JAP (Just Another Powerwall or Jesus Angel's Powerwall)

jesusangel

Member
Joined
May 28, 2017
Messages
277
I do not want to bore everyone with stories of what happened after I decided to build my powerwall so far, I may from now on when I do some updating I will be more detailed but summarizing everything is not fun and I feel a bit lazy today. So let's get down to the facts:

Cells, + - 100 collected here and there and + -700 bought tosabadothrough this forum (120 Dell laptop batteries).


image_vbsklq.jpg


image_aqbffx.jpg


image_xzazhy.jpg


Test equipment 8 tp4056 and 2 Opus BT-C3100.


image_yjjrtu.jpg


Cell holders, I liked the idea of Average Joe, pvc pipesince I work in a company in the sector and had easy access to these materials but it is difficult to find pipe by its interior measurement so I discarded it and bought the holderswe all haveat aliexpress, but the seller had problems and after 12 days since he placed the order as shipped without being I had to cancel it. Finally a friend is making me the holderswith his 3d printer, he says he does not care and it's cheap, we'll see what he thinks after havedone the 42 I need initially.


image_oeuvcu.jpg


About the fuses I also liked the choice ofAverage Joe, so I bought them. I've made my first test with cells with less than 1000mAh.


image_ntsmnh.jpg


Yes I know,the same problem of cold joins, tomorrow I will receive a 100w soldering Iron.


I plan to use the owitte's system because I like it, I have easy access to copper pipes and a free wall :D.

I will try to make some kind of BMS with an ESP32, have enough analog inputs for 14S, at least will monitor it, to balance is another problem...

It took me a month to process the cells, now I only have to give a second pass and see if any lose the charge after a month to discard it.

On panels and inverter I'm still thinking.
 
Looking good :) I like the custom cell holder. Nice and clean. You must like teal a lot for your colors :p

What kind of solder are you using? I notice there is brownish residue left on some of the cells in that last picture. Sometimes the flux residue can actually rust the ends of the cells; so you might wanna take a brass brush and wipe it off to be safe.
 
yep, gota wash the flux off !!
 
jesusangel said:
I do not want to bore everyone with stories of what happened after I decided to build my powerwall so far, I may from now on when I do some updating I will be more detailed but summarizing everything is not fun and I feel a bit lazy today. So let's get down to the facts:

Cells, + - 100 collected here and there and + -700 bought tosabadothrough this forum (120 Dell laptop batteries).





Test equipment 8 tp4056 and 2 Opus BT-C3100.



Cell holders, I liked the idea of Average Joe, pvc pipesince I work in a company in the sector and had easy access to these materials but it is difficult to find pipe by its interior measurement so I discarded it and bought the holderswe all haveat aliexpress, but the seller had problems and after 12 days since he placed the order as shipped without being I had to cancel it. Finally a friend is making me the holderswith his 3d printer, he says he does not care and it's cheap, we'll see what he thinks after havedone the 42 I need initially.



About the fuses I also liked the choice ofAverage Joe, so I bought them. I've made my first test with cells with less than 1000mAh.



Yes I know,the same problem of cold joins, tomorrow I will receive a 100w soldering Iron.


I plan to use the owitte's system because I like it, I have easy access to copper pipes and a free wall :D.

I will try to make some kind of BMS with an ESP32, have enough analog inputs for 14S, at least will monitor it, to balance is another problem...

It took me a month to process the cells, now I only have to give a second pass and see if any lose the charge after a month to discard it.

On panels and inverter I'm still thinking.


Hi, looking at your idea on ESP32 as a BMS. Looking at doing the same, not started, just kicking around some numbers. Would be great to collaborate. Was looking at using python and clustering them...all independent.

I was going to use a mosfet which would dump around ~<400ma for balance, a voltage divider on a analoguepin to measure voltage, and a temperature monitor 1-wire/i2c.?

Software could determine:
rate of charge
discharge time
compensate for temperature

Get back if you are interested.
 
My idea was make something like that.


But as many cells as posible with that ESP32,

https://www.banggood.com/ESP32-Deve...32-ESP-32S-Board-p-1109512.html?rmmds=myorder

it has 14 analog inputs but not 14 additional pins to activate the mosfet, so maybe must use a couple and control 7 banks with each one.

Maybe will add some lcd/tft to show the values, but the main reason to chose esp is to send data to a grafana server, so maybe on site screen is secondary.

I will post when I have the ESP32 and made some test.
 
jesusangel said:
My idea was make something like that.


But as many cells as posible with that ESP32,

https://www.banggood.com/ESP32-Deve...32-ESP-32S-Board-p-1109512.html?rmmds=myorder

it has 14 analog inputs but not 14 additional pins to activate the mosfet, so maybe must use a couple and control 7 banks with each one.

Maybe will add some lcd/tft to show the values, but the main reason to chose esp is to send data to a grafana server, so maybe on site screen is secondary.

