bryan wann :: binary fury

If you’re running Qodem on MacOS with iTerm2, you may notice extended ASCII characters don’t render as lines, but just a bunch of letters like PPPPPPPPPPPPPPPPPPPP. Try unsetting the environment variable TERMINFO_DIRS (unset TERMINFO_DIRS) before starting Qodem. This seems to have fixed it for me.

I figured this out by noticing when I ssh’d from my Mac to a Linux system and ran Qodem, extended ASCII and ANSI escape sequences rendered just fine. However, when I ran Qodem locally from my Mac, all of the line drawing was messed up. I’m reasonably sure it was compiled right by the Homebrew community but couldn’t see any other reason for it to not render correctly. For giggles I did ssh localhost, ran qodem again on the same Mac and to my surprise everything rendered correctly!

I did a little diff’ing of env before and after unsetting things and finally settled on the TERMINFO_DIRS variable that was doing it. It gets set locally, but unset when ssh’ing to a remote system. I don’t know offhand what the difference is in the terminfo iTerm is using vs system terminfo, I just know it fixed my problem.

(For reference I’m running iTerm2 with the Cousine font, and Courier New for Non-ASCII Font, and UTF-8 character encoding. TERM is set to xterm-256color too.)

Before, running Qodem in iTerm2 on MacOS:

Yuck!

After running unset TERMINFO_DIRS before running Qodem:

Much better!!

Terminal.app

This doesn’t seem to be an issue in the standard Terminal.app, as it doesn’t define any sort of custom TERMINFO variables. However Terminal.app does seem to enforce a character and line spacing of “1” which gives ANSI graphics a bit of a screen door effect, which is mildly annoying to me.

Default Terminal.app settings

Wasted

[photos: flickr – Seagate ST-225 repair]

A follow-up to https://binaryfury.wann.net/2023/07/2023-drive-belt-saga/

I wound up buying another “parts only” Seagate ST-225 hard drive off of eBay to attempt to fix my ST-225. This is a 20 MB MFM drive out of the old family 286 computer. Long story short I’m pretty sure my repair worked but ultimately the drive wasn’t recognizable by the BIOS. I think track 0 (on the outermost of the platter) was too knackered up for it to be found. This could have been due to the move.  I don’t know if there’s any software I can use to get any sort of raw dumps since the drive doesn’t even get recognized.

As an aside when working on my 486, I noticed in the 5.25″ floppy drive there’s totally a stepper motor and a split band assembly in there. I’m wondering if I could’ve stolen one off an extra floppy drive, if it was long enough to resize to what I needed.

Donor drive before fiddling

Opening up the new donor drive both of its split drive bands were intact which was great. Upon further inspection I found one of the read/write heads had broken off and was rattling around in the drive, which explains why it was dead. A literal hard drive crash.

Crashed/missing drive head on donor drive

I very carefully disassembled the donor drive bands, taking lots of close up pictures beforehand so I would see how the various bits like washers and ends were orientated so I could do the same on the old drive.

After some very careful threading and tugging, I got the drive bands on my old drive. Tip: mount up the left band first to the two studs, then do the rightmost band. At the end of the band next to the screw hole is an extra hole, put a paperclip or something through this to let you keep everything under tension while trying to thread up the screw.

Bands assembled on old drive

After hooking my old drive back up to the 286 and booting it, I could see the stepper motor spinning and it sounded like it was seeking. Unfortunately POST paused and it coughed up the C: drive error.

After a couple of reboots I decided to throw in the towel. I just had to open up the drive to see if my fix worked. This probably destroyed the heads with dust or something. Turning on the system again, the heads did fully seek across the platters and back. It had a slight ticking sound like maybe it was trying to seek further than it could so I don’t know if maybe there’s an index on the stepper motor and I twisted it out of place when working on it.

Maybe someday I’ll get back to working on it…

A series in yak shaving…

I’ve been working on vintage equipment all summer, mainly from bringing back old computers and media back from Oklahoma with the intent of trying to get the systems running again like an old ’67 Mustang and/or retrieving the data for memories sake.

I’ve quickly learned if it’s not an old capacitor going up in smoke, it’s a good chance a drive belt of some kind has worn out and broke or lost tension. I present four case studies:

TRS-80 model 4

I naively plugged this thing in before reading any sort of restoration guides and within a minute a RIFA power filtering capacitor went up in smoke. A new, modern power supply made by a TRS-80 hobbyist later, it was running.

Problem is, disk 0 lights up but will not spin the disc at all. Based on what I’ve seen of similar IBM XT-era full-height floppy drives, there’s probably a drive belt on the bottom of the drive that has failed. I haven’t seen in anyone’s documentation if it’s some sort of direct drive or if it indeed has a belt, and I haven’t yet taken it apart again to see. Fortunately there’s a nice Youtube video on how to clean and lubricate TRS-80 drives which should be pretty handy in getting it going.

80286 and ST-225 hard drive

Broken split drive belt

[photos: flickr – ST-225 repair]

This was our first IBM clone at home from the mid-1990s. It boots, but the 20 megabyte (!) Seagate ST-225 MFM hard drive isn’t readable. It’s motors are quite loud so I’m pretty sure they’re spinning up and the stepper motor moves around a bit. I tried giving the stepper motor a drop of oil, but that didn’t help. I don’t really expect any data of interest on the disk, maybe some old Lotus 1-2-3 spreadsheets or Windows 3.0, but now I want to try to save it. As a last ditch effort I opened the drive up to see if there was anything obviously wrong like scratched platters before I gave up or decide to send it to a data recovery place.

When I opened it up, I immediately noticed a band that connects the stepper motor to the read/write arm had snapped off its end that secured it to a tiny peg. Without it, it couldn’t seek the heads across the platter. I had also noticed some light chips on the outside of the platter. I suspect this all happened from it’s 1,700 mile road trip and had I turned it on before I left it might have actually worked.

This set me out on a crusade to replace that little piece of metal band. Googling around I discovered it’s called a “split-band drive belt”. It’s metallic because it’ll stick to a magnet I have, yet it’s very flexible. At first I thought it might be some sort of mylar, but given it’s job to constantly wrap around the shaft of the stepper motor and pull the read/write arm around, I can’t imagine it to have much, if any, stretch.

I thought a strip of aluminum can might work, but after cutting it down to roughly the right size it’s clear it isn’t very flexible and wants to stay bent. I don’t have much confidence that would work, but who knows.

I even showed the problem to a semiconductor machinist friend if he had any idea what kind of material it was or what I could use to replace it with. He drew a blank, but suggested maybe the coil out of a thermostat or thermometer could work. The other alternative would be to buy another ST-225 drive off eBay, disassemble it, and steal the part from it. Those are going for over $50 + shipping now, whereas a thermometer at Wal-Mart was $11.

I discovered the temperature coil spring or whatever you call it is actually quite stiff, more so than aluminum. It was however almost exactly the same width as the broken band, so that was neat. The thermometer actually had a secondary needle, for relative humidity. This smaller coil appeared to be some sort of brass, it was the right width and appeared to be very flexible! Winner!

Original, aluminum, bi-metal, and brass

Because it was a smaller coil, maybe 3″ when flattened, I had only enough material to make one replacement band with it. The holes in each end of the original were 1/16th – 5/64 of an inch. I rigged up a jig to hold the material, drilled one hole before the belt in my cheap drill press snapped. And the spare too. I drilled the last hole by hand because I was in an excited rush, which was a bad idea because it was not a smooth hole.

Of course another problem with this approach is that the thermometer metal is designed to expand and contract, so it remains to be seen if this is a problem with such a short piece that could cause it to not align with magnetic tracks.

The original and newly made band

When I went to put on the replacement, I tugged just a bit too much trying to get the band over the last post and the ragged hole ripped out. Crap. So now I have to find replacement belts for my China Number One jeweler drill press and another thermometer to sacrifice.

I catch myself browsing for ST-225 drives on eBay, I may say screw it and get one for parts if it too hasn’t already suffered a similar fate.

Drill press

I bought the worlds smallest drill press to use for my various projects in my apartment. It’s great when I need it, very quiet, I can set it on the kitchen counter, poke some holes, and stick it back on the shelf when I’m done. It’s drive belt is a glorified red plastic rubber band.

In the process of drilling holes for the hard drive repair, it snapped, followed by the spare belt. It took me a while of Googling to find replacements for it because my unit’s part number wasn’t turning up anything, but I eventually found a jeweler supply shop that had some that fit. So that is at least one problem I’ve completely solved.

QIC-80 tapes

[photos: flickr – QIC-80 repair]

By some small miracle we found three of my old QIC-80 backup tapes while cleaning out the house. These are very important to me because they have backups of my old BBS I was running in 1995 and would very much like to get that data back.

This was a whole side yak shaving quest because Colorado QIC-80 tape drives like I had piggyback off a floppy disk controller. There were some external units that connected to a parallel port that the Internets said didn’t work too well, and wanting to have the greatest chance of success, bought an internal drive like I had before. The problem was I didn’t have a single thing anymore with a floppy disk controller, not even the various mini-ITX boards I had. I don’t think I ever even found a PCIe floppy controller either, just an ocean of IDE interfaces. So naturally I eBay’d a 486 system. Ironically the 486 I wound up getting just happened to have a Colorado tape drive in it, so now I have two!

Being the excited person I was, after getting Windows 95 and the Colorado Backup software installed, I stuck in a “new” QIC-80 tape I also eBay’d as a test. Formatting, re-tensioning, a backup and restore all went well. I popped in one of my BBS tapes and it instantly jammed. The tape was all kinked up inside the cartridge and somehow looped between the top of a spool and the top shell.

QIC-80 cartridges have an internal tension belt that wraps around the spools holding the magnetic tape, linking them together. Reading some blog posts it was mentioned that on older tapes these belts give out and cause the tape to slip or cause one spool not to pick up tape fed off the other spool, causing it to wad up. Fortunately it’s all contained within the tape cartridge, unlike a VCR or cassette tape player where it pulls the tape out and gets stuck in the reading mechanism. (Although in this instance, this wouldn’t even be a problem if the QIC tapes didn’t depend on a tension belt!)

Again, being a do it yourselfer, how hard could it be? I have Youtube! I cracked open one of my BBS tapes and one of the “new” tapes to act as a donor. When I took the old tension belt off it just laid there like a limp noodle, it was clearly shot. The belt off the donor tape was clearly tighter, as soon as I had it off the spools it shriveled up, turned inside out into a wrinkly mess and I was worried that I could never get it back on.

I wound up moving the spools containing the tape from the old cartridge into the donor cartridge, that way all of the spindles and belt were known to be in good working order. I noticed the crinkled part of the tape was at the beginning, before an index hole so I thought maybe the data survived unscathed. Re-tensioning the belt turned out to be quite the pain in the ass! I found out the belt has quite a bit of elasticity to it, I was afraid I was going to snap the sucker getting it on all the spools again but time and time again I could pull on it just enough to get it in place. I was using the end of a wooden swab to pull things around as to not damage it more.

After getting the cartridge back together, I inserted it into the drive and with a loud screech it instantly jammed again. I took the cartridge apart again, wound things back the way they should and did the tension fiddling again. It’s quite annoying if you have slack on one spool because you have to find something to fully hold the tension belt away enough to spin the spool to take up the slack, else it’ll never work out. A couple of times I had a twist in the tension belt to work out. I also learned the belt has to be aligned exactly in the middle of the tape, if it’s high or low then it tends to push the tape up or down as it feeds off the spool, leading to it coming off. Also all of the spools and spindles are held in place by the top part of the plastic case, so without it as you wind the spools everything wants to lift upward.

