Even Newer 4 to 24-cell Battery Management System (BMS)

deVries said:
GGoodrum said:
I also have the new LVC/HVC/parallel adapter boards, with the active cutoff module, coming on Friday as well.
Super! :)

What can the LVC LiPo voltage(s) be set at w/o using the CellLogs? 3.3v or 3.4 or ?

Thanks :mrgreen:
Well, usually, when I don't get a reply it means I totally don't understand something... so I go research it a bit more... :wink:

Correct me if I'm wrong Gary...

Your LVC boards are "hard set" to cut-out at 3.0v. There is no tweaking the voltage to other settings such as 3.2 or 3.3v, so there is really no way to take advantage of using a cell-log with your LVC board and set it at 3.2v or 3.3v volts (except by using the BMS-balancer onboard while riding w/CellLog AND alarm too).

Only Knoxie's mod for the CellLogs will do this w/o your balancer along for the ride too, click-here & read posts, which he borrowed some of your design ideas to do it.

Correct or ? :mrgreen:
 
deVries said:
Well, usually, when I don't get a reply it means I totally don't understand something... so I go research it a bit more... :wink:

Correct me if I'm wrong Gary...

Your LVC boards are "hard set" to cut-out at 3.0v. There is no tweaking the voltage to other settings such as 3.2 or 3.3v, so there is really no way to take advantage of using a cell-log with your LVC board and set it at 3.2v or 3.3v volts (except by using the BMS-balancer onboard while riding w/CellLog AND alarm too).

Only Knoxie's mod for the CellLogs will do this w/o your balancer along for the ride too, click-here & read posts, which he borrowed some of your design ideas to do it.

Correct or ? :mrgreen:

Sorry, I missed that post. Anyway, the LVC point is set to 3.00V for LIPo, and 2.10V for LiFePO4. The HVC trip point is set at about 4.30V for LiPo and 3.75V for LiFePO4. The HVC is simply used as a failsafe safety point, so the exact number is not critical. The same thing with LVC. What you are trying to do is catch the cell just as it is "jumping off the cliff", at the end of capacity. It doesn't really matter if the cell is at the edge of the cliff, or halfway down, as long as you catch it before it hits the bottom. I've played with these for three years now, and have tried all variant of the voltage detector chips, from 1.4V on up. With LiFePO4, what I've found is that with the LVC set at 2.1V, if I stop at the first time the LVC trips, there will be about 10% left in the "tank". If I then back off the throttle a bit, I can usually go a couple more miles before the LVC starts tripping with even a little throttle. With LiPo, 3.0V is roughly the equivalent trip point. Even if I was using CellLogs for the LVC function, I'd use the same numbers.

I got all the boards late today. I will ty to get them all up on my site this weekend.

-- Gary
 
I'm planning to rebuild a pair of 12s LiFePO4 packs into a single 24s pack. I charge it frequently away from home, so I think I'd likely want to put the HVC and balancer on the bike, probably the charger too, for that matter.

Is the "Full BMS" one of the boards that's available now (or soon)?
If not, which boards (and how many) would I need to get similar functionality?

-Nick
 
Sorry for the delays, but this time of year, my time is not my own. :roll:

Nick, if you want all functions, physically on the bike, I think the full BMS is what you will need. This is not quite available yet, but should not be too far behind the rest of the balancers, etc. The current issue is how we do the LVC function, opto-based, or with the "ladder" type implementation. It really depends on whether we include a option for active cutoff or not. If we stick with the simple "pull-down-the-throttle-signal" implementation, to remove the load, the opto-based circuit we've always used makes the most sense. We are looking at active cutoff options, but you can't simply cycle the full pack power on and off, as an LVC circuit trips, and then resets, as the controller won't be happy. Instead, we are looking at combo cutoff function that uses the pulldown method on the throttle, to handle cases when the LVC is tripping due to high loads, and then uses active cutoff only for the cases where a controller was left on, for instance, and the cells are slowly being drained down. Anyway, we are close to sorting this out, and as soon as we do I'll add it to the PCB layouts for the full BMS, and get them submitted.

