[rwaudio]

It seems that lots of people use a fixed voltage power supply on new cells to get them all to the same voltage before assembling them into a pack. What if that concept was taken one step further and the assembled pack is charged with individual isolated dc/dc converters fed by a central power supply. I know the system loses efficiency compared to some of the high end chargers, however there is a lot of flexibility in the system, changes to pack voltage would just need a few extra dc/dc converters. Not to mention there is no bms, or manual cell balancing required since each cell is charged individually to a set voltage. There is also a lot of flexibility with regard to the power supply 115v 230v, both. Or even DC to feed the converters directly. Common 3.3v dc/dc converters can usualy be adjusted up 10% which would be ~3.65v which is a pretty safe upper limit for most Lithium cells, but this voltage is also fully adjustable between 3.3 and 3.65V.

Paying retail for these parts would add up pretty fast, however there are a lot of surplus items out there that would make it affordable.

Pros / Cons?

 

[major]

It seems that lots of people use a fixed voltage power supply on new cells to get them all to the same voltage before assembling them into a pack. What if that concept was taken one step further and the assembled pack is charged with individual isolated dc/dc converters fed by a central power supply. I know the system loses efficiency compared to some of the high end chargers, however there is a lot of flexibility in the system, changes to pack voltage would just need a few extra dc/dc converters. Not to mention there is no bms, or manual cell balancing required since each cell is charged individually to a set voltage. There is also a lot of flexibility with regard to the power supply 115v 230v, both. Or even DC to feed the converters directly. Common 3.3v dc/dc converters can usualy be adjusted up 10% which would be ~3.65v which is a pretty safe upper limit for most Lithium cells, but this voltage is also fully adjustable between 3.3 and 3.65V.

Paying retail for these parts would add up pretty fast, however there are a lot of surplus items out there that would make it affordable.

Pros / Cons?

Not a new concept. Victor of MetricMind did this on his Kokam LiPo batteries. http://www.metricmind.com/audi/main.htm Click on _Battery_ to see details. He designed and made a small DC/DC for each cell to take it up to 4.15V. He would use a bulk charger to bring the pack up to about 3.8V/c then switch over to the individual chargers to top off each cell.

Pros----it can work nicely.

Cons----Costly, complex, possible reliability issues, and likely a low charge rate, meaning long charge times.

And you still need a pretty healthy power supply wired to each of your converters.

major

 

[Tesseract]

This is how we charge the battery pack for our dyno now. Specifically, we use a bunch of Meanwell 13.8VDC switching power supplies that go into current limiting rather than "hiccup" mode when overloaded. Even when we pull 800A from the pack and the batteries have sagged down to 7V or so the power supplies dutifully contribute their 10A.

We previously used a big Variac to charge the pack in series, but we have so thoroughly abused the dyno pack that hardly any two batteries are in balance anymore. One would be at 12V while another would be at 15.5V... what a mess. I suppose I could have built balancers for them, but buying power supplies off the shelf took less effort (for me, anyway... ).

EDIT - I also modified a commercial 3.3V/50A switching power supply to individually charge LFP cells. I modified the range over which the output voltage can be adjusted to let me go as low as 3.5V and as high as 4.2V. Bumping up the output voltage does require reducing the output current, so I changed the current limit to 40A to remain within the ~165W rating.

 

[rwaudio]

Not a new concept. Victor of MetricMind did this on his Kokam LiPo batteries. http://www.metricmind.com/audi/main.htm Click on _Battery_ to see details. He designed and made a small DC/DC for each cell to take it up to 4.15V. He would use a bulk charger to bring the pack up to about 3.8V/c then switch over to the individual chargers to top off each cell.

Pros----it can work nicely.

Cons----Costly, complex, possible reliability issues, and likely a low charge rate, meaning long charge times.

And you still need a pretty healthy power supply wired to each of your converters.

major

I'm surprised I missed that, I've read most of the update on their Audi, at least someone else is doing it to. I think I'll go forward with a test on the 4 cell 12v replacement battery and see how that works out. As far as charge rate goes, I'm having trouble finding stuff with LOW enough power output to charge off 115v 15A lines, the common 3.3v dc/dc converters are 15-30A and I'm doing a 90S pack so that's a lot of power. I'll just have to see what the current draw is at different SOC starting points.

