SPOT WELDER WITH PIC16F88 tested - YOUTUBE video

[kalpakchiev]

Spot Welder based on PIC 16F684 with LCD. The firmware is in PBP. Tested with Proteus. I am making the PCB in ORCAD now. When finished I will send pictures and samples. Any suggestion will be welcomed

[SuperDrive]

Nice project, i always want thing like this, waiting for your progress.

One of my interest: http://www.ultrakeet.com/index.php?id=article&name=cdWelder

[kalpakchiev]

Firmware changed:
Added welding time MIN and MAX limit with display
Added welding voltage MIN and MAX limit with display

[pickit2]

I think you need to add Green and Red Led's and a weld only button. in the workshop I see you set voltage and time then forget about settings, then you only need leds to say when you can weld, and then hit the start button.

[kalpakchiev]

Unfortunately there is no free I/O. But when the device is not ready LCD shows "CHARGING"(not ready) and when cap charged LCD shows "READY". When in working mode (MODE=1) LCD shows all the time selected TIME and VOLTAGE so operator can see current parameters. Also in working mode "UP" button works as "WELD" button and I am planing to connect an foot switch in parallel with "UP/WELD" button.
Another option is to  replace 16F684 with 16F88 for example which have more I/Os and then a buzzer and dedicated WELD button can be added.
Your opinion is highly appreciated.

[pickit2]

This would make a good plastic bag sealer project too. I would go for more I/O ports too, as using Just the LCD and not LEDS in the work place, would make it hard on operator in repeat operations.

[kalpakchiev]

Yes probably a bigger pic  will be better for this task - i will remake it for 16f88. Plastic bag welder is interesting idea

[solutions]

Unfortunately there is no free I/O. But when the device is not ready LCD shows "CHARGING"(not ready) and when cap charged LCD shows "READY". When in working mode (MODE=1) LCD shows all the time selected TIME and VOLTAGE so operator can see current parameters. Also in working mode "UP" button works as "WELD" button and I am planing to connect an foot switch in parallel with "UP/WELD" button.
Another option is to  replace 16F684 with 16F88 for example which have more I/Os and then a buzzer and dedicated WELD button can be added.
Your opinion is highly appreciated.

 No need for a bigger PIC - here's how to cheat and get three more pins (BTW, did you forget to pullup pins 11 & 12?)....

You can easily add LEDs and their pullup resistors to pins 4, 11, 12 of the uC. They will also come on when the button (or footswitch) is pushed (who cares if they do? may actually be a good thing to ack the button press...), but more importantly, when the button is not pushed, the uC can turn them on....you can also add a buzzer by simply putting the buzzer in as a load to VCC for the drain of a source-grounded FET and also attach it to one of these three pins along with a pullup.

[kalpakchiev]

But to use leds i have to make the  I/O an output and then to turn on LED in this time I can miss a press of button Am I right or I miss something. Using diodes I can add buzzer find attached file. Best regards

[solutions]

One more observation - how are you limiting the current through the second FET?  Through its Rdson? Counting on the load resistance of the metal being welded?

I get 3700 amps at 24V, assuming a dead short for the weld resistance, and the IRL1004 is only rated for 520 amps...

I think you might need a power choke and its snubber diode in there to current limit the cap dump into the short that is the load (when spot welding copper for instance...)

Posted on: April 04, 2010, 08:37:52 20:37 - Automerged

But to use leds i have to make the  I/O an output and then to turn on LED in this time I can miss a press of button Am I right or I miss something. Using diodes I can add buzzer find attached file. Best regards

You can pulse the LEDs, say 1000Hz, 50% DC, and look at button presses when they are dark - you won't miss a thing.

[kalpakchiev]

i will try it in 10 minutes

[solutions]

Same trick for the FET/buzzer ad reading the buttons, only a very low positive duty cycle for the buzzer to keep it from being audible yet allowing a read of the button at that time to see if it is grounding the FET gate pullup.  You have the option of a quick buzz in firmware using the fet on button presses versus an annoying buzz as long as the button is pressed using the diodes you've shown.  Can't buzz while button is pressed, though.  This one's up to you, of course.

Dibs on a blank board if you found all this useful :-)

[Ichan]

Some thoughts:

1. No need to use high amperage mosfet for charging the cap.
2. Gate driver need to be modified, pnp transistor will be better.
3. Better put the electrode on the Drain not Source.
4. Wiring + contact resistant will be very important, at 24V 0.1 ohm resistance will limit the peak current to 240A.

