfet would short out. Also, it seems like regular power Fets need a gate voltage of 10 volts for minimum channel resistance whereas my controller would put out ~5 volts, meaning I'd probably need a voltage-amp of some sort... that'd probably wouldn't minimize the component count as much as I hoped.

 

[swbluto]

OMG, the solution is (sort of) simple!

View attachment 2

The volt-meters left of the cells represents the voltage samplers that the BMS uses for LVC cutoff. The parallel cell groups are represented by the battery-icons with each having a no-load voltage of 3.3V and an internal resistance of 16 milli-ohm. To drop the voltage of the "cell group" that I'm seeking to replace while retaining the ability to use the battery with its BMS, I just shorted a path around the middle cell group as you can see in the picture. To minimize the resulting short-circuited current, I placed a 1,000,000 ohm resistor in front of the middle. As you can see, this doesn't trip the LVC as the sampled voltage stays at 3.3 volts while the total voltage of the battery drops by the amount you would expect if you just simply cut the cell group out of the battery.

Now onto the charging solution... Hmmmmm... I think I might just think of a way to use a switch to short circuit the resistor while "turning off" the other short circuit, so I can switch it between "charging" and "discharge". But, that would also charge the cell that isn't depleted... so maybe I'll just used an amplified zener for the charger to drop the charging voltage correspondingly, and the short circuit around the middle cell will stay in place. Well, I should model it first before just doing it.. :lol: (Hmmm... it seems I'm having some difficulty modeling a charger.)

So, now that I know(at least the model implies it) that this would work, do you think one cell group or two cell groups in series should be "dropped"? The Ping charger charges each cell to 3.7ish volts or 55.5 volts "hot off the charger" for 15s, or each cell will drop to around 3.4 volts or 51 total volts if left for some time or if slightly discharged - It can get down to 3.35ish volts each exactly, or 50.1 volts, if discharged enough(it's only, like, .1 AH at most). I can build a "quick load discharger" for the battery for just this purpose, if needed - I can't imagine it'd be hard or expensive. Or I could build a voltage-sensing turn-off circuit for the battery charger.

But dropping to effectively 14s would definitely put me "below the limit" but... it's less, I guess. And it entails more work.

(On a side note, I deduced that these streaks of blue dots that appear on my laptop's screen isn't a "connection problem" as a PrintScreen copy of the screen also copied the dots exactly. So it's either a "software problem"(Windows, maybe? Intel's chipset software?) or a not-easily-fixable hardware problem with the video memory, assuming "PrintScreen" might copy the video memory directly, however, that I don't know.)


EDIT:

View attachment BringDownTheVoltageSolutionProblem.JPG

My previous model was flawed. The voltage sampler uses each wire in the middle for both the cell behind's positive and the cell in-front's negative(or the inverse), so it was pretty bad of me to use the voltage sampling wire for the top cell to be placed near the top cell's negative past the 1M resistor - it'd actually be the same wire as the voltmeter's positive below it. And, correcting that, it appears there's still a problem. And I thought I had it solved too! Ahahahaha... more thinking needed.

Further edit: I just revised the model of the battery so now that it's more accurate. Each cell's "volt meter" actually share the neighboring "volt meter's" wires, as depicted. Any ideas that don't require modding the BMS?(Of course, BMS mod ideas are welcomed, as well, assuming it looks like it works and is reversible!)View attachment batteryModel.JPG

 

[swbluto]

Finally! It seems like no matter what I did, I couldn't find a way to alter the middle cells(where I'm repairing the cell group, soon) to get a volt-meter reading that wouldn't cause the battery to dysfunction(i.e., trip the LVC). So, I went to one of the ends, tried out this combination that came to mind, and it works! OMG... Finally! A Solution! And it seems as easy as moving the wire near the end to the cell behind it! OMG, soooooo easy! And my battery is repaired as usual. Of course, I'll still have to drop the charger voltage.



Now... I have to wonder what else the sensor wires are used for. I hope it's just for voltage-sampling because I didn't account for anything else.

