JP spot-welder, FET-switched, timed adj. pulse
- Thread starter okashira
- Start date
spinningmagnets
100 TW
I readily admit that I'm a doofus when it comes to electronics. That being said, I have a couple of questions.
It "sounds like" some of the FETs were damaged from excessive heat, and the cure was waiting between spot-welds longer to allow them to cool. Couldn't this design simply double the number of FETs?
Lebowski arranges his DIY controller FETs in groups, instead of in a long line. Would doing that here for the spot welder reduce inductance? I'm imagining two aluminum bars/plates with three threaded holes each. Each hole can have a FET screwed on from each side. Three FETs on one side, three on the other. Two heat sink bars = 12 FETs. Or...any layout that dramatically reduces inductance simply by a clever physical layout.
Maybe a two sided board? Four FETs on one side, four mirrored on the other (slightly offset so pins aren't conflicting). The Adaptto controller uses larger "package" FETs. Fewer FETs, but higher current per FET = compact layout = low inductance?
If the "turn off" causes cascading voltage spikes (27V was mentioned, from a 12V system), could we use common 60V FETs from an old ebike controller?
It "sounds like" some of the FETs were damaged from excessive heat, and the cure was waiting between spot-welds longer to allow them to cool. Couldn't this design simply double the number of FETs?
Lebowski arranges his DIY controller FETs in groups, instead of in a long line. Would doing that here for the spot welder reduce inductance? I'm imagining two aluminum bars/plates with three threaded holes each. Each hole can have a FET screwed on from each side. Three FETs on one side, three on the other. Two heat sink bars = 12 FETs. Or...any layout that dramatically reduces inductance simply by a clever physical layout.
Maybe a two sided board? Four FETs on one side, four mirrored on the other (slightly offset so pins aren't conflicting). The Adaptto controller uses larger "package" FETs. Fewer FETs, but higher current per FET = compact layout = low inductance?
If the "turn off" causes cascading voltage spikes (27V was mentioned, from a 12V system), could we use common 60V FETs from an old ebike controller?
Most of the board is pretty straightforward to design. The part I would have trouble with is the code for the PIC. I'm not so good at that. Seems like it's fairly simple from a programming standpoint for someone with experience. I have an analog version of a timing circuit that would produce the same kind of output using 3 timer chips but the parts count is a lot higher. The PIC or something similar makes things much cleaner.
One idea I had is if Riba would share his source code for the PIC, someone 'trusted' could make and sell programmed chips and give a percentage to Riba as a license fee. It's not going to make anyone rich but it would make these available again. The rest of the board is easy and an improved design could be made.
I've been very happy with the results I got with the original design Riba made, though I added a flyback diode.
One idea I had is if Riba would share his source code for the PIC, someone 'trusted' could make and sell programmed chips and give a percentage to Riba as a license fee. It's not going to make anyone rich but it would make these available again. The rest of the board is easy and an improved design could be made.
I've been very happy with the results I got with the original design Riba made, though I added a flyback diode.
I got all the parts to build the 3 boards I have. The ones I have are Arduino nano powered and writing code for Arduino is easy. I will post it when I find it.fechter said:
Also I will test the boards and see what I can do to make them switch clean and safe. Mine use 60v rated fets which is most likely for the big overshoot at turn off.
I can redesign these if people want and use something with a better driver. But when you shut off to fast it will cause a very big voltage spike which is likely whats killing most boards on this forum.
okashira
10 kW
spinningmagnets said:
Problem with hv FETs is that they cant handle as much avalanche at turn off.
With the diode circuit and tvs diodes it could work well.
But the irf1234 6x and diode + tvs diodes has been rock solid for us.
Arlo1 said:
How much do Audrio nano boards cost?
I'm still not sure what the best switching speed is. If too slow, the FETs will overdissipate and blow. If too fast, you get big voltage spikes. Either way, the FETs take a beating. Having the external flyback diode has to help spread out the energy. TVS diodes seem like a good idea too.
$3 ushttp://www.banggood.com/ATmega328P-Nano-V3-Controller-Board-Compatible-Arduino-p-940937.htmlfechter said:
its just 2 single pulses If you are refering to the PWM frequency a nano can do that does not mater.
As for the switching speed from on to off and off to on thats controlled with the gate driver and the resistors.
Diodes are a good idea to help cover the diodes in the opposite side of the H bridge.
As well its a good idea to have some sort of low ESL cap on the board.
Banggood? Like it goes bang. Got to love those Chinese names. Like all the lithium batteries with 'fire' in the name.
Anyway, that's plenty cheap. I like it.
Yes, I was referring to the gate driver about switching speed. There seems to be a wide variety of opinions on what is best.
Anyway, that's plenty cheap. I like it.
Yes, I was referring to the gate driver about switching speed. There seems to be a wide variety of opinions on what is best.
okashira
10 kW
fechter said:
I think it doesn't really matter. a slow turn off will be beneficial if there is no flyback circuit
pguk
100 W
So today I replaced the power board - this time using individual 75R gate resistors.
