Off grid workshop - first steps.

[kibi]

I am miles away from taking my workshop completely off grid, but you have to start somewhere and take small steps with some things.
A few weeks ago I did a teardown of my newly acquired inverter / charger. Since then I got to work on preparing the battery to make it go.
Several years ago I discovered an old APC 7.5KVA UPS in the dustbin. It was too unwieldy to transport besides it smelled of burning MOSFET, but it's batteries were modular and easier to transport. I have kept the batteries maintained, waiting for the opportunity to find a good inverter cheap, that's when I found the Victron out of an old ambulance.
The APC had four battery trays in it each tray comprised of eight 12V batteries wired in series for 96V. Two of these trays were then wired in series for 192V system voltage for the inverter. My guess is that at 7.5KVA, a 192V system voltage reduces the current to a more manageable 60A. The APC that I found had another battery pack alongside it which contained another eight trays in a similar configuration.

So, I have a total of twelve 96V battery trays.



Here is one of the trays showing the original wiring.







APC went for two 30A fuses in parallel to protect each tray. I made use of one these fuses per tray in my configuration.



Most of the batteries in the trays still read above 12.5V despite having a top off charge about two years ago. I'm just not so sure of the brand of batteries used. I guess if APC were happy, then they can't be that bad.





So now, I need 12V for my system, so I needed to wire all the batteries in each tray in parallel, fuse it and put some terminals on the end of the tray.

I made some terminals out of threaded brass rod.





I made a jig to assemble the new wiring inside the tray.



...and made up the wiring with a fuse included.



My little JBC iron didn't seem to bothered about soldering an M6 brass nut onto a length of threaded rod, it just sort of got on with it.



Next, the 12V batteries were re-orientated as required and wired up.







Then the tray is closed up again.





Wash, rinse, repeat twelve times. I must tell you, I don't cope too well with repetitive tasks, but luckily I was well behind on my Amphour episodes, so that helped the time tick away somewhat.

Now I needed to make up the leads to connect all the trays together. I went for 25mm² cable for this task as it's what I had to hand.
I cut short lengths and soldered terminals onto the ends. The JBC was not quite up to this task though, so I had to use a torch.



I just heated and fed the solder in through the hole in the top until solder came out the bottom





Then, bolt all the wiring in including the shunt for the Victron battery monitor.





It's ready. The whole lot weighs in at about 220KG, so although you can't see, it's on wheels for manoeuvrability.



That's it for now. Hopefully by next weekend, I'll have the inverter installed and running.

[Kiriakos GR]

By owning one APC Smart 1000 XL,  I can say that I have what you have in a smaller scale. 
My XL version can be connected with four additional battery packages for 24/24 operation.

I had very hard times to find the special male connector (gray color) so to make my own extension cord,
so to add more batteries with out buying any APC battery boxes.
Luckily I found what I was looking in one seller in England.

APC and battery quality ?   My UPS came with batteries (Made in India).
At the third year the batteries had capacity problems, and close to the fourth year they was unable to pass the UPS qualification test.
I got new but the most expensive ones, from Panasonic and made in Japan.
Special part code which is made to last for 10 Years.
They have actually three codes of the same battery for  3Years  or 5 Years or 10 Years.
With 140 EUR I got two from Germany and with stamped production date ( 6 months old).   

About the rebuild of your own battery pack, this is magnificent work, very few they can understand the class of the technician behind those pictures. 
The picture with the flame torch it is 65% successful . 

About making one large battery bank with untested batteries it is a bad  idea.
You could use two automotive spot lights as load, so to find any weak links.

I am crossing my fingers about having made up front, the correct calculations about the needed charging current.
If not, you will have to break this battery bank to two or more smaller ones, which will charge one at the time. 
 
Your next stop could be one AC/DC clamp meter.
So to test and detect weak links under true stress tests, the Brymen BM189 with sensitivity starting from 200mA it would be great for this task.

[kibi]

Yes, these batteries will maybe only last a year or two at the most, but this buys me time to get proper batteries which are suitable for this task.
Managing 96 batteries all in parallel is not easy, so yes, some care needs to be taken.
I should put a 20 or 30A load on and then measure the current out of each tray to make sure it's the same.

Some of my photo's are rubbish. With close focusing distances, DOF gets shallower and my silly P&S camera does not offer manual focus points. I should use the 7D, but the P&S just seems easier at the time.

[G7PSK]

If you can find a forklift service engineer, they very often have batteries that have been removed from forklifts due to one or two bad cells, the rest are good but many operators of forklifts will want the whole pack changed out in a misguided move to ensure reliability. These batteries can very often be had at a very cost effective price.

[kibi]

Unfortunately, this plan did not work
The batteries are too tired. I charged them at five amps for several days until they floated. I then requested about 100A out of them and the voltage plummeted rapidly and steadily. Approximately 37 seconds then the inverter shut down at 11.5V. Fail.
Oh well, I still think it was worth trying to get some use out of the batteries that I had to hand.
Now I'll have to do some hardcore saving and get some decent batteries.