-ichan

[kalpakchiev]

One more observation - how are you limiting the current through the second FET?  Through its Rdson? Counting on the load resistance of the metal being welded?

I get 3700 amps at 24V, assuming a dead short for the weld resistance, and the IRL1004 is only rated for 520 amps...

I think you might need a power choke and its snubber diode in there to current limit the cap dump into the short that is the load (when spot welding copper for instance...)

Posted on: April 04, 2010, 08:37:52 20:37 - Automerged
You can pulse the LEDs, say 1000Hz, 50% DC, and look at button presses when they are dark - you won't miss a thing.

Capacitor charge voltage is limited by firmware to 15V and I hope that resistance of wires and material will be enough to limit the current but if this is not enough I will have to pare two transistors in parallel.

Posted on: April 04, 2010, 10:32:32 22:32 - Automerged

Some thoughts:

1. No need to use high amperage mosfet for charging the cap.
2. Gate driver need to be modified, pnp transistor will be better.
3. Better put the electrode on the Drain not Source.
4. Wiring + contact resistant will be very important, at 24V 0.1 ohm resistance will limit the peak current to 240A.

-ichan

1. I have a lot of them and it is easy for me to use one
2. You are right , I will change it
3. And again You are right , I will change it
4. Voltage is 15V and I think that I will manage to reduce resistance using proper wires
Many thanks for advice

[pickit2]

if you have a good supply of mosfets use one as power resistor in place of 0.1 ohm

[kalpakchiev]

if you have a good supply of mosfets use one as power resistor in place of 0.1 ohm

But this 0.1R resistor is only for simulation in Proteus. In real life it will be formed by wires and resistance of material which is being welded.

[kalpakchiev]

updated with leds and piezo

[Ichan]

I am proposing the circuit below for the gate driver, Q1 need to have Ic >= 2A, low ohm resistor (< 10 ohm) might be needed between this driver to the gate of the mosfet. On your original driver, the turn on time of the mosfet will be very slow, as the Gate-Source capacitance will be charged via 10 K ohm resistor.

-ichan

[solutions]

I am proposing the circuit below for the gate driver, Q1 need to have Ic >= 2A, low ohm resistor (< 10 ohm) might be needed between this driver to the gate of the mosfet. On your original driver, the turn on time of the mosfet will be very slow, as the Gate-Source capacitance will be charged via 10 K ohm resistor.

-ichan

If you want to be totally anal about it, a push pull would be better - now your turnoff has to charge with your R5.  You flipped the problem over, but still have the same gate cap charge problem, but now in discharge....

Real men use FETs, also....

 

[Ichan]

Fast turn off is not required in this case - slow would be better, IMHO.

-ichan

[kalpakchiev]

Which one is the best for this case

[kalpakchiev]

  According to PDF IRL1004 is LOGIC-LEVEL GATE DRIVE so it can be operated by PIC without driver or it is better to use microchip's driver like TC4422 or TC4427?

[solutions]

According to PDF IRL1004 is LOGIC-LEVEL GATE DRIVE so it can be operated by PIC without driver or it is better to use microchip's driver like TC4422 or TC4427?

It's really a matter of how much current your output buffers, or transistors, can drive into and out of the Qg of the device that determines the speed.  It'll get to logic level eventually, but as you transition slowly through the triode region, things may get rather toasty for the FET (think about even a few, let alone tens of, ohms times 500+ amps...).  Ichan and I both expressed concerns about doing this via the 10k resistors, as the RC time constant for turnon (or off....) may cook your FETs.

Posted on: April 05, 2010, 11:28:41 23:28 - Automerged

Which one is the best for this case

I like Driver 1, though I think you may need some bleeder current/bias resistors (maybe equal valued ones from the gate of the Fet, one to gnd, and one to Vcc) to make your bipolars happy during steady state.

[Malcolm]

Driver-1-Gate Drove is best.
Same method used to drive Mosfets in a PWM-EDM power Supply

[kalpakchiev]

"Single pulse" and "Double pulse" added. In work mode (MODE=1) UP button switches to Single mode, DOWN button switches in Double mode.

Today I have to buy the capacitor. Does somebody have a rough idea what will be the max thickness of welded material with 1F capacitor or I should go to 2F or 4F?

Posted by: Malcolm
Insert Quote
Driver-1-Gate Drove is best.
Same method used to drive Mosfets in a PWM-EDM power Supply
I am going to try both DRIVER1 and TC4422 chip when I have the capacitor