Author Topic: Modding old ATX supplies for 3.65v output at high current for LiFePO4 charging  (Read 3939 times)

0 Members and 1 Guest are viewing this topic.

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
I have a 4s LiFeP04 pack that I hope to use very occasionally in short notice. It is stored ~1/2 charged. Unfortunately the pack was originally supplied with mismatched cells - at the I was unable to return them. This makes charging very slow as the cells have different charge curves and a standard charger reduces the charge rate before some cells are close to charged as the inbuilt balance shunts can only handle very low currents.

I'd like to build a charger with 4 independent channels capable of around 20-25 amps per channel at 3.65v and am considering using modified ATX supplies. Is this feasible and is the modification likely to be more complicated than changing a few resistors? Thanks for any ideas!
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Modifying PSU for this purpose is not feasible. But it's completely suitable for powering proper charger, though there might be issues with minimum load requirement for some ATX PSUs.
Quote
capable of around 20-25 amps
You'd be an idiot if used dodgy charger at such current.
 
The following users thanked this post: janoc, tooki, ESXi

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
Modifying PSU for this purpose is not feasible. But it's completely suitable for powering proper charger, though there might be issues with minimum load requirement for some ATX PSUs.

Can I ask specifically why it isn't feasible? Thanks!
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
First of all, constant voltage won't work to begin with. It needs current regulation with current dependent on battery charge. Also it needs to detect when battery is fully charged, overcharging is completely unacceptable.
 
The following users thanked this post: ESXi

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9320
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Have you checked how many banks are high/low? Most likely the easiest way to solve the problem is to build a temporary balancer with a comparator switching a load (such as a car light bulb) using a MOSFET, to be manually connected to a high bank.

You can also modify the BMS to have a higher balance current. I like to use LEDs as the loads to give a visual indication of balancing activity.
http://www.kerrywong.com/2017/09/10/modifying-a-4s-100a-lifepo4-bms-module/comment-page-1/
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
First of all, constant voltage won't work to begin with. It needs current regulation with current dependent on battery charge. Also it needs to detect when battery is fully charged, overcharging is completely unacceptable.

If the voltage is set at 3.65v, and the supply max current is ~25 amps, then I can't see any issues with the basic charging requirements - it's CC to 3.65v and then CV. Overcharge will be monitored. My uncertainty is around the modification being possible at all.
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
Have you checked how many banks are high/low? Most likely the easiest way to solve the problem is to build a temporary balancer with a comparator switching a load (such as a car light bulb) using a MOSFET, to be manually connected to a high bank.

You can also modify the BMS to have a higher balance current. I like to use LEDs as the loads to give a visual indication of balancing activity.
http://www.kerrywong.com/2017/09/10/modifying-a-4s-100a-lifepo4-bms-module/comment-page-1/

It's mainly 1 cell that is very different from the others. I'd need ~20amps of shunt to charge at a decent rate.
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
First of all, constant voltage won't work to begin with. It needs current regulation with current dependent on battery charge. Also it needs to detect when battery is fully charged, overcharging is completely unacceptable.

If the voltage is set at 3.65v, and the supply max current is ~25 amps, then I can't see any issues with the basic charging requirements - it's CC to 3.65v and then CV. Overcharge will be monitored. My uncertainty is around the modification being possible at all.
And how you do you get CC by simple modification? And for final 10% you need to reduce charging current a lot or not charge at high current to begin with.
« Last Edit: June 29, 2019, 01:28:48 am by wraper »
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au

And how you do you get CC by simple modification? And for final 10% you need to reduce charging current a lot or not charge at high current to begin with.

My (possibly incorrect) understanding was that the supply would limit the current to the maximum it could supply at the set voltage. If the voltage is set at 3.65v the current will fall off as the battery reaches that set charge voltage.
 

