1

I seem to have the same issue 'Ali One' asked about a few years ago (How to create automatic dual battery changeover/switching circuit for uninterruptible power?), except I have a twist that a relay is involved too. So the answers to that question don't fully apply.

There are two supplies involved

  • 4 x 18650 batteries
  • an external power supply

When on the power supply, the batteries are configured individually and each has its own charger.
And, thanks to Olin's suggestion, the supply goes through a Schottky diode to power the load.

on charge

When the power supply disconnects, the relay configures the batteries as a 2S2P set to power the load at 7.4V.
I can switch TO the power supply without incident - the diode is quick enough. Going the other way though, I can't find a way to hold the voltage whilst the relay switches.
The load is typically about 1.5A. I tried a cap on the end, but I am up to 11,500uF and it still won't hold.

on battery

Does anybody have a more elegant answer?

P.S. I am just a mess-about hobbyist with this stuff, so I could easily have missed an obvious solution.

enter image description here

As requested, I have added the diagram for the relay, you will have to imagine this as one 5PDT relay - the drawing tool won't allow any annotation! The empty boxes are 'TP4056 Lithium Battery Charger Modules'.

The final output is nominally 7.4V feeding Arduino (which is very forgiving) and +5V to a Raspberry Pi3B (which is very finnicky)

27 Dec
I have a solution to my problem - it works, but I am not confident enough to say that it is safe so would appreciate any feedback.
I acquired some high current super-caps. They are low voltage, so by the time I string a few together in series, it comes out at 5F.
Way over the top, I know, but as a hobbyist you work with what you can lay your hands on.

I have moved this 'voltage-holding-device' into the power supply rail only and isolated with an additional diode as shown.

My reasoning is ....

At the moment, there is no inrush current regulation to the capacitors as the bench power supply can do that nicely.
At power-on, the power supply will be dragged down until the capacitors have some charge (I have set the power supply to 9V 4A) and during this time, the batteries can continue to power the circuit.

Eventually (several seconds, if caps are completely discharged) the voltage can maintain 6V at which time:

a) the relay coil will pick - flicking the batteries to charge mode
b) there are enough volts in the cap to supply the circuits without reboots.

At power-off, the relay will be dropped instantly (the extra diode prevents the cap from holding the relay picked).
The capacitors will continue to power the circuits until their voltage drops below the battery voltage when the batteries take over (after about 20sec) . All taken care of by the fast switching Schottky diodes

Is it safe/prudent/good practice to leave the caps charged?

enter image description here

SamGibson
  • 17,231
  • 5
  • 37
  • 58
jc508
  • 11
  • 2
  • Welcome to Electrical Engineering Stack Exchange. Can you show a circuit diagram with the relays? – Math Keeps Me Busy Dec 09 '20 at 12:54
  • I wonder how available make-before-break relays are... – Hearth Dec 27 '20 at 23:58
  • To answer part of your question, yes it is ok to have capacitors charged for extended periods. I’ve heard it said that electrolytics last longer in continuous use than if they are left idle but I don’t know whether that’s true. – Frog Dec 28 '20 at 00:07
  • 1
    You could have avoided all this mess by keeping the battery in 2S2P configuration and using a 2 cell charger. – Bruce Abbott Dec 28 '20 at 00:13
  • I cant find any make-before-break relays in the wild. (ie local store or ebay). And Bruce you are right it did end up a mess and ironically the original configuration was 2S2P with a complicated plug to charge 2 at a time. Live and learn – jc508 Dec 29 '20 at 00:10

2 Answers2

1

It is actually much simpler than that:

  1. Connect all the cells in parallel
  2. Connect the charger in parallel with the cells
  3. Connect the load in parallel with the cells

The end result is the same as what you wish to achieve with the more complex circuit, but it's simpler and more reliable, because there is no inrush between cells when first connected in parallel (which will degrade them).

charger --+-- (DC-DC converter) ---load
          |
         BMS
          |
       18650 cells

At any given time:

  • If the load is using less current than the charger can put out, the battery will charge
  • If the load is using exactly as much current as the charger can put out, the battery current is 0 (the battery neither charges nor discharges)
  • If the load requires more current than the charger can provide, the battery will discharge

In other words, the battery current will be equal to the difference between the load current and the charger current.

battery current = load current - charger current
Davide Andrea
  • 16,164
  • 4
  • 33
  • 62
  • But - will the charger circuitry correctly handle this? How can it distinguish between current going into the load, vs current going into the cells, and deliver the correct voltage and current that best suits the cells? – Daniel Chisholm Sep 21 '21 at 12:07
  • You are assuming that the charger needs to distinguish where the current goes. No, it does not. Physics cakes care of everything, All the charger needs to do is "be a charger": limit the voltage to the CV setting, and the current to the CC setting. Physics will take care of the rest, as I explained in the 3 bullet points at the end of my submission. – Davide Andrea Sep 21 '21 at 13:05
0

MBB relays are available; maybe not readily, but available. search for "Form D", and you'll find for instance:

Panasonic TQ2-2M (30V 500mA switch: Mouser https://www2.mouser.com/ProductDetail/Panasonic-Industrial-Devices/TQ2-2M-12V?qs=YINDDaGsG3HU5D8r0ZCxZA==)

Panasonic/Aromat NF4EB-4M-12V (30V 2A; mpja: https://www.mpja.com/12VDC-4PDT-Form-D-Aromat-Relay/productinfo/36045%20RL/)

Potter & Brumfield (Mercury Wetted Reed relay; ebay: https://www.ebay.com/itm/264709580430

(instead of using a single MBB/Form D relay, you could also use two "normal" SPDT, one with a very short delay

grid current
  • 1
  • 1