Skip to main content

CD Resistance Spot Welder

This post is for anyone interested in owning or building a CD welder.

I wanted a reliable Resistive spot welder for lithium ion pack construction and or repair. I didn't want to spend 3k or buy from china. I decided to design and fabricate my own. Bellow is my first crack at a Mosfet Switched capacitive discharge mcu controlled spot welder.

This post is a work in progress. If you've stumbled across in hoping to build your own cd welder feel free to ask questions in the comments section. 

I'll provide any help I can. 

My CD Spot Welder 

Here's pics of the first working version from 2018

What is a CD Welder?

There are two main design types for spot welders. One is transformer based and the other is a cd welder. The CD stands for Capacitive Discharge.

It might be more accurate to say there are two power sources for spot Welding. Each has its own advantages and drawbacks. The selected one also heavily influences the design.

For instance my design is Mosfet Switched. If you tried to use it to switch a ac power source you would be greatly disappointed in its performance.

The long and short of it is mosfets are DC switches. You can use two of them in a special configuration to make an ac switch. That's not how mine are implemented.

The capacitive discharge part implies that the power source is a large Bank of capacitors.

I went with Maxwell d cell ultra capacitors.

The ultra capacitors are the blue cylinders visible on the bottom of the board above.

Why Spot Welding ?

Resistance Spot welding is the industry standard method to join lithium ion cells to create battery packs. It's prefered to soldering because the connection formed is less brittle and despite the intense heat of a weld the duration is so brief the temp of the entire cell doesn't rise much. Soldering with even a high powered iron tends to raise the entire cells temp. This can result in decreased pack/cell life and or boom.

If you're a fan of high-power flashlights or vapes you know what an 18650 battery looks like. They and other sizes are used in everything from laptop batteries to electric cars. To join them into a pack nickel(or in tesla's case copper) tabs or wires are welder from cell(battery) to cell to form the pack

I also recently read that SpaceX is using spot welding to attach the hardware that allows the heat protective tiles on starship to stay put.

The little metal protrusions are spot welded on.
 They form clips to hold the heat tile in place.

For a while Elon musk also owned Maxwell (actually I think Tesla bought it but that's similar to the same difference.)

CD Welder Design and Build

This is intended to be DIYish more than diy cd welder. If anyone reading this has questions about building their own cd welder; I'll do my best to help.

There are many ways to design and construct a CD spot welder. Here's how I made mine and some of the theory.

The Capacitor Bank

I went with Maxwell BCAP0350 ultra capacitors. Somebody was selling a box of something like 20 of them on ebay.

Bellow are 12 of them on a balance board.
For my CD welder power source I went with Maxwell "D Cell" ultra capacitors(wikipedia def). 

The name was given because they're roughly the size of an alkaline D cell.

 Each ultracap is 
  • 350f 
  • 2.7 volts max
  • Internal Resistance (IR) of 3.2 milliohms

The 310f model would have likely been Superior but I got these for cheap.I say that because it's IR value is 2.2 milliohms

In the explanation below I refer to a single capacitor as a cell.

I went with 6 cells in series to give me a Max working voltage of 16.2v

The equation for bank max V of series connected capacitors

 cell max V * # of cells in series
2.7volts*6 in series= 16.2vdc

The problems with series connected capacitors

 capacitance of the series string:

Capacitance of 1 cell
 number of capacitors in the string

Or in my 6 series case: 350F/6 = 58.3f

Resistance of Capacitors in Series

Bank ir =  R (ohms)of single cell * cells in series

This is an issue because :

I  * R
(Ohms law)
(Cover the term you want)

E = volts (electromotive force)
I = amps (current)
R = ohms (resistance)

E= I*R

The middle or I equation is the issue here.
To maximize heat of the weld you want current(I).

To get this current  you want to minimize R(ohms) in the path up to the weld.  Starting with a high r value in your power source is no bueno.

the data sheet for my CD welder ultra caps

0.0032ohms * 6 = 0.0192ohms

Which plugged back into I=E/R might still look impressive but that's not including every other component up to the weld site.

The way you lower this R value is place multiple series strings in parallel.

