Ceramics for Cartridge Capacitor Replacements

[cptdipsh]

I'm looking at replacing my cartridge electrolytic capacitors before they leak. They of course are under way less heat and strain and are more stable than the SNES board's, but time will come for us all, including the cartridge's electrolytics.

Solid polymer capacitors are the obvious fallback and will last for two decades and when they fail they won't destroy the cartridge; but ceramic capacitors last forever. A true one-and-done.

Looking at the cartridge board, it looks like the electrolytic is a 22uf, 6.3V that goes from power to ground. ESR shouldn't be a factor in that case.

So the question at hand: there any downsides for replacing the cartridge's electrolytic capacitor with a ceramic capacitor and never, ever worrying about the cartridge ever again (except to clean the pins sometimes, of course).

I'm thinking of a 22uf, ~16V 1206/1210 XR7 capacitor. Based on some divinations I did a while ago, I found that the 1206/1210 size would be sufficient to go from pad-to-pad, but that was a while ago and I didn't write anything down about those measurements so it might be wrong. The ~16V is to achieve aggressive derating. 6.3V is too close to 5V - and there are physical actions such as pushing the cart into a (maybe?) live system that could cause more-than-double instantaneous voltage spikes; so 16Vish.

Love for some input on this.

[lidnariq]

Because MLCCs have a much lower ESR, that means both that you get much tighter voltage control, but also higher inrush currents and risk of oscillation.

Lower ESR also means you may not need a whole 22µF of decoupling, although it's hard to guess how much you do need.

MLCCs have been apparently found to be unsuitable specifically for the audio amplifier in the Game Gear, producing the threatened oscillations (causing massive self-heating in the audio amplifier chip, bad distortion, and eventually self-destruction of the amplifier chip)

[93143]

If you could find out the resistance of the old part, perhaps you could put an equivalent resistor in series...?

[Ben Boldt]

Ceramics act a lot different than electrolytics. Another major problem is that the effective capacitance decreases as DC bias increases. This issue is not helped by a higher voltage rating of the cap either. A higher voltage ceramic follows the same capacitance curve, just fails at a higher voltage. So 6.3V is adequate. I want my caps to fail and short out if I have more than 6V in my Famicom!

It is possible that in some cases, the effective capacitance got reduced this way and that would be the cause of oscillation. It might be interesting to revisit that with this in mind.

For a long-lasting replacement in a prized cartridge, you could use a Tantalum cap. (Stripe = + on these!) But I can’t generally recommend that because it is expensive and because Tantalum is a conflict material. Using Tantalums helps people keep fighting wars and violate human rights, and pretty silly to support that for your video game.

Keep in mind, there is nothing requiring any sort of precision about the value of an electrolytic cap in a cart. I think if you do some homework with effective capacitance curves, you might replace a 10uF electrolytic with a 22u or 47u ceramic for example. Even if you used a 10u, probably still totally adequate.

[Markfrizb]

A ceramic 22uF cap is rarely 22uF. I’ve tested a lot of these and it’s capacitance falls with the +/- 20% but always is less than rated value. More like 10-15uF. If you want 22uF actual, then get a 33uF. But these caps are just bulk capacitance, and Nintendo believed that every single cartridge needed it. So maybe it’s there to compensate for power supply ripple or even cartridge connection brown outs? I have come across several electrolytics that have leaked and caused damage to carts so I have replaced many many caps with ceramics. 0805 packages, sometimes 1210. And seeing that these carts seemingly run fine without them also I question if they are really needed.

Not an electronic expert by any stretch, just my 2cents.

[cptdipsh]

Data point from Markfrizb that the carts don't even need them to operate (logical, the carts are in the center of the power filtering on the main board) and that they've replaced them with ceramics. That real-world testing really confirms I can get away with it.

Good note on how ceramics only undercut the rated.

I think Ben Boldt has a strong point on the DC bias issue, here is a datasheet for a relevant ceramic cap: https://web.archive.org/web/20250712202 ... E_Spec.pdf page 14 fig 2-7-1 shows that at 5V DC (which the ceramic will be operating in) it will lose ~80% of its capacitance. 22uF (at best) -> which leaves us with 4.4uF.

You'd need a ~110uF ceramic cap to get ~22uF capacitance at 5V DC bias. And that's expensive and huge. A 6.3V, 47uF ceramic is reasonable and comes in a 1210 package and yields ~10uF including the DC bias drop. That might be the way.

Reading on ceramic caps, https://web.archive.org/web/20250712203 ... 3134#43134 it seems that below 10V the margin can be like 25% or less, compared to the usual 50% on electrolytics. Agree on 6.3V is adequate.

lidnariq noted that lower ESR may reduce the required 22uF, I'll see if that's divinable. Maybe it'll reduce the required uF more, but there aren't any options between 22uF and 47uF. Might be that 47uF is it and we just assume that 10uF works for this application without ESR.

This application isn't near audio (I have seen that the non-CP0/NP0 ceramics vibrate at audio levels and cause problems) so that issue shouldn't be a problem here.

Only outstanding question is newly introduced oscillations, but Markfrizb's real-world data points of not explicitly needed/ceramics have worked makes me hopeful it's a replacement that's without trouble. 93143 did suggest a ~0.1 ohm resistor in series, that would be hard to do physically, but would fix the oscillation worries since ESR is restored. But at that point, should just use a solid polymer cap I think (easier to do).

Thanks for all of your inputs! I'm going to dig into "lower ESR may reduce required capacitance" but if anyone knows more on that, please do share!

[Markfrizb]

I think you are over-thinking this.
It’s just bulk capacitance. It’s on the main power rails. It’s not even essential for operation.

[cptdipsh]

Definitely am over-thinking; but I wanted to do some sort of due diligence on this since cartridge caps will become a larger and larger problem in the future and people will want a solid answer on ceramics since they'll be the easiest forever solution for cartridges.

Reading on the LM7805, it will not oscillate based on ceramics. So I think the power oscillation issues are moot here.

Taking into account the ~80% DC bias loss at 5VDC, the answer (for now) is a 47uF, 6.3V X7R 1206/1210 ceramic capacitor (at 10% tolerance if you can swing it) to yield around 10uF which should be enough since these caps are clearly not essential and ceramics have very little ESR which should also help make up the 22-to-10uF gap (if it mattered greatly, which it does not).

I do not touch tantalums because of the conflict material nature and their explosive failure; the other viable replacement is a solid polymer (not hybrid) capacitor with a 22uF/6.3V rating voltage with leads (not SMT/surface mount), but they will fail (without destroying the board) at some point in a few decades. Remember that these are polarized like the electrolytic that it is replacing. Ceramics are not polarized.

Next step would be oscilloscope readings that show no oscillations induced and/or circuit simulations; but I don't think that effort is relevant for this.

Thanks again y'all for the input, I'm gonna implement ceramics in my cartridges. Will report back if I see disaster, which I am confident from this discussion that I will not

[Ben Boldt]

Yep, we’re over-thinking it but it’s never a bad thing to gain a better understanding. A great many of us enjoy nerding out over non-issues.

I am guessing the connector and traces are already at least 0.1 ohms.

The basic point of having the small 0.1uF caps next to each chip is to have very low ESR. The small cap can react very fast and strong to small high frequency noise.

Consider the big picture, these caps are doing nothing really specific, so their exact value doesn’t matter a whole lot. You just want good practice and general rules of thumb:

Place a small ceramic for each chip, 0.01-0.1uF
Helps high frequency noise because it is low ESR.

Place 1 bigger cap on each remote board (like a cart), 1-100uF
Helps low frequency stuff like resistance of wires or traces getting there. Because its purpose is to help low frequency stuff, ESR doesn’t matter for this cap.

It will work with NO caps at all, yes. The caps can help make things less glitchy and crashy so it is a good idea to have them.

[cptdipsh]

I was looking at ceramic cap brands and found that some have their own sims. KEMET and Murata both have them, Murata's is a bit clunkier https://ds.murata.co.jp/simsurfing/mlcc ... d5ac92705b but KEMET's worked smoothly https://ksim3.kemet.com/capacitor-simulation

Here's the results for a 47uF/6.3V, 47uF/10V, and 22uF/16V: https://files.catbox.moe/o9rs7z.png

47uF/6.3V -> -45.6%, 25.6uF @ 5VDC bias
47uF/10V -> -40.8%, 27.8uF @ 5VDC bias
22uF/16V -> -28.0%, 15.8uF @ 5VDC bias

Murata's has a 47uF/6.3V dropping to 24.8uF @ 5VDC bias (and lowest ripple). Harder to check them out though there and it seemed a bit buggy for some.

Apparently Samsung's ceramic caps aren't as good as KEMET's or Murata's since both don't loose 80% at 5VDC even with a 6.3V rated cap.

But an important note! More voltage rating == less DC bias voltage impact. Can't get a 47uF/16V ceramic cap from KEMET, though, so there's limits.

Anyway, the 47uF/6.3V is plenty fine to get around 22uF. Get KEMET or Murata caps for this, likely whichever is cheaper.

On Ben Boldt's mention of "Because its purpose is to help low frequency stuff, ESR doesn’t matter for this cap." I got to revisiting the RC circuit. ESR looks to matter because it limits the upper frequency at which a capacitor can react. At low frequencies the small electrolytic ESR is fine, at high frequencies you need ceramics and their very negligible ESR to decouple those high frequencies. So I hopefully am interpreting correctly that for this application, they're targeting low frequencies, no ESR ceramics are fine, and the low ESR ceramics will also catch higher frequency deviations that the electrolytics couldn't.

[lidnariq]

As another data point, USB devices have a maximum permitted inrush current, which in turn limits the amount of decoupling you're allowed to put in a USB-powered device.

The bulk capacitance might well end up less important than the smaller tighter decoupling on each IC.

[Ben Boldt]

But an important note! More voltage rating == less DC bias voltage impact. Can't get a 47uF/16V ceramic cap from KEMET, though, so there's limits.

That is interesting because I have heard that the voltage rating doesn't generally help the DC bias curve. I will have to ask some people at work about this.

As another data point, USB devices have a maximum permitted inrush current, which in turn limits the amount of decoupling you're allowed to put in a USB-powered device.

The bulk capacitance might well end up less important than the smaller tighter decoupling on each IC.

We should characterize the maximum allowable inrush of a cartridge for the FC/NES console. This might actually shed some light why Hi-Def NES boards are blowing up. Mine blew catastrophically and I never could fix it; LOTS of stuff got wiped out whatever happened. I probably DID have carts with way too much cap in them.

[cptdipsh]

But an important note! More voltage rating == less DC bias voltage impact. Can't get a 47uF/16V ceramic cap from KEMET, though, so there's limits.

That is interesting because I have heard that the voltage rating doesn't generally help the DC bias curve. I will have to ask some people at work about this.

I think you're correct. The real answer is it seems like there's "series" of ceramic capacitors, if their sim is to be trusted (it should be, they're a well-regarded manufacturer and this is supposed to be actionable data). The "series" have different DC bias characteristics, and changing voltages often changes the "series" of capacitor (and thus the DC bias characteristics).

https://files.catbox.moe/l4u7gs.png

This plot shows (pink and blue are on top of each other):

22uF/6.3V pink at -6.2% and 20.6uF @ 5VDC
22uF/6.3V AUTO edition teal at -21.5% and 17.3uF @ 5VDC
22uF/10V blue at -6.2% and 20.6uF @ 5VDC
22uF/16V yellow at -27.9% and 15.9uF @ 5VDC
22uF/25V green at -24.7% and 16.6uF @ 5VDC

So clearly pink and blue are from the same "series", and likely the other 3 are also a "series". Maybe a set of formulation that doesn't work at high voltages and the AUTO edition is moved to the more resistant-to-voltage formulation?

Anyway, I guess refining farther, look at the sims; I can get a specific KEMET capacitor that's 22uF/6.3V X7R with 10% tolerance that gives 20.6uF at 5VDC and then I can rejoice. (The 10V one seems to be cheaper though, might go with that) Close enough and I don't run the risk of over-capacitancing as lidnariq pointed out is a possibility and Ben Boldt noted it can actually overload things. (I'm not sure how long a ceramic cap acts like 47uF, presumably until it's had its fill and reached its nominal DC bias, and caused a larger inrush of current than a 22uF would in the process!)

So now, going with a specific 22uF/6.3V cap (C1210C226K9RACTU or C1210C226K8RAC for 22uF/10V) that the detailed sim (in lieu of a datasheet) shows is clearly adequate and checks all other boxes. As noted earlier, it might be good to have the cap fail at above 6.3V - so I'll stick with that for an extra $0.50.

Again, thanks y'all for helping me learn the ins-and-outs of ceramic caps. Haven't got it all yet obviously, but better than before!

[cptdipsh]

Got the time to do it, I learned that the 1210 is a good fit - but maybe even too tight. Smaller would work (maybe even an 0805 size), if that’s what you can get ahold of.

Found out that battery cartridges have a second 22 uF cap (labeled CB near the battery). I replaced that one too. I suppose the 22uF will be much lower effective since it’s not on a 5V rail but rather 3V.

The snes light gun game (something like falcon’s revenge or something) had an SMD cap. I did the ole’ “needle nose pliers pointing down, push down lightly, and twist lightly clockwise/counterclockwise till the leads break off” for that. Worked here and for all the caps on the SNES itself, so that method works 100% of the time, every time.

None of the caps leaked, not even the falcon’s revenge game which did have its battery leak (bought used with light gun, must’ve been kept in a hot garage or something). So in 2025 this is still not a priority compared to the caps in the SNES itself.

All cartridges work, tedious but not hard. At least I’ll never have to look in them again (except for the battery ones…).

Again, thanks for all the feedback and discourse on how to do it right.

[Ben Boldt]

Are you sure it’s a good idea to replace the caps though? I understand, it feels good to replace old caps. But there are some pros and cons you should weigh. I didn’t replace any caps in my carts unless I was already making lots of modifications or something.

Reasons to do it:
- old caps could be going bad, though generally we are not seeing that happening.
- ceramic caps should last forever
- you might have the opportunity to find other problems or do a good cleaning inside the cart while you are in there.

Reasons not to do it:
- ceramic caps are different than the original design intended
- inrush is higher, could cause problems for the console
- capacitance may effectively be lower; guessing game to match the original caps
- could accidentally cause physical damage just opening the cart or changing the cap, some small risk here
- don’t fix what ain’t broke; you could wait until there is a problem to change the caps.