Shuttle Swap Mess

So, yeah. Got a bit of a mess on my hands.

Back in 2024, a certain person handed me what was the gateway drug to my new retro addiction: A Shuttle XPC SS30G2. Well, kind of. It was a franken-Shuttle because the chassis belonged to a different system, but the board just slotted right in and worked. Unfortunately that board--prior to me owning it--had blown capacitors, and it was recapped by someone else. While it kinda worked, it still acted a bit quirky, leading me to pulling it apart, reflowing the capacitors (because they were on a HUGE ground plane, the solder joints looked a bit iffy) and that seemed to fix it.

That Shuttle became my Windows XP Shuttle. I really started adoring these systems because, well, they're the practical alternative to the towering behemoth boxes I usually run, and are easy to pack up and put away when I'm not using them. While the XP Shuttle didn't have the performance of my Core 2 Quad overkill XP box, it was easily good enough for some rounds in Unreal Tournament 2004.

Prior to VCF West 2025 (where this Shuttle was featured), it was running a Pentium Dual Core E2180. Seeing as this was just a stripped down Core 2 Duo, I rolled the dice (with Shuttle's documentation saying that this board might not work with it) on a Core 2 Duo E4500. For all of like, $5 plus shipping on eBay. Yeah. Dice roll. Yep.

To my utter amazement, the SS30G2 fired right up, recognized the new CPU, and we were off to the races. The SS30 performed admirably at VCF, remaining rock solid and stable throughout the two seven hour days of nonstop use.

Later in 2025, things would take off in a big way, as I'd get four more Shuttles (and a Shuttle equivalent). A good friend had three Pentium 4-era Shuttles in his garage and he gave them to me so I could either use them or give them good homes. One of them, an SB61G2, became what I call the "practical" Windows 98 machine, a foil to the ultimate Windows 98 machine I had on my VCF table earlier that year. Where the ultimate 98 box was big, loud, and heavy, the practical SB61G2 was quiet, less hot, and way lighter.

I digress, however, as the Pentium 4 trio doesn't have anything to do with the mess that I described at the beginning.

Toward the tail end of 2025, another interesting Shuttle walked into my life: a Tricaster Pro 2. This was just a Shuttle SG31G2 with a custom front panel and a PCI card to enable all the Tricaster stuff. Past that, there was nothing really bespoke about it. The board was standard, the frame was standard, you name it.

Unfortunately, this Shuttle was in dire straits when it got to me: Blown caps everywhere, on the motherboard and power supply. However, me being impatient...I decided while I was waiting for the caps to show up, I swapped in the PSU and board from the SS30G2. (And yes, in the process, I did dump the Tricaster's hard drive so I have access to the drivers/software/recovery partition.)

Imagine my shock when the Tricaster mostly sprung to life. Given that the Tricaster's software is pretty much just a customized XP install, I fully expected it to crash and burn with a BSOD. It did crash and burn, but in trying to load what seemed like Explorer. Turns out by default the customized install sidelines Explorer and sets the Tricaster software to start in its place, so you only ever see Windows XP proper if you enter Admin Mode.

The problem: Because the Tricaster can't see the Video Toaster card, it locks up. Because the chipset drivers aren't installed, and thus, the Video Toaster can't be detected. I was able to boot into Safe Mode and set it to start Explorer by default, so now whenever the Tricaster booted it would boot into "Admin Mode". (Airquotes because it's really just the XP desktop. NewTek really wanted to abstract the Windows OS away as much as they could.)

Thankfully, on a reboot, XP picked up the chipset of the SS30's board and chugged right along installing drivers. Given the system change it DID deactivate, but a quick hit of the offline XP activator took care of that. I jumped in, fired up the Tricaster software, and hey! It was working! I didn't have any video cameras or devices to check the inputs, but it fired up and detected everything, so I assumed we were good to go.

Now, given the Tricaster really is a one-trick pony (as in, I'm not going to Bringus Studios this and install games to it) and seemed to mostly lean on the GPU for acceleration, I made the decision to just keep the SS30's board in it. Yeah, it "only" had a pokey 2.2GHz Core 2 Duo, but that was fine enough for what the Tricaster was meant for.

The SG31 board that was originally in the Tricaster, though? Oh, this thing has potential, and this is also why I wanted to put it in my more general purpose XP machine. Not only does it have a better chipset (Intel G31), but it supports all Core 2 Duos. The SiS 662 in the SS30 felt like a Pentium 4-era chipset with bodged-on Core 2 support (and as such it only supports the earliest ones).