I found the best way seems to be to first pull the belt over the black spool that interacts with the capstan, feed it through between the two tape spools, pull it taught to make sure its relatively untwisted, bring it down across the left spool of tape, loop it over the left white spool, while holding tension to keep it flat and in place. Then pull on it with some force using a end of a wooden swab to get it temporarily over the top of the post of the right spool. Assess if there’s any twists, any slack in the tape media, and make sure it’s aligned. Finally, get a good grip on everything, use the same wooden stick to pull tightly (yet gently) to stretch out the belt enough to get it all the way over the right hand white spool. Check things over again for any twists, slack, and alignment.

On second attempt it fed a little while and then jammed again. I don’t know what the RPM of the tape drive is, but when it goes south it goes in a hurry. It’s maddening because I can spin the tape by hand a couple dozen revolutions when the cover is off and it all works perfectly. I repeated this tape disassembly and assembly two more times, all times resulting in a jam. The tape was getting pretty wrinkled now from all the wadding up, I can’t imagine there any good data on that section. To add insult to injury, I dropped the cartridge at one point and the spools went flying to the floor. Several feet of tape unspooled, instantly catching the attention of one of my cats. Completely frustrated, I gave up at that point. I got all the tape wound back up, the tension belt on once again, and quit. I’m going to outsource this data recovery job to somebody else. Just maaaybe having it wound up and under tension again may help smooth some of the wrinkles for the next person.

I give up

Starting in high school English I and onward we had to spend like 10 minutes every class period writing in a journal that we had to hand in at the end of the nine week period for grading. Presumably this was to get us to write more and use our words, and we’d be graded on the number of pages. I absolutely hated this exercise because who does this teacher think she is, reading our private thoughts? It got more than one classmate in trouble when the subject of said writing was about teachers.

A while back I found these journals buried in a box in Oklahoma and dug them out to go through them again. I apparently hated this work so much I wrote about it daily, my hate of school, how bored I was, and other teenage angsty things.

However it did capture the time period when I started using bulletin boards, Linux, and the beginnings of my ISP (which I started in between my junior and senior years of high school). I was also raising sheep in a pasture in our backyard, so there were lots of entries about taking care of them, building fence, and other sheepy-things.

The earliest entry was the fall 1993 semester so I had just started using a modem that summer. Oddly enough I barely mentioned my PC work or upgrades which I was deeply into at the time, if I had then that may have helped bump up the page count.

Jan 12 1994: Got a copy of the Wildcat! BBS 2.6 Test Drive and installed it for the first time

Mar 17 1994: Got my own phone line, a root canal, and about to turn 15

Dec 5 1994: $70 phone bill!  Notes show that long distance calling after 6 PM was costing me 17 cents a minute.

Dec 20 1994: $30 CRIS/Concentric Research Net/BBS Direct bill, which provided 1-800 access to larger bulletin boards and SLIP/PPP Internet access

Jan 11 1995: Got a MSI hat, a shareware collection CD, and a card from Rick Heming at Mustang Software thanking me for answering a bunch of tech support questions about Wildcat! on the message boards over the holidays on the Mustang Software BBS

Jan 24 1995: Installed Slackware Linux for the first time on my 386SX

Jan 31 1995: Got Wildcat! BBS Multiline 10 up and working

Feb 6 1995: Went to a OneNet “convention” in Red Rock, OK to hear about the development of a statewide Internet for schools, first time I encountered a Sun SPARCstation. (Red Rock was a very wealthy school district that had state of the art everything including a 56k or T1 line)

Feb 9 1995: First time installing X-windows on my Linux box

Feb 14 1995: ran up 967 minutes of long distance calls, $80 CRIS bill

Feb 15 1995: Got my hands on some old VT-100 terminals from a surplus donation at school, realized how dumb terminals worked

Mar 3 1995: First time I’ve written HTML and uploaded a public web page to the CRIS web server

