Wrapping up build and bench testing of the 1130 MRAM memory replacement board

DID A FULL TEST OF THE INPUTS, PARITY GENERATION AND OUTPUTS

Inserting a 1 on each bit position of the MRAM data lines produced a change in the associated parity output, just as it should. Each pad was touched to validate the behavior. I then tacked a wire on each of the MRAM data lines and injected a 1 or 0 into the B register (data inputs from the CPU). When I asserted the Storage Write line the value showed up for the duration of the 100 ns pulse that also triggers the MRAM write operation. 

When I tested each output I discovered to my chagrin that I had put AND gates in the output lines, which produced a positive going 100ns pulse for each 1 injected into the MRAM data pads. The CPU expects a negative going pulse, however. I ordered open collector NAND gates and will swap them into the board as soon as they arrive. 

MEMORY CHIP INSTALLED 

The MRAM chip is soldered into place as the last step, once all the circuitry that drives and supports it appears to be working properly. At this point the board is functionally complete, although I am still waiting on the proper value capacitors in surface mount packaging and for the NAND gates I just ordered. 

Above you can see the board with the temporary capacitors on the right and a wire tacked on to replace the 3.3V regulator that was going to be on the board. In the picture I had begun to install the SLT pins where the cable from the CPU will plug into the board. 

SLT PINS INSTALLED

The last set of pins will be installed on the board so that the SLT ribbon connectors T1, T3 and T4 between the 1130 and this board can be plugged in. That will complete assembly, after which testing must be done with the 1130 system. 

1132 printer motor is spinning freely with almost zero resistance

CHECKED ON MOTOR THIS AFTERNOON AND WAS PLEASANTLY SURPRISED

I was expecting that I would have to disassemble the motor to clean out the bearings but when I touched the pulley I found that it would spin with little force. The torque required to turn the pulley, lets call it 10 when I first tried the motor, is now down to somewhere less than 1 on the same scale. 

The clock oil did its thing, seeping in between the old lubricants and the metal surfaces. Within a couple of turns it was as good as when the motor was new in the 1960s. I will do one final test with AC applied to be certain that it starts and spins, then it goes back into the printer. 

More testing of the 1130 MRAM memory replacement board - part 2

HOOKED UP ARDUINO DUE TO SIMULATE THE CPU SIGNALS

An Arduino Due will develop the three input signals at the proper timing. I coded this up using the timers on the Due SAM chip, which are handled differently than the ATMEL chip timers in the other Arduino boards. I used the half speed clock timer1 running at 42MHz, then set up the compare register A in the timer to 38 and compare register C to 76. 

Starting the timer in wave generation gives us a 50% duty cycle square wave at 560 KHz or about 1.8 microseconds per cycle. It interrupts when hitting these register values, allowing us to flip digital pins 2 and 3 on and off to form Storage Read and Storage Write. Pin 4 is Storage Use. We run with pin 4 off for six seconds then flip it on for the duration. 

We can display several signals on an oscilloscope to verify that this circuitry drives E, G, W and the two buffer control signals at the intended times and only for memory cycles with Storage Use asserted. I had to tack on wires to drive the three control signals on my board, plus some wires to detect the output of E, G, W, sense pulse output and buffer enable. With four channels, I set one to Storage Read and monitored three at a time on the other channels. 

The timing for various actions on the board is show in the diagram below. As soon as we start the read part of the cycle, the chip does a read and leaves the data bits on its output pins (purple). The timing chain does a delay of 800 ns (green) and then emits an 80 ns pulse (turquoise). If the data bit from the MRAM read is a 1, the pulse is passed through as a sense bit. In the write cycle, we gate the B Bit signals to the MRAM chip during the entire write cycle. The write delay (red) takes 800 ns then a pulse is emitted (yellow). This is passed to the MRAM chip to command a write operation (orange). 

The chip enable signal E should turn on whenever we have +12V to the board and the Storage Use signal is asserted. Signal G to set the MRAM chip data lines to output mode should be turned on whenever we are in a read cycle and E is enabled. Signal W requests the MRAM chip to write the data on its data pins into the location configured by the address bits; this signal is turned on by the write control pulse from the timer chain during a write cycle. 

The buffer chips normally sit in high impedance mode, not driving any data on to the MRAM chip data lines. When we have the write control pulse during a write cycle, we activate the buffer to pass the data to the MRAM chip. We are activating the buffer at the same time that we start the write in the MRAM chip, so there is a bit of settling of the data values immediately but the memory chip only samples later in the write operation when the buffer outputs will be stable. 

The sense output pulses are generated by the read cycle timer chain with the pulse at 800 to 880 nanoseconds into the cycle, when the data from the memory location was read and became stable about 45 ns into the cycle. It is 'sampled' at 800-880 ns where we either have an 80 ns open collector pulldown to ground if the data bit value was 1 or we have the sense output staying pulled high. 

IMPROVED TIMING WITH CHANGE TO CAPACITORS

I substituted some different valued capacitors until I got the timing chains to behave closer to my target. The final 680 pf choice gave me just about perfect delay into the midst of the read and write cycles, while a 33 pf part reduced the pulse width down to 100 ns. That is close enough that I will lock in these values and move on. 