With a recap done, that board got right up and started working flawlessly. Even pinched a Core 2 Quad Q9550 from my overkill XP build to stick in it and sure enough? Happy as can be.

So to recap, in case I lost you: 

The SS30's motherboard (the one previously in my general purpose XP Shuttle) is now in the Tricaster.

The Tricaster's original board (the SG31) is now in the general purpose XP Shuttle.

You'd think that'd be the end of it, right? Oh, you know how my luck is by now.

Houston, We Have A Problem

I didn't get around to recapping the PSU until a few weeks after the board. Partially because I recapped the SG31 board, put it aside for a bit, then went to give the PSU the MikeTech test (connect some dying HDDs to it and force it to turn on by grounding out the green pin) and while it started up and ran, it smelled awful. Cracking open the PSU's case revealed the cause: Several freshly-blown caps. Oops.

 

Fortunately for me, all the caps were of a single value, so I was able to grab a multipack for cheaper than it would be to buy them individually from somewhere like Mouser. I recapped the PSU, stuck it back in the Tricaster (with the SS30's board) and...it was doing the same thing the SS30 did when it first came into my possession.

 

Specifically, what it does is it'll get stuck in an endless power cycle loop. Not a hard power cycle either, a soft power cycle. You'll hear the fans spin up, then down, then up, then down. I initially thought this was due to the chipset (SiS 662) not getting along with more modern video cards (like the Quadro K4000 I had in it) but this was now happening even with era-appropriate cards. Something was amiss. And I already thought I fixed this problem!

 

"Wait a second," I thought. "Last time I tried this, it was outside of the case. Maybe if I pull it out of the case...?"

Sure enough, that was it. At least I thought. Outside of the case, with a tower cooler sitting on the CPU so it doesn't bake to death, it fired right on and POSTed. Even with a video card. Great! This must just be something in the case shorting. I looked at the board. Nothing looked amiss. So I just went absolutely nuts and coated the entire bottom of the Tricaster's chassis with kapton tape.

 

That didn't fix it. I double checked. No standoffs where they shouldn't be. Everything lines up. Nothing should be shorting. I then remembered the one thing I did differently when the board was out of the case: Using a different heat sink. So I pulled off the stock Shuttle cooler, laid the tower cooler onto the CPU, hit power and...there's the POST screen. Huh.

 

Experimenting a little more: If I screw the heatsink all the way down, board acts dead. If I only screw it in a couple turns, it's fine. Still, nothing is shorting, so it can't be that. I grab a bit of cloth as to not shred my fingers, and press down with the tower cooler on the CPU socket. No POST.

 

I check the LGA 775 socket. Only one bent pin on the LGA, and it's a common VCC pin, so that won't be causing it. Everything else checks out. I tried getting the old Pentium Dual Core out of the drawer and slapping that in. No change. I pull the board back out and notice the most subtle of board flexes right along the plane the chipset sits on.

 

Yeah, uh, this isn't looking good.

 

I pull the heatsink off the chipset, flux it up, and try to reflow it with my hot air. The good news is I didn't make the problem any worse, but the bad news was that I didn't make the problem any better.

 

Worse, I guess all my (gentle) handling of the board finally did it in, because eventually, it finally just...died. Can't get anything out of it now, even outside of the case. Talking with some people on Discord, apparently the SiS 662 wasn't known for its reliability, either. If I had to hazard a guess this poor board was on its way out all this time and moving it about finally tipped it over the edge and it died for good.

 

At this point, I just put the Tricaster back together and let it be a pretty corpse. Could I just swap boards back? Sure, I guess. But the Tricaster is a single-purpose device. Whereas I use the XP Shuttle quite often. To me, it's more important that the XP Shuttle remains operational, whereas the Tricaster can be a future me problem. It doesn't do me super well to have an operational Tricaster but a nonfunctional XP box.

 

That said, I haven't given up on it. I do have feelers for an era-appropriate replacement mainboard so I can get it up and running. I wanted to have the Tricaster running for VCF West 2026, but at this point I'm planning my table as if it wasn't going to be here at all. I'd rather have a more general-purpose fun machine than something that's purpose built for something not many people are going to care about.

 

I do hope to have it running soon, it would be nice. But I'm not going to lose any sleep if it takes a minute. At least the HDD is backed up and the parts that make it actually a Tricaster are still good.