This is how we charge the battery pack for our dyno now. Specifically, we use a bunch of Meanwell 13.8VDC switching power supplies that go into current limiting rather than "hiccup" mode when overloaded. Even when we pull 800A from the pack and the batteries have sagged down to 7V or so the power supplies dutifully contribute their 10A.

We previously used a big Variac to charge the pack in series, but we have so thoroughly abused the dyno pack that hardly any two batteries are in balance anymore. One would be at 12V while another would be at 15.5V... what a mess. I suppose I could have built balancers for them, but buying power supplies off the shelf took less effort (for me, anyway... ).

EDIT - I also modified a commercial 3.3V/50A switching power supply to individually charge LFP cells. I modified the range over which the output voltage can be adjusted to let me go as low as 3.5V and as high as 4.2V. Bumping up the output voltage does require reducing the output current, so I changed the current limit to 40A to remain within the ~165W rating.

I'd like to think I'll be nice to my pack, but with a warp 11 HV and Soliton1 I'm sure I'll be drawing 10-15c on occasion from the headways. I'd like to make sure that each parallel cell group gets charged properly each and every time. And from the looks of it a 1500W 288v nominal (90s) charger would only cost less than $900US, and going to 3-4kw would only be a few hundred more and about 7.5kw would be the peak of the dc/dc converters but require 230v 40-50A line to power it. There are some nice industrial PFC power supplies that would probably do the job nicely and can take from 85-268v AC on the input.

Thanks for your input guys.

 

[rwaudio]

Well after testing with some dc/dc converters on hand I've decided to take the plunge and build a 3kw 90 cell (328.5v) dc/dc converter based charger.

I will be using 90 Synqor 36-75v input (48v nominal) 3.3v output 25A modules adjusted up 10% which should give 3.65v @ 22.6A give or take. (if I could feed them enough current on the input, and the cells draw that much) These will be powered by a pair of Power-One industrial 48v power supplies. They will be configured to put out about 1500w if fed by 120v and put out 3kw on 240v. (they take any input between 85-264v AC) I will make an auto detect circuit that will only power on half of the modules at 120v and everything at 240v. There is no need for an auto shut off since the voltage is capped at 3.65v but there will be a shutoff triggered when the current draw drops below a certain point, or possibly a timer, or something else.

I am also designing a PCB that will hold 15 dc/dc converters that will go along with my 48v (15s7p) packs, so there will actually be 6 individual "chargers" all fed by the same 48v supply to charge my 288v nominal pack. This makes it easy to change the cell count in the future, since the charge board will be directly connected and mounted inside the battery boxes. I might also make a couple of boards configured as high current 3.65v chargers with all dc/dc converters in parallel, which would be 3.65v @ 340A to quick charge 6+ cell parallel headway packs. I would probably only do this for testing to see how the cells hold up close to their "maximum" charge rate.

This is the data sheet for the DC/DC conveters:
http://www.synqor.com/Datasheets/PQ60033QGA25_Datasheet.pdf

And the data sheet for the Power supply, it's the HPF5.
http://www.datasheetarchive.com/pdf/getfile.php?dir=Datasheets-20&file=DSA-391095.pdf&scan=

I found a place to buy 100pcs of the dc/dc conveter for $525 shipped. And the power supplies $315 shipped for a pair. There will be ~100-150 worth of PCB's since I'll be getting such a small quantity made, plus connectors/ferrules/wire/resistors/capacitors/fuses to hook it all up. But this will replace the charger/bms and be part of my guage setup as well. (the voltage of each cell will be monitored during charging/discharging)
A 90 cell BMS would cost between 1200-4000 alone, plus a flexible charger capable of hooking directly up to 120/240 and putting out 1.5kw/3kw is also expensive.

With a large enough 48v supply the setup would be capable of 7.5kw, I simply don't have access to 240v 40/50A so no need for it anytime soon. It could also be hooked directly up to 48v (36-75) dc if I had a large enough battery bank at home, charged by solar perhaps?

 

[dimitri]

Lots of heavy gauge wires to go from DC/DCs to each cell, it would probably be a little messy. Many points of failure, too. One of 90 DC/DCs fail and you are risking to lose a cell if you don't do cell level monitoring. If you do cell level monitoring, that essentially brings BMS cost back.

Even though the idea is solid and probably would make sense for a bench pack, I am not convinced this is a better way in a car than a single string charger with simple cell level BMS.

It would be interesting to see you go thru with it, please post details and lots of pictures.