I will post when I have the ESP32 and made some test.



Yeah, for each battery pack you could have a 1-wire 2405, (opto isolated)Mosfet and temp probe (DS18B20), with another wire to measure voltage. It wouldbe nice to have an isolated voltage sensor on a 1-wire....so only a single pair string across all cellsexpandable.

With theESP32 proving the smarts and a web interface to manage the environment.

Voltage measurement would beproblematic when measuring multiplevoltages across a string ofcells, soneed to think about this more on the easiest and most accurate solution.

Like the YouTube that you provided, I was thinking on similar lineshowever with a ESP8266, it would handle temp 1-wire/i2c with no additional parts, analogue input for voltage measurement, plus the driver pins for the mosfet. Isolation is not needed as each unit would be stand alone, plus the ESP8266 is super cheap. Power would come from the lipo pack.
shrinkable / expandableand cheap per unit.

Both approaches have there advantages and disadvantages. The biggest issue for me is voltage measurement for each cell if done centrally, and how this can be done...easily.
 
Yesterday I made a test with single cells pack and some components I already have, Arduino Uno, 2x 4WR82J resistors in series and a 75N75 mosfet.
This mosfet has a Drain voltage 40A at 10V, as the arduino only can feed 5v that is too low so I only discharge a current of 0.25A from battery.

Maybe a mosfet like IRL640 that drain 10A at 5v is more appropriate for us, moreover ESP32 have 3.3v logic.

About analog readings with only the ESP32 you have to make voltage dividers and make some test and calibrate it.

I do not think we need isolation while we feed the whole system with 7s so the negative of the first series is the negative for all, use a step down convert to power the ESP32 and voltage dividers so the last series that may have 29.4V can be measured with ESP32 in a range of 3.3V.

After all test I do not rule out that the best solution is an ESP8266 per series. Maybe converting that high voltages to 3.3V range will lose accuracy.

Next I have to make an order of components I will buy mosfet and resistors more appropriate to test a 7s bank.

Regards
 
jesusangel said:
Yesterday I made a test with single cells pack and some components I already have, Arduino Uno, 2x 4WR82J resistors in series and a 75N75 mosfet.
This mosfet has a Drain voltage 40A at 10V, as the arduino only can feed 5v that is too low so I only discharge a current of 0.25A from battery.

Maybe a mosfet like IRL640 that drain 10A at 5v is more appropriate for us, moreover ESP32 have 3.3v logic.

About analog readings with only the ESP32 you have to make voltage dividers and make some test and calibrate it.

I do not think we need isolation while we feed the whole system with 7s so the negative of the first series is the negative for all, use a step down convert to power the ESP32 and voltage dividers so the last series that may have 29.4V can be measured with ESP32 in a range of 3.3V.

After all test I do not rule out that the best solution is an ESP8266 per series. Maybe converting that high voltages to 3.3V range will lose accuracy.

Next I have to make an order of components I will buy mosfet and resistors more appropriate to test a 7s bank.

Regards

With a mosfet you need the gate voltage to be1-2v +/-difference from the source depending on the depletion type to turn the FET on. So unless you switching batteries in and out of circuit, the ground with respect to your driving circuit becomes an issue.

So say a4 battery pack will have a source voltageof say 12V to ground, so to switch this on you will need a difference of 1-2 volts. You could do this with say an opt isolated version of the FET...BUT as the voltage dynamic changes the current (power) across the loadresistor will also change. So at 60v the power dissipated will be larger than when its 4V.

So if you are intending to go down your suggested path, each cell will have a floating earth, with its own discharge circuit, opt-o isolated with a n-channel mosfet (depletion mode / drive the gate voltage-2V to drain to get the GS negative.).

I have been kicking around the idea of a capacitor based switch load, which will switch between cells to lift ordischarge cells, more energy efficient with less issues ofheat dissipation. Still playing with the idea.
 
So, if we need the same GND to Gate (esp32) and Drain, we need at least all outputs to be pwm so we can send less voltage to mosfet connected to high series, there are bigger resistors too so can manage more watts, but at these point I wonder if wouldn't be easiest use relay boards, they are optoisolated and only have to worry about the resistor connected to NO and send a high or low to port and you get a fix current drain.
 
jesusangel said:
So, if we need the same GND to Gate (esp32) and Drain, we need at least all outputs to be pwm so we can send less voltage to mosfet connected to high series, there are bigger resistors too so can manage more watts, but at these point I wonder if wouldn't be easiest use relay boards, they are optoisolated and only have to worry about the resistor connected to NO and send a high or low to port and you get a fix current drain.

I have been sniffing around for designs.

For me, I'm goingwithsomething like this. An ESP per battery pack. The circuit provides a 1.2W dump, and voltage sensor. If I want to add temp ill go with a Dallas 1-wire and which can be hooked directly to the MCU with a 4.7k pull up.


image_tnabxa.jpg


To your question a relay board Yes and have seen this been done that way, or go with aopto isolated FET.