All the signals looked perfect. I carefully studied the MRAM chip control signals for gating the data out on the bidirectional pins, for requesting a write of data coming in on the bidirectional pins and for disabling the chip so it is insensitive to other signals. The signals that pass the incoming data through to the MRAM chip and initiate a write looked exactly as they should. Finally, the signals were right on time to create a pulse for any data bit whose value was 1, sending those out to the CPU. 

INPUT AND OUTPUT BUFFERS INSTALLED

The chips that take the data input from the CPU and connect it to the memory chip must leave the memory chip pins in high impedance except for when we are doing a write and it is the appropriate time to drive the data input signals onto the memory chip pins. They were all soldered onto the board and carefully checked for shorts/solder bridges. 

We validated the control signals in the prior test, but tomorrow I will double check that no signals are driven into the memory chip except when we intend to.

The chips that output a pulse on the sense lines should only drive a pulse (open collector gate pulled to ground for the pulse duration) when Storage Use is true and it is the proper time. Further it should only send a pulse when the bit retrieved from memory is a 1. I will test this tomorrow as well. 

The Arduino will continue to drive the control signals in place of the CPU. We will inject values as if the memory chip were installed, in order to test the output buffers. We will probe the inputs where the memory chip will be installed to verify that no 1 bits arrive except those we introduce to the input side of the input buffer chips. 

PARITY GENERATING XOR CASCADE INSTALLED

A group of chips will generate a parity bit P1 or P2 based on the eight bit values being output from the memory chip during a read. P1 and P2 also connect to the output buffers to the sense lines - these are to be checked tomorrow by inserting some ones on the empty memory chip pads and observing the sense line output. 

Freeing up 1132 printer components - part 4

WORKING ON PRINT CLUTCH LATCH LEVERS FIRST

I found that I had good access to the Print Clutch Latch levers looking into the machine from the front. The angle is a bit uncomfortable but sitting in a chair made it tolerable. I worked on these for about an hour and was able to get the first 24 of them moving well. Another few hours of work and they will all be restored to operation. 

REMOVED MOTOR TO DEAL WITH ALMOST FROZEN SHAFT

I unbolted the motor and dropped it out of the printer, moving it to the workbench. It is a GE 1/4 HP 208/230V AC motor that rotates at 1725 RPM. The system of drive belts and pulleys lowers the speed to 300 RPM at the print mechanism. 

This motor has oil caps to add oil into the bearings. The bearings have a felt pad that oils the bronze bearing material. According to a plate on the motor, it would not need reoiling under light use but if used heavily might need retopping every few years. I guess five decades sitting idle wasn't in their plans. You can see how much dust and grunge was being absorbed over that long sleep:

I added some clock oil to the oil wicks and began to force the shaft to turn. In just ten minutes, the resistance had decreased noticeably. It is still not spinning freely or even close, but if the original torque needed to budge the shaft was a 10, it had dropped to about 6 just from the initial work. I still don't know if I can get this spinning better without having to disassemble the motor but one way or another I should be able to return this to service. 

Freeing up 1132 printer components - part 3

FINISHED PRINT CLUTCH RESTORE LEVERS

I put in a couple of hours to free up all the parts of the Clutch Restore Lever assembly. They were stubbornly frozen and had to be handled a few at a time, using copious oil and lots of repetitive movement. By the end of the session I had them all behaving properly.

These levers are pushed by the lobe of each Clutch Print Cam. They push the Clutch Trigger and Armature Knockoff levers back into their idle position. 

WILL MOVE ON TO FREE UP THE PRINT CLUTCHES

The Print Clutch rotates freely around the Print Clutch Shaft as long as the tooth of the Print Clutch Dog is held away from the flutes of the rotating Print Clutch Shaft. In the idle state, the dog is held by the Clutch Trigger lever, sitting on the Armature Knockoff lever after being restored. A small notch on the outside of the Print Clutch is held by the Print Clutch Detent levers, so that they don't rotate around from vibrations even though the dog tooth is not engaged. 

The Print Clutch discs are stuck to each other by congealed lubricant. Some of the dogs have slipped off the Clutch Trigger, with their tooth jammed onto one of the flutes of the Print Clutch Shaft. The dogs pivot on one end and are pulled into to engage with the flutes by a spring. 

I need to lubricate the Print Clutch Dog pivots, then free up the clutch disc plates so that they all spin freely when not engaged. I have decent access to them from the top of the printer without disassembling this part of the mechanism. 

However, some will not remain free of the Print Clutch Shaft flutes because the Clutch Trigger is frozen in the wrong position (tripped). I will have to loosen up the triggers. Not sure yet how I get access to them - whether I need to remove the print wheels which is a big hassle. 

More testing of the 1130 MRAM memory replacement board - part 1

BOUGHT SMALL 3.3V REGULATOR AND HOOKED UP TO CONTINUE TESTING

I discovered that I had downloaded a footprint for the inductor from a third party that did NOT match the inductor part I bought. I searched and found an accurate footprint, which matched the real world inability for the part to be soldered onto the board. I would have to relocate nearby components and make other adjustments to be able to fit the inductor onto the board. 

