A "Virtual Battery" for Mobile Home usage

Cherry67

Member
Joined
May 13, 2018
Messages
518
I am in the process of building up a preliminary version of the "Battery enhancer" i described in

https://secondlifestorage.com/t-Project-capacity-enhancer-for-mobile-home

Thanks to all who havel helped with questions and input.

Extracted from the mentioned thread is the functionality i need, means replacing a second home battery (the first is AGM 90 Ah) for "something better" (at moment called"unit")

[concept]
The "Unit" supervises the (first) home battery and does one of the following three things (as long as its own capacity lasts):
- on 12.8 Volt of the home batt it does nothing
- below 12.7 V of the home batt the unit starts to supply increasing current to the home batt with a linear correlation to lower voltages. In effect t takes the load which "obviously" drops the batts voltage, limited to 10 (just a figure at moment) Amps.
- over 12.9 unit starts to surge energy from the home batt (as well limited to 10 Amps), as well with a somehow linear correlation, because the original home battery is "obviously" fed.
- If the unit is "empty", it does nothing anymore on discharge, the AGM has to supply the load alone, and will be discharged then - as the last in the row (good for a lead battery to stay full at all times)
- If the unit is "full", it doesnt surge energy anymore, and the charger can fully load or "occasional balance" the AGM as usual.

So, in "Normal Use", the AGM stays mostly full (which it likes), the capacity is taken to and from the unit, cycling it. That increases lifetime of the AGM.
Shorttime high loads (inverter) need help from the AGM over 10 Amps, discharging it, but it will not be much SOC spent of the AGM.

In summary, i need just the Enhancer batt/the unit and a bidirectinal inverter, controlled to above rules. I have no complex connection to the mobiles circuitry, the circuitry or its parameters must not even be changed, but just a connection to the AGM, in parallel. There are some losses of energy, because it goes through the units inverters twice.
(But, i do not have limited resources of energy, i have limited resources of STORED energy, so no problem there).

In thinking about the system i found the name "Virtual Battery" for the concept, because it behaves like a battery, seen from the inside.
It supplies 12.8 Volt, with CV and CC on 10 Amps.
If i externally rise the voltage by a supply ( a "Charger"), the unit starts to surge energy to load the internal battery, which can be whatever type. The surge current is limited as well. The charged voltage does NOT have to match the battery full voltage, as every usual charger has to do (for decreasing the current).

I have not come across this concept ever, so it would be interesting for me if i have invented something already known (as it i the usual case... rotfl)



It looks like i can get 2-3 kWh (4-5 Pedelec Blocks per 36 v 18 Ah) plus two standard Alipress inverters with an Arduino in a PC case, and have 3-5 times the energy of an AGM for half the weight.

Currently i am building up a preliminary version with smaller convertesr i have in stock, that will be an about 3 Amp version.

Battery will be 36 Volt 18 Ah blocks coming from Pedelec ( bicycle) Batteries. These are used as they come out of the box, means not torn apart and repacked, because they are next to flawless as they are, plus very good mounted mechanically.

Each Block will get an own new BMS, the bluetooth version. At first i wanted to use a common BMS for all planned 5 blocks together, but that gives a hassle of BMS wires and connections which is a security risk by being damaged, so i made my mind up.

The converters will get their place in the old PC Mains adapter, gives it wonderful as a modular concept.

More informations and pictures will follow.


Update (re Redpackets notes below) The voltages might not be perfect correct, he is right, that has to be slightly adjusted for stable operation.
 
Just wondering if those voltages are correct? They look low to me?
I would have thought a LA battery would have about 13.8V nominal, higher = xfer to VB, less draw from VB.
 
Redpacket said:
Just wondering if those voltages are correct? They look low to me?
I would have thought a LA battery would have about 13.8V nominal, higher = xfer to VB, less draw from VB.

I do not care at moment if the voltages are correct down to the last 10ths of Voltage. That may be fine tuned in reality anyway. I have a lead battery pair charged by a small Solar panel, i see the voltage drop after loading down to 12.8 by itself, and easily down to 12.5 if i discharge even slightly. That matches the generally known value of 2.1 Voltage (full) for a single cell.
But, many words, short answer - yes, you may be right.
 