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
You need to find out what cell type specs and it's MAX charge currents - both boost and top (as others have mentioned).
You will HAVE to do the current limiting yourself, don't rely on the P/S, or you could possibly shorten / kill the cell life.
Hello <tap> <tap> .. is this thing on?
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
You need to find out what cell type specs and it's MAX charge currents - both boost and top (as others have mentioned).
You will HAVE to do the current limiting yourself, don't rely on the P/S, or you could possibly shorten / kill the cell life.

They are able to be charged at least 25 amps safely. Cheers :-)
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv

And how you do you get CC by simple modification? And for final 10% you need to reduce charging current a lot or not charge at high current to begin with.

My (possibly incorrect) understanding was that the supply would limit the current to the maximum it could supply at the set voltage. If the voltage is set at 3.65v the current will fall off as the battery reaches that set charge voltage.
ATX PSUs don't CC, they shut down if current limit is exceeded. Or fail, if PSU is crappy and without proper protection.
 
The following users thanked this post: tooki

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9320
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
It's mainly 1 cell that is very different from the others. I'd need ~20amps of shunt to charge at a decent rate.
Connect that bank to the 5V output of the PSU through a high current diode and a length of wire to act as a current limiting resistor. Plug the PSU into a one shot timer (available at home improvement stores or salvaged from an appliance) so you won't accidentally leave it running for too long. Keep a watch on the voltage with a multimeter.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Online magic

  • Super Contributor
  • ***
  • Posts: 7453
  • Country: pl
^do it outside and far from civilization if you go that route :-DD

4 independent channels capable of around 20-25 amps per channel at 3.65v and am considering using modified ATX supplies.
You cannot change the ratio of output voltages on one PSU because it's determined by turn ratio of secondary windings on one common transformer. You also cannot current limit each rail independently, for the same reason. So you are looking at converting one PSUs per channel.
Each needs to have a rail rated for 25A or you are looking at upgrading secondary diodes, chokes and capacitors, pretty much everything.
The channels will all have earthed grounds, hope you are OK with it.

Is this feasible and is the modification likely to be more complicated than changing a few resistors?
Reducing output voltage is a few resistors. CC mode would take adding shunts, comparators and driving the optocouplers yourself.
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
^do it outside and far from civilization if you go that route :-DD

4 independent channels capable of around 20-25 amps per channel at 3.65v and am considering using modified ATX supplies.
You cannot change the ratio of output voltages on one PSU because it's determined by turn ratio of secondary windings on one common transformer. You also cannot current limit each rail independently, for the same reason. So you are looking at converting one PSUs per channel.
Each needs to have a rail rated for 25A or you are looking at upgrading secondary diodes, chokes and capacitors, pretty much everything.
The channels will all have earthed grounds, hope you are OK with it.

Is this feasible and is the modification likely to be more complicated than changing a few resistors?
Reducing output voltage is a few resistors. CC mode would take adding shunts, comparators and driving the optocouplers yourself.

Thanks for the input - it is helpful :-) I have found this voltage adjustment and current limiting circuit which is not too complicated so I might give it a try https://weekly-geekly.github.io/articles/257427/index.html
 

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 9330
  • Country: fi
Can I ask specifically why it isn't feasible? Thanks!

Because battery charging requires a current-limited supply. Even though a modern switch mode PSU does have current limiting, their control ICs implement, on purpose, modes that prevent the supply from running in the current-limited mode. It's typically either a "hickup mode" or "latch mode", but either one prevents it from charging: the power supply provides the constant current for a few milliseconds, and then just errors out.

If they did use a control IC with disableable over-current error modes, it would do the trick easily, but this is almost never the case: even though allowing the continuous constant-current output mode is trivial for a control IC manufacturer, the demand for such controllers is low enough that these ICs are fairly rare. The ability to do constant current is a huge "value add" for a manufacturer, even if it comes for free.

