Flying pack topology

My idea of a basic flying cell/pack balancing system...
It is kind of designed for my setup, needs to be able to handle high(er) currents.
I want it to havea bit intelligence build in, to detect low loads on the system so itknow when to startbalance (only at night?)
In my mind, the switching frequence has to be slow on a bigger battery, even very slow, something like 15min/step.

The idea is to get a Arduino cycle the packs, messure and log voltage, and connect the high and low cel in turns to a flying cell...
Currentsto and from the flying cell can be messured, if no current flows ==> balancing between static and flying packis done ==>next step....

Like to know if there is somebody doing this kind of balancing on a system in use? or is this not going to work and a big waste of time??
Planningto use relays because i like the clicking of them (and easy to isolate)

I have the Batrium BMS, so adding a few longmons is a option also, but that isway to easy... i wanne haveit complicated this time.....

You wouldn't need the voltage divider as your packs are all parallel and thereby will only get to a max of 4.2V. So you can read the voltages directly.
Maybe the flying pack should be a bit beefier, maybe 50p. You don't want the cells in that pack getting hit too hard by current flow.

You could even build a solid state relay. But, u said u wanted the clicking noises.

Make sure there is at least 500ms delay between 1 relay powering off and the other one powering on. You should be able to get DT relays instead of having to use 2 relays per pack. This actually could work out since you could put the relays for each pack on the pack. Then you'd only need to run 3 wires to each pack (1 Pos, 1 Grd, 1 Relay trigger; Grd is shared at the relay and flying pack)

I have a fewArduino mega's laying around, these have 3,3v adc's.... so i gonne need the voltage divider.
In my drawing i have already 1 DT (if thismeansdouble pole) relay drawn per pack, and i insist on those clicks ... maybe even some leds blinking
The ss relays which handle 50Aare rather big and expencive, and i need two of them in parallel to make a double pole, i think, have to google them a bit.

And yep, delays are a must, i like delays, they give me time to think.

Here's a quick and easy solid state relay build: http://www.instructables.com/id/DIY-Solid-State-Relay-1/

And here's a quick diagram of one:

wim said:

In my drawing i have already 1 DT (if thismeansdouble pole) relay drawn per pack

Click to expand...

Usually DP means Double Pole andDT Double Throw, DPDT both combined.

But if you change your drawing to DPDT-relays, it becomes much more elegant. Connect the first(lets say top one) DPDT relay so that it'sNO pins connect to first main/static cell (+ and -), and NC pins to next relay (which connects to next static cell) in chain and so on till end of the pack. This way you cannot by accident ever parallel the main cells, if you accidentally switch two relays there won't be a current path between the main pack cells.

You'll also need some more fuses. I would add a fuse to every output from main pack cells, and you'll most likely would want them in both positive and negative terminals. Don't take it wrong but your current drawing setup is bit of a time bomb if that arduino ever switches wrong. There will be smoke if say first and last relays are both on at the same time.

But as you see, you'll end up needing same number of switching elements and same amount of control with one flying cells for the whole pack vs one flying cell for each main pack cell. Also the balancing will be much slower and the flying cell is not really contributing in pack capacity positively. It is actually also making things a bit worse in some cases as when it is connected to one cell that cell will charge and discharge out of sync with rest of the pack causing imbalance which this whole exercise was trying to avoid. So I don't think it makes sense versus more flying cells.

---

Korishan said:

And here's a quick diagram of one:

Click to expand...

In my limited understanding themain disadvantage in using a triac to switch DC loads is inability to switch it off after switching on, but I might be wrong.

Actually it occurred to me, that even using single mosfet like in original thread is unclear if it fully works as intended. See every mosfet has a body diode, so they will always conduct power to one direction even when switched off. With static and flying cells higher voltage can be either size, so I feel double opposed mosfets are needed for every terminal. This means 2 mosfets for every connection, for two current paths, for each flying cell terminal, so 8 mosfets per flying cell which sounds pretty crazy, expensive and complex compared to one DPDT relay. Off course it permits fast switching speeds, but I am not sure if it is worth the complexity. Plus ticking off relays is very nice and comforting sound in retro kind of way.

I follow this idea closely. Why have multiple flying packs?

I was going to suggest this on the original discussion.

However. My idea (though I have no idea how to implement it) would be to bridge each pack using some kind of isolated DC setup.

I have thought about this a lot. However I have not come to any sensible and practical way to implement it.

Geek said:

I follow this idea closely. Why have multiple flying packs?

However. My idea (though I have no idea how to implement it) would be to bridge each pack using some kind of isolated DC setup.

Click to expand...

Having multiple flying cells, preferably same number as main pack has, leads to flying cells being connected in parallel to main pack most of the time (like 99%) and this way contributing positively to main pack capacity. That way their capacity is not lost nor wasted, so there is no problem making them pretty big say 20-50% of the main pack capacity. If we do not have same number of flying cells and doing the same, the pack will be unbalanced as different levels will have different capacity and when charging discharging imbalance will naturally result.