I'm just about done with the instructions and the BOMs for all the different CellLog-based balancers. In addition to the 12-channel version there will be variants for 8, 16 and 18 channels. I hope to get this finished up before the end of today, or at least before the end of the weekend, and then make them available.

As I've mentioned, I'm also looking at adding an HVC capability back into the LVC/parallel adapter boards, to handle the case, primarily for LiPo setups, where the cells are still fairly well balanced, so just a bulk charge needs to be done. I'm really only comfortable doing this, however, as long as the HVC function is present, to act as a failsafe against damaged cells overcharging and running the risk of thermal runaway. For my first cut at this, I thought I'd kill two birds, and implement Randomly's active cutoff circuit, and make use of his matching LVC circuits. I did this, and added HVC to this as well, but there were a couple of hiccups. The problem is that Randomly's "ladder" based circuit that replaces the optos, requires that the first and last channels be different than the rest. I was able to do the 6-channel boards in a way that allowed any of these boards to be used as the first board in the string, the last board in the string, or one of the ones in between. To do this, jumpers are used, a couple parts added and a couple are omitted. If you are making one big pack, this is not really a problem, but it became a showstopper for my own setups. On my main bike, I have two identical 12s3p Turnigy packs, that are connected in series on the bike. I have a third 12s3p pack that will eventually become part of an identical 24s3p setup on my wife's bike. For now, I've been rotating the three packs on my bike. In any case, it isn't really practical for me to designate one 12s3p pack as the "first" one, and the second as the "last" one. I want to keep these identical.

The other "hiccup" I have is like I've mentioned above, I just don't think it is wise to cycle power on the controller when the LVC trips, so I want to wait until we have a suitable "dual-mode" active cutoff function, before making use of this on my own setups. In any case, I was back to square 1, in terms of still wanting to add an HVC capability for bulk charging without the CellLogs/balancers connected. I still have some of my previous LVC.HVC combo boards, so I could use those, but frankly I just don't feel like ripping my packs apart again, to swap out the LVC-only boards for the LVC/HVC variants. Instead, my laziness has generated an idea to do a very simple, HVC-only board. What I will do is uses a couple of these and attach them to one of the 14-pin VAL-U-LOK plugs that is the same as what I use with the 12-channel balancer. I'll shrink-wrap this assembly all together and simply plug it into the same connector coming from the pack I use when with the balancer. The other end of this will be the 2-wire HVC pigtail that will attach to the MeanWell charge controller (or to a standalone Charge Controller...). Here's what these boards look like:



These will work equally well with any of the existing LVC boards, and also with Geoff's new SMD-version of the same LVC circuit. For new setups, I'll also do an updated version of the combined LVC/HVC boards, both 6-channel and 8-channel variants, and I'll probably do an 8-channel version of the HVC-only board as well, just for completeness. Functionally, it makes no difference if the HVC is combined, or if it is separate. When balancing is required, the CellLog unit provides the HVC function.

Anyway, back to work... :roll: :)

-- Gary
 
I've been following Gary's design evolution (like the rest of us) with interest. It seems clear to me that, at least for the short term, LiPo packs developed for R/C application deliver the most "bang per buck" for e-bike use. The convenience factor in handling these packs also is important, as is their high power and energy density. However, managing a permanently-mounted multi-cell pack in a bike has its challenges, and Gary has been forthcoming with them as they appear.

Here are my requirements for a battery pack:

1) 2P18S configuration, to be permanently mounted in the bike, using the ubiquitous 5AH 6S LiPo packs from Hong Kong / Hobby King

2) Per (parallel) cell LVC that is always present, for both momentary cutoff during operation, as well as to prevent discharge damage if I leave something running. That is, both passive and active cutoff.

3) Per (parallel) cell HVC that is always present, to enable simple non-balanced charging (Gary's experience here has convinced me)

4) Simple nightly plug-in charging, with cell balancing and monitoring function as needed


Clearly, Gary's experience (which he has so graciously shared with us in this forum) has influenced my requirements.

When I first started looking at the HK LiPo cells, I naively thought that the existing R/C balancing chargers and battery pack monitors could somehow provide these functions. But the more I dug into it, not to mention read this thread, I realized that those chargers were designed for single pack charging, and certainly not for serial pack charging.