I hooked up the scope to 4 adjacent gate pins and captured some images from a few weld fire cycles:
Here's the image of interest - gate voltages at turn on. Individual gate resistors should now allow those gate voltages to do what each FET wants to do. I set the 0v of the traces with an equal spacing of 4 grid notches - there does seem to be a little variation there, the curves look to be rising slightly differently.
Here's the turn off image. We can see here a clear variation in the shape on ch2.
This is the same image as above but zoomed out. You can see the ringing that happens on the gates when the magnetic field of the weld cables collapses and hammers the FETs with about 27v plus back emf.
That's all I had time for today - progress I think
I hooked up the scope to 4 adjacent gate pins and captured some images from a few weld fire cycles:
Here's the image of interest - gate voltages at turn on. Individual gate resistors should now allow those gate voltages to do what each FET wants to do. I set the 0v of the traces with an equal spacing of 4 grid notches - there does seem to be a little variation there, the curves look to be rising slightly differently.
Here's the turn off image. We can see here a clear variation in the shape on ch2.
This is the same image as above but zoomed out. You can see the ringing that happens on the gates when the magnetic field of the weld cables collapses and hammers the FETs with about 27v plus back emf.
That's all I had time for today - progress I think
Two rightmost fets have longer current path... May not be fatal, but not good.
If you are, of course, willing to take my humble advice. I know I'm not your welder guru, but still :lol:
Also, long wires for gate signals - random soldering... Not looking too good. Also those resistors could act like antennas...
If you are, of course, willing to take my humble advice. I know I'm not your welder guru, but still :lol:
Also, long wires for gate signals - random soldering... Not looking too good. Also those resistors could act like antennas...
destro23
1 W
I Origionally posted this in an AUS-Spot welder Thread and realized i'm making a North American list of parts that we can get from Mouser. So i'm reposting this list here and will ask to have the other deleted.
Below is the closest mouser parts I can find.... The only issue I have is possible differences and how it'll effect the final welder.
We can use this a starting point for a North American parts list.
Everything underlined is the closest i can find on mouser...If i can get a full list that everyone can agree on maybe i can ask to update the file on github for us on this side of the pond.
Mosfet IRF1405 55V 169A IRF 1405 1.15 € 9.20 € Infineon / IR IRF1405PBF
Mosfet Driver MCP14E10-E/P MCP 14E10-E/P 1.70 € 6.80 € Microchip Technology MCP14E10-E/P
Poti 6mm Linear 50K RK09K113-LIN50K 0.99 € ALPS RK09D1130A1L
7 Segment Display SC 56-11 RT 0.84 € Kingbright SC56-21EWA
1x20pol female header BL 1X20G8 2,54 0.26 € Harwin M20-7822046
1x36pol male header SL 1X36G 2,54 0.17 € 3M Electronic Solutions Division 929647-01-36-I
PCB connector set 2pin PS 25/2G WS 0.29 €
IC socket 8 pole GS 8 0.03 € 0.12 €
diode 1N5817 1N 5817 0.06 € 863-1N5817RLG
elektrolytic capacitor 1 µF RAD 1/63 0.04 € 598-SK010M063ST
ceramic capacitor 100pF KERKO 100P 0.06 € Xicon 140-50P2-101K-RC
Resistor 221 Ohm METALL 221 0.08 € 0.16 € 603-MFR-25FBF52-221R
Resistor 100K-Ohm METALL 100K 0.08 € 279-LR1F100K
LED 5mm green LED 5MM GN 0.06 € 696-SSL-LX5093GDTR
Just realized again i'm in wrong thread again.. this is the for adruino spot welder in instructibles.
Below is the closest mouser parts I can find.... The only issue I have is possible differences and how it'll effect the final welder.
We can use this a starting point for a North American parts list.
Everything underlined is the closest i can find on mouser...If i can get a full list that everyone can agree on maybe i can ask to update the file on github for us on this side of the pond.
Mosfet IRF1405 55V 169A IRF 1405 1.15 € 9.20 € Infineon / IR IRF1405PBF
Mosfet Driver MCP14E10-E/P MCP 14E10-E/P 1.70 € 6.80 € Microchip Technology MCP14E10-E/P
Poti 6mm Linear 50K RK09K113-LIN50K 0.99 € ALPS RK09D1130A1L
7 Segment Display SC 56-11 RT 0.84 € Kingbright SC56-21EWA
1x20pol female header BL 1X20G8 2,54 0.26 € Harwin M20-7822046
1x36pol male header SL 1X36G 2,54 0.17 € 3M Electronic Solutions Division 929647-01-36-I
PCB connector set 2pin PS 25/2G WS 0.29 €
IC socket 8 pole GS 8 0.03 € 0.12 €
diode 1N5817 1N 5817 0.06 € 863-1N5817RLG
elektrolytic capacitor 1 µF RAD 1/63 0.04 € 598-SK010M063ST
ceramic capacitor 100pF KERKO 100P 0.06 € Xicon 140-50P2-101K-RC
Resistor 221 Ohm METALL 221 0.08 € 0.16 € 603-MFR-25FBF52-221R
Resistor 100K-Ohm METALL 100K 0.08 € 279-LR1F100K
LED 5mm green LED 5MM GN 0.06 € 696-SSL-LX5093GDTR
Just realized again i'm in wrong thread again.. this is the for adruino spot welder in instructibles.