I have been thinking other options, probably much easier than having isolated DC setup would be just have one AC-DC charger per cell (I mean cell level, however many parallel individual cells that is)controlled externally as needed by voltage levels sensed. AC is trivial to isolate compared to DC, and most of the powerwall type of packs have AC output already, also called an inverter.

noaano said:

Usually DP means Double Pole andDT Double Throw...
... Connect the first(lets say top one) DPDT relay so that it'sNO pins connect to first main/static cell (+ and -) ...
You'll also need some more fuses...
...
... themain disadvantage in using a triac to switch DC loads is inability to switch it off after switching on ...

Click to expand...

- DP and DT ... did not know that, learned something, Thanks
- More fuses... yes i know, just did not think of drawing them ... i can havethe arduino even detect a broken one...
- A triac is for AC use, a Mosfet is used for DC and has no problem switching high currents on or off... just not easy to isolate, needs a optocoupler.

noaano said:

Having multiple flying cells, preferably same number as main pack has ...
...have one AC-DC charger per cell ...

Click to expand...

Noaano, this is the reason i have started a new thread on this topic, Brians idea of a kind of flying string of packs is, in my mind, not usable in a bigger and high current setup. But very smart and efficient onsmaller systems... https://secondlifestorage.com/t-18650-equaliser

If the flying cells (packs) are connected (your 99%) to the satic cellswhen the battery is in use, the relays, connections and wiring has to be able to carry the loads, in some cases (like mine) 50-60 A... only then there will be the gain in capacity where you are talking about.
Those heavy wiring is not easy at allto install on a 14s setup like mine, and will be quite expensive.
If you "disconnect" the flying cells when the system is under load, wiring can be less heavy butthere will be no gain in capacity... you are only beshifting it.

To do balancing with a single flying pack, the idea is to change the capacity of the packsactive...
Connect the flying packto the lowest static packwhen discharging and on the highest cell when charging, this way you can bring the packs closer together on voltage.
It is possible to have the balaning done while the system is in idle ( ieless than 20A draw or charge current )
I aggree, there will be no capacity gain... but there is also only just 1 flying pack needed.
I like to see it moreas a add-on to a real BMS, to prevent to much passive balancing when the cells get older.

I realy think this is different as Brians idea, i do not claim to be the first to think about a flying cell balancer, lots of examples out there, just wanne see if there are others on the forum (or elswhere) using it on a powerwall type of system.

A AC/DC charger for each pack, is again something different ... i am not sure about this idea ... too much losses in converting from DC to AC and back to DC.

wim said:

- DP and DT ... did not know that, learned something, Thanks

- A triac is for AC use, a Mosfet is used for DC and has no problem switching high currents on or off... just not easy to isolate, needs a optocoupler.

If the flying cells (packs) are connected (your 99%) to the satic cellswhen the battery is in use, the relays, connections and wiring has to be able to carry the loads, in some cases (like mine) 50-60 A... only then there will be the gain in capacity where you are talking about.
Those heavy wiring is not easy at allto install on a 14s setup like mine, and will be quite expensive.
If you "disconnect" the flying cells when the system is under load, wiring can be less heavy butthere will be no gain in capacity... you are only beshifting it.

A AC/DC charger for each pack, is again something different ... i am not sure about this idea ... too much losses in converting from DC to AC and back to DC.

Click to expand...

- If you go with one flying cell, try drawing your diagram with chained DPDT-relays, flying cell to common terminal, static cell to NO-terminal and next relay chained to NC-terminal and repeat this till you reach end of the pack. You'll see it eliminates a lot of error possibilities. I wish I had a decent drawing program, but I do not currently.

- Mosfet only blocks current one way, and needs a driver to work properly. Relays are much easier and cheaper.

- If you worry about current, size the flying cells accordingly to be small enough. More flying cells wont matter, as they will all "see" the same current, with even one cell you have same problem, just much longer wiring that still has to be sized the same.

Say you need 150A max, 10% of this is only 15A, no biggie. Almost any relay will handle this fine, 2.5mm wire is enough (1.5mm also is enough, as this is not professional installation). Plus if you have flying cells only connecting to two static cells, wiring can be kept pretty short. I just don't see any problem in doing this, and even 50A is easy with cheap and reliable automotive relays, and we don't need to go that high. Of course disconnecting balancing while high draw solves this, but still I don't see it as a major problem. Maybe in EV with 1000A currents, but not inpowerwall use.

-AC/DC -losses might be high in percentage, but overall energy needed to balanceis so low, that it ends up being pretty insignificant. When not balancing, AC-DC chargers are not using anything. Inverter is still needed right?

