You are viewing a single comment's thread from:
RE: A detailed look at my Solar System setup
I have always built my own controllers, but this looks interesting. I must have the RS485 so the individual battery banks can report, and be under control. One of these on each bank might fill the bill. I have several hundred left over Pic processors, but haven't started that yet, so I will look at these processors first.
Are you saying that you built your own charge controller? Do you not currently have shunts for monitoring?
Sorry, I was not clear. The solar charge controllers are cheap, using several of those. I am building a power bank, that is composed of multiple sets of LiIon batteries (24 volt) in parallel. I will have a resistor under the cathode side of each battery stick, with a Tmos power FET switching ground under processor control. This will allow me to set the current each stick provides. If a stick drops voltage (bad cell) I will turn it off, and signal the failure. Then I can pull it, and repair it. With 350 sticks, a battery problem will not affect the available power. I will build 30 sticks in a 5 gallon bucket, buried in the ground to cool them. Each bucket will provide 1800 watts to the buss, through shottkey diodes (to reduce the diode losses). That is why I need the RS485, so I can poll single buckets. The controller I was talking about would be a simple SBC to controll the Tmos Power FETs; I/O is the problem, I need 30 analog inputs, and 30 digital outputs that I can PWM, per bucket. It looks like this SBC has 40 general I/Os to begin with, it might be eaasier to add some I/O instead of scratch building the whole thing. Might even need to look at a PC-104 brick to get the hoursepower I may need. I can just plug in the needed expansion onto the top of that unit.
BUT, once done, I will have batteries that are rated for 15,000 full cycles, that are repairable as needed.
At 0.05 ohm shunt I will waste .31 watt at full power. and the feedback voltage will be 0.125 volts. The MTP 3055E Tmos power FET will drive directly from the processor Output. I may need a better FET, but will test first; the Rds on is 0.15 ohms, so the FET will disapate 0.94 watts (1.5% of the total switched power). A MTW45N10E will only waste 0.188 watts when switching, looks like a better choice to me. That will be 65 watts lost to switch 21000, and with the sensor loss total control cost is under 200 watts. Not sure if I wll need to amplify the 0125 Vdc, until I see how good the analog inputs are.
Wow, I think I understand most of that. You should do a post with pictures.
Instead of doing a standard power wall, with hundreds of batteries in parallel where I can not see a single battery if it goes bad, I am making a stack of cells that will give me 24 volts (at 60 watts), and combining them through a special (shottkey) diode bank that has lower losses. This way, if I loose a cell, I loose 60 total watts, when I turn it off, with a very low risk of auto ignition. If the auto Ignition occurs in a normal power wall, the offending battery can not be isolated, and the adjacent cells provide the power to ignite, then they ignite from the heat...Poof, all gone.
In my battery, if one does burn, the surrounding plastic tube melts and allows water to enter. Water will not extiguish the fire, as it provides its own oxidizer, but it will lower the temperature below the level needed to sustain the fire. Since the battery (5 gallon bucket) is buried in the ground, and not in my house, my house is not burned down. With about 350 of the battery sticks, I get the needed power from the parallel of these sticks. This also lets me limit the current supplied from each battery stick, which will give me the longest possible cycle life.
The last paragraph above is the power cost of using these controls. The FET acts kind of like a relay, and closes (Drain to Source) when a signal is applied to the gate, except it works VERY fast. That way, I can regulate the amount power allowed to flow, by turning them off and on very quickly (70,000 times per second). By leaving them on, more that the time they are off, the output is increased. This is called Pulse Width Modulation (PWM) and is so useful that most processors include the function.
I really like this post on the rasberry pi, with python software. It may save me a lot of design work on the processor control setup. Since I can't afford to pay myself for the work, this may be a good route, LOL!