Author Topic: kWeld - "Next level" DIY battery spot welder  (Read 47803 times)

0 Members and 5 Guests are viewing this topic.

Online technix

  • Super Contributor
  • ***
  • Posts: 2069
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: Guesses on what I am attempting here?
« Reply #26 on: June 14, 2017, 09:34:01 PM »
I find it funny though that there is such a huge difference between SOT227 and SOT223 ^-^
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline P90

  • Frequent Contributor
  • **
  • Posts: 646
Re: Guesses on what I am attempting here?
« Reply #27 on: June 14, 2017, 09:39:37 PM »
i guess it's safe to say your planning on switching a big load on and off...
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: Guesses on what I am attempting here?
« Reply #28 on: June 14, 2017, 09:49:04 PM »
more to add to the riddle: I need the die size, definitely need the Rdson, but don't need the power handling capability. And because it's only AWG8, to keep the wiring short.
« Last Edit: June 14, 2017, 09:54:31 PM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: Guesses on what I am attempting here?
« Reply #29 on: June 15, 2017, 07:22:43 AM »
sorry, next update not until before Friday, I had a more urgent problem to solve as you have seen a few posts earlier. Mission accomplished :phew:



DETAIL A

« Last Edit: June 15, 2017, 07:24:36 AM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Online blueskull

  • Supporter
  • ****
  • Posts: 7804
  • Country: cn
  • Power Electronics PhD Candidate
Re: Guesses on what I am attempting here?
« Reply #30 on: June 15, 2017, 08:08:26 AM »
1. The answer has been given before #20.
2. It's not about the 2kW motor.
3. It's not a railgun (40V Vds_max).
4. You need low L/R as well as high die energy handling capacity, but don't care about continuous power dissipation.

I think my guess was correct.
SIGSEGV is inevitable if you try to talk more than you know. If I say gibberish, keep in mind that my license plate is SIGSEGV.
 

Offline DBecker

  • Regular Contributor
  • *
  • Posts: 181
  • Country: us
Re: Guesses on what I am attempting here?
« Reply #31 on: June 15, 2017, 08:51:50 AM »
Given the lack of freewheel diode, it's unlikely to be powering anything even slightly inductive.  That rules out a motor controller, and points to a contact welder.  Or a fuse/breaker tester.

I expect the lack of snubber or capacitor at the device terminals will make this unreliable unless the turn-off time is quite slow, or the current decays on its own (e.g. it's a big capacitor, the easy way to build a single-shot battery terminal welder).

The main point of that $20 package is that it is isolated yet still has good heat transfer.  If you don't need the isolation, you can get the same die in a $4 TO247 and bolt it to a chunk of aluminum.  Although for a one-off you might decide the screw terminals are worth the extra money, especially if you buy a snubber module that mounts directly to the terminals.
 

Online blueskull

  • Supporter
  • ****
  • Posts: 7804
  • Country: cn
  • Power Electronics PhD Candidate
Re: Guesses on what I am attempting here?
« Reply #32 on: June 15, 2017, 09:02:42 AM »
Given the lack of freewheel diode, it's unlikely to be powering anything even slightly inductive.  That rules out a motor controller, and points to a contact welder.  Or a fuse/breaker tester.

I expect the lack of snubber or capacitor at the device terminals will make this unreliable unless the turn-off time is quite slow, or the current decays on its own (e.g. it's a big capacitor, the easy way to build a single-shot battery terminal welder).

The main point of that $20 package is that it is isolated yet still has good heat transfer.  If you don't need the isolation, you can get the same die in a $4 TO247 and bolt it to a chunk of aluminum.  Although for a one-off you might decide the screw terminals are worth the extra money, especially if you buy a snubber module that mounts directly to the terminals.

These are MOSFETs, not BJTs. They are avalanche rated.
When BJT breaks down due to avalanche effect, highly concentrated carriers from emitter inject to base region and causes current multiplication, eventually triggering a secondary breakdown (RBSOA). That's why snubbers are mandatory for thyristors, BJTs and IGBTs.
When MOSFET breaks down, since it's a majority conductive device, there's no base region and current gain, hence no RBSOA. Since the drift region, source region and reverse doped channel region form a parasitic BJT, there is still a chance for secondary breakdown to happen in MOSFETs, but modern devices are designed to reduce hFE of this parasitic device, and for trench MOSFETs, this BJT is intrinsically very weak, compared with flat MOSFETs.
SIGSEGV is inevitable if you try to talk more than you know. If I say gibberish, keep in mind that my license plate is SIGSEGV.
 