Rather than trying to desolder the chip, with all the nearby components being disturbed as well, then go through the attempts again, I went a different direction. While I tried to be ultra efficient with the power supply for this board, it is peripheral to the real functionality of the board. It isn't worth battling this when the extra heat and power consumption of a linear regulator is insignificant in the larger picture of the 1130 system. 

I cleared all the parts off the board from the power supply in preparation for moving forward with the next section of the board's construction and testing. As soon as Amazon delivers the regulator I can wire it into the board and test what I have installed. 

CONTROL CIRCUITRY INSTALLED AND PARTLY TESTED

The heart of the functionality is the control logic. This consists of two chains of timers, each with two SN74LVC1G123 timer chips. These are interconnected with a number of logic gates to produce the gating and control inputs to the memory chip and buffers. 

The timer chips are also extremely teeny, an 8-VSSOP package. This has the same pad width and pad spacing as the ultra annoying LFQN power supply chip footprint, but this has leads sticking out from the side whereas the power LFQN part has no leads so the solder must flow underneath to bond. It was much easier to get this wee little part soldered correctly. 

Due to the small size of the pads - .25mm width - no Weller soldering iron tip is narrow enough to fit on these; it would take a tip that was about .15mm to .20mm wide to properly fit, but those don't exist. Equally, my probes on the meter are gargantuan in comparison and don't fit down in the angle between chip and lead. At least for the leads I used thin wire as an extender for the probing. Hot air reflow of solder paste did the job for the soldering task.

Since the timer chains trigger on the rising edge of the Storage Read and Storage Write signals, I tested the timing first with a debounced button on a prototyping tool I own. The timer waits 620 ns then emits a 115 ns pulse. During the pulse, on a read, the -Sense Bit x line is pulled to ground if the bit read from memory was a 1, otherwise it stays high. During the pulse, on a write, the memory chip is told to write the data that was incoming from the -B Bit x signals at the address set up by the Storage Address Register bits 3 to 15. 

The intent was to achieve an 800 ns delay then emit a pulse that is only 80 ns wide, so I have some tweaking to do with the resistor and capacitor values in the timing chain. I will experiment with this until I get close to the target values. Note that the board would work properly with the values I observed, but I would prefer to achieve my design times. 

The board receives three control signals from the 1130 CPU. Storage Read and Storage Write alternate each lasting 1.8 microseconds. A memory cycle on the 1130 is a storage read followed by a write, thus the cycle lasts 3.6 uS. Each cycle either has Storage Use on or off for the 3.6 uS - when it is off, we should be ignoring any data or addresses from the CPU and should not be driving any signals on the sense lines back to the CPU. 

We want to see the appropriate state for the memory chip E (enable), G (drive output with contents of memory) and W (write input data to memory). We also want to see that the output buffers to the sense lines are activated at the appropriate time, and that the input buffers to put data onto the memory chip pins are activated at their appropriate time. 

CODING THE TEST DRIVER TONIGHT TO CONTINUE THE TESTING

I have an Arduino Due which uses 3.3V signals, compatible with my board. I will write code to generate the exact pattern of alternating 1.8 microsecond Storage Read and Storage Write assertions. It will run with Storage Use turned off for several cycles and then turn on Storage Use for the remaining test time. 

Began assembly and test of the core memory replacement for the 1130

ALL PARTS ARRIVED FROM DIGIKEY AND THEN THE BLANK PCB SHOWED  UP

The last of the orders from Digikey was delivered on Tuesday. All that remained was waiting for UPS to deliver the PCBs which had been fabricated by JLCPCB. It took a few days to get here from China but I had them in my hand by Thursday. 

PIECEWISE ASSEMBLY AND TESTING OF THE BOARD

I decided to construct logical sections of the board one at a time, completing testing of its functionality, rather than soldering together all the components and attempting to debug the entire thing at one time. There are quite a few very tiny surface mount parts, each of which has the potential for unconnected pads, solder bridges or other shorts. Building 'bite size' portions increases the chance I will completely check over each connection and part both visually and with electronic tools. 

POWER SUPPLY CIRCUITRY INSTALLED AND TESTED

Building the power supply was the obvious starting point. This takes 12VDC as an input on a jack and produces 3.3V rails for all the logic chips. The LT8635S device is a mere 4mm x 4mm with 24 pads around the periphery plus four ground pads underneath and four corner tabs to anchor the chip. 

It will be challenging to solder this on as there are barely any external pins - mainly the connections are underneath as is the potential for solder bridging between pads. I will get the chip installed before any other part, so that I have the best chance to test for connectivity and for shorts. Once that appears good, all the resistors, capacitors and the inductor will be arrayed around the chip to complete its circuit. Connector J1 gets put on the board and then I can apply 12V to verify that it is producing 3.3V. 

SOLDER, REMOVE, SOLDER, REPEAT

The pesky LQFN sized chip has only .25mm pin to pin separation so the potential for solder bridges or shorts was very high. Time after time, I would use the hot air tool to melt the solder and try to get the chip positioned perfectly. I would then test for shorts using a continuity tester, find one or more, and remove the chip. 

CHIP SEEMED PROPERLY CONNECTED SO I ADDED THE OTHER COMPONENTS

In finally got the chip so that there was no short on all the pads connected to the part. I finally got to the inductor and realized it didn't match the footprint on the PCB. In fact it was too big, not enough room from the chip to the pad. 