 Parallel in this case is; connect six in a set in series, make additional sets of six in series and connect those sets plus two plus minus to minus. the equation for Bank capacity is multiplied by the number of strings in parallel

the internal resistance of a series string is divided by the number of strings in parallel.

In the testing videos I'm using only the bank above. 
Aka two parallel strings of 6 in series series 

Short hand notation is 2p6s

Later I added a third parallel string of six.
3p 6s.

 Capacitance of S&P Capacitor Bank:

(Capacitance of a single cell)
/ (cells in series)

* (#of parallel stings)

(350f/6 in series)*3 parallel= 175f

Internal resistance of S&P capacitor Bank:

((Internal Resistance single cell) * (cells in series))
(#of parallel strings)
(0.0032ohms * 6 ) / 3 = 0.0064 ohms

Hypothetical Max

 if plugged back into I=E/R 

gives a whopping 

2531.25 amps @16.2v as a theoretical Bank Max.

Aka:  41006.25 Watts or 41 KW


IIRC the equation for heat is 


if I could figure out subscripts on blogger life would be a lot better.but the state has insisted that I shouldn't even have a laptop so Android phone and blogger... This site used to be hosted on a WordPress that was self hosted on the floor of my apartment.

 In practice it won't deliver this. For a cd welder this number is not a bad place to start.

The two parallel rows are upside down underneath the string on top. It's not visible in the photo but there's a piece of foam tape supporting center. Without it it works just fine but in the off chance the middle of the board deflected enough to short it would be a hell of a short.

Balance Board

You might notice in the examples above I've heavily coated the traces with solder. This is supposed to help increase current capacity, and decrease resistance. At very least the extra metal gives you increased thermal mass/less probably of vaporizing your traces on weld.

Why do you need a balance board?

I'm entering this on a phone or I plug in more equations for you. But when it comes time to charge your capacitors or ultra capacitors in series that R value per cell starts to become a pain as well.

More so with these ultra capacitors and their relatively high IR. 

Especially so when you want to go towards their Max voltage.

Put simply the risk is some in the middle or on the far end will reach their full voltage before the others and at that point they're in an overvolt condition. If you like shrapnel don't worry about it.

It's also worth noting that the real life IR value of any given ultra capacitor is slightly different than the factory rating.

If you do a lot of plugging with the ohm's law and E equals IR you might see that in series there's a possibility that one of the cells reaches its maximum voltage before the bank is fully charged or the ones around it is. This can be a problem because all of a sudden you are over volting that capacitor. 

To manage this risk you use what's called a balance board. This is the same with lithium ion batteries in series configuration.

A balance board is basically a mosfet a little bit of logic or clever use of diodes and some power resistors. When the cell in series reaches its maximum voltage during a charge the mosfet switches on a power resistor.

If you want to look it up more than that you're on your own. All the theory is out there but it's not worth going into here.

On a random note: the 3p6s config jump-started a minivan in 2021. Somebody had stopped to help before me and was trying to use their truck. with the cables attached for 10 minutes while I waited and watched it only produced a click on the van.

My wait was knowing I had the bank in my trunk and debating what are they going to think when this guy walks up with this random white box?

How do I explain it if they want to know what it is? I'm glad I thought about that because they did.

I said basically if you know what a car audio cap is this is 350 times more powerful.

I instructed them to disconnect the minivan and connected the cap bank to the truck (to charge it up). When the balance board LEDs lit up we disconnected the truck and connected it to the van.

It's sprang to life on the first try. I believe it was negative 10° f that night. 

it put quite the load on the truck alternator as well. The engine actually revved to counter it. It took something like a minute to charge it. I wasn't actually counting though it just seemed inordinately long.

Low side mosfet switch

I needed this fast have no formal EE training. I was also relatively new to Circuit design in general. The easiest thing to implement was a low-sided mosfet topology. I went with six powder fets in parallel.

But here's what it looks like in this most basic implementation

I went with this because it simplifies a lot of v(gs) issues in design (I'll explain this more later)

The drive signal lines correspond to the output from either a mosfet driver, a discreet drive circuit or sometimes just a microcontroller.

