Wednesday, July 4, 2018

Continued work restoring the ASR 33 printer unit


After clearing some sticky residue on the code bars and selector mechanism, I discovered that a spring is missing for bit 4 from the selector, leading to that bit always registering as SPACE or 0. I look throughout the entire printer hoping it was stuck somewhere. Amazingly, I found it jammed behind another spring.

Seems simple, just reattach the spring. The two endpoints are deep in the machine and the loop at the end of the spring is teeny. I first had to detach the carrier return dashpot (shock absorber) just to get better access. Essentially, I need to be in two places at once - pulling from the bottom with one spring hook and pulling from the top with another spring hook - just to get the two ends attached.

I spent a very frustrating 90 minutes trying to set the machine on its side at a stable angle, shine light and get the tools in from both top and bottom simultaneously. Nothing worked, other than dropping the spring many times and having to conduct tedious searches each time.

If I could get my fingers down into the mechanism to make the first attachment, it would seemingly be easier to use a springhook to make the second connection. I removed the connectors to the selector electromagnet, but still couldn't reach.

Area where I have to work to reattach the spring
Another hour of failed attempts went by. I really need a pusher type of spring hook, one that fits into the tiny loop at the end of the spring and pushes it away from me and hopefully wrestles it over the tight fit of the tab on the lever arm. I ordered some but won't get them until the weekend.

Ordinary springhook, good for pulling springs
I did come across a tool in the garage, among my selectric typewriter objects, which can be used to grip the far end of a spring and maneuver it over a post. Since this is the first step I want to take, I brought it in and prepared for another stab at attaching the spring.

New tool, with retractable gripping hook
The hook pushed out - it will retract when pressure is removed
I had to use forceps to hold the two electromagnet wires out of the way. I used the new tool and managed to hook one end of the spring over the top post. I couldn't remove the tool or the spring would fall off, so it was important to now push a springhook up from the bottom, grab the other end and stretch the spring out.

Top attachment point is circled
Even this seemingly minor operation took 15 minutes to accomplish, since the spring is angled outward by the new tool making it hard to get the springhook into the other end's loop. Finally, I had it and could release the new tool while holding the spring extended.

The base on which the printer unit rested was slowly rotated while I maintained spring tension, until I could see and begin working on the bottom attachment point. The springhook itself has a relatively large diameter, mostly filling the loop on the spring end and making it impossible to attach. I therefore used a smaller springhook and transferred the spring over from the larger tool.
Bottom attachment point is circled
I then maneuvered the end over the lower attachment point! All that was needed now was to carefully detach the springhook and the job was done. Alas, somehow the top attachment came loose as I gently wiggled the springhook on the bottom, causing the spring to 'spring' somewhere and I was back to zero.
After a half hour looking in vain for the spring, which I think is still in the printer mechanism somewhere, I had to put it aside. The spring could also be somewhere in the kitchen nearby, depending on its 'spring' action when it let loose. More wasted time ahead.

I hooked up a power supply to the selector magnet, allowing me to drive the MARK condition as long as I want. The magnet takes 500ma, but I could get it to reliably operate with the current limited to about 300ma. The power switch allows me to control SPACE or MARK as I wish.

The next idea I had was to count the fan vanes on the motor during one entire selector cycle. The protocol transmits 11 bits during that cycle, allowing me to predict how many vanes take me to each of the data bit positions. I can switch the selector magnet off and on at the appropriate times and produce the code bar results I want. 


  1. Sometimes - not always - you can tie a tail of silk thread to those damn little springs so that they can't fly away quite so far when they launch themselves. Happy Fo(u)rth!

    1. Excellent idea - wished I had thought of it.

      Once I find it again, I will do that and anchor the other end of the thread to part of the machinery.