New Open Source Antispark Switch

I’m in need of new antispark switch, and most of the switches are either too expensive or will suffer the same failures

As pointed by @b264 the main problem is the inrush current killing the MOSFET’s as time goes on

Here are the requirements:

  • all componentes from , this way we can get everything on the same shipping as the PCB from for ridiculously low prices *
  • no micro controllers
  • 100A continuous
  • pre charge circuit
  • up to 12S
  • max size (need to look at my board what can fit, I think 30x50mm)

*the MOSFET’s may need to come from another place

Any help appreciated, never designed anything for power switching and higher voltage like this, only done mostly low power audio stuff

What I have so far is two candidates for the main MOSFET’s:

  • HY4306B6 from HuaYi Microelectronics - datasheet
  • NCEP85T25D from Wuxi NCE Power Semiconductor - datasheet

I need to research more if they should be trusted, but from all the specs they look really similar to the IRFS7350

For the pre-charge the plan is to use a 555 timers (or maybe two if needed)


I’m working on a way, but so far can’t eliminate the MCU from the design… it may be better to just use an ATtiny10 or something rather than try to come up with an elaborate analog circuit


If I can’t find a away I may go that route, I would rather avoid it to open it to as much people as possible

About the linear region operation, more FET’s in parallel would help alleviate the problem? I say that because I’ve never blown my Vedder antispark, it has been in two board so far with a lot of mileage, it’s only a matter of time? I need to study more on the subject

1 Like

The problem is that sometimes the user cannot See on the switch whether it is Turned off or on. Then it can Happen easily that the as switch is connect to the system in Turned on State which kills it. Most commonly offers free shipping on every order no minimum purchase.

Would you be able to source parts from there?

Also, the whole inrush current problem is solved with the LTC7004.

page 11 of the datasheet.|602dc2b7711e4c16b88d2f5de19e1ff6|42d88dfcf3ce4bfd8d401d1ee3b6d229 this is what I have been using. It will charge 3mF without exploding, haven’t tested 6mF, but it should survive, the RC timing circuit has a 300milisecond rise time. NOTE: you need to bypass the ground plane with blue wires, (that is connect negative to negative) there are too any traces that impede the ground plane.

edit: oh bleh, the pcb is too wide, if you change the PTH parts to SMD and move the header, it might fit in 30x 50mm.

And maybe use the TPW1R306PL,L1Q instead of the IPB014N06NATMA1, you could fit three and still make it smaller.

Including the PCB shipping cost, it cost me about $70 to make three of these (well two, I screwed up the pins on one of the ltc7004s, but I can just order more from Arrow with free 1 day shipping).


wooo, nice find. thats super interesting.

Edit; this is what @Gamer43 is referring to 06%20PM


Implementation is continued on the next page :).


@Gamer43 I took a look at them and really liked, what kills here is shipping, Mouser, Farnell and Digikey were all in the 40 USD no mater how little I ordered

Really nice find of the LTC7004, I was giving up and just adding more MOSFET’s to the Vedder switch, and you already having and schematic made my day

Thanks a lot, I will definetely look into it

Making a new PCB and changing the components are no problem, will post here once I have more compact layout


Glad I could help. :slight_smile:

Some passive values may need some tweaking, just realized I didn’t quite follow the datasheet recommendations exactly… (i.e. the rise time may have been significantly longer than 300 miliseconds…) Actually, nvm, what happened was I changed some values on the fly and didn’t update them in the schematic.

Also, is Arrow still charging shipping for you? I remember when I ordered parts internationally, they still offered a free option.

No, Arrow gives me free shipping, and a way better selection of components than LCSC

I will take a good look before ordering anything

I’m a bit lost on the connection to the switch on the header 1, the switch itself connects to pins 5,4 and 2 and an LED to 1 and 3?

1 Like

My apologies, I didn’t label them in the schematic, pin 1 is Normally Closed pin 2 is Normally Open pin 3 is Common pin 4 is LED cathode pin 5 is LED anode.


This is why I would want an auto powerdown after like 8 hours of inactivity

I was gonna ask if it’s not a good idea to have a pull down resistors but it’s build in

Thanks for the diagram

do you think that 3 TPW1R306PLL1QCT would be enough for the 100A goal? price wise they are cheaper than the IRFS7530 and llower Rdson

Another challenge I"m looking at is finding the right fuse, ideally I would like to mount it on the PCB, but once we go to higher current the automotive fuses start to get too big

I remember Batman from @raphaelchang used tiny PCB fuses in parallel, I may go that route

They should be able to handle 100A but at those currents you also need to make sure the pcb traces can handle it as well. The entire PCB will need cooling if those currents are sustained. Cooling can be passive, like with a modest heatsink.

What I use for fuses is I inline an autolink fuse using ring terminals and bolts in my wiring. Trying to incorporate it on a pcb becomes too cumbersome.

I’m thinking is ditching the Toshibas since they are really small to work with

I’m really really space constrained that is hard to fit any of the common fuse types

So far I’ve found this one form Littlefuse that I really liked, but expensive

Spent a lot of time thinking about this after @b264 explained the issue. Ended up going with a mcu attiny85 to control everything.


They have the same footprint as an SOIC-8, and unlike the DFN MOSFETs from ON Semi, they actually have leads, you shouldn’t have a problem working with them; I used a standard soldering iron to switch out the MOSFETs on a FSESC 4.20 with them a little while back.

Are you sure you wouldn’t be able to inline something like this? It has almost the same profile as 10 gauge wire once everything is bolted together.

@deltazeta Do you have the firmware for that? Is it on github?

will upload once i’m back from class

1 Like

At least in my case it will be only peak, I doubt it will sustain anywhere near that

@deltazeta will take a look in what you did, thanks

Nice to hear that, I can use an SOIC-8 and just extend the pad to the bottom of it? Couldn’t manage to find any component in EasyEDA that has the same package

The Midi ones I may manage to fit, even more if I go with the Toshiba ones that are slim