 

[rwaudio]

Lots of heavy gauge wires to go from DC/DCs to each cell, it would probably be a little messy. Many points of failure, too. One of 90 DC/DCs fail and you are risking to lose a cell if you don't do cell level monitoring. If you do cell level monitoring, that essentially brings BMS cost back.

Even though the idea is solid and probably would make sense for a bench pack, I am not convinced this is a better way in a car than a single string charger with simple cell level BMS.

It would be interesting to see you go thru with it, please post details and lots of pictures.

Thanks for the comments, I will have an arduino based voltage monitor (optocoupler and multiplexer to read each cell), hopefully that catches any problems before they become failures. I will be integrating the charge boards into the pack itself, so ideally the wires from dc/dc to cell are only 3-6" long with good placement of connectors on the PCB to match the pack layout. I would say that the weakest link is the normal price of the parts required, as surplus the prices are great, however at retail this setup would cost 5-10grand easily. I'll keep posting my progress, I'm not convinced it's a better solution either, but i'm going to give it a try and find out. In theory it has a lot of advantages, and really only as many failure points as a charger + shunting BMS solution. But a failure has a worse outcome, and the chance of an over discharged cell would be greater without a backup monitoring system. However the chances of an overcharged cell go down.

 

[rwaudio]

Update from my blog:

The basis of the charger design is 90 quarter brick DC/DC converters, so I got 100 just incase. I may use the same converters to charge/balance the 12v system battery as well. Each DC/DC converter will accept 36-75v input range and by default puts out 3.3v, however this output will be adjusted up 10% to put out 3.65v which is the perfect CV charge voltage for many LiFePO4 cells.

The 630 cell battery pack will be broken down into 6 individual battery boxes each containing 105 cells and one charger board which will contain 15 dc/dc converters. I’m still deciding if I should use connectors, or wires soldered directly to the board to connect from the board to the cells. The inital board design is shown below (not complete, just to get something visual), which will be a 4 layer PCB roughly 6″ x 12″ which will be mounted above the headway cells, with very short connecting wires from the PCB to each buss bar. There is no longer the need for a BMS because each cell is effectively charged individually to the preset 3.65v.

Each battery box will be roughly 6″ x 12″ x 24″ which will make it fairly easy to mount and integrate into the car. The DB25 connector will connect each battery pack to the Arduino which will monitor each cells voltage. There are fuses onboard that will protect the pack and wiring from shorts or other problems.
The power supplies can accept between 85-264v AC so I simply need a 3 prong grounded plug for the charge port, the standard connection for this voltage and current range is the NEMA L6-20, so behind the fuel filler door will be something like this:

This will allow me to plug into any available 110/220v outlet to recharge, it would also be technically possible to plug directly into a 48v DC power supply or battery (36-75v actually due to the wide input range of the dc/dc converters). This would bypass the AC power supplies and require a high current 48v DC connection.

 

[rwaudio]

I've been working on this for quite awhile now, with the biggest problem being too much current draw from a low cell. With 96 of these chargers forced to share a ~1500W plug it would blow a breaker in no time. Some people use resistors or long wires that act like resistors to limit the current. But in the end that limits the charge current through the entire charge cycle and as soon as the voltage comes up on the cell a little bit to get over that brief high current period then they are charging at 1/2 current or less. I wanted to turn each of these dc/dc converters into a true CC-CV charger, I went through various feedback methods and thought I had it solved when I used a feedback loop that would vary the output from 3.3-3.6v, but then I plugged in a low cell and 3.3v wasn't low enough to limit the current to the preset level. DAMN... back to the drawing board. What I needed was the full swing to adjust the trim voltage below the 3.3v nominal output of the converter. This involves bringing the trim pin from full positive to close to or full negative to get the output voltage in check. Now I have a system that is easy to adjust the CV voltage, then you can adjust the CC current independent of the CV voltage.

There is a 0.001 ohm shunt attached to the negative terminal of the charger, the whole circuit is referenced to the charger negative. When current flows it creates a positive voltage across the shunt which is fed into an op-amp to bring it up to a usable voltage (adjusting the gain is also the method of adjusting the CC current). The output voltage is then stabilized by an R/C network (without the cap the loop would oscillate) this feeds directly into a Pic microcontroller that turns the input voltage into serial data that controls a digital pot. The pot is attached to the +/- and trim terminals, however between the pot and + terminal is a normal trim pot that sets the CV voltage.

