My First 48V 6.5KWh Li-Ion Powerwall šŸ„°

Could it be that the MOSFET was overheating when the opto was Off?
The opto reverse max voltage is about 6V, so if you apply 12V, it cannot block the too high voltage and let current through until the MOSFET gate is at 6V.
This MOSFET at 6V is only partially On, so would heat up.

Not entirely sure about all this reverse bias stuff... it's generally best avoided.
 
Uhm, 100% I did all the heating tests with opto On and checked the MOSFET was cold before starting the test. As soon as I finish at work I'll try inverting the opto again; that's the good thing of soldering components without and boards, it's up&running in 10 minutes!
 
that's the good thing of soldering components without and boards, it's up&running in 10 minutes!

This might be a stupidly obvious suggestion... but you do have these plastic breadboards and bunch of jumper wires, right? They're super cheap and awesome for just trying things out quickly. I keep a set, just in case.
breadboard.jpg

But after that, factory produced PCB with SMD components are the way to go. Simply better in almost every way. I recently stepped up my game and started manually soldering a 4x4mm UQFN-16 package Microchip. Just the size of grain of rice, yet runs at 32Mhz with RAM/ROM/EEPROM and all sorts of modules inside!
SMD.jpg
 
manually soldering a 4x4mm UQFN-16 package Microchip. Just the size of grain of rice
Wish the gear I soldered today was that massive! lol ( i have little else to add to the conversation :p )
 
Wish the gear I soldered today was that massive! lol ( i have little else to add to the conversation :p )
There's always a bigger fish.. eh.. smaller chip.. I'll have to leave the smartphone and laptop repairs to the real Pros for now... but I'm learning all sorts of interesting stuff from two of my favourite channels right now about repairing various computer components:

Makes me a bit upset now thinking of all the defective computer components I threw away in the past, and half of them were probably easily fixable :-(
 
Love Alex, he really needs to stop taking in jobs he can't possibly complete in a reasonable time frame & old Luis is getting a bit too much - he needs to get back his roots..Learned much from both that's for sure!
 
Love Alex, he really needs to stop taking in jobs he can't possibly complete in a reasonable time frame & old Luis is getting a bit too much - he needs to get back his roots..Learned much from both that's for sure!
Yeah, some of Alex's repairs hardly make economic sense. I guess his thinking is that those average out with the simple fixes, and it also builds his reputation. And of course there's the income from YouTube. But above all, I think he does it just for the joy of completing a challenging repair, judging by his grinning afterwards.

I think Louis has a team of ~15 working for him. So instead of spending time repairing Macs, publicly fighting for "Right to Repair", and indirectly promoting his company/brand probably makes more sense for growing his business. Somehow I see him going properly into politics... one can hope.
 
This might be a stupidly obvious suggestion... but you do have these plastic breadboards and bunch of jumper wires, right? They're super cheap and awesome for just trying things out quickly. I keep a set, just in case.
yes, but since I've been using on-the-fly soldering, without breadboard and wires, I go a lot faster. Plus I noted that breadboards get destroyed by TO-220 ICs and pins start getting lose. I do use them though
But after that, factory produced PCB with SMD components are the way to go. Simply better in almost every way. I recently stepped up my game and started manually soldering a 4x4mm UQFN-16 package Microchip. Just the size of grain of rice, yet runs at 32Mhz with RAM/ROM/EEPROM and all sorts of modules inside!
That's a wow! Surely the best way to go for a final board. I tried soldering SOT-23 packages last year with no success at first go. INA219 current sensor, @Wolf was talking about them in another thread, was my first (failed) attempt. But I'll try again.

For a top quality result I'll surely go for a PCB manufacturer. I did not until now because I must learn how to use EAGLE, I've been using a quick-learning-curve software called Express PCB which doesn't create Gerber files.
 
That's a wow! Surely the best way to go for a final board. I tried soldering SOT-23 packages last year with no success at first go.

For a top quality result I'll surely go for a PCB manufacturer. I did not until now because I must learn how to use EAGLE, I've been using a quick-learning-curve software called Express PCB which doesn't create Gerber files.
For SMD, solderpaste and hotair station is a game changer. So much easier and delivers consistent good results, even if it's applied with just a toothpick and with a $30 Chinese station. I have a small test project right now - will try to film it.
A bit more pricy, but use a solder stencil and hotplate/oven, and the product is basically indistinguishable from factory units!

I use KiCAD. It's free, increasingly popular, and has improved significantly in the last years, and is under very active development.
Another very popular one is EasyEDA - developed by the group behind both JLCPCB (PCB manufacturer) and LCSC (major parts distributer, like Mouse/RS/Digikey). I haven't used it, but I hear it's easy and works extremely very well if you also want them to manufacture the PCB and solder on the parts.
 
Have a look at this. It's really that simple. Some sloppy smear of solder paste, apply hot air, and it all just flows into the right spot due surface tension of the solder, capillary action between SMD and PCB, and the repelling from the solder mask (the greenish paint). At least that's my theory.

YOUTUBE Hot Air soldering
 
Last edited:
Very interesting need to get a hot air smd solder station and more solder paste. the Solder paste i have is old starting to dry out. Any recommendations for a SMD rework station?
Later floyd
 
Solder paste i have is old starting to dry out. Any recommendations for a SMD rework station?
I take what I need out of the solder paste tube, and mix in a drop or two of liquid flux, trying to get the consistency of Nutella. Liquid enough to easily stick to the PCB/SMD, but not enough to flow all over the place.
I keep the tube in a zip lock bag inside the fridge. Supposedly extends the shelf life.