 

[swbluto]

My wimpy 30W electric soldering iron wasn't doing the job(It sucked at soldering the 10 gauge wire!) so I searched the forums using the neat search feature and came upon Reid Welch's video and went to the Radio Shack downtown on his suggestion. As I arrived, I cruised along the soldering irons in their section and I was kind of disheartened to not see it: I reasoned, "Hey, Mr.Welch bought this thing five years ago. Maybe they took out due to discovered liability issues?", but I started browsing the product-price labels just in case they were out - I noticed one said "butane"(The first one I found) but yet the box was empty with a paper sign on front that read "see a assoc."(The only product with this apparent label) - I went to see the clerk, and they went in the back and retrieved the mythic item and it was it! I asked him about the sign and apparently it's a frequently stolen item - A soldering iron? C'mon. It turns out that it also functions as a blow torch, so I guess I can see some appeal there.

Took it home and, wow, it's a dream. It melts solder like no one's business and it even functioned that way on the lowest setting possible! Apparently its "low setting" is around 530 degrees Fahrenheit, so that's understandable. I got the wire soldered onto the correct pad, checked the battery's voltage, and it lowered by 3.4 volts as expected to within range! I also reasoned that the BMS cut out at 3.9 volts/cell as was my previous observation with its charging so I didn't think I would have to actually create a voltage dropper - It'd automatically stop charging when appropriate. I plugged it directly in using the newly created positive wire and it charged normally! So, now everything's a go! The only thing I want to do now is change out the positive wire(10 gauge home depot stranded wire) to a bit softer one as I'm finding it's actually quite stiff in comparison to the wire that was originally the positive wire leading to worries about yanking on the tabs that I'm quite worried about being brittle or easily torn... I think I'll try to locally source some "silicon" or soft wire tomorrow - Tomorrow I'll be heading towards the "Seattle Robotics Society" club meeting and there's a Fry's in the area, so I'll make sure to take advantage of the opportunity.

So, summing it up, the battery is a go and all I had to do was add a wire to reduce the voltage!

 

[nomad85]

I would like to try this as well. Your pretty diagrams are greek to me...Can you post some actual pictures of how to do this?

 

[swbluto]

Hmmm... It's now shortly after my move and I seem to be unable to find my SD card reader which allowed me to transfer pictures from the camera to the computer, so sorry about not providing pictures! I was wanting to originally, but it seems it still eludes me. I may have to order another one.

Basically, the Low-Voltage-Cutoff works by measuring the cell group voltages by the way of small sensor wires. If you rip apart your ping pack(take off the duct tape) - which I did so I could diagnose its problem of only providing 7ah out of its designated 10Ah and I still need to repair when I have time and the interest - you should see multiple colored skinny wires heading out the cells: these are the sensor wires. In the picture,the circle with a big V inside is the "voltmeter" used to represent the BMS's voltage detection system. The lines represent wires.. and, so on. As you can see, the last picture makes sure that the cells are measured to have the correct voltage(above the LVC cutoff) while also reducing the total battery voltage by the expected amount.

Anyways, practically speaking, it is possible to do what I did by desoldering the original positive cable(the red "power wire" that has a sensor wire attached to the same joint. The black negative power wire doesn't have a sensor wire soldered to its joint, so you know the microprocessor uses the black wire as the "ground" node, and you don't want to move that! And, you'd also have to install another sensor wire but the voltage would be incorrect for the BMS... so that wouldn't work.) and soldering it further up the series of cells - In this case, only one cell group was desired to eliminate, so basically it'd be moving the red wire to the other terminal of the cell group it was attached to. In actuality, I didn't want to desolder anything, so I just added a new "positive power wire" to where it should go and I'm disciplining myself to ignore the original red-wire - Of course, if the battery is fully charged, I can go back to the original red wire and use the battery as originally designed(With full voltage and all) which is what makes this solution so simple, elegant, efficient and functional.