I start with a few pictures of the packs i use, and how i want to use them.
(Sorry, it will take awhile until i have posted everything, i write it in smalll packets as i have time to do)

I am taliking about the infamous sealed bike blcks, which are so difficult and time consuming to rip apart:


image_snjjup.jpg


A 60 cell (10s6p 18 Ah) block, sealed with a smooth silicone-like plastic. On top a BMS in a separate small frame.
The BMS is quite sophisticated, as far as i have figured out (with help from German websites) it has a one-line serial communication, which is used when discharging (Pedelec) and in the loader as well.
For the discharging, obviously status information about SOC and so on are transferred, and in both cases something worse is used: An ecrypted Challenge-Answer protocol, which must be correct otherwise the BMS stops all actions.
Means, as long as this encrypted data transfer can't be simulated, the BMS is of no use at all and must be replaced.
(It might be possible, it wouldn't be the first encryption i was able to crack, but the outcome is not guaranteed, and it is unsure if the BMS really gives all possibilities i would like to have)

The frame is fixed with 3 or 4 small screws, and can be removed. Unclipping the BMS-Connector and removing the screws must be followed by some dosed brutality, because the sealant has connected the frame more or less to the pack as well. Starting at one end, levering with a screwdriver, careful to the side of the cells, and accepting some bending of the frame (the BMS may be damaged, but it is for no use either, i need the frame only) will soon be successful.


image_dritux.jpg


Sorry, Pic is bad.
The printed circuit board is seald into the frame with a transparent sealant, which is sort of slightly flexibel same as the black one sealing the cells together.

Getting the Board out is quite easy, just use a small but stable screwdriver past the side of the board ( there is about 2 mm room) down to the base of the frame, close to an edge of the board, and start to lever the Board out. The sealant comes off the frame quite easily. Now, there are 2 (or 3 ?) standoffs coming up from the frame to hold the board during sealing, let them rip off, the damage to the frame is niche.


image_qhnzqk.jpg


You see the transparent sealant has comme completely off the frame, i had peeled some of the board later to look some of the electronics more easily. On the edges you see the room you can use for the screwdriver around the baord.


image_wldvwe.jpg


The underside of the board reveals at least 2 bigger chips, looks its quite high sophisticated for whatever task. The balancer resistors are very small, in the beginning i didnt find them, i guess now ist the 2*5 array right of the six wires going into the middle of the board.
That again shows, that professional BMS for batteries (usually) have astonishing small balancer current capabilities, which makes the sometimes very big currents discussed on this boards disputable.

The bluetooth BMS i will use has 50 mA, which i proved works quite fine so far, i will talk about that later.

The new BMS is just a bit smaller than the original one, it fits into the frame.


image_fztuyz.jpg


In the picture is the first test of the BMS mounted in the frame on one block, 10s6P.

I wanted to test if the metallic side of the BMS distributes heat on balanicing, it does hardly. There may be some heat from the Mosfets as well, but since i will use only about 10 Amps i see hardly any heat dissipation as problem. This is fine, it gives the chance to mount the BMS upside down in the frame (Aluminum to frame bottom) which makes the wiring much easier and shorter:
- The BMS plug can face the side where the Block plug is,
- ground is on the same side as from the battery
- much shorter BMS wires
- the BMS can be fixed by 4 scres wich go into provide theradings in the edge of the aluminum

Next descriptions will be about the behaviour of the BMS, and the resulting plans how may BMS i want(ed) to use.
 
Some thoughts on the blocks.
Older member of this board know, that we have had a discussion about ripping those blocks apart. I didnt follow the crowd, in saying that i would NOT rip them apart as long as i can use them as they are. Means, i even will test them as the original package.

This is even more my intention for the usage in the mobile home. The packs are what companies sell as a professional package for pedelecs, where humidity/rain/shock is a significant problem - plus a quite high occasional loading. Why rip them apart and the repack them ? Nope.

So, i started with the plan of paralellizing several of those packs for the project. Recently i saw the idea on this board of using PC cases (I have had this just 2 weeks before as well), and so such a case it should be. This solves so many problems, since one of my saying re DIY is:

The problem of electronics is always the mechanic.

The PC case has just a few nice features, it is cheap, available in various sizes, and it has the place and the structure of the PS unit - a nice feature here because i can modularise inside of the case:
- the Blocks with all the BMS apparel and separate fuses
- and the Converter package inside a PS case.

That would make updates easy, because most of the behaviour defining electronics are inside the PS case, which can be changed easily.


So i took a case, and checked aout how many packs i could get in. By chance it was a quite broad case, giving room - on the ground level - for 5 blocks 10s6p. Since that was quite what i had in mind I took it at this, in spite of the fact that nearly half of the case would stay empty.

Now, how to arrange 5 blocks electrically ?

My first idea in the beginning was 5 paralelled blocks, one common BMS ( that should be the Bluetooth BMS i talk about here). Effectively making it a 10s30p.
For safety the positive line of each block gets a fuse. So far, so bad.

Soon it dawned me, that parallelizing the blocks onto a common BMS means that the blocks are coupled at each stage to all others - over the BMS connectors and wires, which where by far not able to stand the current that MIGHT be able to try passing them. It meant as well that i have 5 times 9 balance wires ( 90 feet of wires), two connectors per line, same number of solder points, filling every crevice around the Blocks to connect to the BMS.
There is NO room for wiring failures, obened connections or chafing faults in this concept.
Additionally - what happens when only ONE fuse of one block opens? The rest of the block stays connected fia the weak BMS lines to the others.