I know that Gary's work is going this direction, and I'm grateful. But my question is, can I achieve my requirements with a configuration of the current systems available from TPPacks, or do I still need to wait?

Thanks again, Gary, for all the experience you share.

Wanders
 
Thanks for the kind words, but clearly at least half-credit needs to go to Richard, my partner-in-crime. :)

To answer your questions, the bottomline is we are very close. FWIW, my requirements exactly mirror yours. I have 24s3p setups on my larger bikes, and 18s2p on at least one of my folding bikes. I want to do "safe" bulk charging all of the time, and periodic balance charging, as needed. I also want the LVC function to pull down the throttle signal, for LVC trips under load, and also cut power completely if there is a slow drain on the pack.

The CellLog-based balancers are ready, except for completing the instructions and BOMs. I'm working on these today. For the pack-mounted portion, what is needed is a low-drain, combo LVC/HVC/parallel adapter board, and a "smart" active cutoff module. We had that, based on Randomly's circuit (see pages 23-25 of this thread...), but it had one show-stopping flaw for me, which is that is isn't very modular. Richard has come up with a new solution, which appears to solve these remaining issues, so I will be doing a new set of LVC/HVC/parallel adapter boards and a new active cutoff module, this coming week. In the meantime, I will continue to use the LVC-only boards that are already in my packs, and will use the "add-on" HVC-only modules, shown above, that will plug into the same plugs used by the CellLog-based balancers. I'm getting some of these HVC-only modules made now. Once the new LVC/HVC boards are ready, along with the new "smart" active cutoff module, I'll then switch out my LVC-only packs.

-- Gary
 
Glad to see you are stil beavering away Gary. This thread should be called "the Zen of BMS development". Like wanders, I find watching the process of development fascinating (and also frustrating in approximately equal measure). I was about to ask a dumb question about why you prefered this split-out design over a permanently on-baord all-in-one solution, but looked back in the thread a little instead. http://endless-sphere.com/forums/viewtopic.php?p=330520#p330520 is an excellent explanation of the whys and wherefores for anyone who is lost in all these different boards and possibilities. Read that, then ask questions if you are still confused.

So there I was getting really excited that finally everything I needed (8S versions of LVC and charger/HVC) was going to be ready within days, but now you tantalise me with the idea that even cooler stuff is coming along just behind: A combined low-current LVC/HVC with storage cutoff to live on the bike. To wait or not to wait... (and there's geoffs SMD LVC boards too just to complicate matters)

And, as I think I've mentioned previusly, what I _really_ want is an all-in-one set of stuff that can live permanently on the bike. I guess this is where scooter owners have slightly different emphasis from ebike owners: there is more space to hide extra stuff like the heat-generating balancer board on a scooter - it doesn't have to live right in the pack. But that requires the balancer board to be either sufficiently low current or self-disconnecting, and I'm not sure that's a feature yet, nor is it clear when that might become one. I wonder if a manual switch to disconnect the balancer but leave it fitted is a practical option? Does that need to happen on all 16 channels or just on the cellog supplies in order to get standby down low enough? In practice onboard LVC/HVC plus external balance charging at home only will work just fine, so I guess that's what I'll do for now.

A couple of other minor things I notice:
1) The meanwell add-on board sticks up rather if you only have one meanwell supply (not the two in Gary's example pic). This refinement would make it rather slicker. (Is there in fact room to fit it inside the box to keep it out of harms way?
2) In all this rather nifty engineering I've not seen any mention of actually turning the charger off completely - i.e disconnecting it from the mains. Mine uses 6W when it's finished and doing nothing. That puts it outside the forthcoming EU regs for PSUs in standby (1W), and more importantly, outside my own limits for 'things that get left on'. What would be really cool was that once the charge-controller told the meanwell to shut down due to charging being completed it would disconnect the mains end, with a triac or a relay, resulting in no standby current.
Clearly this is a refinement rather than a major requirement, but it would be nice. Are there technical reasons why that is difficult or could we (you :) do it?

Looking forward to the next exciting installment (is this the longest endless-sphere thread yet BTW?)
 
wanders said:
When I first started looking at the HK LiPo cells, I naively thought that the existing R/C balancing chargers and battery pack monitors could somehow provide these functions. But the more I dug into it, not to mention read this thread, I realized that those chargers were designed for single pack charging, and certainly not for serial pack charging.

