At VCFMW 13, I saw some blue metal box at the auction that was described as a paper tape reader and punch.
Long story short, I won the auction, and walked away with this odd looking blue box with zero documentation.
Since then, I've searched for documentation, reference material, or any sort of useful information on it
to no avail. Thing is, I know atleast three others are out there.
Mine has a Decitek branded reader mechanism, and a Remex branded punch mechanism.
The brand of the actual box is unknown, and I've found myself referring to it as "The Decitek",
but that's probably not an accurate name.
One in particular has been the subject of reverse engineering by a pair of experienced EE hobbyists.
Download the reverse engineered schematics on
The unit is mainly composed of 7400 series logic, and includes no microprocessor or UART.
The external interface is a DA-15 connector with only two pins wired: ground and signal.
What signal does it speak? Why, half-duplex/simplex 20mA current loop.
The only switch anywhere on the device is a single pushbutton to advance the tape, which only punches
sprocket holes on the tape as it advances.
This means that there are no obvious ways to switch between reading and punching modes.
My efforts have been focused on figuring out whatever protocol it speaks.
Using a basic current loop interface, I've been able to adapt from TTL and RS232 serial levels
Thus far, I've had my best luck at about 1200 baud 8N1, but I'm not convinced that's right.
I've tried all sorts of baud rates, protocol variations, and punching a wide variety of data to it.
Modern Windows machines, 80's MS-DOS machines, the Cactus, and even an arduino have been used to
generate the serial signals that feed the machine, but at no point has the tape output been congruent
with the data input. I haven't a clue how to force the machine into a read mode, but I'm betting
that knowing whatever punch speed it expects will help in figuring out whatever commands it accepts.
That is, assuming it accepts commands.
Unfortunately, I created a short-to-ground during the process of my tests, releasing a bit of
magic smoke, and preventing the unit from punching or advnacing tape.
The specific components damaged are still being investigated, but a trace was torched, and a TIP-125
transistor part of the drive circuitry is suspected to be fried.
I've also identified a failed 74LS197 (U13) that may or may not be related to thep previous failure.
Here's a view of the interface board with a colorful overlay of the traces from chip to chip.
No less than dozen people have assisted me in some what during the course of my attempts to get this thing
going again. For that, I am grateful.
If for some reason you have information on this device, please feel free to contact me because I'm
certainly eager to learn more about this blue punch/reader.
June 2022 Update!
I took another crack at figuring out this beast after showing it to my friend QuantX.
He started analyzing the asynchronous logic on the reverse engineered schematics, and was able to discerne
what signal to check to verify the baud rate which is 31250! Odd baud rate, ain't it?
The data is formatted in groups of 13 bits:
1 1 0 [8 data bits] [reader active bit] [punch active bit]
Working through our debug process, we got it to punch data bits, but only data bits -- no sprocket holes.
It also would not advance the tape, so debug focus was on the drive transitors.
Then it dawned on us "what if the solenoids themselves are fried?" which is a scary thought.
I have a spare punch mechanism from Facit, identical design, and I used it to get a baseline of
continuity and expected resistance.
I compared against the Remex installed in the unit, and noticed that 4 signal lines had no continuity.
Turns out that the little green axial components in series with each coil are 2Amp protection fuses!
The sprocket hole punch fuse blew, along with all three of the drive solenoid fuses.
Confident in the circuitry being repaired on the board, I swapped in the spare Facit mechanism, and ran
a test of the unit.
We gave it instructions to punch 0xAA, and it punched that!
There's more work to be done, but we're finally making positive progress.