#dot matrix printer

Behold the Delta-10, a machine so fast it can print the alphabet six times in a single second, assuming, of course, you actually want six copies of the alphabet in a font that looks like it was stitched together by a caffeinated woodpecker. In 1984, this was the peak of "red-hot efficiency," which is a polite way of saying it sounded like a robotic chainsaw massacre was happening in your home office.

The ad treats the printer like a suspect in a noir film, placing it under a harsh desk lamp as if trying to get it to confess where it hid the missing perforated edges of the paper. And let's talk about that 8K buffer. It’s adorable how we used to brag about memory capacities that couldn't even hold a modern low-res emoji today. But hey, it was "fully compatible" with the IBM PC, which meant you were officially part of the beige-box elite.

No updates. No patches. No microtransactions.

Only fun.

Games that taught us timing, punished mistakes and rewarded skill.

Retro Dream

signal-scroll // dot-matrix caption experiment

Today I ran the full signal // NOISE caption-scroll through the Oki ML3320 — eight pages of continuous fanfold, about 10" × 11", which happens to sit perfectly inside a 12" sleeve.

The idea was simple: take the “clean” text document (the long caption essay for the album), send it via the Send to Ezra script, and let the dot-matrix chew it out as a physical scroll to wrap the cassette and drop loose into the LP package.

The reality: the Oki is running on ancient drivers via CUPS and clearly has its own ideas about encoding. Certain characters don’t exist, lines overprint, glyphs get swapped, and entire sections smear into dense bands of nonsense. It’s still recognisably the text I wrote, but half of it is misaligned, misrendered, or partially redacted by accident.

I was going to fix it — reformat the source, strip characters, force it to behave. Then it clicked: this is exactly what the record is about.

APD in hardware:

signal goes in clean

the system “hears” it through a compromised path

output is there, but unstable: hard to parse, full of dropouts and substitution errors

the only truly legible version lives elsewhere, if you know where to look

So I’m keeping it as-is.

Each scroll will ship with a handwritten note across the top margin:

SIGNAL-SCROLL: printed via misaligned dot-matrix driver. Output left intentionally corrupted to mirror how signal arrives here. – DSP

I’ll hand-write that line on every copy, slightly differently each time. Human caption layered over machine mishearing.

For the LP, the scroll will sit as a loose insert. For the cassette, it’ll wrap around the case like a continuous, damaged subtitle track.

Anyone who gets the physical edition will also have access to a clean, readable version of the scroll online— the delayed caption, arriving after the broken “live” output. This also includes a download link to the five digital appendix tracks.

researcher!Ford using dot matrix printer... save me...

save me researcher!Ford using dot matrix printer...

Not that one specifically, that one is beautiful and in great condition, but one like it — Apple ImageWriter II. Specifically mine is the ImageWriter II/L variant, the last revision of the ImageWriter II line, but it looks like this one.

Or at least it did once upon a time.

My family acquired this printer second-hand in the late 90s along with a Mac Classic. It got used regularly for school reports and letters and business documents and tax forms for a few years until we finally were able to get a new computer with a color inkjet printer.

Long story short, like the computer that went with it, ultimately this poor printer ended up sitting in storage without air conditioning in East Texas heat and humidity for nearly twenty years. It's a sad story of slow decay.

My ImageWriter is now yellowed and scuffed and scraped and rusted and missing a piece or two; just a dim reminder of its former beauty. Given the state of it, what hope do we have of ever again hearing it sing the song of its people?

Well, I'm not going to let it go without a fight. Time to dig in and see what we can ...

... oh. Oh dear that won't do at all.

A good rule to follow when working with these 30+ year old systems, is to (carefully!) open and inspect before applying any power. In this case I'm very glad I did. Three large filter capacitors on the power supply have very obviously swollen and burst, spreading their corrosive bile all over the neighborhood.

The back side of the power supply circuit board was a wasteland of rotting solder mask, corroded traces, and displaced silkscreen. The electrolyte has eaten its way down the leads, through the solder, and left carnage in its wake all across the bottom of the board.

First order of business is getting those old capacitors removed from the board so cleanup can begin. If you've never worked with hardware of this vintage, a fair warning — make sure you're working in a well-ventilated area. Sure the solder has lead and the flux ain't great for the lungs, but the big concern here is the unholy stench of heated capacitor electrolyte hitting the nostrils like the revenge of Poseidon's refuse bin. The local fish market has nothing on these things.

The old solder, especially when mixed with the electrolyte, tends to behave in a very un-solder-like fashion. It will refuse to melt and when it does it will slump around like wet sand rather than flow like liquid metal should. While it may seem counter-intuitive, the best way to get rid of it is to add more fresh solder to it. On these single-sided boards with large components like this, a spring-action solder pump works well for getting the old parts removed, and then some solder braid will clean up the pads well.

Once the old parts are out, I like to thoroughly clean the area with isopropyl alcohol to remove the electrolyte and years of grease and dirt and pet hair that may have cemented itself to the board. In this case I also needed to use a mild abrasive to remove that damaged solder mask where it had bubbled up off the corroding copper traces. I was lucky here that none of the traces were actually broken or corroded through completely. Clear nail polish works well for protecting the now bare copper (just make sure it's not the UV-cure gel stuff).

From here I turned my attention to the case, because the power supply is the last item removed and first installed when conducting a complete tear down of this printer, and it didn't make sense to put my newly cleaned power supply into a dirty old case.

I'm not really a fan of retrobrite, and these large case pieces would be a real challenge anyway. So all I want is to clean up the dirt and rust and as many scuff marks as I can. As far as I'm concerned, the rest is just part of the history of the item. Each mark tells a story of how this item was used, not just put up on a shelf to be looked at. And if I didn't have any interest in using the machine until it completely falls to pieces, I wouldn't be bothering with going fishing replacing old capacitors.

This is a good point to do some testing. There may still be more wrong with that power supply. Output voltages could have drifted out of spec from other components aging, or maybe I installed capacitors that don't quite match the originals. The ImageWriter II/L power supply has three outputs — +5VDC, -5VDC, & +26VDC. With no load on the power supply, I measured the outputs at around ±7V and 30V. That seems high, but it's not outside of what I would expect for a power supply that's not actually driving anything. This would be a good point to use an adjustable test load, but since I don't have one of those, I'll just have to move forward with my "well it seems fine'

Spoiler: it was not fine.

As part of its startup sequence, the ImageWriter exercises all of its stepper motors to get everything to a known state. This high current draw immediately after power on was more than its old power supply could give. There's clearly more than bad capacitors on the supply, but identifying what exactly is still beyond my current skill level.

So in the interest of getting the machine working (because I have plans for it), I opted to try replacing the power supply with something more modern. The catch here is the odd assortment of voltages the original supply provided. It's easy to find a ±5VDC supply, but 26V is virtually unheard-of.

Apple's documentation for the printer mentions the +26V supply is for driving the motors. I suspected that the 26V supply was less carefully regulated and probably targeting something more like 24V. Sure enough, the stepper with the highest voltage rating on its label was 24V. With a little extra current capacity available, I figured the printer would function just fine with a 24V supply.

The catch is, 24V & ±5V is not a common configuration. There are plenty of 12V & ±5V supplies, but that won't do here. I settled on a Mean Well 24V & 5V supply with a -5V inverter ... And promptly ordered the wrong part. I had a nice new 12V & 5V supply. That's ok, once I got it in hand it was a bit too large to fit in the space I had anyway.

So I got a different Mean Well 24V supply and a separate 24V-5V DC-DC converter. It's a bit of a mess all crammed into the bottom of the case, but it should give all the right voltages (or near enough).

I did have to remove the power switch from the old supply though. That particular part has long since been discontinued, and compatible replacements proved difficult to find.

Now that it's all assembled, it's time to test. This is the part that always makes me nervous, especially when dealing with mains voltages. There's so much that can go so very horribly wrong.

I started out with a smoke test — switching on power briefly to make sure there were no direct shorts that might cause an explosive failure. No smoke is a good sign, so check the voltages. With no load, the new supply rails read 23.99V, 5.00V, and -5.55V. That's about as good as I could ever ask for. So now there's only one thing left to test … does it actually print?

Success!