Why so few surface mount ICs in NES/FC carts

Discuss hardware-related topics, such as development cartridges, CopyNES, PowerPak, EPROMs, or whatever.

Moderators: B00daW, Moderators

Post Reply
User avatar
Bregalad
Posts: 8036
Joined: Fri Nov 12, 2004 2:49 pm
Location: Caen, France

Why so few surface mount ICs in NES/FC carts

Post by Bregalad »

I am asking myself why Nintendo (and 3rd party developers of Famicom cartridges) used almost exclusively through-holes components in NES carts : Except the MMC3, MMC4, MMC5 and MMC6 chips, and the SRAM chip in the extremely rare HVC-UNROM-04 board, all components in production cartridges are through-holes.
(I'm excluding epoxy blobs carts because that's yet a completely different technology)

Not that this is bad in any way, but by using surface mounted SRAMs and 74xx logic circuits, they could have reduced the PCB area and production costs, as in seen in this UNROM-compatible board with the same area as a NROM board that Memblers made.

As far I understand the only real advantage of through-hole is that it is easier for soldering, measuring and routing, but there is few real advantages for production batches.
lidnariq
Posts: 10677
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Why so few surface mount ICs in NES/FC carts

Post by lidnariq »

Nowadays, package costs don't vary all that much from QFN up through DIP, and I believe the difference 25 years ago (of packages available then) wasn't large. If you're going to optimize for cost of PCB area—which is comparatively inexpensive given the volumes they bought NES-UNROM-xx boards in—they may as well go for the extra price savings of epoxy blobs. Modern costs of PCB area seem to be around 25¢/in² (0.06EUR/cm²) (plus a fixed overhead per mask and per board); it was likely the cheapest component at the time, since ICs were much more expensive.

To make it worse, it tentatively looks like SMT mask ROMs may have cost a premium at the time, given that they weren't used in the SNES era until later, and their use in the DMG could be explained by size constraints.
User avatar
Bregalad
Posts: 8036
Joined: Fri Nov 12, 2004 2:49 pm
Location: Caen, France

Re: Why so few surface mount ICs in NES/FC carts

Post by Bregalad »

Ok I understand package cost was the same and still is the same between surface-mounted and through-hole components, and that PCB area cost is negligible, but then that still doesn't advocate through-hole components, it makes both choices legitimate.

It's definitely bugging me that only a single UNROM PCB used a surface mounted SRAM.
lidnariq
Posts: 10677
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Why so few surface mount ICs in NES/FC carts

Post by lidnariq »

Bregalad wrote:but then that still doesn't advocate through-hole components, it makes both choices legitimate.
Well, SMT also makes routing harder, because it's not as trivial to put traces between legs.
The CAD software they used (since it's clearly not hand-routed) may have not supported SMT (at least at first). And later designs are easier/safer to do as a modification of earlier ones.
And, as I pointed out before, it looks like SMT mask ROMs could have been more expensive than DIP mask ROMs at the time.
drk421
Posts: 329
Joined: Sun Nov 14, 2004 11:24 am
Contact:

Re: Why so few surface mount ICs in NES/FC carts

Post by drk421 »

I'm more surprised by the fact the the SNES were both through hole. Only a few SNES games were surface mount (mostly the ones with special chips).
I'm even more surprised that the N64 is almost entirely through hole. I guess DIP still wasn't that expensive in the mid to late 90s and early 2000s.
yogi
Posts: 133
Joined: Sun Nov 17, 2013 8:14 pm
Location: Bowie, Maryland

Re: Why so few surface mount ICs in NES/FC carts

Post by yogi »

I'm guessing some of the reason is the manufacturing infrastructure at the time. Most assembly lines were setup for through hole till the mid 90s and by then there was no point to switching over to a SMT design (re-design cost for a system reaching end-of-life). Re-tooling pick an place bots and moving from wave soldering to reflow ovens slowed the acceptance of SMT.
OTOH, maybe Nintendo just wanted to keep things hacker friendly :)
Yogi
tepples
Posts: 22345
Joined: Sun Sep 19, 2004 11:12 pm
Location: NE Indiana, USA (NTSC)
Contact:

Re: Why so few surface mount ICs in NES/FC carts

Post by tepples »

Or perhaps it's because Nintendo had franchised service centers in major cities until switching to its current service-by-mail model, and some of these might not have been fully equipped for troubleshooting boards with surface-mounted components.
lidnariq
Posts: 10677
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: Why so few surface mount ICs in NES/FC carts

Post by lidnariq »

drk421 wrote:I guess DIP still wasn't that expensive in the mid to late 90s and early 2000s.
DIP is still not appreciably more expensive ... at least, for things that you can get in DIP at all.
tepples wrote:franchised service centers [...] might not have been fully equipped for troubleshooting boards with surface-mounted components.
But gameboy games were all SMT. At the time, did they handle them differently?
User avatar
Bregalad
Posts: 8036
Joined: Fri Nov 12, 2004 2:49 pm
Location: Caen, France

Re: Why so few surface mount ICs in NES/FC carts

Post by Bregalad »

But gameboy games were all SMT. At the time, did they handle them differently?
That. Surfact-mount was a possibilitiy for Nintendo as :
1) MMC3, 4, 5 and 6 are done in surface-mount
2) HVC-UNROM-04 uses a surface mounted SRAM
3) GameBoy games uses only surface mounted components, but they didn't have a choice for those cartridge dimensions, so pehaps they went that route even if it was more expensive

It's true DIP is easier to route, and usually leads to a simpler PCB by avoiding vias which are unavoidable when surface-mounted components are used, but is also eats more area, even including the "via cloud" area around SMT components that can't be avoided which makes it take virtually more space.
Post Reply