Offline macona

  • Contributor
  • Posts: 17
Re: Guesses on what I am attempting here?
« Reply #33 on: June 15, 2017, 09:59:43 AM »
Well, if you are building a capacitive discharge spot welder I would use a hockey puck SCR instead.
 

Online blueskull

  • Supporter
  • ****
  • Posts: 7804
  • Country: cn
  • Power Electronics PhD Candidate
Re: Guesses on what I am attempting here?
« Reply #34 on: June 15, 2017, 10:01:22 AM »
Well, if you are building a capacitive discharge spot welder I would use a hockey puck SCR instead.

The T shaped RC battery connector tells me it's gonna be powered by an unprotected Li-ion pack (aka. RC battery).
SIGSEGV is inevitable if you try to talk more than you know. If I say gibberish, keep in mind that my license plate is SIGSEGV.
 

Offline LukeW

  • Frequent Contributor
  • **
  • Posts: 581
Re: Guesses on what I am attempting here?
« Reply #35 on: June 15, 2017, 11:02:05 AM »
The little Deans connector looks really out of place on the giant wire.

Consider an Anderson connector, XT90 or a directly bolted connection onto battery terminals.
 

Offline buck converter

  • Regular Contributor
  • *
  • Posts: 143
  • Country: us
Re: Guesses on what I am attempting here?
« Reply #36 on: June 15, 2017, 11:20:01 AM »
high voltage Molotov cocktail. if that makes any sense
Just me and my scope.
 

Offline LukeW

  • Frequent Contributor
  • **
  • Posts: 581
Re: Guesses on what I am attempting here?
« Reply #37 on: June 15, 2017, 01:54:25 PM »
So those beer bottles are supposed to be HV insulated standoffs?
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: Guesses on what I am attempting here?
« Reply #38 on: June 15, 2017, 03:50:47 PM »
So those beer bottles are supposed to be HV insulated standoffs?
:-+ ;D

The T shaped RC battery connector tells me it's gonna be powered by an unprotected Li-ion pack (aka. RC battery).
They are okay for the same reason that I don't need a heatsink on the MOSFET.

Well, if you are building a capacitive discharge spot welder I would use a hockey puck SCR instead.
Compare the power dissipation of an SCR (at probably also higher price) with that of a MOSFET when switching high current, low voltage loads. Also, a capacitive discharge spot welder violates my requirement/assumption of this being a new idea.

the screw terminals are worth the extra money, especially if you buy a snubber module that mounts directly to the terminals.
That was exactly the rationale: how to attach AWG8 to TO247? Although it's a one-off - it it works well then I hope that it will not only be built by me.

I expect the lack of snubber or capacitor at the device terminals will make this unreliable unless the turn-off time is quite slow, or the current decays on its own (e.g. it's a big capacitor, the easy way to build a single-shot battery terminal welder).
An avalanche (or UIS, unclamped inductive switching) rated MOSFET can even be used to switch an inductor without any additional protection. They behave like zener diodes during breakdown (for a 40V device maybe at 50-60V), and just dissipate the stored magnetic energy. All you need to make sure is that their avalance energy rating is not exceeded, which is 5J for this one. This is because the energy is dumped into the die fast enough such that there is no heat transfer to the case. That is very useful for a flyback transformer for example, if it was not for EMC you could just remove the snubber network.
« Last Edit: June 15, 2017, 04:45:55 PM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: Guesses on what I am attempting here?
« Reply #39 on: June 17, 2017, 10:10:20 AM »
@blueskull: razor sharp deduction  :-+

It's a bit late now, but I had to see it generating the first current pulse  :phew:

The scope screenshot shows the voltage across the AWG8 wire (35cm, approx 700 micro ohms). I couldn't believe it, but the peak does read 2.76V. That corresponds to 4kA  8)

I was a bit too scared for a full test by shorting the output leads, this one was taken with only a few litz strands twisted together. They vaporized in less than 100 microseconds... Which also means that I probably haven't seen yet what the circuit is capable of doing.





