NOTES 20210815

MagCard (MC/ST) Observations

The card reader was full of foam but has been de-foamed, like the other equipment. It's in better condition than the MC/ST II. Its Selectric also appears to be mechanically complete, but severely gunked up, full of IBM foam, and frozen. After working the power switch back and forth and reseating the connectors, it powered on. It was stuck in a tab cycle, but mineral spirits cleared out the gunk and got it to idle normally.

The card reader has not been powered on yet, but appears mechanically intact except for two small PCBs on top which each have a piece snapped off in different locations (see photos near end of this page as of Oct 30th 2022). A service history log with a mag card tucked inside was found inside the lid, it's sticky due to all of the rotten foam. Currently we believe this unit will be easier to restore.

We were able to actuate one of the solenoids beneath the Selectric using an external power supply set to something around 24V (with successful actuation as low as 14V).

Shifting was demonstrated to work on the MC/ST's Selectric unit. Also, by manually actuating the character selection solenoids, we were able to get the Selectric to print arbitrary (albeit incorrect) characters. The typeball is too gunked to return properly, and the typeball is misaligned, but otherwise it was able to put carbon on paper. We recorded several slow-motion shots of it actuating.

The following photos depict labels and components of interest in the MC/ST as found on this initial visit.

MC/ST II Observations

The card feed/eject wheels inside the reader are rotted, but only the outside rubber. The remaining metal cores, while corroded, should be salvageable with some brushing and wrapped with rubber bands or heatshrink or something to replace the rubber. However its Selectric unit is missing an entire clutch assembly and possibly other components, making it impossible to safely operate at this time.

To open the MC/ST II's card readers cabinet, the headshell must first be removed. The process for doing this is everything but obvious, therefore is documented here.

Do this by sticking a thin flat screwdriver into the bottom of the middle vent of the head unit and push the hook inside out of the way, and lift the rear of the head shell.

Then place the cabinet on its left side and *LOOSEN* the two screw on the right side of the bottom panel.

These retract two clamps from the farthest vent holes along the bottom.

Then the sidepanel can be removed, it will need some wiggling, easiest to pull up from both top and bottom.

NOTES 20210821

We liberally spirited the MC/ST I's upper keylever arms and their tiny clips. Most of the keys binding on press was due to the clips being glued in place by old grease. These move freely now and most of the keys move freely now too, except a couple that are just a little bit sluggish, but this is a minor problem and they are good enough for now.

Identified the tab solenoid, and successfully demonstrated that it doesn't work. It appears to be sticking on the escapement. Cleared that.

Found that the third mystery tape has fallen off the right-side pulley. It appears that the little plastic clip that goes over it has caught under the tape and threw the tape off. Tape survived with minor deforming along its length, luckily it was not destroyed. Kept the little plastic piece off for now to eliminate the possibility of that happening again, because it immediately happened again. May keep it off permanently since it's an extraneous part whose problems seem to outweigh its value.

Cleaned up the ribbon mechanism because the release lever was glued in place due to grime.

Toured someone around who was interested in keyboards, then got lunch at Walt's, then came back to put the carrier back together and find out whether the A.B. Dick Magna II unit was compatible with the MC/ST I.

Turns out the plugs are completely different, it uses it own auxiliary power cable, and it is far heavier. At this time it is being disregarded.

The carrier still has a little bit of difficulty moving and the typeball is still misaligned, but we can now reliably print garbage text and advance the carrier.

At this time we believe every solenoid is supposed to engage properly when given 24V according to the Adjustment Parts Manual (see Ephemera folder). They were found to usually engage with as little as 14V. The carrier return solenoid has trouble engaging even at 24V though, we believe this is due to grease buildup elsewhere.

NOTES 20211114

Ordered a replacement drive belt for the MC/ST (and for my own IBM MT50).

NOTES 20211204

Developed a belt repair procedure for the MC/ST and performed it, available HERE (this page is not yet made).

During this procedure, a long time was spent trying to get the shaft to go back in only for it to utterly refuse to do so. Gave up for the day.

EDITOR'S NOTE 20230909: this was the beginning of cycle clutch hell.

REPAIR LOG 20220304

Returned to the warehouse to continue working on getting the MC/ST's cycle shaft back in after replacing the broken drive belt. After a lot of fiddling and cleaning other parts that needed it, eventually discovered that the cycle clutch's spring is deformed and bulging out to the side, which is what's stopping the cycle shaft from going back in. The errant spring coil is sticking out between the spring sleeve and the plastic part on the shaft that interfaces with it. I suspect this was deformed by accident during our previous fiddlings with the machine.