I tried to solder leads to the inductor and somehow get it into circuit just for testing, but it broke. I grabbed an inductor from my parts bin - not the right value but something I could try to test with. As I tacked it onto the pads I did a short test. The output, one side of the inductor, was shorted to ground!

SOMEHOW THE CHIP IS NOW SHORTED OUTPUT TO GROUND

When I first tested it, the chip wasn't shorted. I put down all the components then it went downhill. I had to walk away and work on something else. In the interim I can modify this to accept 3.3V regulated input so that I can proceed with testing the rest of the board. Later I can figure out how to get this power supply working or do something different for power. 

Freeing up 1132 printer components - part 2

FINISHED UP WITH PRINT CAM DETENT LEVERS

The usual formula - lots of Nye 140 clock oil and careful repetitive manipulation of each lever - broke up the sludge keeping the parts from moving. Once the levers were moving smoothly and restoring under spring power crisply, I set them aside to move on to the next group of parts in the printer.

ORDERED MORE OIL

My clock oil is down to about 1 ounce and there remains quite a few parts to loosen up. I am going to need to get more. I placed an order and hope to have it within a week.

REMOVED DRIVE BELTS TO WORK ON POWERED PARTS

I loosened a pulley to release tension on the belts and take them off. One was previously identified as shot due to a cut but the other one looks a bit rough and is deformed from sitting in that position for decades. 

CHECKING TO SEE WHAT PARTS CAN BE ROTATED

The first drive belt ran from the motor to a pulley. The second drive belt goes from that pulley up to the Print Clutch Shaft. It has a gear on the end of it which meshes through a neutral gear to drive the Print Wheel gear. With them all connected, I couldn't get anything to budge. 

By disconnecting the pulleys I found that the Print Cycle Clutch shaft and the Print Wheels turned relatively easily. It was the motor itself that had enormous drag. I could barely turn it at all by hand. My suspicion is that the grease has degraded in the motor bearings, but we shall see after I pull it off the frame and examine it more closely.

STARTED WORKING ON THE PRINT CLUTCH RESTORE LEVERS

These where the most frozen parts of all in the machine. I had to flood the slots and pivot points, then force the part to begrudgingly move before I could work the old lubricant out. It was one by one, sometimes two at a time, a very slow process. Once freed up they moved nicely. 

Freeing up 1132 printer components - part 1

MAGNET ASSEMBLY WORK

I worked on the Magnet Assembly with my trusty Nye oil. That has the Link and the Armature Knockoff Lever components on it, plus the plate with the print magnets that will pull on the Link. I oiled the pivot points and the slots that align the parts. Initially they were frozen solid. I would get a small group of them moving each time, at first barely moving at all but eventually crisply snapping back under spring power.

About 1 in 10 of the levers would still be sticky, restoring very slowly when I move it. It took extra time and manipulation to get those to perform properly. The picture below is at a point where I was working from right to left through the levers but hadn't finished. 

The armatures of the print magnets were all working well already, as they had not been previously lubricated. Thus once I had the Magnet Assembly restored, it was time to move on to the rest of the print mechanism. Looking down from the top, you can see the Print Cams and related parts visible.

In the picture above, you can see the Print Cam Detent parts engaged with the notch in the Print Cams above. The Print Cam Detents are held in place by a machined bar sitting across the front of the printer frame. 

The Print Cams rotate around the 18 flute shaft - just visible in the picture below taken from the right side of the cams. 

MISSING PRINT CAM DETENT IN COLUMN 120 - PREVIOUS FAILURE

I had previously noticed a gouge in the platen over on the right side. Seeing the missing Print Cam Detent and partially disabled Print Cam tells me that a part failed, it jammed into the platen, and somebody shuffled the good parts to the left so that only column 120 is dead. Above you can see the missing parts on the right. Below you can see the gouged platen. 

MORE DAMAGE DETECTED - NEED A RUBBER BELT

One of the two rubber drive belts had a big cut in it. It will fail soon and needs to be replaced.

The pulley that this belt turns rotates the Print Cam Shaft. A passive gear just below it couples the rotation onto the drive gear that turns the 120 Print Wheels. 

The drive gear turning the Print Wheels has the timing disk on the end of the shaft. A lamp shines through the disk onto photocells that detect holes in concentric circles as the disk spins. The outer circle has 48 holes evenly spaced around the disk which provide timing pulses for the arrival of each character on the Print Wheels. The seven inner circles have holes that define the code for the character so that when the CPU reads the code it knows which print magnets to fire to put that character on the paper.

REMOVING BAR WITH PRINT CAM DETENT LEVERS FOR RESTORATION

The machined bar for the Print Cam Detents is secured to the frame by two bolts and one pin on each side. Unfortunately, the passive gear is right in the way, blocking removal of one of the four bolts and one pin. 

I had to remove the circlip and nudge the passive gear partially off its shaft in order to get the machined bar out of the machine. It will need to be worked carefully with the Nye oil to free up the 119 levers. 

Below that machined bar sat another one, holding the Print Cam Restore levers. It also has four bolts and two pins holding it in place, but in addition had two socket head bolts through a slotted guide plate above.

