Over-discharge protection for a ~30A max load

Hi all,

I'm in the market for some over-discharge protection for a 12V battery bank and was wondering what you'd suggest.

The application in this case is for a solar-powered server cluster.  The cluster consists of 5 servers (8-core Intel Atoms), a battery monitor PC (ARM), border router (PC Engines APU2), an ADSL modem and a 24-port gigabit switch.  The whole lot runs direct off two 105Ah AGM batteries wired in parallel.  No inverters, the servers run off Mini-box PicoPSU power supplies with Micrel MIC29712 LDOs, the 240-12V PSU in the 24-port switch was also replaced with a MIC29712, the APU2 runs off a LM2940CT-12, the ADSL modem runs of a LM-2596 adjustable switch-mode PSU set for 5V and the battery monitor PC, being an industrial-rated PC is happy running directly from 9-30V DC.

On the roof are 3 12V 120W solar panels wired in parallel to a fuse box under the roof of our back deck using 8GA figure-8 cable, and from there it goes to 6GA figure-8 cable approximately 10m to the solar controller.

Detail can be seen at https://hackaday.io/project/10529

In the beginning, I ran a Redarc BCDC-1225 controller; I had the change-over relay set up so that I could either run it off a 12V 40A switch-mode PSU (Powertech MP-3809), or from solar panels, however I've found that while the BCDC-1225 will happily run off either if powered down prior to switch-over, it isn't so good at switching from solar back to mains "live".

I ran it a little while purely as a solar controller with a Xantrex charger, but the Xantrex had a fault and cooked the now damaged batteries.  (Come home 2AM one morning, open the garage door… last thing you want to be thinking at that hour is "why do I smell sulphur?")

I've since procured a Meanwell HEP-600C-12 which is doing the mains charging, whilst a Powertech MP-3735 is managing the solar.  The Meanwell does a good job, and can be controlled from a 5V TTL signal to boot, so I have a small ATTiny24A controller monitor the battery voltage and it turns the mains charger on when the battery gets low.

The MP-3735 though, has been utterly hopeless.  For comparison, this is the battery voltage throughout the day when the BCDC-1225 is hooked up:

https://cdn.hackaday.io/images/8385941538624384520.png

On a similar mostly fine day, this is the MP-3735 (I've simply disconnected the solar input on the MP-3735 and hooked that up to the BCDC-1225 -- I have SB-50 connectors throughout).

https://cdn.hackaday.io/images/6595561540890474853.png

So I'm looking at giving the MP-3735 the boot and putting the BCDC-1225 back in permanently.

Now, the only thing the MP-3735 offers me right now, is that it has built-in over-discharge protection.  My load hovers around the 8-9A, but can spike over 10A.  For the sake of being conservative, let's call it minimum 20A … anything connected in series between the batteries and the load must be able to carry 20A continuous.  I need to cut power when it gets below about 11.8V (much below that, and the IPMI BMCs on the servers will moan that their 12V rails are sagging).

Jaycar do have one that looks promising, but I'm also aware that this is the same place the MP-3735 came from and this is a single point of failure.

It isn't difficult to rig up a little LM311N comparator to a beefy SSR, or make a small microcontroller handle switching a SSR, but I was wondering if there was a suitable over-discharge protection system that might do the job.

As always, I'm open to suggestions, and if there are queries about my set up, I'm happy to answer them.

Regards,
Stuart Longland

Hi Stuart,

Thank you for your enquiry.

We certainly have the ability to produce a voltage sensitive relay, providing you with the ability to control your loads.
The standard VS12 has an adjustable "turn off" voltage threshold, with a hysteresis of approximately 0.7V. (E.g. if it is set to de-energise the relay at 11.8V, it will energise it at approximately 12.5V). If you require a smaller hysteresis (e.g. approximately 0.3V), this may be possible also.
As the VS12 has a current limit of 10A, you would then use this to switch a continuous duty solenoid such as the SBI12.

If you do not require custom voltage threshold hysteresis, using the VS12 and SBI12 in conjunction is suitable for your application.
If you require a different hysteresis, I recommend contacting the Technical Support Line on (08) 8322 4848, where someone will be able to assist you in furthering the process.

Best Regards,

Kahlan

Hi Kahlan,

Many thanks for the quick reply.  I was looking at the VS12, but as I say, 10A is a bit anemic for my needs, but as you say, the output can be used to control something bigger.

Regarding the SBI12, is that relay solid-state or electro-mechanical?

Reason I ask is that the big electromechanical ones usually require a pretty hefty amount of current to keep the coil energised.  (Not to mention the back-EMF you get off the coil when you disconnect it.)

The datasheet mentions a 5mA standby current (presumably coil de-energised), but not the current drawn by the device when the load is connected (coil energised).

Are there any specifications on how much current the SBI12 needs to keep its contacts closed?

Regards,
Stuart Longland

Hi Stuart,

The SBI12 is an "electro-mechanical" solenoid, with ~600mA required to turn on, and ~160mA to "hold". Additionally, the VS12 has a load of <1mA when on standby, and ~30mA when energised.
The Smart Module on the SBI12 has been developed to suppress the back-EMF created when collapsing the field.

Best Regards,

Kahlan

Ahh, so a roughly 2W continuous load… with a spike of over 7W when it initially actuates.

It seems like a lot actually, but having a look at the solid state relays, I see many of those have a 350-500mV voltage drop which would equate to a continuous power dissipation in the order of 5-6W at the full rated current.  So they're actually worse in terms of power dissipation.

I'm just looking at various P-channel MOSFETs… the Infineon AUIRFS3004-7P looks tempting as it has an Ron of 1.25mOhms and a continuous drain current of over 200A.  If I assume a worst-case of 30A current, I'm looking at a little over 1W dissipated in the MOSFET.  I'll have to clamp the sucker to a nice big heatsink, but that is no trouble.

I have no highly inductive loads (e.g. fridge compressors, etc) to worry about in my application.

I'd be hooking the source up to my battery supply and the drain would go to my load, I then just need something to drive the gate pin to turn it on/off.  I'm thinking the VS12 could fulfill that function, and is probably more finely tuned than a LM311 with a trimmer pot.

If I understand the operation of the VS12 correctly, I should be able to ground the relay common and hook the nomally-open contact to the MOSFET gate pin?

I use these Victron units, very low power draw.

https://www.victronenergy.com/battery_protect/battery-protect

Hi Stuart,

Correct.
You will indeed be able to utilise the VS12 for a trigger on the gate of a P-Channel MOSFET, as it's normally open contact will switch 12V in its energised state.

Best Regards,

Kahlan

Wiring_Solutions wrote:

Hi Wiring_Solutions,

Ahh, much appreciated, knowing a few alternatives is always worthwhile.  A nice feature of that particular unit is over-voltage protection too … i.e. disconnect load when Vin ≥ 16V.  It just so happens that the LDOs I'm using on the servers have a maximum input voltage of 16V.

Whilst over-voltage hasn't been a problem so far, it's useful piece-of-mind to have some sort of protection in place.

Kahlan@REDARC wrote:

No problems then.  I'll do a bit more research, but this gives me some possible avenues to investigate. :-)

Many thanks to all.

Regards,
Stuart Longland