Hmmm, well I've rethought this statement. Now, there is no doubt that the work that Gary and Richard have done will yield a more optimal (e.g. faster charging and balancing) solution. But for off-the-shelf stuff sourced from Hobby King / Hong Kong, as well as the LVC boards from TTPacks.com, how about this solution?

View attachment Battery Setup.jpg
If I understand the way these packs and charger/balancers work, they can simply be daisy-chained, as long as you are careful to avoid a ground loop. Or is there something that I don't understand about these devices? Note that the laptop power supplies that I would use have simple AC cords (no ground wire). I don't show the connectors that must go between the LVC boards, the battery pack power terminals and the charger/balancers, but at least I indicate their general area.

Is there any reason that this would not work? Or is this just the somewhat obvious way the people set up their packs?

Regards to all,

Wanders
 
Actually, that would work, but only if the 16V laptop supplies have isolated inputs and outputs. If there is a common ground between the input and output sections of the supply, magic smoke will ensue. :eek: :roll:

I have finally updated the instructions and the various BOM files for the 8, 12, 16 and 18-channel versions of the CellLog-based balancers, and have made all of the versions available now. The packages include the balancer PCBs, the custom lids/end plates for the Hammond extruded aluminum cases, the special 7 or 9-wire JST-XH pigtails for connecting the CellLog units and the special 2-wire pigtails used for the HVC opto signal. The only other parts that are "non-Mouser" are the nuts and M3x8 bolts for the shunt transistors. The rest of the parts can be automatically placed into a Mouser shopping cart, with the correct quantities, using the Mouser BOM Import Tool, and the formatted lists in BOM Word files.

Richard and I are still working on the low-power LVC/HVC/parallel adapter boards, with matching dual-mode LVC with active cutoff module. I do have the HVC-only boards coming early next week. I plan to use these as an interim solution for bulk charges.

Once we get the new VC/HVC/active cutoff working, I'll then finish up the full BMS options as well.

-- Gary
 
GGoodrum said:
Actually, that would work, but only if the 16V laptop supplies have isolated inputs and outputs. If there is a common ground between the input and output sections of the supply, magic smoke will ensue. :eek: :roll:

Yup, isolated supplies are the key. I gave up (magic) smoking 20 years ago. :wink: The laptop supplies (which are dirt cheap in the surplus market) are isolated. Each is about 60W (typically 16V @ 4A) and these would match up well with the HK ECO6 50W charger ($19 plus S&H). Should be able to charge a 667 WH (2P16S LiPo) pack from dead in ~4.5 hours, or a 1000 WH (3P16S) pack in ~7 hours.

So, I finally figured out my battery pack and can order the whole shebang in one order from HK, plus the LVC boards from TPPacks. Now, if I could just get all the LVC parts from a single supplier to avoid double S&H charges, it would be sweet.

Wanders
 
FWIW, there are a bunch of 90W 19V supplies out there for Compaq and HP laptops that can easily *become* isolated, but are not to start with. They run an actual ground wire, even colored green/yellow-stripe, from the ground pin on the AC input receptacle to the ground plane at the DC output thru-holes. Wire runs right across teh top of all the component side of the board just inside the plastic case. Snip the wire and now it's isolated. ;)
 
Can you tell me How do you isolate them in an easy way wolf, with off the shelf parts
thx pal

amberwolf said:
FWIW, there are a bunch of 90W 19V supplies out there for Compaq and HP laptops that can easily *become* isolated, but are not to start with. They run an actual ground wire, even colored green/yellow-stripe, from the ground pin on the AC input receptacle to the ground plane at the DC output thru-holes. Wire runs right across teh top of all the component side of the board just inside the plastic case. Snip the wire and now it's isolated. ;)
 
The easiest way is to break the ground pins off of the power cables.

Just make sure they don't get used for anything else, as it's a pretty major safety hazard with some equipment.
 