Disassembling 1132 printer mechanism to free up thousands of stuck parts

PRINTING MECHANISM IN 1132 PRINTER

The printer has 120 wheels suspended from the top on a pivot rod. The wheels have teeth around the edge that engage with a gear the entire line width, below the print wheel. That wide gear spins all the print wheels continuously at 112.5 RPM. The print wheel has 48 characters formed on its rim spaced evenly around the wheel. 

A cam on another mechanism will bump into the left side of the print wheel hanger, so that it pivots to the right and strikes the paper on the platen. A spring pulls the hanger back to its neutral position as soon as the cam lobe rotates past. 

A spinning clutch shaft to the left of the print wheels has 18 flutes on it. Suspended around the shaft is the print cam with its clutch mechanism that holds a tooth (print cam dog) away from the spinning clutch shaft. A print cam detent lever fits into a notch in the cam to hold it from rotating. The cam has one lobe on it, the lobe will eventually push a print wheel to type a character.

In the diagram above you see the print cam sitting in its latched position. The Print Cam Detent is engaged in a notch near the top of the cam which holds it so its lobe is near the bottom - hidden behind the latch mechanisms in the diagram.

The Print Clutch lever is holding the Print Cam Dog outward, against its return spring, so that the tooth on the Print Cam Dog is not in the path of the flutes on that Clutch Shaft rotating inside the cam. The Print Clutch lever is held in this position where it blocks the Print Cam Dog movement, by the Armature Knockoff Lever whose rightmost tine presses up to hold the Print Clutch in that position. 

When it is time to print a character, a Print Magnet is activated, attracting its Armature. The Armature pulls on a Link which tilts the Armature Knockoff Lever, causing the Print Clutch to fall into the space between the two tines of the Armature Knockoff Lever. 

Because the Print Clutch Latch has fallen off the Print Clutch Dog, its spring pulls it down and the tooth on it engages with a flute of the Clutch Shaft. The cam begins rotating clockwise. When it is about 3/4 around the lobe strikes the print wheel hanger to push the print wheel into the platen. 

Meanwhile, the lobe of the cam has moved the Restore Lever which pushes the Armature Knockoff Lever back into position and sets up the Print Clutch Latch to hold the Print Clutch Dog tooth out of the flutes. This stops the cam thus we get just one rotation and one swing of the Print Wheel into the platen. 

 

All of these levers and springs must move freely. The Print Clutch Dog and the cams must move freely. So to the print wheel and any other parts that contribute to printing a character in a column. For each of the 120 print columns, there are 18 parts involved, most of which must not be glued in place. 

The oils and grease used by IBM in the 1960s breaks down with time, as well as absorbing dust in the air. The result is a sludge or glue that forms keeping every moving part from moving. It keeps every spring from compressing and extending. It keeps every rotating part from rotating. 

The restoration requires me to free up all these tiny parts for all 120 columns - a total of 2,160 items I have to free of old lubricant and get moving properly. I use an ultrafine clock oil (Nye oil) that will seep in between the sludgy old lubricant and the metal surfaces. By carefully manipulating every part, I can work it free. 

To make this happen, I have to partially disassemble the printer mechanism. The parts from the Print Magnet up to the Armature Knockoff Lever are in a removable Magnet Assembly. This in turn has a plate with all the Print Magnets that can be removed from the larger Magnet Assembly. 

I have the Magnet Assembly out of a workbench and the plate with the Print Magnets is sitting by the printer read for me to restore those parts. 

Magnet Assembly minus Print Magnet plate

Plate with Print Magnets

The Print Wheels with their hangers and the remaining clutch parts are still in the machine where I can access them to some degree. 

In the last picture we are looking up towards the clutch and its parts, with the print wheel parts farther behind that. I will first work on the Magnet Assembly and Print Magnets. 

Completed rewiring of power supply into the IBM 1132 printer

FINISHING UP TERMINAL BLOCK 2 WIRING

This side was slightly complicated because there were two wiring looms plus a single wire that had been attached. IBM laces up bundles of wires so that the ring terminals line up precisely with the positions they will be connected to on the terminal blocks, making it very easy to get the reattachment done correctly. However, with two looms there is some relative positioning to work out. 

Some of the terminal block 2 (TB2) screw positions had two wires attached from a single loom, but other had even more wires as both looms contributed ring terminals to the screw location. My pictures prior to disassembly don't always clearly show all the layers of wires and from which loom they came. 

To resolve this, I beeped out the connections using the schematics so that I confirmed that the wires on each TB2 screw were connected on the other end to the proper circuit or component. 

TERMINAL BLOCK 1 WIRING COMPLETED

Terminal block 1 (TB1) is much more complex, having many different cables attaching to it, rather than mostly wire in a loom. I do have pictures to help me with the routing of the wires, but in this case I definitely needed to trace all the ends that I tentatively assigned, ensuring they went to the proper destination. 

TB1 handles the higher voltage AC power that will drive the motor, gate blower, usage meter power supply, and convenience outlet. TB2 carries low voltage mostly DC power to the carriage motor, solenoids, logic gate, carriage control tape PCB, and timing disk lamp. 