- Windows paint rocks ...
- Relay's rule... (no mosfets for me, i am not smart enough for those)
- Currents are a big concern for me, beleave me, they are the only concern.
- It is 250A - 270A max, and you have to be able to handle atleast 25% of that ( 3 strings static packs and 1 flying pack ) 260A/4 = 65A ... will need a 16mm2 wire... possible yes, but for sure not easy.
Might be possible if you can design asetup with flying packs before building it, as saydmy setup already is build... sotrying to find the next best thing.

The ac/dc thing you have to explain a bit more... not sure what you have in mind.

wim said:

65A ... will need a 16mm2 wire...

Click to expand...

short note,for 65A 10mm2 should be just fine. It is even officially allowed for 230 AC permanent installations with wiring exposed (ie. no hidden wiring inside walls etc) with current up to77A in Europe.

And AC/DC was just the idea of having a inverter AC driven 1 cell charger per cell to do the balancing, sure the losses are high in percentage but in actual Joules/Wh minimal as we are talking about mostly balanced pack, right?

- 10mm2 or 16mm2... did not checked, but a bit of overhead might be a good thing to have on DC.



A AC/DC on each pack, hmmm, still needs to be a isolated charger ac or not ... i remember a other thread on this topic ...https://secondlifestorage.com/t-Charging-Low-Capacity-Pack-Is-This-Safe&highlight=isolated+charger is this what you have in mind ?

if you gonne have losses, then passive balancing might be the simplest way to go.
Not sure what you mean with Joules/Wh ? if it is energy,it is Wh or Joules ... same thing.

wim said:

- Currents are a big concern for me, beleave me, they are the only concern.
- It is 250A - 270A max, and you have to be able to handle atleast 25% of that ( 3 strings static packs and 1 flying pack ) 260A/4 = 65A

Click to expand...

How often you plan to draw 250A? You could simply stop balancing when current >100A. Or have more than one balancing string in parallel. How many parallel cells was in your planned flying pack? If say 80p, split it to 4 x 20p and current drops accordingly? These separate flying cells could then be switched with different topologies, making balancing even more optimal. Just need more relays, they are dirt cheap and small ones even cheaper, and more relays always sounds better!

If it is often you draw 250A, you would benefit from higher voltage pack...

---

wim said:

10mm2 or 16mm2... did not checked, but a bit of overhead might be a good thing to have on DC.

A AC/DC on each pack, hmmm, still needs to be a isolated charger ac or not

Not sure what you mean with Joules/Wh ? if it is energy,it is Wh or Joules ... same thing.

Click to expand...

10mm2 should be fine and there is not much difference in AC/DC current if we are talking RMS values and well behaved loads (PF near 1).

AC or DC is the same, just that AC is a) about 1000x easier to isolate and b) already available from inverter right? Actually it leads to a new idea, most basic crude power supply with current control to balance the cell. Rectify AC at each cell level, even half bridge will do, maybe add a capacitor to smooth worst of the ripple out, chop and limit current with pwm output connected to mosfet with control coming from that Arduino with feedback sensing from cell's voltage level with Arduino ADC. Add a small resistor in series to act as shunt and monitor this also with same Arduino for precise current feedback as well... But this is a different idea altogether, lets think this flying cell balancing through first!

And yes, by Joules or Wh I meant energy, though technically Wh is a fair bit more than 1J

Idea of chained DPDT relays:



I only drew the positive terminal for clarity, but negative terminal would take another pole on the DPDT relays and be wired exactly the same.

This makes pretty sure flying cell on the right is only ever connected to one static cell on the left, or am I missing something obvious?

Unfortunately there is no MS paint in this Mac

noaano said:

How often you plan to draw 250A? You could simply stop balancing when current >100A. Or have more than one balancing string in parallel....

...10mm2 should be fine and there is not much difference in AC/DC current ...
...lets think this flying cell balancing through first!
...And yes, by Joules or Wh I meant energy, though technically Wh is a fair bit more than 1J

Click to expand...

- 250A will not happen often, but it is possible ... so everything on the load side has to be able to handle it. Stop balancing on loads > 20A ish was the plan.

- 10 mm2 should be ok, Yes...
- Flying cell thing first ... lets do it !!
- 1 Wh is about 3599Sec more than 1J


Now i get what you mean with a chained DPDT relaisstring, great idea, will add to safety a lot ... will update drawing.
I am sorry you have to do with a Mac, but stay positive, one day ....

wim said:

I am sorry you have to do with a Mac, but stay positive, one day ....

Click to expand...

I was gonna say get real PC (which, Apple has notoriously claimed they are not a PC, but different)

Korishan said:

I was gonna say get real PC

Click to expand...

Does my couple of Fedora/Debian boxes/laptops count as a real PC?They all can run indeed run Gates/Ballmer/Elop-creationsin KVM/Vbox if feeling really desperate...

To be honest difference is pretty minimal nowadays. Chrome looks pretty much the same on every platform. Of course months or even years of uptime is something that I would not easily give away.