I have a cheap Chinese "8586" unit, and it's perfectly fine for my occasional use.
When first assembling a PCB, I use the "946C" hotplate to heat the whole PCB at once. But that has to be done before adding components that stick out the back, or are temp sensitive.
 
Sounds good, I have liquid flux and The solder was old when i bought it so maybe 4 drops of flux hopefully that will soften the paste up enough to use it.

Later floyd
 
Parked everthing for a week because I'm changing project at work, learning everything again (and again and again!).

1) Schematics for heating/cooling: pins 3&4 of optocoupler are good :rolleyes: IN any case I slightly changed the schematics to use a BJT transistor to switch MOSFET on. Share as soon as I draw it.

2) My first reshrinking is done! Supa Result! Have no specific tools, so I just used my wife's hairdryer on max heat (with low/medium heat nothing happened); and medium air flow; and small magnets to keep cell standing (like @hbpowerwall showed in one of his YT videos).

Rewrap 1.jpeg Rewrap 2.jpeg Rewrap 3.jpeg

To remove old wrapping just used a really thin blade, I took it off a razor blade; doesn't scratch or damage cell, only a bit dangerous to hold, they're so small:

Rewrap cutter.jpeg
 
Last edited:
Ah, also broke a Peltier cell. I didn't expect it to brake. They make ever so much heat on the hot side and for my tests I'm just using a CPU heatsink. After a couple of minutes it just stopped working. I found out that red wire must have got some damage, if I move it it starts working again. I'll need real big 1.5cm high (2/3") aluminium plates.

WhatsApp Image 2021-07-25 at 20.46.24.jpeg
 
Yeah, they're easily damaged by overheating. The problem often is not the heatsink, but the poor heat conduction from the Peltier. Need thermal grease and pressure, or thermal adhesive tape.

Dave Jones from EEVblog just released a video on the freewheel/flyback diode, and the potential high voltage (700V !!!). Perhaps a bit hard to follow for beginners, but as always very interesting.
EEVblog 1409 - The DANGERS of Inductor Back EMF
 
Yeah, they're easily damaged by overheating. The problem often is not the heatsink, but the poor heat conduction from the Peltier. Need thermal grease and pressure, or thermal adhesive tape.
Ehm yes, my test setup had pieces just sitting one on the other... :whistle:
Dave Jones from EEVblog just released a video on the freewheel/flyback diode, and the potential high voltage (700V !!!). Perhaps a bit hard to follow for beginners, but as always very interesting.
EEVblog 1409 - The DANGERS of Inductor Back EMF
Oh very nice, I'll go check it.

In the weekend I found a very good answer on stackexchange about calculating dissipated power on flyback diodes... I must say, I was wondering how to choose the correct diode.

Answer by user Nick, here: https://electronics.stackexchange.com/questions/110574/how-to-choose-a-flyback-diode-for-a-relay

[StackExchange] Nick: "If we solve for the power, something very interesting happens... P=((1/2)L(I2)R)/(5L) Here, L cancels out and P=1/10(I2)R. We know that R is the resistance of the diode in conduction and I is the current flowing through the diode during the discharge. But now, what is the diode current during discharge?"
^^ Loved it!
 
Wow I am learning lots reading through this thread. Very timely since I just spent much of yesterday trying to replace/upgrade a blown MOSFET on a PWM LED dimmer so that it can run a long 24vdc strip light that is just barely over spec. Fried lots of components and had some fun but still getting too much heat buildup even switching to MOSFETs that have less than half the Ohm on resistance. So now it is time to dig even more into mosfets and or get some nice heatsinks and thermal tape.
 
Wow I am learning lots reading through this thread. Very timely since I just spent much of yesterday trying to replace/upgrade a blown MOSFET on a PWM LED dimmer so that it can run a long 24vdc strip light that is just barely over spec. Fried lots of components and had some fun but still getting too much heat buildup even switching to MOSFETs that have less than half the Ohm on resistance. So now it is time to dig even more into mosfets and or get some nice heatsinks and thermal tape.
I got up to 12V 3.5A (load) without any heating at all from the MOSFET, I thing I can go even higher with no heatsink. How many amps do you need for your leds and at what voltage?
The trick, as said by our most expert friends, is to give the gate enough voltage. So, in my case, a 12V (I tried up to 18V with identical results) on the gate gave me good results. Lets see if I can post an updated schematics, dunno because I'm getting ready to go on holiday so my wife keeps calling me for this and that LOL
 
I got up to 12V 3.5A (load) without any heating at all from the MOSFET, I thing I can go even higher with no heatsink. How many amps do you need for your leds and at what voltage?
The trick, as said by our most expert friends, is to give the gate enough voltage. So, in my case, a 12V (I tried up to 18V with identical results) on the gate gave me good results. Lets see if I can post an updated schematics, dunno because I'm getting ready to go on holiday so my wife keeps calling me for this and that LOL
I am trying to PWM dim about 8A of light strips running off 24vdc nominal so about 29vdc max. I figured design it for 10A for some buffer. I think part of my problem is I was trying to just upgrade the MOSFET on a burned dimmer I have running off a 555 timer. I was able to read the full 28vdc on the gate which seams problematic and I don't know the switching frequency or gate charge capacity and was reading a significantly higher ohm resistance across the MOSFET when on.

Now after some more learning about MOSFETS I am thinking it is time to build my own circuit to run of an ESP32 so I can add remote control. Of course that becomes a big project and I need to decide what I want to use all the extra pins on the ESP32 for :)
 
Back
Top