Aug 25 1995: Installed Windows 95 for the first time

Sep 1 1995: CRIS bill was $156

Dec 6 1995: Learned to configure named to set up a caching nameserver

Jan 3 1996: Compiled SOCKS 4.22 proxy server and kernel for first time (I forget what I was doing with this, maybe I had put my Linux box on my PPP connection and proxied my Windows 95 apps through it)

Sometime around this time period I had sold a 10Base2 network and some new computers to the school. They had a free PPP connection and I had rigged up things to share it along with DOS files on the LAN

Jan 8 1996: Got an IPX/ODI packet driver stack working on MS-DOS so TCP/IP and LANsmart could co-exist

Jan 23 1996: Get an account with a new ISP in the area, discover the wide area calling plan our telco offered did not cover it. This disappointed me so much and planted the ISP seed

Jan 29 1996: At school, built an ARCnet to Ethernet router with KA9Q NOS so older computers on old LAN can use Internet

Feb 29 1996: Declare my plans to start an ISP

March 5 1996: Went to a school board meeting to try to get them to get a leased line connection, tried to give an Internet demo

Apr 1 1996: Talking to Galaxy Star Systems (an ISP in Tulsa) about quotes on a 384k frame relay circuit

Apr 1996: Passed my Netware 3 Certified Novell Associate (CNA) cert; started searching for office space for the ISP, getting equipment quotes

I started my ISP in the summer of 1996, I want to say the 56k frame relay connection went up on June 18, 1996.

The 1996-1997 journal covers the growing pains and trying to break even before running out of runway.

Jan 6, 1997: Got an extra loan to build own office building to escape high rent

Jan 8, 1997: Got first Cisco 2501 router, learned how to configure it to be primary ISP router. I needed this because I was bringing in a second frame relay T1 to open my first POPs in other towns.

Jan 29, 1997: Southwestern Bell techs say T1s have been turned up at new building, moved all of our gear over, only to find out the other end of the T1 hasn’t been connected to the frame relay switch. Moved everything back that night, a several hour outage.

Feb 4 1997: Hit 160 monthly dialup customers, almost breaking even!

Mar 20, 1997: Turned 18, wrote hot checks to cover Southwestern Bell bills!

Mar 31, 1997: Finally turned a monthly profit of $100 for the first time. Went directly to Southwestern Bell to read their tariffs to see how to set up call forwarding to expand access, options on T1 contracts. Setting up largest PoP (2 or 3rd?). This is the start of me being a real thorn in the side of telcos.

Not a whole lot of journal entries after this. We were on a new very-not-good English teacher and guess she was pretty lax.

To this day I still have dreams where I’m an adult and am still in English class, like it never ended. Then I wake up and thank god I really passed and finished years ago.

This is going to read like one of those annoying-ass online recipes because I’m gonna write down some of my experiences with the TRS-80 before I start getting into what I’ve been doing to get it going again. However it is Father’s Day and this was my Dad’s computer so it’s kind of fitting.

After sitting in storage for 25+ years I finally brought home our old TRS-80 model 4 from Oklahoma. It probably hadn’t been turned on since the late 1990s. I didn’t want it for the longest time because the IBM PCs I built would run circles around it, I really didn’t have room for it, and plus I always flew back home so shipping it would be a pain. After I drove out to start cleaning out dad’s house, I brought it back to California to see if I could get it up and working again because retro/vintage computers are “in” now. Fortunately it was never abused in storage or sat outside, so I thought it had a decent chance at survival.

TRS-80 model 4

Way back in 1983 our first home computer was a TRS-80 model 3. I recently found the sales receipt for it, around $2500 for the computer, printer, some cables, around $7500 in 2023 dollars! At the time my family owned the natural gas company that provided gas to our small rural community with 100 or so customers. Dad wanted to use the computer to generate + print monthly bills for customers instead of doing it by hand. He was primarily a school teacher teaching business and not really a programmer, so he wound up taking night classes at college to write in BASIC, and gradually learning what he needed to do. I also recently found printouts of his class assignments, starting off with 10 line equivalents to Hello World and then adding in calculations and various outputs.