[Kiriakos GR]

I will pop in, this link so to motivate you a bit.  

http://www.youtube.com/watch?v=gom38SWkFU4&list=UUTNV_Oh5QAi1Is9r9Aw5bpw&index=33&feature=plcp

[MJLorton]

Unfortunately, this plan did not work
The batteries are too tired. I charged them at five amps for several days until they floated. I then requested about 100A out of them and the voltage plummeted rapidly and steadily. Approximately 37 seconds then the inverter shut down at 11.5V. Fail.
Oh well, I still think it was worth trying to get some use out of the batteries that I had to hand.
Now I'll have to do some hardcore saving and get some decent batteries.

Aaaah...pity to hear that after all the effort. I hope you manage to get some at a good price...or free!

[kibi]

Well, I have seen a jolly good deal on six Trojan T105 6V batteries. These could be wired in a 2S 3P configuration for a 675Ah 12V bank.

[SeanB]

At least you can scrap them for some money to pay for the new ones. Check them and use the better ones to drive the solar lighting when you install it.

[kibi]

This evening I finished connecting up my BMV600 battery monitor.
It's mounted in a box that is remarkably similar to Martin's!



I have also wired up the battery monitor to a computer. This was rather challenging.
First I had to find a computer with "Windows" on it, apparently made by some company that call themselves "Microsoft". OK, I have one knocking about, but it also needed a serial port on it, another unfortunate rarity these days.
Once this was done, one is supposed to shell out £100 for a cable that connects the BMV600 to the serial port. I guess Victron's idea is that this is pocket change to people who own a yacht and they'll just buy one, but not I.
After a bit of scouring through Victron's own publicly available documents I found out the connections, the protocol and the signal levels.
So, I knocked up my own cable using a MAX3232. The BMV600 powers the cable via it's own 3.3V line.





....and it works well with their software which allows you to log to a CSV if you wish. Leaving a computer (with "Windows" on it) switched on the whole time is silly, so I'll have to figure out how to port this data to my Thingspeak channel.



Edit:

Oh ja, I forgot to mention that I ran the battery bank for two hours at a rate of 30A, then it wanted to die, probably would have got another hour out of it.
I don't think I'm going to put much more effort into this idea, just get some decent batteries.

[SeanB]

This is my solar cell monitor, simple and cheap R5 meter from the OneHungLowMall. I was going to use it as a light meter, but it was not successful, so decided to turn it into a simple voltmeter instead.

Measured the meter movement resistance, and noted both the voltage across it and the deflection. 140 indicated and it had a 945R resistance. I then calculated I needed around 134k to get FSD on it, so went looking for resistors on scrap boards. A 100k unit later and a test on 12V and it was in the right range ( near 18V or so) so added a 33k unit in series, too low, so added 47k in parallel to the 33k and it was close enough to correct. Simple meter movement, and a lot better than the idiot diode that turns on when there is enough light to have more than 5V across the panel.

[kibi]

Right, so I've got everything installed and running now.
I had to wait for a time delay (S) RCD switch to come around a sensible price. They are quite expensive understandably. I eventually found a good 100mA time delay RCD for a price I was willing to pay. This switch is needed to replace the original 100A incomer and is wired upstream of the inverter. 100mA protection is acceptable for non-contact circuits.
The time delay aspect is to provide discrimination for the downstream RCBO's. This is not just for convenience it's absolutely essential for my application. It prevents you from waking up and finding yourself with two or more arses, which apparently, is never an ideal situation.
So, imagine that the inverter is not running and it has switched the incoming mains through to the loads. If there were to be a fault current to earth (via a human for example) and there was a 30mA fast acting RCD ahead of the inverter that responded first, the mains to the inverter would fail and the inverter would then switch on thus fault current would continue to flow through the person. There would be an extended delay whilst the earth relay in the inverter closes (the relay that connects neutral to earth to facilitate the operation of downstream RCD's) and then the downstream RCBO on the circuit to which the person is connected to would finally have a chance to disconnect. By this time the person (luck being on their side) would be feeling like they'd been hit by a train, a steam one to be more precise.
Now it's all be wired up correctly and the disconnect times need to be measured.

The 100mA time delay switch operates in 250mS at 100mA fault current up to 80mS at 500mA fault current. Good.
I then tested all the downstream RCBO's beyond the inverter with the inverter off (grid passthrough). They all disconnected within 20mS regardless of the fault current between 30mA and 500mA. The upstream time delay RCD never responded at all in this test. Good.
I then repeated the test on the downstream RCBO's this time with the inverter running (grid disconnected) and again, around 20mS disconnect times. Good, the gods are pleased. 

This installation of the inverter is now complete and I can run some experiments using my existing batteries until I get hold of some good ones.

[Kiriakos GR]

I have a strong disagreement about electrical installations over wood.
But I am not aware if plasterboard's are everywhere available, even today.   

[Monkeh]

I have a strong disagreement about electrical installations over wood.
But I am not aware if plasterboard's are everywhere available, even today.

Wood is actually not as flammable as you might think. It's also a hell of a lot more durable than plasterboard, plus less messy. Cement board is stronger, but still messy, awkward, and not as long-term durable as wood. Plastics, of course, have this horrific tendency to melt and make things worse.

If you can think of something equally durable and less flammable than plain pine board or chipboard which is easy to work and equally available and affordable, I'm all ears.

[SeanB]

There are literally thousands of old houses with wooden backing boards for power meters. Only in the last 30 years has there been a move to plastic and GRP housings for them, there are sill many using wooden backboards. The hotel next to my work had k&T wiring with bare wire running on wood until recently when the roof was replaced., along with many power line poles still being wood.