Bench test of usage meter power supply, control circuitry and meter; began reinstallation of PS into 1132

USAGE METER BOX TESTING

I applied 230V to the meter supply to verify it produced 41VAC which drives the usage meter on the printer. The usage meter control box js inside the printer frame, with the wires that would normally be connected to TB1 terminals 3 and 4 available for connection. Inside the meter box I can check for 41VAC across terminals 3 and 4 of the terminal strip inside. 

At one time IBM charged customers by the hour on rental contracts, thus the usage meter was very important to generate revenue for the company. After the anti-trust lawsuits by the US government, IBM no longer rented products nor charged by the hour of usage, but the meters remained on the equipment for some time after. 

I tested that a small logic signal applied to the usage meter supply/control box would switch the 41VAC to the meter when pulled to ground and remove it when the logic signal rose to 6V or was disconnected.

To test this, I hooked 6VDC to terminal 6 of the strip and then connected terminal 9 to ground to test activation of the relays. With a voltmeter across 3 and 5 of the terminal block, I should see 41VAC when I ground terminal 9 and no voltage when the terminal is not grounded. 

Everything worked perfectly - the box is ready for reinstallation.

BEGAN WIRING UP POWER SUPPLY BOX

I bolted down the power supply box and then the usage meter power/control box, now that everything has been checked out. I then began the rewiring process, starting with terminal block 2. I am about half done with TB2 but have to stop working because I have a condo association meeting tonight.

Alternatives to replace reed relay in 1132 printer carriage control tape PCB

REQUIREMENTS OF THE CIRCUIT

The relay coil is activated by +48VDC driven by the circuit below through the coil to ground.

The contacts are in the circuit below which switches +6V power to a bank of eight SCR circuits in parallel that act as latches to record when a brush contacted the commutator under the carriage control tape due to a hole in that channel. 

The contacts must handle 6VDC and a total load of 435 ma if all eight SCRs are latched at the same time. At a minimum we need a SPDT contact plus a SPST (normally closed) contact, although practically that is implemented as a DPDT switch. 

The relay will break the contact on one side of a pole before making the contact on the other side. This is essential for correct operation of the circuit as it must reset the SCRs but interrupting power at the moment it switches. 

OPTION 1 - REED RELAY

I found a Panasonic relay (TQ2SA-48V) that supports 500ma contact current, is DPDT and has a 48V coil. The only downside is that it is a surface mount part, so I could have to create some kind of interface between the pins on the IBM PCB and the pads of this replacement relay. It is about $5 so cost is swamped by shipping cost. The real issue is the interface. If I need to build a custom PCB, with the current tariffs and shipping costs, I probably have about $30 to build the minimum batch of five boards. 

OPTION 2 - INTELLIGENT POWER SWITCHES

A device such as the STM TDE1798DP switch can act as an SPST switch, handling 500ma at 6V. I would need two of these to replace the SPDT action of the reed relay, plus another lower power switch for the additional SPDT switch inside the reed relay. The STM part costs $8 each plus the other parts. More importantly, a means must be devised to achieve the break before make behavior. 

If I use a capacitor to slow the rise of the enable signal to the gates, it delays the make. Because the intelligent power switch uses a differential comparator trigger with E+ and E- inputs, I can have an RC network drive the E+ so that it slows the rise with an inverter driving the E- nearly instantaneously. This gives me instant break and delayed make for each switch. 

I whipped up a circuit using a diode, an inverter, a resistor and a capacitor to drive the intelligent power switch. These are fed by the 48V coil voltage through a voltage divider to produce 5V when on and 0 when off. An analog switch capable of handling up to 10ma is used for the additional contact since the circuit in the 1132 drives about 5ma through the switch. 

The analog switch is about $5.25, the inverter is about $0.30, the diodes, resistors and capacitors all together don't come to $0.50, so the total would be a bit over $22 plus tax and shipping to substitute for the reed relay. I would use some kind of breadboard to wire it together and a way to mechanically anchor it on or near the PCB. 

OPTION 3 - EBAY USED RELAYS WITH SOME ADJUSTMENTS

I found a relay on eBay that with a 5V coil but SPDT with 500ma current capacity in a DIP form factor. I could buy two, with tax and shipping, for $18.40. I would use a pair of resistors to drop the +48V coil supply for RR1 down to 5V and put the four parts on a breadboard/perfboard then mount it near or on the PCB. 

DECISION MADE - GOING WITH OPTION 3

The relays are ordered. I have the perfboard on hand, thus I just need to work out the resistor values and solder it all together when it arrives. 

Checking the 1132 Carriage Control brush latch circuits

LATCH PCB

The large printed circuit board that is attached to the left rear frame of the 1132 printer does multiple things, but its main purpose is to detect the holes in the carriage control tape as the paper moves up in a skip operation. It uses SCRs to latch when a brush makes contact with the metal roller upon which the carriage control tape loop moves. These are then released by a relay once the paper has moved to the next line position - a signal +Carriage CB is generated when reaching each line. 

The normally closed terminals of the relays will feed the +6V supply to the SCRs which will latch on if they receive a positive voltage on the input gate. The relays have to activate to release the SCR(s) that latched. The use of SCRs allows the various tape channels to be detected at any point during the movement of the carriage, even if they don't occur at the same instant or if they disconnect before we reach the next line position. 