A brief attempt at reforming the spring only made it worse, and in the interest of avoiding undue friction from uneven coils and this never being a problem again, a replacement shall be sourced from ebay as this part is not known to be available as a 3D printable model.

Also the gear pack was disassembled (as much as it can be, anyway), cleaned, and relubed with the new marine grease consisting of the same spec as the original IBM grease.

The entire cycle shaft was also removed and cleaned as a side effect of the deformed spring discovery. The plastic piece on the cycle clutch, which is screwed onto the cycle shaft, was found to have some age-related deformation and cracking, and so a replacement was also ordered (in the same kit as the spring, NOS cycle clutch set).

REPAIR LOG 20220310

Continued cleaning and lubricating the cycle shaft.

Made the unfortunate discovery that the new spring is a visually indistinguishable but critically different specification than the old, which means the entire surrounding assembly needs to be replaced with the new parts, not just the spring itself.

Old parts to be kept together for historical accuracy and maybe tucked inside the chassis - this clutch design appears to be uncommon. A 1962 Selectric 71 will be checked to see if this is an obsolete clutch design used by early Selectrics.

Used a 1/16th allen wrench to remove the two setscrews holding in the plastic (sacrificial?) gear at the right end of the cycle shaft.

Used a 3/32nd allen wrench to remove the screw holding the spring clamp onto the cycle shaft - this worked on both the original and the new clamps, which seem to use the same screw despite everything else being different.

EDITOR'S NOTE 20230909: these set-screws probably needed Bristol wrenches instead of Allen wrenches.

The new spring and clamp fit perfectly in place of the old one when the entire assembly is used - do not mix n' match parts between the old and new clamps, the old spring will not properly fit in to the new clamp, and the old clamp is too large to clamp over the thinner new spring.

Installed the new cycle clutch onto the shaft and performed final cleaning on it.

Freed two stuck rollers in the selection block that the screwdriver gets wedged into. Cleaned some other old oil and grease out too.

Couldn't complete cycle shaft reinstallation due to an interruption for emergency tech support, will have to wait until next visit for another chance; gearpack area of frame still needs cleaning and lubrication.

NOTES 20220408

Attempted to reinstall the cycle shaft but discovered the cycle clutch's metal cylinder design *also* changed when the spring assembly on the shaft itself changed. Need to get a replacement that's compatible with the new plastic spring clamp on the shaft, which has a thin collar that the original did not. Could steal one from another machine, but would prefer to get an actual replacement.

Brought over a tube of JB PlasticWeld and the tiny vise to clamp the PCB with the snapped contact limb back together. Need to get a piece of street bristle or some other stiff bracing material to also epoxy onto the back of that, because it will still be very weak upon reattachment, ever after soldering the severed trace back together. Could perhaps JB SteelStik attach some paperclip fragments side by side if a more ideal piece can't be found. There are probably better ways to do this so any actual attempt will wait until after considering alternatives.

Left the tiny vise and the PCB to set over the course of the week, over by the other parts, which were moved onto the big table instead of on the cabinet's roof.

NOTES 20220413?

Returned to see if that JB Weld attempt worked, it appears to be successful, the PCB is in one piece again but fragile and needs reinforcement. Retrieved the tiny vise for use in JB Welding my MT50's broken keys.

NOTES 20220618

Ordered another replacement clutch spring assembly but mistook the new style for the old style and it too does not fit. Gave up after months of trying to track down an old-style clutch on ebay and elected to cannibalize one of the unsalvageably ruined SIIs in the warehouse. Cannibalized CSII with SN ??? (the blue one with the DO NOT OPERATE warning) for its cycle clutch collar.

THE REASON WHY THE OLD-STYLE CLUTCH COLLAR CANNOT WORK WITH THE NEW-STYLE CLUTCH SPRING ASSEMBLY: The new-style clutch assembly is largely identical, except that the plastic lobe piece that the collar interfaces with has a lip running around the inner circle that the old style one does not. This lip prevents the old-style collar from sitting flush against the new-style plastic lobe piece.

Also used the CSII's cycle shaft as an adjustment reference for the MCST's, since the CSII's has not been tampered with. DO NOT OPERATE THIS CSII, IT MAY DAMAGE OR DESTROY ITSELF OR INJURE THE OPERATOR! Reassembled the CSII as completely as possible, minus the cycle clutch collar.

Installed the collar into the MC/ST, it fits properly, but now the arm that controls the clutch is missing the sleeve when it's supposed to engage it. The sleeve is an identical outside size as the old-style, so this is likely an adjustment problem with the arm. Holding off on installing the remainder of the cycle shaft's left-side components until the clutch is demonstrated to be working.

After some adjustments, learned completely how the cycle clutch works and got it installed.

HOW THE CLUTCH WORKS: When the clutch solenoid beneath it is activated, it pulls a wedge down from beneath the clutch locking arm. This allows a spring to pull the clutch locking arm away from the clutch, which allows the clutch sleeve to spin freely. When it is allowed to spin, the spring inside the sleeve wraps securely on the spindle, the friction of which allows the motion from the operational shaft to be transferred to the cycle shaft, thus driving the gearpack and the filter shaft. The solenoid is only active for a moment - when it shuts off, it comes back up and catches the locking arm, which is pushed back into position by the plastic lobe on the cycle clutch. It then gets trapped by the solenoid's wedge until someone presses a key that activates the solenoid again.

The solenoid was manually engaged using a screwdriver and after some adjustment, it appears to work properly, or at least close enough to properly for a test run once the cabinet is also ready for a test run.

NOTES 20220701:

TODAY's GOALS:
GOAL1: adjust the MC/ST's cycle clutch into proper alignment.
GOAL2: at least plan a repair for the remaining broken cabinet reed switch PCB.

Pulled out the MC/ST to adjust it, and discovered that based on comparisons against the CSII, it may already be aligned, because the CSII's clutch has not been tampered with and it's sitting in the exact same location as the MC/ST's after a powered cycle.

Manual cycling versus powered cycling seems to put the clutch and left-side mechanism in different resting positions, presumably because of the different amount of force/inertia the mechanism has between the two motivators.

Checked the MC/ST II since it **should* be a good reference, but it appears to be severely out of alignment and powering on the Selectric required powering on the cabinet, which I did without thinking for a brief second. Leaving the MC/ST I's clutch alone for now, moving onto the cabinet's broken PCB.

NOTE TO SELF: compute the pitch of the broken plug from the taken measurements to find a similar plug on digikey.

The repaired reed switch PCB is still holding together by itself, untested as to whether it will survive the forces of the wire being attached. Possibly the solder will help, but that's irresponsible and solder traces are not load bearing. Best idea so far is to epoxy paperclip rods to each face of the tab to brace it. Probably can't do that for the other one though, whose screw mounting hole is broken.

Turns out I probably can; formed a structure out of a paperclip to epoxy to the back of the PCB to replace the missing fragments of its mounting hole. Just need to score the surfaces and find out what the best epoxy for holding it on will be. Also made a brace for the PCB with the broken tab; need to score and epoxy that too. Neither fix has been attempted yet, pending further brainstorming for better ideas.

NOTES 20220715ish

Realized that it would be much better to model a plastic brace in CADD and 3D print it, allowing the JB Weld to stick properly to both materials and for the braces to fit perfectly around any obstructions. Need to return to the warehouse to take more measurements.

MEASUREMENTS FOR THE BROKEN PLASTIC MOUNTING HOLE ON THE MOUNT BLOCK FOR THE MENDED FINGER PCB:

Reference Point: Looking straight down the bore of the screwhole at the backside of the PCB were it properly installed. There are no accompanying photos at this time (because it didn't occur to me to shoot any of it).

X-width of thread-tip-to-thread-tip of the PCB mounting screw: 2.75mm
If we're lucky, a regular computer screw can replace the missing one.
Original thread is unknown.
Appears to have 1.5threads per millimeter.

X-width of broken screwhole frame: 8.00mm
Y-height of broken screwhole frame: 6.5mm
The screwhole frame originally had a flat roof.
Z-depth of the broken screwhole frame: 3.00mm (approx, measuring this one was tricky)

BRACE IDEA 1

3D printed plastic brace extends up against the back of the PCB where the text is.

Brace also extends down into the 9.85mm area measured earlier to replace the missing hump arch.

This block will be the same thickness, or even a zillionth of a mm thicker, to bear the pressure of the screw so the PCB doesn't have to.

And finally the brace has wings to be epoxied onto the area where the pencil markings are.

No epoxy will be used anywhere except on those wings.

Oughta put "DO NOT TOUCH ME I WILL SHATTER" warnings on both of these. 

BRACE IDEA 2

Same as above, but the rough edges of the PCB will also be dremeled out so the lower block and wings can epoxy along the Z-axis as well as the X-axis for a stronger attachment.
Might be the strongest version, but this will also be invasive to the PCB and since there is no pressure put on the PCB, shouldn't be necessary.

The screwhole mount for the PCB with the broken base has enough surviving thread and hole that it doesn't need a brace (though a little reinforcement wouldn't hurt since old nylon is so bad).