okashira
10 kW
Arlo1 said:
Yall are right but it's actually not a huge deal because it's single pulses
Yeah if it was a motor controller at 10khz it would blow up in 200 milliseconds
Increase the gate resistor to 150 ohm or so to calm things down
Its much more important to run a flyback circuit on the power side
pguk
100 W
The power board sure isn't gold standard from any point of view - but for a DIY build it's okay, it shows that it's quite easy to build a board that works using copper clad board and a dremel. For those who don't want to be getting busy with pcb computer design (kicad).
Code was developed in the arduino. The language is like simplified C, but in the background, there is powerful software underpinning the whole thing, so when necessary one can write more complex code, even modern C++. With this project it isn't necessary to get more complicated. I spent some time playing with the examples, getting led's to blink and playing with switches (cheap switches are very noisy - fix = modify an example switch debounce piece of code). Then I worked on how to run an lcd dislpay so I can see and change the 3 parameters - pulse 1, delay, 2nd pulse. After that I worked on getting the microcontroller off the arduino board and onto my own design pcb (kicad). All this took time but you get there in the end. To flash the chips you can do this with two arduinos, one has programmer code on it, the other has your chip to be programmed. So I would say, don't be afraid to have a play with the arduino - it's fun to get there - I got there and I'm no lebowski, just a fellow with some interest and a little spare time.
Code was developed in the arduino. The language is like simplified C, but in the background, there is powerful software underpinning the whole thing, so when necessary one can write more complex code, even modern C++. With this project it isn't necessary to get more complicated. I spent some time playing with the examples, getting led's to blink and playing with switches (cheap switches are very noisy - fix = modify an example switch debounce piece of code). Then I worked on how to run an lcd dislpay so I can see and change the 3 parameters - pulse 1, delay, 2nd pulse. After that I worked on getting the microcontroller off the arduino board and onto my own design pcb (kicad). All this took time but you get there in the end. To flash the chips you can do this with two arduinos, one has programmer code on it, the other has your chip to be programmed. So I would say, don't be afraid to have a play with the arduino - it's fun to get there - I got there and I'm no lebowski, just a fellow with some interest and a little spare time.
Quokka
1 kW
destro23 said:
I think you may have the wrong poti. Pretty sure you need a 10k version
destro23
1 W
Quokka said:
I actually don't have a potentiometer listed above (sorry formatting is a little messed up but the one listed is from the github parts list for the arduino spot welder)
I couldn't find a close enough one on mouser so i don't have a link after that listing.... i'll dig around for that and update the thread once i can.
ALPS RK09D113000D is the closest i can find right now
Mouser Potentiometers, Trimmers & Rheostats
Jorge Rocha
100 µW
- Joined
- Feb 24, 2016
- Messages
- 9
Hello everyone,
I have bought a spot welder to Ribas, and everything worked well, 'till now, since the last week i'm having problems with the mosfets, i blow 2 of them 1 week ago and i replace them, now yesterday i just blow more 3, i don't understand why is that happening 'cause i'm doing everything normal (i'm welding 18650 samsung cells), does anyone know what's possible going on to blow the mosfets?
Thanks
I have bought a spot welder to Ribas, and everything worked well, 'till now, since the last week i'm having problems with the mosfets, i blow 2 of them 1 week ago and i replace them, now yesterday i just blow more 3, i don't understand why is that happening 'cause i'm doing everything normal (i'm welding 18650 samsung cells), does anyone know what's possible going on to blow the mosfets?
Thanks
Same thing with me..........the FETs weaken with every weld it seems and then they go. If one goes bad and you replace it you still have weak sisters that msu not share the load and they go...Others have said here...replace them all when they go bad. Other subtle factors are strength of battery, resistance and onductace of all leads and connections, fly back transients......
destro23
1 W
destro23 said:
Just updated origional thread with a few more items... hope someone has time to compare/ check these i guess i'll order a bunch of pcb's soon. No one in North America wants to build this thing? Am i the only nerd lol ?
i'm having some trouble with what the 2 pin connector falls under or if there is a satandard name for it.
Jorge Rocha said:
Wow, it burn the leg off. I'd have to say that's probably from too much current :wink: Maybe try a lower voltage or shorter pulse time.
The only other thing I can think of is if one of the other FETs was loafing or not working, all the load goes to the remaining ones. Using a matched set when replacing might make them last longer. Check the gate drive on each one.
markz
100 TW
How do you match MOSFETS?
Just get them from the same batch. You could test them, but there is more margin for error in that process than just using ones from the same batch.