Sometimes i started an hour search on the net about LiIon fires, and found enough reports of them for pedelecs. There where hardly any exact investigations about the reasons, but the summary of guessings concentrated to one point: Something "has gone wrong" with the loaders, a shortcut, wrong charger, BMS failed, and so on. I found NO reports gussing on " LiIon just wanted to blow up", all reports sounded like "they where mistreated".

That gave me thoughts, and soon i decided just to thrash my concept of parallelizing the BMS lines via the connectors. Just too much (long) wires, solder points and connectors.
But what else?

I didnt find any appalling concept until i had the Bluetooth BMS tested. It does everything you can wish for, nearly, for 30 Bucks.

Now the concept looked easy:
One BMS per block, directly attached to it. Shortest possible wires, every wire failure is detected, and so on. OK, i need 5 BMS, but what ?
I do NOT want to sit in the Womo hearing the fire warner go havoc...

This is what you have seen above in the picture, as preliminary test. And 5 of them in parallel, each get an own additional fuse for further safety.

Now, this would work, and for now i will work along this plan.
But in the meantime i have had to more ideas to modify that.

Idea 1:
I put 2 Blocks together and connect the BMS Connections not via the Plugs, but thick wires soldered to the nickel strips. This double block as 10s 12p gets an own bluetooth BMS. So i need only 3 BMS for 6 packs.

Idea 2:
I dislike the fact that i have 12p this way, without the safety of a fuse-structure. So, why not modify the 10s6p block with fuses, without ripping it apart? I just have to remove the nickel strips from all positive ends of cells, and install a small "Bar" as solder point for the fuses?
That would have the advantages of
- using the blocks "as is", without ripping them apart
- Proper/safe connection of the parallel cells
- Fuse-Structure whith its known safety against cell breakdown.

Idea 3:
@DrBacke has made a most interesting test of fire protection - i will use his result in chosing the appropriate material insde of the PC case to separate the blocks from each other - even the ones connected as pair. That should be a safety against a complete runaway of all blocks.

Summary, for the moment i will go with single Blocks per BMS, make a test with a 2 Block 2BMS structure ( 3 BMS have arrived), and will decide later on what to do probably its the Idea 2 +3 - construct which will be my final choice.

The next post will deal with the Converter structure and a test.
 
While i had ordered some converters, i started to use some i had in stock, which have lower power capabilities, just to demonstrate the function.

I investigated how they behave, when the stay connected eithe on input or output, their own consuption, possible Current shunts in the ground line ( which makes things difficult.

Update: During the last days i have been testing the voltage conecpts of the "virtual battery". I have been feeding a Lead Acid batt on 12,5 Volt with a voltage converter from the LiIon constantly, and have discharged the Lead Acid unregularly and with different loads. Partially i have charged the lead with a Solar Panel during day.
It works fine as expected, this converter even can stay connected, as long as the battery vortage is highe than the adjustmend of the converter it consumes only its internal consumption of 4 mA. ( means it could be used without disconnecting the converter on higher voltages). The converter could be used up to 6-8 A, but has no current limiting function.
Another converter i bought could go up to 14 Amps, but behaves not nicely durng rising the voltage over its own adjustment - it starts to charge up to 400 mA to nowhere util it seems to switch off internally.

Next will be the charging, as well with a smaller converter up to 6-8 A /12 Volt. This would result in quite slow charging, but it has as well nice behaviour because the output can stay connected to the Liion without internal consuption.

See what happens, stay tuned.
 
I have been testing the setup i will have in the womo by simulating feeding and surging from a lead acid battery.
The feeding is obviously the easy part, just have the buck converter asjusted to the desired Voltage ( which seems to be slightly higher thatn the 12.5 Volts i have been talking about).

The rest does a CV, limiting the feeding current independatly how far the LB goes down during discharge.

The surging issue, which is charging the Liion, is a bit more complicated. As can be expected, on charging situations with low currents the setup can start oszilllating on the time constant of the capacity to voltage slope of the lead acid. Basically its a situation which is very similar to the MPPT Tracking of a solar to converter Situation: Feeding no more power than the source is able to supply (MPPT is searching the best Voltage as well, which is not necessary here). A hysteresis can be the first measure, a more sophisticated function in an Arduino, or as last resort just measuring the current are the tools to manage this.

Except this, all works well, i have been thoroughly testing the blocks in question which will be used, plus the BMS behaviour as well. The chinese bluetooth BMS works fine so far.

I have decided I will be testing everything in a simple configuration in the real car, means a 1.5 kWh Block in a manually switched fashion between chrage and discharge, to get more data how this behaves with a real 90 Ah Battery ( My test battery here is a near dead one, say 12 Ah, this shortens testing times but is not "hard" in Voltage on load).

