I think you are deliberately ignoring my point.
I’ll spin two PCB designs, one with SMD and no thermal reliefs, and a second one with all through hole and thermal reliefs and possibly with a different kind of MOSFET.
This way, someone can mass produce the SMD PCB and have it work perfect, and anyone who’s up for DiY can spin the through-hole PCB for themselves.
Also, small run assembly can still run up $5-10 dollars per board in 10 quantity Dx.
3 Likes
Okay, I finally got around to redoing the PCB, how does this look? @shaman @AlexBE
Dimensions are ~55mm x 39mm.
This is some A+ via stitching xD.
Any traces I should make thicker or clearances I should increase beyond 10mils?
Schematic for reference
1 Like
I would
- make the main 4 solder pads much larger
- Reroute Batt- -> NO to not cut the top copper in half
- Not sure what the P1 connector is, but if it’s simple header pins, I would change to a locking RA connector.
- Similarly I would move H1 and P1 down to level with R6 to take them out of the high current path
- Then because I’m being picky - The entire board can be made much smaller and thinner (RA connectors)
- LED connector should be locking also, so much vibration in this environment. I suppose most people would solder those connections though.
The headers are intended to be soldered with JST XH connectors, which I believe are locking (at least to some extent).
I thought the pads were large enough to solder 10 AWG, if not, I’ll make them 250x250mils.
Not sure how practical this would be, but I was also intending that the two BATT+ pads be bypassed with external wires, as that is what I have done on all my prototype antisparks, or better yet, don’t run the positive current path through the antispark at all (i.e. only use one of the BATT+ pads).
Just realized I had the silkscreen mixed up on the headers xD, fixed that.
EDIT:
@AlexBE, okay I fixed a few things that were dumb
1 Like
Did you have the silkscreen box correct for those connectors? I don’t think they look big enough. You also have different size holes for the two connectors. I believe a JST-XH connector will interfere with R4, it will at least be close. Not sure what software you are using, but I always find it is useful to import all the 3d models.
They are large enough, but why not extend them double the length towards the centre of the board?
Yes this is a good way of doing it, depending how neat you are. I don’t use anti-spark, but if I did I would connect to both batt+ pads for neatness.
Oh, you’re right about the silkscreen, my bad.
I’m using EasyEDA, not the best, but works for getting something up quickly.
I don’t want to make the pads too big, because I don’t want to put vias in them; from my experience, soldering on vias can sometimes get the solder to drip and pool down on the other side.
Ok so now I’m nickpicking, because that’s my job.
- You will get pinged for silkscreen on copper on the connectors (manufacturer might fix it for you)
- Would be much better to be single sided if you can squeeze it in.
- Ideally the NO NC C etc labels would be visible with the connector mounted.
1 Like
How does this look? Can’t fix the silk screen, manufacturer will most likely do that.
If this design works, it should be bulletproof. If anything fails, it’ll be the zener diodes, which are cheap and easy to replace. And if they fail, it won’t be catastrophic.
4 Likes
Looking good, I would make the output pads a bit wider so two wires cans be soldered, onde for each vesc directly from the switch
1 Like
Who wants to be the guinea pigs for a prototype run? 
By my estimates, in 200+ quantity, this shouldn’t cost more than $10 per switch to make through a turnkey PCB assembly service. (with 3oz. copper and ENIG :O)
Is the $20 100A 60V antispark (that doesn’t die) finally here? Better question, can we get 200 people to buy this? X_X.
10 Likes
is it possible to go for smaller batches?
It would cost more, like twice as much unfortunately :(.
199 left to go. 
4 Likes
I have enough MOSFETs to make three prototypes, I think I’ll order the PCBs sometime next week, who wants to be the first ones to try out the new antispark?
Why isn’t there a precharge circuit on that?
1 Like
The switch controls the precharge circuit, MOSFETs turn on when the voltage difference is below 4V.
By connecting Common and Normally Open, the resistors bypass the MOSFETs, charging up the load capacitance. Once the voltage difference between the input and output is less than 12 - (7.5 + 0.7)V, Q3 turns on and activates the primary MOSFETs.
Connecting Common and Normally Closed disconnects the pre-charge resistors and shorts MOSFET gate to ground, turning it off.
Which resistor is the precharge happening through?
R4 and R6, I’ll repost the schematic down here for reference.
4 Likes