But, I would caution about decreasing the voltage too much while leaving the charger unmodded - I haven't really tested it but it'd seem possible that the charger could be burnt out or something, but I'm rather ignorant about the charger itself - if it does current limiting by, say, switching means... than it isn't a problem(I think?) - but if it's linearly regulated, which I'd guess it is since it has a noisy fan and everything to dissipate heat(I don't know), then decreasing its output voltage would increase its heat which may exceed its original design. I wouldn't think a voltage drop upto 10 volts or so would be a problem(with the heat factor and everything - Don't know about the other parts), but don't sue me! :lol:

It's just that large leaps in changes from the original design tends to necessitate other changes in electronics. But, someone may get lucky.

 

[nomad85]

Still greek.. need pics(most cameras have a mini usb port, you could access the files that way ). Are you just moving one of the sensor wires to another cell group giving that group 2 sensor wires, sounds like you are using a different output positive wire.

 

[swbluto]

Still greek.. need pics(most cameras have a mini usb port, you could access the files that way ). Are you just moving one of the sensor wires to another cell group giving that group 2 sensor wires, sounds like you are using a different output positive wire.

My camera's mini-usb port is "special" I've found and I don't have the "special" connector, so sorry about that. I understand how this would be much more easily understandable if the notational symbols were known or if there were pictures, but it seems I'm not really helping in either regard.

Basically, a different output positive wire was used(I'm actually using speaker wire because it seems to lay much more flatter than the standard home-depot stranded wire). By "different", I mean I just soldered on a new positive wire where I wanted it to go on the battery(in a way that eliminated one "cell group"; Ping arranges his cells so they're parallel-ized in groups and then he strings those groups up in series, so I just "eliminated" one of the cell groups that were in series by "skipping" over it.).

And, no, I didn't even touch the sensor wires. As far my thinking goes, you definitely want to leave those in place so you don't inadvertently trip the LVC(which is much more likely than not what would happen - However, if you're more creative than I am, maybe there's a clever hack involving moving the sensor wires that doesn't remove unwanted functions?).

 

[nomad85]

If the cells you are skipping are problem cells that trip the LVC wont that still happen if they have their sensor wires? (eventually they will lose their charge right?)

 

[swbluto]

I should really get pictures.

My "problem cells" are still in there in the middle of the pack somewhere(they're not near the end) that'll still trip LVC at around 7 AH rather than 10 Ah like it would if operated as designed, and I've just left them in there. I'll replace them with brand new shiny cells at some later date when I feel like it, and that replacement will be identical so they're won't be any "mods"... just cell replacement.

This mod happened at the front of the pack, where the positive red wire attaches to the cells.

 

[swbluto]

I just ordered a new SD card reader so if you're willing to wait a week or so, you'll become totally sure of what's going on.

 

[nomad85]

Sweet I'll be looking forward to it

 

[swbluto]

To fully view the images as the right part gets cut off, you might have to right click the image and select "view image" or something similar.

This is what a ping battery looks like when you take off the duct tape. The wire with the cloth around it closest to the yellow-wrapped BMS is the negative wire, the thick red wire is the original positive wire in its original position.

View attachment Exposed battery.JPG

Here's a close up of the newly added wire - It's the clear speaker-wire. I chose speaker wire because it was easy to bend and wasn't so stiff it yanked at the cell's tabs under its own weight. Notice it's on the opposite side of the original red positive wire: It was done that way so that it'd skip the cells in between which lowered the total voltage by the amount of volts contributed by a cell group(~3.3-3.4 volts). By connecting the clear wire and the original black wire, I get a total voltage that's within limits. I also charge with this wire as charging with the original positive wire would cause the battery to turn off too early due to the undischarged cell group in the front(And its voltage would reach the voltage threshold much earlier than others.).

View attachment NewPosAdded.JPG

 

[nomad85]

Thanks for the pics I tried to work on mine, but its a lost cause, I cant solder the silver tabs... I am putting the good cells from my pack up for sale in the for sale section.

 

[swbluto]

Thanks for the pics I tried to work on mine, but its a lost cause, I cant solder the silver tabs... I am putting the good cells from my pack up for sale in the for sale section.

If the iron's not hot enough, just get a better one!(Butane ones seem to be much more effective than electrics for this heavy-gauge work)

In order to solder mine, I sand-papered away the the surface of the solder(it was dirty before, trust me) and cleaned it with isopropyl alcohol - solder doesn't work on dirty surfaces.