I can test each circuit by first applying the +6, -3 and ground power to the board, then applying a +6V value through a 1K resistor. I can sample the output which will be at ground when unlatched or at 3V when the SCR has fired. I then shut off power to clear the SCR state before the next one it tested.

The board is wired in through edge connectors but I can attach clips to the parts on the board to perform the test. For example, -3V is attached to an appropriate component such as one side of R18 on the PCB. 

The board contains three reed relays RR1, RR2 and RR3 that are used to manage the carriage skip and space signals plus related logical conditions. 

RR1 turns on when the program starts a carriage skip and remains on until the program issues a stop skip. While RR1 is on, it keeps the SCRs powered to record any carriage holes they pass and sets the +Space Interrupt Allow to low. Normally, as the carriage reaches the next line it allows a space interrupt, but during a skip we block that. 

RR3 turns on when the microswitch on the carriage closes, indicating we have reached the next line of the paper. Its only purpose is to activate relay RR2. Thus RR2 stays on for the duration of the contact interval, lagging RR3 by a brief amount in both turn on and turn off. 

The two relays are used to form an edge detector for the carriage microswitch. The logic diagrams and schematics for the 1132 are particularly horrific to decipher, with circuits spread across multiple pages as seemingly isolated chunks. I grabbed a few sections and moved them together just to create a somewhat clearer circuit for the reader to view.

The relatively long duration carriage microswitch pulls the left end of coil RR3 down towards -3V causing it to close the RR3 contacts. In the lower right, the RR3 contact will energize RR2. Then the RR2 contacts on the top will open up, so that the carriage microswitch is isolated and the output is pulled up to +12V by the 6.8K resistor. 

The output -Carriage CB drops to a low voltage with the voltage divider of the 6.8K, 1K and 240 ohm resistor pulling it down to a slightly negative level in the instant after the carriage microswitch closes. The contacts move to close in RR3 as it energizes. This then starts the contacts of RR2 to move as the RR3 contacts energize RR2. As soon as the RR2 contacts have opened, the -Carriage CB output is pulled back to +12V even though the carriage microswitch will remain closed for a while longer. 

This creates a sharp edge pulse of about 10 microseconds. By comparison, the carriage microswitch is closed for 1500 to 2500 microseconds. 

RESULTS OF THE TESTING

When I saw the rusty covers of the SCRs, while everything else on the board appeared pristine, I expected problems with the SCRs. However, they all latched and released exactly as they should. 

I wanted to check out the reed relay circuits, thus I applied +12V to the coils and checked the conductivity of the contacts in both the energized and rest states of the relays. RR2 and RR3 were working perfectly. However, no response whatsoever from RR1. I pulled it off the board and checked directly. Open circuit for the coil. 

FIRST TRY TO LOCATE A GOOD END OF THE COIL WIRE

I did a bit of excavating to see if the wire snapped off due to corrosion, allowing me to reconnect it. It didn't look promising. Several fragments of enameled coil wire broke off as I looked. I will need to find a replacement reed relay or create an equivalent circuit.

FUNCTIONALITY CONTROLLED BY RR1

Relay RR1 is only turned on and off by the signal from the processor that activates for a skip operation. It is not high speed with critical timing like RR2 and RR3 which define the edge detection pulse for every line that is reached by the carriage. 

RR1 has three sets of contacts. Two of them form a break then make pair such that when the relay is changing state, both are unconnected for a brief instant. This is used with the circuit that resets the SCRs for detecting a brush passing over a hole in the carriage control tape. 

When RR1 is not energized, the +6V to the SCRs is fed through RR2 contacts. Thus the SCRs have power but are interrupted by the carriage microswitch activation. This means that the carriage hole detection is reset at each line allowing the program to sense only the holes at the current position of the paper. 

When RR1 is energized because we are in a skip operation, the SCRs remain continually energized. They are therefore collecting all the holes that are passed during the entire skip duration. When the program tells the printer to stop skipping, RR1 turns off.

The two sets of RR1 contacts ensure that the SCR is turned off when the skip starts and again when the skip starts. The bottom RR1 contact opens up as the relay begins to energize, but the top RR1 contact is not connected until a bit later as the relay fully energizes. Similarly, when the skip is stopped, RR1 breaks the top contact earlier than it makes the bottom contact. In both cases there is an interruption that lets the SCRs reset. 

The third RR1 contact pulls the +Space Interrupt Allow line low when RR1 is energized. 

I am relatively free to substitute relays or modern electronic switches for these functions, but have to provide the brief interruption of +6V when switching states. I will be thinking about the most straightforward solution so I can get back to restoring the printer. 

Checked out magnet firing board in 1132 printer

MAGNETS (SOLENOIDS) DRIVE PRINT WHEELS TO STRIKE PAPER FOR EACH COLUMN

120 spinning print wheels ride a common axle, driven by the main motor. These have 48 characters arranged around their periphery and all wheels are locked so the same character is ready for printing on all 120 columns. A magnet will pivot a print wheel forward so that it strikes the print ribbon, making an impression onto the printer paper behind the ribbon which is held in place by a platen. 