I totally agree. If the equipment is not "double insulated" ie. isolated from ground in its design you can be creating a major hazard if something inside goes wrong. ie. any metal external parts could become live.
 
novembersierra28 said:
Can you tell me How do you isolate them in an easy way wolf, with off the shelf parts
thx pal

amberwolf said:
FWIW, there are a bunch of 90W 19V supplies out there for Compaq and HP laptops that can easily *become* isolated, but are not to start with. They run an actual ground wire, even colored green/yellow-stripe, from the ground pin on the AC input receptacle to the ground plane at the DC output thru-holes. Wire runs right across teh top of all the component side of the board just inside the plastic case. Snip the wire and now it's isolated. ;)
 
I've replaced the LVC-only boards on my site with the LVC/HVC version: http://www.tppacks.com/products.asp?cat=26. For those wanting LVC-only still, there are a few parts per channel that can be omitted. Another option, for those who want LVC-only is to get the pre-assembled/tested version that Geoff is offering.

The purpose of doing this is so that bulk-charging can be done, without balancing, but still having the "safety valve" of the HVC charge current limiting that keeps any cell from exceeding the HVC trip point. The LiPo version of this sets the HVC at about 4.21V, and for LiFePO4 it is set at around 3.72V.

Still coming is a new "smart" active cutoff module that will pull down the throttle signal, for LVC trips under load, and then will actively cut pack power for cases where a "trickle" charge slowly drains a cell, or cells, down to the LVC trip point (3.0V for LiPo and 2.1V for LiFePO4...), such as the case where a controller was left on. After I get this module going, I will go back to finishing up the full BMS board, which will now have the smart active cutoff module as an integrated option.

-- Gary
 
Can we safely use a cheap, non lipo ($65) bulk charger like i posted below with your LVC/HVC? We would not leave it charging all night long ofcourse.

My setup is a 16s lipo. I also have the 208B icharger for balancing when needed.
 

Attachments

  • 60 volt charger.gif
    60 volt charger.gif
    115.5 KB · Views: 2,717
If it puts out the right voltage, and/or has an adjustable voltage trim pot, it should work fine. If you have a 16s LiPo setup ideally you would want to have the bulk charger set to 4.15V x 16, or 66.4V. The HVC will keep the cells from going over 4.21-4.22V, so you want the charge voltage to be a bit under that. The consensus these days is that it is easier on the cells to charge to something around 4.15V, instead of the "max" of 4.2V per cell, like the RC chargers do.

-- Gary
 
GGoodrum said:
If it puts out the right voltage, and/or has an adjustable voltage trim pot, it should work fine. If you have a 16s LiPo setup ideally you would want to have the bulk charger set to 4.15V x 16, or 66.4V. The HVC will keep the cells from going over 4.21-4.22V, so you want the charge voltage to be a bit under that. The consensus these days is that it is easier on the cells to charge to something around 4.15V, instead of the "max" of 4.2V per cell, like the RC chargers do.

-- Gary

Great news, thanks Gary. The charger is from "TNC scooters" and charges to 68volts, with no trim pots inside, I'd like to use it for emergency charging on longer distances, it only weighs couple ounces, easy to carry on the bike. Thanks again for the help.
 
In the never ending quest for a better BMS, I came up with an alternative to using optocouplers for the output.
I love optocouplers and they have many advantages, but there are some drawbacks:

If driven by a TC54 or similar chip for LVC protection, the pack drain will go from a few microampres to around 1ma when the pack hits the LVC level. This will further drain the cell, which happens at a fairly slow rate. When the cell gets down to around 1.4v, the TC54 gets into the "Undefined Zone", which is their way of saying it doesn't work anymore because the volatge is too low. Eventually, the drain would become near zero when voltage drops below about 1v, as the opto emitter takes that much to light up.

Nomally this wouldn't be a big deal, but if you had one weak cell or left the pack for a long time without charging, there is a chance the low cell could get below the 1v point and the opto could no longer signal a low voltage condition and allow full power to be discharged through the weak cell, reversing it, and causing cell death.

Due to this weakness, there have been other signalling methods suggested, like the transistor ladder circuit proposed by Randomly. A similar approach is used on some commercial setups, including the Seiko Instruments BMS chips. While Randomly's setup was elegant and simple, it was a bit challenging to apply in a modular fashion. It also had some limitations in the number of series cells it could support.