Except that, i have not finally decided if i use the pedelec Blocks - in real parallel - as they are, or if i add fuses. I will put a technical question to this problem in the "General" folder.
 
Short report, i will have an first test with 2 paralleled Blocks with one BMS. The blocks are tested and are well in tune, means in synch.
The converters are not the final ones, but what had been in stock. That resulted in max 8 amps Feed in 12 V, and about 5 Amps surge on 12 V for the charging.
This is enough for a first practical test.
The Control ist extremely simple, the Buck stays connected all times, the boost ist fed only above 13.2 V, its output stays comnected.
An additional Switch can Switch Off the boost manually.
That was enough to cycle through feeding and surging, measuring temps of the units. It needs a small fan, which should be switched, but in the beginning i do permanently.

I put as little effort into control as possible, because i have made my mind up programming an Arduino for all controls.
There are some losses due to quiescent currents atm, but that will be fixed in the Arduino Version.

The voltages are set to 12.8 V for feed, and 13.2 V up for surge and load. I'll see If that works with solar charged AGM battery.
 
Cherry67,

That is quite a read. It will be interesting to see it all come to fruition. Good work on describing your process and progress.

Wolf
 
Any interest on some circuit scetches ?
 
Cherry67 said:
Any interest on some circuit scetches ?

Certainly as I will be in Austria on a learning Holiday (not Vacation for Sean's benefit) I will pour over many diagrams and play with my arduino during my spare time. Don't worry mountain climbing and other "holiday" activity will take precedence. :D

Wolf
 
Austria ? less then 1000 km away from me.....
 
Cherry67 said:
Austria ? less then 1000 km away from me.....
Yep flying in to Munchkin er I mean Mnchen (Munich) in the AM on the 16th from there to Salzburg on the train and then on the bus to my home town of Bad Ischl. :D :D :D
WhenI get there Iwill have myFrankfurter (Ein paarwrstel bitte) with kren,good mustard and a Semmel at Zauners.
Can't wait.

Wolf
 
Just to Note, system ist running as i am in vacation with it.
 
Wolf said:
Cherry67 said:
Austria ? less then 1000 km away from me.....
Yep flying in to Munchkin er I mean Mnchen (Munich) in the AM on the 16th from there to Salzburg on the train and then on the bus to my home town of Bad Ischl. :D :D :D
WhenI get there Iwill have myFrankfurter (Ein paarwrstel bitte) with kren,good mustard and a Semmel at Zauners.
Can't wait.

Wolf
I love your Austrian & German food
 
Just to give an Update, i have Made a 10s18p block out of 3 Pedelec batteries. The converting in both directions simultaneously rund smootly. Its 36 V about 50 Ah.
4 Weeks Test in A mobile Home successful.
 
Real life experience has driven the combo in lots of strange but possible real life conditions.

And then the converter failed, without other notice (like fuses, smoke).

Failed hat the buck converter, its functionally dead, with life Mosfets, but obviously the IC is dead, and the input capacitors have blown....
They are marked as 50 Volts, so anything/anybody has put significantly more than 50 V there.

The usual suspicion goes to the Boost, which supplies there, but whatever i checked, i could not bring him to "confess".

In the end, it was the Buck (!!) which transferred energy from official Output (12 V) to official Input (36V), up to 0.3 Amps and up to 70 Volts.
Means it transferred energy from the ouput to the (Higher Voltage) Input - like a Boost.

Who would have considered that ??

The answer to the secret is the IC used in the converter, a LM25116, which is capable of "Syncronous Mode". And this was enabled....
For the moment, i could disable the feature, and use the "Diode Emulation mode".

And soon i will check if the Syncronous Mode can deliver enough energy for me "Upward" - then i would have my bidirectional converter i have dreamed about.
 
All issued solved. System ist runing smooth and perfekt. Total supply until now about 15 kWh.
Totally compatible to every 12V lead supply.
 
Cherry67 said:
All issued solved. System ist runing smooth and perfekt. Total supply until now about 15 kWh.
Totally compatible to every 12V lead supply.
Super Cherry67 good to hear from you again.


Wolf
 
Hi folks, quite awhile since i was here.

Next inverter in work, the first one still running smoothly.

What a simple system came out of this mad idea. Rare succes.

I reduced the battery down to a 36 V 12 Ah one, the big tripleblock had just Too much (!) stored energy. I was not able to empty it during 16 Days in late winter.... full Mobile home supply, Lights, Heater, TV, Radio. frigo contol.

The inverter ist easiest, 2 chinese inverters, one little Circuit changes... in effect just setting a bridge on both.

anybody interested ?

PS: the syncronous up direcion inverting worked, but only with 1/4 of the necessary current, so its useless.
 
Back
Top