« Last Edit: June 17, 2017, 10:16:35 AM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline Bendba

  • Regular Contributor
  • *
  • Posts: 170
  • Country: au
Re: Guesses on what I am attempting here?
« Reply #40 on: June 17, 2017, 10:25:04 AM »
I thought lipo's internal resistance was in the order of 10 milli ohm. How do you get 4kA out of them?
Stop dreaming your life, start leaving your dreams.
 

Online blueskull

  • Supporter
  • ****
  • Posts: 7804
  • Country: cn
  • Power Electronics PhD Candidate
Re: Guesses on what I am attempting here?
« Reply #41 on: June 17, 2017, 10:33:23 AM »
That's not 4ka. That's induced voltage. To accurately measure fast rising current, you need a current clamp or a high bandwidth coaxial current shunt.

I do a lot "few ka/us" measurements (my day time job is to design wide band gap power devices application and testers), and learned this the hard way. Sometimes even very small 1206 shunt can give you nasty inductive voltage.

The poor man's current shunt can be made out of a piece of coax cable, and short circuit inner and outer conductor on one end (current in), while on the other end, connect your scope between inner core and outer shielding. Current goes out from outer shielding.

What you get from scope is only resistive drop on outer shielding. Inductive voltage will be cancelled out by inner conductor.
SIGSEGV is inevitable if you try to talk more than you know. If I say gibberish, keep in mind that my license plate is SIGSEGV.
 
The following users thanked this post: evb149, MK14, tatus1969

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: Guesses on what I am attempting here?
« Reply #42 on: June 17, 2017, 06:05:44 PM »
That's induced voltage.
That also immediately came into my mind, but I thought that the numbers would never add up for that - and skipped doing the math. But they do: Estimated loop inductance (low frequency, wires packed tightly) is 200nH. MOSFET switching time is 1.4us, so dI/dt would be 700A/us when assuming 1kA. That means, L*dI/dt = 140V. Now I would need to add skin effect calculations, but that is enough for me to close the case here.

high bandwidth coaxial current shunt
Thanks for that great trick!!

I thought lipo's internal resistance was in the order of 10 milli ohm. How do you get 4kA out of them?
That is valid for 18650 cells, but these are high current race LiPos rated at 40C discharge or 200A. Total circuit resistance should result in ~1kA. The second cursor in the screen shot is at 640mV, which corresponds to ~ 900A. That is good to know, because 4kA would just have been way more current than what I wanted to achieve.
« Last Edit: June 17, 2017, 06:12:52 PM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: low cost DIY battery spot welder [guesses on ... - solved]
« Reply #43 on: June 18, 2017, 11:47:33 PM »
The control circuit is finished and works well after a few tweaks. A 555 timer generates 8V gate pulses from 1ms to 100ms. The current into a short is ~ 1kA (700mV across my "shunt").

So far so good, but after I slowly increased the pulse width, the transistor died. Its case didn't get hot at all, so I think that either the die or the bond wires must have given up.

I didn't expect that from the datasheet figures:
Id = 660A (chip capability)
Idm = 1800A (limited by maximum junction temp)
Il = 200A (external lead current limit)
Pd = 1040W
Rdson = 0.85mOhm

The power dissipation during the pulse should be ~850W, which the chip should be able to handle continuously. Am I right that the external lead current limit suggests that the bond wires probably failed?
« Last Edit: June 19, 2017, 03:34:01 AM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: low cost DIY battery spot welder [guesses on ... - solved]
« Reply #44 on: June 19, 2017, 01:55:58 AM »
Ordered 4 x AUIRFSA8409-7P  FDB0105N407L for the same price as one of the broken one. Arranged in parallel they can handle 1840A continuous.

EDIT: updated to 6 parts in parallel. It is quite difficulty to find reliable information on package current limits. The AUIRFSA8409-7P datasheet mentions 360A (wire bond limited), which should mean that the bond wires will just not melt at this continuous current and theoretical package temp of 25°C. I hope to be on the safe side with 167A per chip, assuming reasonably well load sharing.
« Last Edit: June 19, 2017, 06:20:23 AM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: low cost DIY battery spot welder [guesses on ... - solved]
« Reply #45 on: June 22, 2017, 09:14:41 AM »
OK, I decided to make things right this time. The FETs arrived, and their summed lead cross section exceeds that of the 300A ANL fuses that I also ordered. Lets hope that the cross section of all the bond wires is not too far from that. At least the fuses should be okay for a 1kA pulse of up to 0.5 secs.

