Google Mini 1u Supermicro Server

So I picked up a Google Mini 1u server from a local electronics recycling goldmine-of-old-stuff. I’m not a huge Google fanboy by any means, it’s kind of like self-inflicted spyware — or some sort of trap. But I figured this computer looks cool and it would be fun to re-purpose into other tasks.

The machine is pretty old. 250 watt power supply (Ablecom), Supermicro motherboard that has PCI and PCIe-64 bit slots, and a 3Ghz P4 era CPU as I recall. In the future I can see myself swapping in something faster and more power efficient. But I’m not going to worry about that now.

Someone else provided some instructions about how to reset the BIOS as there was a password on it. It’s two half-moon pads near one of the large ICs near the PCI slots. Thanks for the info, that got rid of the password. I ended up mounting two 1TB disks in the thing, which required buying a 2nd right angle SATA power connector and I have a right angle data cable on the way for good measure.

I used a utility knife to carefully cut across the top front of the sticker that covers the entire top of the computer. This is the one with the google logo. I used a dremel tool to cut a straight slot through the security screws so I could remove them with a normal flat blade screwdriver.

Installing CentOS on the thing was the tough part. With the default BIOS revision of 1.1, it didn’t seem to want to boot off of USB. I burned a CD with the CentOS 7 ISO and the initial boot screen would come up but the OS didn’t seem to load. Memtestx86 from that CD just locked up. This was using an external BD-R drive connected via USB. So I downloaded the version 1.1A revision of the firmware and that added much better handling for USB booting. It did wipe out the custom Google Mini splash screen, though. Once the USB thumb drive was connected it was possible to go into the BIOS and set the removable USB device in the boot order. Note — one of my hard drives had some sort of boot block that just gives a cursor and a freeze so that complicated things a tad. There is a setting to slow the blower down which helps with the noise. It’s not as loud as the 3u Supermicro I have so it isn’t much of an issue to me.

Disks are setup with software RAID. I have a 4 port 3ware left over from an old 757.org server and I contemplated getting a 1u 64bit pci riser card so I could put the 3ware card in, but decided against it. I’ll probably regret this at some point in the future but at least the disk IO will be faster.

In the end, it’s up and running CentOS 7, which is kind of gross. It’s this new systemd that most people seem to dislike and they’re doing everything they can to make it look like Windows. I installed OwnCloud to evaluate it for a personal project but realize it’s not what I need and will revert to a long configuration of Apache + mod_dav + LDAP with 389 Identity Server behind it.

Yamaha C1 Music Computer Laptop

Most vintage computers I fall into. But there was one machine that has been on my radar since I first saw one in a store when I was little. That is the mysterious Yamaha C1 laptop. It’s a 286-12 MHZ laptop that has 11 midi ports and SMPTE time code in and out. I hesitate to call it a laptop because it’s so heavy, and there is no option to battery power it. It feels heavy enough to have a battery in it, but it’s just built fairly sturdy.

I’ve had an eBay trigger setup for a few years, but it mostly triggers on people trying to sell advertisements for the machine or ECUs from Yamaha motorcycles that have C1 in the description. But finally, one came up with books and disks and the original bag, and I bid and snagged it.

The caveat is that it has a functional issue. The unit always says it’s sending video to the external display. There is a dip switch on the rear that switches between the internal LCD, or an external CGA monitor. But in the case of this machine it always says it’s going to the external. So I have to fix this.

In the meantime, I am going to scan all the documentation, archive all the floppies and have already read all of the eproms into files. I’ve also documented the power supply connector, and in the documentation have the pinout for the expansion connector slot (but not the hard drive controller header.) So I will be getting all this info together and posting it for the world at large.

In the meantime — back to the issue of the LCD not working. Randomly, perfect timing, my friend Chris Favreau sent me two huge boxes of Amiga 500 goodies. In this lot was an Amiga 1084S monitor which can do CGA. So I wired the Yamaha over to the Commodore monitor and as I expected, there is no output. I never hear a floppy seek so my suspicion is the computer isn’t booting at all.

It’s been sitting open on the kitchen counter for a good while next to a few other projects getting attention. Last night I started really looking around the motherboard, and something I looked for originally but missed is the case. The green electrolytic capacitors look to have leaked some electrolyte. This means two things. First, the capacitors have probably lost their properties and might be causing the machine to not function properly (I assume they’re noise filtering but not sure.) The second — worse part — is the electrolyte is often corrosive. I think I can see a trace that is damaged by this. So the next step is going to be to replace all of the 47uf 25v green caps with replacements, and if the one trace is indeed bad I’m going to have to jumper it if it goes anywhere important. It looks like it just goes to the expansion interface slot — so if it dead ends at the expansion slot connector it’s not a big deal. But I won’t know until the motherboard has been removed. So. Many. Screws.

In the end I’m glad to have found something that looks like it could be the issue. I heard of someone else with the exact issue (External CRT always no matter switch position, and non functioning perhaps.) Electrolytic caps are probably the #1 failure of all electronics so not totally unexpected.

—[ UPDATE 7/5/2015 ]—

I replaced the 9 x 47uf 25v electrolytic caps that were leaking in the system. I cleaned up the board and tested the traces that looked like they have some corrosion from the leaked electrolyte and they seem to still function. The bad news is the system still seems to not be posting at all. No floppy seek and it’s still stuck saying external CRT when the dip switch is set for internal LCD.

The rear panel dip switch #1, on one side is connected to ground. The other side hits resistor pack PA12 near the keyboard connector, then goes on to IC52 which presumably reads it. So it’s probably read by software.

I sent an email to Yamaha USA asking about schematics. This normally does not work with any large corporation, but we will see. Anyone that has one of these systems working, you’re at risk of damage from capacitors leaking. This system I think was stored sitting upright, so the electrolyte managed to go through the board and run down the back.

Next up is I don’t know. I might look into the inverter stuff for the LCD, the sub-board. I will also wait to hear from Yamaha.

—[ UPDATE 10/2/2015 ]—

Okay this thing has been on the counter too long. I re-capped the LCD inverter board, still no signs of life. There is no disk seek on startup, which makes me think the computer isn’t executing code. I found online someone else talking of having the same issue. I looked at the motherboard more, trying to inspect for damage from the traces that show slight corrosion from the electrolyte leaking. But nothing really stands out.

Another thought that hit my mind is I’ve heard that some EPROMs lose data over long periods of time. So perhaps the BIOS chips are corrupted? It might come down to having to find a working machine to compare. Looking in the data it does appear there could be corruption. The name Phoenix Technologies has the first letter as a C in one ROM, another character in the other. Bitrot might be the issue.

I dug out the o-scope and looking at one of the BIOS EPROMs I see activity on all data lines, all address lines and the OE pin constantly. Not sure if this is normal, since no working unit to compare against. The lack of any normal ISA slot means there is no way to plug in any of the POST code debugging boards that existed for ISA slotted computers of this era.

————————————

Here are some pictures of inside the Yamaha:

View post on imgur.com

And new blog started

I’ve been hammering through a ton of projects recently and decided I need to kick off a blog about them — mainly for my own notes. But some of the information should be shared with the world at large.

I used to post on the 757labs hackerspace blog, but since that’s mostly archival now this will be the new repairs and info spot.

I still need to figure out some workflow stuff — mainly posting of pictures inline. I moved to using Imgur more than Flickr as I don’t like the newer Flickr interface. Not sure if Imgur will do resized direct linked images. We’ll see!

Also I’m not on the current theme. In time will tweak it. For now tho, content.