After a bunch of thinking, I came up with a scheme that combines most of the desirable features of the above systems.

First of all, the power consumption in the active state is on the order of 10uA, which is way less than the self discharge rate of most cells. The system is modular, that is each cell circuit is the same. It uses a single wire between cells, if the wire breaks or goes open, the LVC will be activated. When the cell gets low enough that the TC54 goes into the twilight zone, there will still be a narrow range where the LVC won't be asserted, but it will once again activate when the cell voltage drops below around 1v, protecting it from reversal/destruction.

Each cell circuit has what amounts to a logical "AND" circuit with an open collector PNP output. If the TC54 is high AND the cell above is high, the output is high. If either the cell above or the TC54 goes low, the ouput goes open and the signal cascades like a row of dominos to the bottom cell where this is detected. Since each transistor 'sees' only two cells worth of voltage, there should be no limit to the number of cells that could be combined using this scheme. Each circuit is identical but there needs to be a jumper on the top cell.

By using higher gain transistors and bigger resistor values, the drain could be reduced even further, but there would be the risk of false triggering by noise. Based on my breadboard test, I added a small capacitor to each one to give some noise immunity. A larger cap could be used but will increase the delay time, which could be interesting in a long string.

The same cascade setup could be used for signalling HVC as well, either combined or separately from the LVC. HVC can still use optocouplers as well, since power consumption is not an issue for HVC.

Here's an example of how it could be applied to a LVC circuit (only 3 cells shown):
Cascaded LVC concept.jpg
 
fechter said:
In the never ending quest for a better BMS, I came up with an alternative to using optocouplers for the output.
I love optocouplers and they have many advantages, but there are some drawbacks:

If driven by a TC54 or similar chip for LVC protection, the pack drain will go from a few microampres to around 1ma when the pack hits the LVC level. This will further drain the cell, which happens at a fairly slow rate. When the cell gets down to around 1.4v, the TC54 gets into the "Undefined Zone", which is their way of saying it doesn't work anymore because the volatge is too low. Eventually, the drain would become near zero when voltage drops below about 1v, as the opto emitter takes that much to light up.

Nomally this wouldn't be a big deal, but if you had one weak cell or left the pack for a long time without charging, there is a chance the low cell could get below the 1v point and the opto could no longer signal a low voltage condition and allow full power to be discharged through the weak cell, reversing it, and causing cell death.

Well this actually happen to me. The cell did not die from draining, but it actually somehow just became a shortcut, showing no voltage at all but conducting current like a busbar. The 2.6 version did not see this fault, it was first at closer inspection with the Cellog that I found this faulty cell.
I know another guy who hade the same fault with a cell of the same type(not the same batch or factory though.) It was a TS LFP 160 ah cell that just turned into a very expensive busbar.

Since that happend I have been running BMS free. But its nice to here that you have been thinking about this problem. And that you have a solution for it. This ladder approach isn't that pretty much what Dimitri did in his Mini BMS ?

Best Regards
/Per
 
Dimitri's ladder is also using optos, but in the opposite logic level. The optos stay on all the time, until something trips. That will drain the pack at a rate about 100 times greater than the low power transistor ladder approach.

What needs to go with this, in order to prevent killing cells from leaving a controller on, for instance, is the "smart" active cutoff module we are also working on. With this, the throttle signal is still pulled down, for LVC trips under load, but it will cut power if the voltage slowly drops.
 
Sorry !

I got it mixed up with something else then. Actually did not even bother to go thruo the whole thread of miniBMS to find out again.
But your approach sounds right for keeping the BMS from doing damage to the Cells instead of slowly killing them.

Regards
/Per
 
GGoodrum said:
Dimitri's ladder is also using optos, but in the opposite logic level. The optos stay on all the time, until something trips. That will drain the pack at a rate about 100 times greater than the low power transistor ladder approach.

Actually, the drain would be nearly 1000x lower than Dimitri's setup since he was running a LED as well.
If any cell goes too low, it will break the chain and trigger the LVC signal.
It would be possible to use this as a combined HVC/LVC trigger like Dimitri's, but there are some advantages to keeping those two separate.
 
Back
Top