The next problem was to find a way to connect the FETs to the AWG8 wiring. A PCB trace would certainly not be enough, and I don't want to spend money for PCBs with solid copper inlays.

The idea now is to use solid brass bus bars (8 x 12 mm) very close to the transistors, which will be able to carry the current, and hopefully also provide enough heat transfer to keep the solder from melting.

The first one is finished, that took again quite some time. But I started this, and I will succeed  :box:



I also started designing a decent controller with the following specs:
- pulse energy control instead of time control, this will hopefully provide more consistent welds
- ANL / MEGA fuse holder
- 1kA switch for max 0.5sec
- 10 - 24V power supply is a separate unit, I will be using LiPos but capacitors could also be connected
- STM32 micro
- 1x8 LCD with backlight
- beeper
- input for foot pedal switch
- potentiometer for quick adjustment of pulse energy (I don't really like these 'modern' HMIs with push buttons)
- size approx 100 x 100mm
- most important: as low cost as possible, I estimate something around 50-70€

I would be happy to receive comments or feature requests. I want to make this available for everyone once it is stable, maybe run a small batch at Elecrow.

EDIT: circuit diagram outdated, see update here http://www.eevblog.com/forum/projects/guesses-on-what-i-am-attempting-here/msg1240839/#msg1240839
« Last Edit: June 24, 2017, 10:33:01 PM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: low cost DIY battery spot welder [guesses on ... - solved]
« Reply #46 on: June 24, 2017, 06:14:12 AM »
The design is finished, boards and parts are ordered  :phew:
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline bktemp

  • Super Contributor
  • ***
  • Posts: 1623
  • Country: de
Re: low cost DIY battery spot welder [guesses on ... - solved]
« Reply #47 on: June 24, 2017, 06:25:10 AM »
The control circuit is finished and works well after a few tweaks. A 555 timer generates 8V gate pulses from 1ms to 100ms. The current into a short is ~ 1kA (700mV across my "shunt").

So far so good, but after I slowly increased the pulse width, the transistor died. Its case didn't get hot at all, so I think that either the die or the bond wires must have given up.
I can tell you why the mosfet failed:
Look at Fig. 2 on page 8 of the datasheet:
At 8V Vgs it can not supply more than 700-800A. Above that the voltage drop increases rapidly and also the losses. You simply exceeded the maximum power dissipation during the pulse.

If you use high current pulses, set the gate voltage as high as possible.
 
The following users thanked this post: tatus1969

Offline tatus1969

  • Super Contributor
  • ***
  • Posts: 1047
  • Country: de
  • Resistance is futile - We Are The Watt.
    • kick scooter electric conversion kit
Re: low cost DIY battery spot welder [guesses on ... - solved]
« Reply #48 on: June 24, 2017, 06:54:30 AM »
I can tell you why the mosfet failed:
Look at Fig. 2 on page 8 of the datasheet:
At 8V Vgs it can not supply more than 700-800A. Above that the voltage drop increases rapidly and also the losses. You simply exceeded the maximum power dissipation during the pulse.

If you use high current pulses, set the gate voltage as high as possible.
You are absolutely right that this is the second mistake that I made. But as the lead current limit is also underrated, it remains open whether the leads or the die failed. As I don't have the required 15V available (12.6V Lipo), this MOSFET was the wrong choice anyway. And I have immediately checked, the new parts should run nicely down to 5.5V.  :phew:
« Last Edit: June 24, 2017, 06:56:33 AM by tatus1969 »
Your living is determined not so much by what life brings to you as by the attitude you bring to life; not so much by what happens to you as by the way your mind looks at what happens.

https://www.kicksurfer.de - kick scooter electric conversion kit
 

Offline MagicSmoker

  • Frequent Contributor
  • **
  • Posts: 752
  • Country: us
Re: low cost DIY battery spot welder [guesses on ... - solved]
« Reply #49 on: June 24, 2017, 09:23:22 PM »
Oops... too late now to implement one critical thing you omitted: a diode in between the LiPo battery and the PCB circuit with a decent amount of capacitance after the diode to provide "holdup" during the pulse firing.

The other suggestion is simply a matter of software: the best CD spot welders deliver two pulses of about 1-20ms width spaced about 1-20ms apart. The first pulse preheats and cleans the joint and the second pulse does the actual welding.

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf