![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
The tornado watch reaches all the way up to here, though all we've had is rain with a little thunder. Looks like Texas and Oklahoma are taking another plastering though. I heard a storm expert on the radio at noon saying that this is already the worst tornado season for fatalities in the US since 1953. In fact, it will take only another 19 deaths to surpass that year.
Since the weekend I've been wanting to boot up the DEC Alpha in my office space at home to test a couple of things. When I went to do so on Sunday, I found that the UPS supplying it had died. Probably just a dead battery, but it will take a couple of weeks to revive that or replace the unit with some other restoree. (I have three other dead ones sitting around that I know just need batteries.) Dreading the job of crawling under desks in very cramped, dark space to try to rearrange power connections, I let it slide. This morning I finally gave in, and with a couple of bruises managed to replace the UPS with a simple surge protector bar. "That should do it," I thought, and turned on the console, drive stack, and finally the CPU.
Beeps, whirs, and clicks sounded normal, but nothing appeared on the console. Thought I'd knocked a cable loose, so I shut down and rechecked everything. Nope, it was all intact. Turned on the seldom-used monitor that provides a graphical display interface for the Alpha, and tried it again.
It was going straight to the boot sequence for Windows NT, which ignores the serial console. Small Alphas of that era (mid-1990s or so) were made to run Windows NT or, at the user's option, a choice of NT, Unix, or VMS. I had in fact tried NT briefly on the machine to see what it was like (pretty primitive by today's standards) but the normal console boot goes to a prompt and waits for the user to choose their boot options. It wasn't doing that. Nothing I could come up with to try would keep it from going straight to the Windows NT bootup sequence. Even after I disconnected the keyboard, mouse, monitor, AND disk drives, it still wanted to boot NT, with no other options offered.
Finally I turned to the net for an answer, since DEC documentation (at least the surviving materials to which I have access) said nothing about this possibility. It turns out, though, that if the CMOS battery or clock battery in one of those Alphas goes dead, this is the typical symptom. OK, no surprise that the battery may have died. I've never changed it. I don't know if anyone else ever had either, and the machine was made in 1997 or so. Lacking clear information about what battery was needed, and fearing it might be a model with one of those difficult to find "SmartWatch" type clock chips with built-in battery, I unhooked everything and opened the case. Of course what I needed to check was completely obscured by the power supply and disk drive cages. Thought I had a major disassembly to undertake for a few minutes, then discovered that the CPU portion of the motherboard slides out from under everything once you flip the right levers to release it.
Lo! It was the familiar CR2032 coin cell. Nothing hard to replace at all, though I didn't have one on hand so I reassembled it all and put it back under the desk until I could get a battery. Then I went searching again on the net just in case there is some trick for getting out of the NT BIOS screen and back to the console monitor firmware. Turns out there is, though I don't think it is documented in any DEC materials I've seen. So, for posterity, here is the procedure:
Alpha PWS433au console escape to get back to SRM from AlphaBIOS for Windows NT
CR2032 lithium cell on motherboard dead? Causes machine to go to AlphaBIOS (Windows NT) on powerup. Getting back to SRM using the serial console is tricky, but can be done. There may be small variations from one Alpha desktop model to another, and these instructions are based on a PWS433au with the MIATA motherboard.
Unplug PC keyboard and mouse (if any,) and turn on serial terminal on COM1. Then power on. Wait a while, and eventually the AlphaBIOS boot options for Windows NT appear on the serial terminal. (Actually some of the prompts say "Windows 2000" but as far as I know, Microsoft and Digital had parted company before 2000 was actually released.)
The Fn keys shown on the screen do NOT correspond to the keyboard function keys on your serial terminal. Use these:
CTRL-B for F2
CTRL-F for F6
CTRL-U for F10
CTRL-V for Insert
CTRL-W for Delete
CTRL-H for Backspace
CTRL-[ for Escape
Keyboard + and - to set options (not the keypad)
TAB key to move around in menu
Press F2 (CTRL-B) to get to Setup
TAB to "CMOS Setup..."
Press ENTER
Press F6 (CTRL-F) to get "Advanced CMOS Setup"
TAB to "Console Selection"
Use + to change to "OpenVMS Console"
Press F10 (CTRL-U) to get back to "CMOS Setup"
Press F10 (CTRL-U) again to save changes and exit
Press ENTER to confirm.
Use the RESET button to restart, and you will arrive at the chevron prompt. Now you can boot Linux or VMS, but will have to set date and time. Don't bother to set them from the AlphaBIOS menu because the format saved is not compatible.
[Edit May 26] Just a postscript. Replacing the CR2032 coin cell did the trick. Everything is back to normal.
Since the weekend I've been wanting to boot up the DEC Alpha in my office space at home to test a couple of things. When I went to do so on Sunday, I found that the UPS supplying it had died. Probably just a dead battery, but it will take a couple of weeks to revive that or replace the unit with some other restoree. (I have three other dead ones sitting around that I know just need batteries.) Dreading the job of crawling under desks in very cramped, dark space to try to rearrange power connections, I let it slide. This morning I finally gave in, and with a couple of bruises managed to replace the UPS with a simple surge protector bar. "That should do it," I thought, and turned on the console, drive stack, and finally the CPU.
Beeps, whirs, and clicks sounded normal, but nothing appeared on the console. Thought I'd knocked a cable loose, so I shut down and rechecked everything. Nope, it was all intact. Turned on the seldom-used monitor that provides a graphical display interface for the Alpha, and tried it again.
It was going straight to the boot sequence for Windows NT, which ignores the serial console. Small Alphas of that era (mid-1990s or so) were made to run Windows NT or, at the user's option, a choice of NT, Unix, or VMS. I had in fact tried NT briefly on the machine to see what it was like (pretty primitive by today's standards) but the normal console boot goes to a prompt and waits for the user to choose their boot options. It wasn't doing that. Nothing I could come up with to try would keep it from going straight to the Windows NT bootup sequence. Even after I disconnected the keyboard, mouse, monitor, AND disk drives, it still wanted to boot NT, with no other options offered.
Finally I turned to the net for an answer, since DEC documentation (at least the surviving materials to which I have access) said nothing about this possibility. It turns out, though, that if the CMOS battery or clock battery in one of those Alphas goes dead, this is the typical symptom. OK, no surprise that the battery may have died. I've never changed it. I don't know if anyone else ever had either, and the machine was made in 1997 or so. Lacking clear information about what battery was needed, and fearing it might be a model with one of those difficult to find "SmartWatch" type clock chips with built-in battery, I unhooked everything and opened the case. Of course what I needed to check was completely obscured by the power supply and disk drive cages. Thought I had a major disassembly to undertake for a few minutes, then discovered that the CPU portion of the motherboard slides out from under everything once you flip the right levers to release it.
Lo! It was the familiar CR2032 coin cell. Nothing hard to replace at all, though I didn't have one on hand so I reassembled it all and put it back under the desk until I could get a battery. Then I went searching again on the net just in case there is some trick for getting out of the NT BIOS screen and back to the console monitor firmware. Turns out there is, though I don't think it is documented in any DEC materials I've seen. So, for posterity, here is the procedure:
Alpha PWS433au console escape to get back to SRM from AlphaBIOS for Windows NT
CR2032 lithium cell on motherboard dead? Causes machine to go to AlphaBIOS (Windows NT) on powerup. Getting back to SRM using the serial console is tricky, but can be done. There may be small variations from one Alpha desktop model to another, and these instructions are based on a PWS433au with the MIATA motherboard.
Unplug PC keyboard and mouse (if any,) and turn on serial terminal on COM1. Then power on. Wait a while, and eventually the AlphaBIOS boot options for Windows NT appear on the serial terminal. (Actually some of the prompts say "Windows 2000" but as far as I know, Microsoft and Digital had parted company before 2000 was actually released.)
The Fn keys shown on the screen do NOT correspond to the keyboard function keys on your serial terminal. Use these:
CTRL-B for F2
CTRL-F for F6
CTRL-U for F10
CTRL-V for Insert
CTRL-W for Delete
CTRL-H for Backspace
CTRL-[ for Escape
Keyboard + and - to set options (not the keypad)
TAB key to move around in menu
Press F2 (CTRL-B) to get to Setup
TAB to "CMOS Setup..."
Press ENTER
Press F6 (CTRL-F) to get "Advanced CMOS Setup"
TAB to "Console Selection"
Use + to change to "OpenVMS Console"
Press F10 (CTRL-U) to get back to "CMOS Setup"
Press F10 (CTRL-U) again to save changes and exit
Press ENTER to confirm.
Use the RESET button to restart, and you will arrive at the chevron prompt. Now you can boot Linux or VMS, but will have to set date and time. Don't bother to set them from the AlphaBIOS menu because the format saved is not compatible.
[Edit May 26] Just a postscript. Replacing the CR2032 coin cell did the trick. Everything is back to normal.