So basically it measures the current (by measuring the voltage across the shunt) The output voltage of the op-amp indirectly sets the trim voltage on the dc/dc converter, so as the current rises the voltage is trimmed down to keep it under control, and as the current drops the voltage goes up until it reaches the CV set point.

I still have to tweak a few things but it's working very nicely now. I've been charging an A123 20ah cell from almost completely dead at 3A, something that was not otherwise possible because of the low internal impedance of the cell, it would try to draw as much current as possible.

The next step is to decide if I want to make the current adjustment remote so that a single pot near the charging port could be used to adjust the charging current of all the chargers on the fly when you plug in making it easy to adjust for different voltage inputs allowing higher charge current, or adjust for shared plugs much like the Manzanita chargers can do. In this case the actual charge current at the cells doesn't matter, but an A/C current meter would allow easy adjustment of the charge current to make use of all the current available from a plug and reduce the total charge time.

Total cost of an 8kw capable charger (only 3kw for the moment based on the 48v power supplies I'm using) is about $1300. I have a source for some larger supplies that would let me charge at 8kw, but I would want them mounted in my garage, and not on the car due to weight/size. I would then plug in a high current 48v cable instead of an A/C cable. Cost for those is around $550 (surplus) Add in an anderson connector and some meters and it's probably 2 grand for an 8kw charger that can't overcharge a cell, and is completely adjustable for CC and CV. (LVC still needs to be handled)

Unfortunately retail price on something like this would be $10,000+ using new parts and covering the design and prototyping costs.

 

[bruceme]

It seems that lots of people use a fixed voltage power supply on new cells to get them all to the same voltage before assembling them into a pack. What if that concept was taken one step further and the assembled pack is charged with individual isolated dc/dc converters fed by a central power supply. ... [snip] Pros / Cons?

Sometimes copper is cheaper and more practical than silicon. The price to balance/monitor/manage is a function on voltage. The price of a charger is mostly a function of kW (power). So often for high voltage systems (I'm thinking > 80v), you want the per/cell cost as simple and cost effective as possible.

That said... I do sometimes ponder what it might be like to install a simple/stupid charger atop each cells and run 110/220vac all the way around in series with the batteries.

-Bruce

 

[dtbaker]

That said... I do sometimes ponder what it might be like to install a simple/stupid charger atop each cells and run 110/220vac all the way around in series with the batteries.

-Bruce

well..... assuming you could find simple stupid single cell chargers where you could set or calibrate the finish voltage, they are small and light enough to build into a battery box or stack up in a space not bigger than a single central unit, and they could pump out some reasonable amps, and they cost less than $10 each... then you'd have something.

but... you'd also have increased the points of failure by n x cells and made the failure catastrophic instead of a benign filure to charge; if any single mini charger failed, it would kill the cell probably on the next use UNLESS you also had cell level BMS to red light an incomplete charge on any single cell.

I think rather than a sophisticated (expensive) BMS, if we had cheap simple cell level monitors where the first cell to a set high voltage could flip a relay that turned off (any) charger, and perhaps turned on led or something so you could see which cell hits first.... thats all we need.

 

[rwaudio]

Well it's definitely been awhile since I started this, but here is the first look at my dc/dc converter charger. The design has changed and it can be used for multiple applications. Shown is an 8 cell 25A charger that will automatically balance the cells. This needs to be fan cooled, I will post more pictures when I finish the fan mount.

 

[rwaudio]

My DC/DC converter/charger is finally finished. The board is designed to hold a raised bracket with a 120mm low profile 12v fan.

I started by milling a copper clad PCB to the desired shape and size to hold the fan. (this is the strongest thinnest material that I had on hand).

Mounted the fan to the PCB with the included fan hardware.

Finally mounted the milled PCB to my DC/DC charging board PCB using suitable stand offs.

I used a low speed (very quiet) low profile computer case fan to keep the height in the desired range. I still need to find a grill or filter to keep fingers and other things out of the fan blades. I have an extra set of 12v terminals on the input side of the PCB, so initially I will be running the fan off of the 12v signal that turns on the board itself.

This can charge and balance 8 series LiFePO4 cells or be used as a 36-75v or 72-150v to 13.6v 660watt adjustable dc/dc converter. (as a dc/dc converter it will charge and balance the 12v battery if it is made from 4S LiFePO4 cells)