Cheers
 
Never tested but I've readed maximum adc voltage for ESP8266 is 1v, so that voltage divider with R3 4 and R2 10 at cell max of 4.2V put 3V max at ADC pin. Zener is ok for 3v3 max too, ESP8266 have internal protection to overvoltgage until 5.8V, but over a 1V you will get 1024 readings.

About BSS308PE it is P channel and makes me wonder if the design was wrong, but found this:

https://www.baldengineer.com/p-channel-mosfet-tutorial-with-only-positive-voltages.html

and when is better use a P channel instead of N channel, good point.
 
jesusangel said:
Never tested but I've readed maximum adc voltage for ESP8266 is 1v, so that voltage divider with R3 4 and R2 10 at cell max of 4.2V put 3V max at ADC pin. Zener is ok for 3v3 max too, ESP8266 have internal protection to overvoltgage until 5.8V, but over a 1V you will get 1024 readings.

About BSS308PE it is P channel and makes me wonder if the design was wrong, but found this:

https://www.baldengineer.com/p-channel-mosfet-tutorial-with-only-positive-voltages.html

and when is better use a P channel instead of N channel, good point.

Thanks for that, I'll redo the divider. Will also add a LED.
 
As we use to say here I started the house by the roof so I already ordered the inverter

MPP Solar MPI Hybrid 4k

http://www.ebay.es/itm/4000w-Hybrid...e=STRK:MEBIDX:IT&_trksid=p2057872.m2749.l2649

Delivery date is next thursday 07/18, supposedly sent from Germany to Spain.

Next question is the place, I will install it at garage that is the basement of my house and has a 8 meters wall and a good place for the inverter at the end of a little stairs. I'm building my powerwall like Oliver

http://secondlifestorage.com/t-There-s-a-new-wall-in-Germany?page=1

I hope next few days have first 14s finished and wall busbars installed.

That's how I plan to do it.


image_xvnwwa.jpg


About solar panels, I think the next summer or maybe before if I'm bored (I want to build bms, and graphana server before and keep collecting cells) meanwhile I've moved to time discrimination rate with my energy provider and decreased the power to 2.2Kw. My plans is to charge at night with cheap rate and feed part of the house at day, Iwill move more parts of the house to the inverter when I had more storage capacity. By now only have 5 KWh of procesed cells and 1.5 kw packed.

Stay tuned for more updates :D
 
On your packs where the wire curves over the edges and joined and soldered in the copper pipe, maybe you could take a zip tie and strap that last bend to the pack. This would give extra strength right there and help keep from stressing the fuse wires when you're connecting/moving the packs.

Nice setup, though! :)
 
Thank you Korishan

Yes, zip tie will help, my pack will be connected vertically with negative bus bar down, so most the weight will be on that side that it is very strong, positive bus bar will give some flexibility and will make it easy to screw, somes fuses already broken while handling.

In last packs I've changed from 2 AWG 10 (6mm2) twisted cables to only 1, for a maximum of 20 amps per pack (10 per cable) is enough and easiest to turn, solder, etc
 
yes, your packs look a little 'fragile'... I guess the weight is like mine, around 1kg per pack. You should not underestimate that. I fixed my terminals with zip ties, too, and most of them broke overe time, because I zipped them with too much tension. If you make them a little loose, they will hold the weight of the pack and you'll still be able to adjust the terminals on the bus bar bolts. If you want, I can show youdetailed pictures of how I did it.

Have sun!
Oliver
 
owitte said:
yes, your packs look a little 'fragile'... I guess the weight is like mine, around 1kg per pack. You should not underestimate that. I fixed my terminals with zip ties, too, and most of them broke overe time, because I zipped them with too much tension. If you make them a little loose, they will hold the weight of the pack and you'll still be able to adjust the terminals on the bus bar bolts. If you want, I can show youdetailed pictures of how I did it.

Have sun!
Oliver

Of course Oliver, any tip will help, today I've made the busbars and tomorrow I will screw them to the wall and make the electrical installation waiting for the inverter that yesterday left Germany by dhl not dhl express so I supose tomorrow it will not be here as ebay said when I bought it.

Today I've bought too a bag of zip ties.
 
I love that you are making 20p packs. I am also. It seems the smaller packs are not very common. I like that I can add as many parallel packs to the initial 14. So I can make a second string of 14s20p or 14s80p or whatever I want. I just wanted small packs for ease of handling and a lot less re soldering if I have a bad cell and need to rebuild a pack. And I'm poor! A smaller first string is more manageable within my budget. My accountant (read wife) is very keen to get a return on investment! Lol!

Thank you for sharing! Please keep updating. I love how this community is so open and freely giving of good advice.
 
Back
Top