Actually found a tiny fragment of the screwhole frame for the mended-finger PCB that has thread on it, will attempt Plastic-welding that back on at some point. Low priority because it's not really needed.

Could possibly be other fragments inside the machine, but it's unlikely. Shame whoever broke the PCBs didn't save the parts in a baggy or something.

NOTES 20221218

It occurred to me that I could probably just cut or dremel out a brace from some spare lexan, rather than take an eternity getting around to learning CADD and 3D print one.

NOTES 20230225

Wound up going with the "take an eternity to learn CADD and 3D print one" route. Designed support braces in FreeCAD and 3D printed them a few days earlier than this time of writing.

NOTES 20230304

Test fit the braces onto the PCBs, trimmed and filed them to fit better, then set them in JB PlasticWeld.

Setting time is only 5-10 minutes, but I left it overnight to ensure it sets.

NOTES 20230305

Returned to check on the epoxy, but it failed to cure and remained goopy.

May have been a bad mix, it definitely wasn't a 1:1 ratio because the yellow half failed to pour reliably while the clear half did.

Brought all epoxy pieces home to retry and to cure them in a warmer environment to hopefully get a good bond this time.

Wiped off the old epoxy with q-tips and 91% isopropyl, then reset the PCB with the broken base and its accompanying PCB mount, which has a fractured frame-mounting screwhole.

Forgot the PCB mounting screwhole fragment at the warehouse though, and I forget which mount it goes to, will recheck upon returning next weekend.

Also need to get a longer screw to mount the PCB with the prosthetic base, since the base doubles its thickness.

Also need to re-measure the broken plug in the cabinet with the new caliper since it can measure inside diameters and depths, unlike the other.

NOTES 20230306

The second epoxying appears to have worked better, the epoxy appears to be securing the fracture on the mounting frame, and the prosthetic mounting hole on the PCB, as intended.

NOTES 202303xx

The screw pressure on the mounting frame re-cracked that part.

Took both PCBs and mounting frames home to make additional braces and re-attempt epoxying on the broken part.

NOTES 20230323

All epoxying is complete and the PCBs are finally reattached to their mounts.

NOTES 20230401

Reinstalled both mounts and PCBs, realized I forgot to resolder the broken one, brought that one back home for soldering.

NOTES 20230826

Fell off the face of the Earth for a few months, then began modeling a replacement plug for the broken one in the MC/ST's cabinet, using one of the intact plugs in my MC/ET as a reference for additional measurements.

NOTES 20230828

After some trial and error, got a working version of the plug modeled and successfully printed.

NOTES 20230903

Returned to the warehouse to test fit the replacement plug, and one of the wires clicked securely in.

Also mapped the physical routing of the broken plug's loose wires to their respective components on the Selectric to compare against the MC/ET's plug's component pinout at home.

NOTES 202209xx

Mapped out more of the MC/ST's wiring via comparison with the MC/ET's analogues.

Achieved 5 wires known, 4 wires suspected, and 3 wires still total mysteries.

NOTES 20230917

Returned to the warehouse to retest magcard playback with a card inserted in the correct orientation.

Machine did not throw an error and did begin attempting to print something, though it seemed to get stuck printing the same character over and over, currently unknown if that was a malfunction or the card actually had just the same character saved repeatedly.

Notably, during the first playback, it seemed to rotate the typeball twice, sluggishly, before beginning the above machinegunning all the way to the right where it hit the margin and just kept slamming the same character over and over in place. It was not observed, or at least I do not remember observing, any shifting attempts at the start of subsequent machine gunning.

During testing, the rotate tape fell off because the right rotate pulley arm was so badly gunked it couldn't return fast enough to keep tension.

Unfortunately I damaged it during reinstallation, though it seems to still work well enough for now. To be replaced.

I then disassembled the rotate pulley arm and associated components for cleaning, and though the oil soak pad was misplaced, the pulley arm now swings freely as designed.

I also found a spring laying down inside the frame, which I believe fell off during the test in a location suspiciously close to underneath one of the clutches got stuck grinding while testing.

Found it belonged to one of the clutch's stopper bars (for lack of a better term), that normally stops the clutch after a cycle.

Reinstalling the spring got the clutch to behave again.

Other internal components were also found to be severely gunked and thus either slow or completely unable to return to rest, and are likely causing mechanical malfunctions with the printer.

Now that we know the cabinet not to be entirely braindead, the Selectric is due to as thorough a cleaning as I can give it.

Also the Backspace and Carrier Return wires in the new plug may be transposed and may need to be swapped around, need to investigate that further.