A printing company custom printed the tractor-fed postcards that were mailed out to customers. I remember the printing company had a cool ruler that had stencils for the tractor feed holes and line/character spacing of a dot matrix printer. You drew up exactly how you wanted the card to look, then made all the LPRINT statements in your code line up with the fields on the printout. (This came in handy later when I wrote a basic 1040-EZ tax program that would fill out a 1040-EZ form for you!)

I wasn’t really interested in the billing program, it was hours of work to run the bills each month. There wasn’t any sort of data store, arrears, last month and the current month’s meter readings had to be input for every customer. Further, all of the customer names and addresses were hardcoded into the program! The program didn’t do much input sanity, and if you made a mistake it threw things off and was a hassle to fix. So after messing up bills, that relieved me of the responsibility of running it.

However we did dabble in other software. We had Visicalc, Superscripsit, and a few dozen games. Some were on cassette, others were on double-sided 5.25″ floppy disks. Superscriptsit came with a tutorial on audio cassette, which taught me about proportional space fonts and some other word processor-y things. BASIC programs, I have no idea where they came from, probably hand made bootlegs that got passed around. There was a long piece of paper that told us kids which disk, which side, had a given game on it like CHICKEN.BAS and that was enough for us to LOAD it.

I was around 4-6 at the time and Dad showed me what he knew of BASIC. Things like everything had line numbers, do a PRINT to print it to the screen, LPRINT to make it go to the printer, IF andGOTO statements and what not. It wasn’t long before I was making piddly little programs. Back then a lot of BASIC programs were distributed in magazines and books. The full source code would be printed on a single page or if you were unlucky, multiple pages. It was up to you to type in the whole damn thing, perfectly, else it didn’t work, you got the dreaded Syntax Error? or it had bugs. I was also not a touch typist so it took me forever to type one of these things in one keyboard peck at a time. Fortunately once it was in and working, you could save it to a floppy and never have to type it in again. The upshot was that typing it in line by line you got to understand what the program was doing as it went along, picking up syntax and style.

As an aside, PC/Computing magazine also distributed little MS-DOS utilities the same way in the 90s, several paragraphs of hex values you that plugged into DEBUG.COM!

I don’t remember why, if it was related to the needs of the software or what, but we sold the Model 3 at some point. I was pretty upset at this, because now where would I play games? It wasn’t too long after that we got the Model 4. I seem to recall some stuff we had that ran on the Model 3 wouldn’t run on the Model 4. There was some sort of TRS-DOS conversion program that did something, I wish I could find it or at least the manual to understand what it did.

It feels like I was always on it every day. I was either playing games or fiddling around with other games. Sometimes I’d just be in a TRS-DOS shell goofing around and manage to get TRS-DOS to crash (probably buffer overrun of the command interpreter) in a spectacular fashion with all sorts of garbage spit out on the screen. I spent the majority of my time in BASIC, it wasn’t until way later that I realized TRS-DOS actually could do things other than load BASIC.

At school we started off with Commodore CBMs and then went on to Apple IIe in the computer room in the late 80s/early 90s. The Apples were much nicer to use than the TRS-80 and I got to fiddling with them instead. In the early 90s our school switched to IBM PC clones (8088s w/ monochrome screens, 512 K memory, MS-DOS 3.3), and shortly after we got a 286 at home. I pretty much abandoned the TRS-80 and spent all my time on the PC, altho Dad still used the TRS for billing until they sold off the gas company. After that, it just sat there on the desk under a dust cover.

Ironically years later when I started the ISP I too would wind up having to write my own billing system. Commercial ISP billing systems just didn’t exist. What started off as a bunch of copy-pasting of customer addresses in Excel quickly got out of hand. I had to go learn Perl, MySQL, PHP, and how to parse RADIUS logs or else not make any money!