This is sad, since the power stage electronics might be able to do the trick easily, but given they are built for a specific purpose, i.e., supplying constant voltage to a load with limited capacitance, they won't handle constant current loads (big capacitors, batteries, directly connected LEDs, big motors...) for more than a few milliseconds.
« Last Edit: June 29, 2019, 10:40:49 am by Siwastaja »
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
Can I ask specifically why it isn't feasible? Thanks!

Because battery charging requires a current-limited supply. Even though a modern switch mode PSU does have current limiting, their control ICs implement, on purpose, modes that prevent the supply from running in the current-limited mode. It's typically either a "hickup mode" or "latch mode", but either one prevents it from charging: the power supply provides the constant current for a few milliseconds, and then just errors out...
....., supplying constant voltage to a load with limited capacitance, they won't handle constant current loads (big capacitors, batteries, directly connected LEDs, big motors...) for more than a few milliseconds.

Thanks for the detailed explanation! I have have powered leds from very small SMPS supplies in the past and just let them run at max current which was close to the led rating and that worked fine. I shouldn't have assumed it would be the same for an ATX supply though...
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
It seems like a few people have successfully defeated OCP/UVP on the control IC http://planetimming.com/atx_mod/atx_mod.html http://burnt-traces.com/?p=159
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
I don’t think it can be emphasized enough how dangerous lithium battery charging can be. There are many purposes in electronics where jury-rigging something works fine, but lithium charging most decidedly is not among them!

FYI, PC power supplies are designed to provide large amounts of current at low cost. They do not provide particularly clean power nor precise voltages, because the assumption is that any computer part that needs clean, precise power will use local regulation to produce it from the “bulk” power from the PSU. Just bear this in mind whenever you’re tempted to use a PC PSU for any other task.
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
It seems like a few people have successfully defeated OCP/UVP on the control IC http://planetimming.com/atx_mod/atx_mod.html http://burnt-traces.com/?p=159
For the love of god don’t do this, especially not for lithium charging!
 
The following users thanked this post: wraper

Online magic

  • Super Contributor
  • ***
  • Posts: 7453
  • Country: pl
It seems like a few people have successfully defeated OCP/UVP on the control IC http://planetimming.com/atx_mod/atx_mod.html http://burnt-traces.com/?p=159
For the love of god don’t do this, especially not for lithium charging!
Lol, indeed.
You need additional current regulation reacting and slowing down the PSU before OCP activates, not to disable OCP :-BROKE
But that's no longer a trivial mod. Output voltage adjustment is simple, but constant current regulation doesn't exist in those PSUs and would need to be built from ground up.
 

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
Thanks for all the thoughts :-) Fear not, even when I'm doing things somewhat on the cheap, I'm very aware of the safety issues around lithium cells. If I manage to get something working, it will have over volt alarms and cutoffs.
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Thanks for all the thoughts :-) Fear not, even when I'm doing things somewhat on the cheap, I'm very aware of the safety issues around lithium cells. If I manage to get something working, it will have over volt alarms and cutoffs.
:palm: If you want to get something really working, build proper charging circuit powered from PSU. Not deciding adding 6th feet to a dog if it is still moving after adding 5th feet.
 
The following users thanked this post: tooki

Offline fixy88Topic starter

  • Contributor
  • Posts: 43
  • Country: au
:palm: If you want to get something really working, build proper charging circuit powered from PSU. Not deciding adding 6th feet to a dog if it is still moving after adding 5th feet.

Facepalm seems a bit harsh - it is common in commercial chargers and batteries to include these safety measures.
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
:palm: If you want to get something really working, build proper charging circuit powered from PSU. Not deciding adding 6th feet to a dog if it is still moving after adding 5th feet.

Facepalm seems a bit harsh - it is common in commercial chargers and batteries to include these safety measures.
It's not the point. Point is that adding those things to ATX PSU is like adding 5 and 6th leg to the dog. It won't be easy, and most likely won't work properly anyway. Especially considering you didn't even know they don't have CC.
 
The following users thanked this post: tooki


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf