N64 Electrolytic Caps & Video (Modding Journal)

The place for all discussion on gaming hardware
Post Reply
User avatar
RIP-Felix
Posts: 140
Joined: Sun Mar 31, 2019 7:54 pm

N64 Electrolytic Caps & Video (Modding Journal)

Post by RIP-Felix »

TL;DR:
1) Does Voultar's N64 THS7374 RGB amp depend on console caps at all? If the console's electrolytic caps degrade, will it affect the RGB signal?

2) Does anyone know where I can find electrical diagrams/schematics for the various N64 motherboard revisions? Specifically I'm looking for the purpose of each of the electrolytic caps on the NUS-CPU-08 and NUS-CPU-08-1 revisions in my Funtastic models. I thought I heard these were leaked, so they should be available. I just can't seem to find them.

The TL:
I have a problem buying broken consoles off ebay and trying to fix them. This time I bought a Jungle GRN Funtastic N64 (Motherboard Rev. NUS-CPU-08). This board revision has the MAV DAC/video processing chip making RGB bypasses more difficult/expensive, but results in "better" Composite/S-video compared to the previous DAC (Assuming that's the only reason, there are probably more). Comparing it side-by-side to an RGB modded Charcoal Grey NUS-CPU-04 (Voultar's THS7374 amp > RAD2x, no filter > Startech USB3HDCAP > HTPC via Stream Catcher with lossless settings), the image is brighter. BTW, this capture setup when viewed back on the same TV, looks the same to me as if I were playing directly via HDMI from the RAD2x. I trust it, but it's never going to be perfectly lossless. So long as I don't change the capture settings, they should be directly comparable between runs, even if they don't produce as high of fidelity signal as you would see on screen and in person. Now, I didn't recap my Charcoal Grey, so this may not be a fair comparison. I don't know if the electrolytic caps on the board are completely bypassed by voultar's board or not, hence question #1 in the TL:DR.

I have noticed a huge improvement on my recaped "Funtastic" Jungle GRN (NUS-CPU-08) vs 2 other Funtastic models that have not been recaped yet. I recorded the In-Circuit ESR of the electrolytic caps before the replacement using a cheap MESR-100 100KHz In circuit ESR Tester. I would like to know which is/are most likely responsible for the differences I'm seeing. Namely, it's brighter and it seems cleaner/sharper. I should mention I'm talking about Composite/S-video. I do not have any mods installed on this board.

I'm holding the other 2 Funtastic models in reserve to experiment on, as I recap to test the hypothesis' I come up with. I recorded the ESD of their caps as well. interestingly, it appears that all 3 had a significant number of electrolytic caps test bad or marginal. The Jungle GRN model I recapped had 60% test bad and did not boot, but looking at the caps that were bad compared to the other 2 consoles that do boot, did not identify a culprit. I can post the ESD by Cap/console if anyone is so inclined, but this post is long enough as it is. Suffice it for now that the Grape NUS-CPU-08-1 had 40% and the Watermelon Red had 47% test bad. Both function, but it's hard to make out details in dark areas. For example, darker lit caves & underwater sections in DK64 were too dark to enjoy/navigate. The recaped Jungle GRN brightened these up nicely. If the ESD from 3 NA consoles a cross section does make, then the N64 capacitors are in dire need of replacing. Note, these were late model releases in the N64 lifespan, so earlier models are probably in worse condition due to nothing more than age, let alone other factors. I think it's safe to assume the same for the power supply's caps.

What's my point? Part of the difference people see in RGB/HDMI mods are amplified by how much the Composite/S-video signal has degraded by age. Composite/S-video is getting a worse rap that it deserves, especially S. I compared S-video on my recaped "funtastic" Jungle GRN to RGB on my modded Charcoal Grey and the difference at 480p is underwhelming (As it is with most consoles, it's just more so with the N64 thanks to VI blur). I certainly wouldn't say it's worth the cost/headache of an RGB mod unless you planned to go beyond 480p, at which point you're probably in the sharp pixel camp and an UltraHDMI mod looks more attractive.

If you read this far, I'll end with this. You're a nerd!
User avatar
RIP-Felix
Posts: 140
Joined: Sun Mar 31, 2019 7:54 pm

Re: N64 Electrolytic Caps & Video (Modding Journal)

Post by RIP-Felix »

Anecdote: Testing revealed a Bug in DK64 when using Poregon's APS/IPS patch

Last night I noticed something strange in Donkey Kong 64 while capturing video to document the before comparison (OG caps @ 40% bad). When you start the game and press start to get past the DK Rap, the video demo begins. It tells you to press start to begin the game, but because I want footage I can synchronize side-by-side with a slider mask, I captured this demo. The first time you boot the demo is always the same, DK kills an enemy, jumps into the water, climbs up a root to a plateau with a palm tree. This is where the bug becomes apparent. Normally he goes left around the palm tree, stops at a clifs edge to time his jump, jumps onto a swinging vine, jumps twice to cross 2 more vines and clear the clif between platforms, the rolls into a barrel that shoots him up further. The screen "DK fades" in RARE fashion, before the next demo plays.

When I went back over my footage, I realized that sometimes DK hits the tree and slides to the right. He stops, jumps twice, falls off the cliff and performs a mid air attack (B button press) just before hitting the water. The camera clips through the cliff a bit and the video studders some. This is pretty glitchy for a pre-recorded video RARE would place as the first Demo that plays everytime you start the game. Then I remembered that resources are tight on 64 carts and DK64 is a big game. This is most likely not pre-rendered video. This must be running procedurally in engine. It struck me as odd, but I assumed it must be a revision difference between my physical copy (NUS-NDOE-USA) and whatever version I have on my everdrive 64.

Later, while testing the everdrive with/without Poregon's APS patch, DK went around the palm tree on the left like in the original cart! I enabled the patch and he glitched to the right again. So it's the patch causing this behavior. Something about how Poregon's patch disables the first pass software Antialiasing disrupts the demo. Unless I'm mistaken, isn't this patch supposed to just be the gameshark code? I tried manually entering the code on my everdrive (properly), but it just hangs on boot (Black screen). I have an actual Game shark, but this is the first time I've busted it out in 20 years. It just displays an 8 and refuses to load, not sure what that's about.

I think the inputs are the same, but DK is in a slightly different position, causing him to veer off course. He jumps off the cliff, but had he gone around the tree on the left, he'd have grabbed onto the vine and jumpd twice in rapid succession. I don't see the last jump press, because DK is falling. But when he is supposed to have landed on the platform on the other side, there is a B button press to roll into the barrel. I think that's the in air double fist attack just before he splashes down into the water, on the glitched version.

For some reason the patch causes DK to move more to the right in the opening demo, but just that part. The other demos seem the same. This is just niche enough for me to think that it may have never been reported before, but I haven't done extensive searching. Regardless, it's one of those suprising things that only pops up when you do a bunch of testing and pay close attention. I just assumed, everything else would remain the same. It is an inaccuracy on real hardware though. It's using an unoficial patch that has could have unknown consequence besides this. I wonder what else is different and if any of the differences could be more severe.
User avatar
Link83
Posts: 342
Joined: Tue May 21, 2013 2:39 am

Re: N64 Electrolytic Caps & Video (Modding Journal)

Post by Link83 »

Not much to add to your findings, but just wanted you to know there are actually two different manufacturers of the MAV-NUS video encoder chip used on the later models N64's:-

'MAV-NUS' ROHM BU9906FS
'MAV-NUS' Ricoh RS5C382

I haven't seen any comparison between them online (Most people are only interested in RGB) but just thought you should know that this could be a factor in your comparisons.
Last edited by Link83 on Mon Jun 29, 2020 11:12 pm, edited 1 time in total.
User avatar
RIP-Felix
Posts: 140
Joined: Sun Mar 31, 2019 7:54 pm

Re: N64 Electrolytic Caps & Video (Modding Journal)

Post by RIP-Felix »

Good to know. The Green NUS-CPU-08 says (MAV-NUS BU9906FS 938 162). I'll check the others and be sure to record that. Thanks for the heads up...Edit: the Grape NUS-CPU-08-1 has a MAV-NUS RS5C382 and Watermelon NUS-CPU-08 has another BU9906FS. I should be able to compare them side by side to see if it makes any difference. Ive captured enough before footage. Next step is the recap.

I read an old post you made on the Nintendo Forever forums which mentions that circuitry built into the Multiout itself might also have an effect. I compared the composite on the RAD2x and official Nintendo Multi-out (Via Tink2xPro) and they were identical. I'm using a cheap ebay unshielded s-video cable (wired correctly for S. It's not just passed through composite) to the TINK2xPro to compare S-video. It's significantly better, so if that's the worst case for S, it's still one hell of a case. So at least with the cables I've tried, the differences appear to originate in the console/mod alone. So that just leaves capacitor condition, chipset, output method/mod.
Post Reply