R1 is what's known as a pull down resistor. It basically functions to ensure that unless the line is intended to be at drive voltage it's at 0v/gnd. You could also conceptualize it as a bleed resistor.

 R2 is somewhat optional. Actually some mosfet drivers will specify do not use a series resistor.

Load is where your electrodes are in the circuit.

The  signal path was opto isolated. I intended on developing a multi-purpose or multi-process controller in the future, could also help to ensure that whatever happens the logic side doesn't burn out.

Signal path was mcu->opto->mosfet driver->mosfet.

Microcontroller (MCU)

I'll add this section soon. For now anyone pursuing this with atmel / arduino in mind should check here

If you want reliable timing you need to look into coding with port register manipulation

Fabrication of the CD welder

Weld head/Electrodes

In the videos of the testing I'm using handheld electrodes that are 0awg single conductor wire ground to a point

Later I built a proper weld head
I used a short linear rail to allow it to move vertical y

I ended up remaking the electrode holder on the right. I was working with $100 drill press which really really struggled to drill copper bus bar.

To further improve my cd welder I sourced copper alloy electrodes.

A mock up up the stand

Part of repeatability of reliable welds from any CD welder or spot welder is:

Pressure/force applied during the weld.

By using this linear slide rail configuration I planned on being able to add weight to the top of the platform to have a consistent down force

Control board PCBs

The diy pcb fab article uses my resistance Spot welder as an example. So if you want to see more in-depth on pcb fab using this board click here

I ended up having to move after this board was etched. So I attempted a daughter board with through hole.

The integrated circuits present are a ADC assembly (vertically oriented)a  flying cap driver(in between the yellow wires) and a differential i2c converter(left of the three two pin connectors). 

The particular converter had an AUX pin. Which I'm using to drive the enable on the flying capacitor driver.

This board covers reading bank voltage (adc) and bank discharge through two power resistors in parallel (the connector on the right)

I used a depletion mode power fet to control Bank discharge. The idea was a design that could discharge the bank even with the device unplugged.

I could have used relays as well but as everything else was solid state I though I'd strive twords keeping it that way.

I wasn't quite  comfortable enough with board CAD to send out for it. Today would be a different story.

Initial Testing 

The first time I hit the foot switch I blew  a hole in the nickel

After getting voltage and timing dialed in I managed to get a solid weld; no hole. 

Testing a Weld To Failure 

It's hard to see but the metal tore around the weld nugget. Which is exactly how a real/good weld fails 
See heat affected region

Pulse Length Adjustment via rs232

The Enclosure

I built the enclosure from plexiglass chemically welded with methylene chloride.

After thoughts:
I had a more ambitious design goal in mind. Multi purpose electric fabrication controller. It's not fully baked into this iteration but the basics are there. I was in a rush to get this working but had the larger vision in mind. 

If you are pursuing your own spot welder design I'd highly recommend using a faster mcu than I did. It worked but it was right on the edge of the timings it could reliably deliver. I'm writing this 2 years after I last had time to work on it but iirc the min reliable timing was something like 6  μs.


  1. I'm the owner of a car repairing garage in Miami. Last week I purchased a Millermatic 252 welder for my garage. I'm using this welder for a few days and now I can say this welder is value for money.


Post a Comment

Popular Content

XT - 30 Motorcycle Phone Charger

This post might help anyone looking for a slim line solution to 5v or USB ports on a motorcycle. Not a huge achivment by any means. But made life better for a client.  A perspective client approached me with an issue. His solution for charging his phone via USB on his bike was taking up too much room under the seat 

UV Lithography For Circuit Board Creation

In this multi part series on diy circuit board manufacturing/fabrication I'm going to cover my diy, vacuum pull down UV LED exposure box. If you are looking for a description of the entire fab process click here UV Exposure Rig The photos bellow are of the  uv lithography exposure box I cobbled together. It also features vacuum pull down.  The source: A 4.1w High luminous efficiency 365nm UV LED. Sourced from digi key mid 2016. Model  LZ1-00UV00

Smok Nord 4 Unboxing and Review

 Today I'm going to review the Smok Nord 4 ecig.