The 1132 printer generates a clock pulse when each of the 48 characters reaches the correct position to be printed, during the rotation of the wheels. The 1132 has previously fetched a string of bits from the 1130 memory that indicate any columns that contain this character and thus need to have the magnet fired. The board we are testing takes the clock pulse and delivers +48V to all the magnets. The selected magnets (columns to be printed) are pulled to ground so that only those magnets conduct. 

This circuit accepts the +Print Disk Clock pulse which is generated from a photocell that detects the positioning holes on the timing disk that rotates on the end of the axle on which all 120 print wheels are mounted. The amplified pulse is cleaned up and output as +CB Clock back to the print controller logic inside the 1130 CPU. 

In the CPU, when the +CB Clock signal falls, the printer control logic requests a cycle steal (direct memory access) to begin fetching the bits that indicate which hammer magnet(s) to fire. When the cycle steal is done fetching eight words that define all 120 columns, it raises an interrupt on level 1. Software must read the value of the next character coming up on the print wheels, which are also encoded by holes in the timing disk and read by photocells. 

I can do a rudimentary check of the condition of the six transistors and diode on the board just to validate the diode action but with them in-circuit that won't be good enough. I don't want to burn out the hammers, if the +48V stays on steady, so I feel that I have to test this further.

I will power the components through the TB2 terminals, supplying +6, -3, +48 and ground for power. I will hook up a resistor divider load across TB2-14 and TB2-18 where the scope can observe the positive voltage that would be supplied to the magnets. I will also put a scope trace on TB2-19 to observe the generated +CB Clock signal. As input, I will connect a transistor to TB2-15, set up as an open collector, driven by short pulses that should cause the input to be mostly pulled to ground with the short intervals allowing the line to be pulled up to indicate the print disk signal. 

RESULTS OF TESTING

This circuitry performed perfectly. Both the switched 48V and the 6V signal logic level output moved up and down with the input that will come from the photocell on the clock portion of the timing disk. 

Bench test of 1132 printer power box - part 3

COMPLETED REWIRING OF POWER SUPPLY BOX

Most of the wires had ring terminals on each end. It was easy to put new terminals on new wire of the same length. However, one wire was soldered at one end and ran directly into the transformer on the other end. It was the 20VAC supply that was adjustable by the potentiometer it was soldered to. That allowed the lamp brightness to be adjusted to have the photocells on the timing disk produce the best pulses. 

I had to splice a new wire onto part of the existing wire, the section that came into the metal box from the transformer. The other end was soldered onto the pot. This was the last damaged wire to be replaced. 

I had disconnected about nine wires in order to get the metal plate to tilt out far enough, so I had to very carefully validate where they were to be reattached before doing so. At the same time, I reinstalled the contactor. Because one of its two coil terminals was corroded and fell off, I had a wire coming out instead of a terminal to attach the two ring terminals from other parts of the circuitry. I installed a small terminal block with just one position, put a ring terminal on the loose wire from the coil and thus all three were attached to the new block.

BURNISHING CONTACTOR CONTACTS BEFORE CLOSING UP METAL BOX

I used a burnishing tool to establish a good electrical connection for the contactor points. The K1 side connects the 208/230VAC from the CE Switch to the fuses F3 and F4. The K3 side connects the +48VDC filter capacitors to the rest of the circuitry that uses 48V. The middle section, K2, is unused on this machine. 

I purchased some new screws in order to reassemble the metal box in preparation for power on testing. The reassembly was uneventful with the new hardware. 

TEST 48VDC AND 20VAC POWER SUPPLY OPERATION

Hooking 230V to the power supply main transformer, with the main motor and usage meter isolated, let me check for proper 48V to operate all the solenoids and that lamp voltage was developed by the 20VAC side. 

There are five fuses on the power supply box. F1 and F2 control power to the utility outlet. F3 controls power to the main motor and the usage meter power supply. F4 controls power to the transformer for the power supply portion we are testing. F5 controls 48V power from the power supply portion we are testing out to the carriage motor. 

For this test only F4 was inserted. The output is measured across TB2-2 and TB2-4 for the 48V supply and across TB2-5 and TB2-6 for the 20VAC supply. The contactor had to be energized to route the 48V out of the power supply; pushing down on it with a nonconducting tool closed the points without requiring me to hook to 24VAC. 

The outputs were right on the money - 48VDC and 21 VAC. The latter is adjustable via the potentiometer R5 in case we want a bit less or a bit more than 21V for the lamp that shines through the timing disk into the photocells. 

CLOSED UP THE POWER SUPPLY, READY TO PUT IT BACK INSIDE THE 1132 PRINTER

I put the metal enclosure together since the power supply is functioning correctly. A picture below shows it with the 230V input still attached via white power cord, before the metal mesh was installed prior to movement. 

Bench test of 1132 printer power box - part 2

DISCOVERING MULTIPLE WIRES THAT HAD BEEN GNAWED

As I was preparing to reinstall the contactor, having epoxied the new wire into place on the side of the coil, I noticed a wire whose insulation was gnawed off leaving a long section of bare wire. I could see what appeared to be more bare sections of wire below. The box had to be opened up to allow me access to check all wiring and replace any that have exposed wire strands. 

I had to drill out one of the screws because it was too rusted to break free, the others could be removed after some percussive rotary moment was applied to them to start them moving. I could tilt out one of the sides of the metal box. 

To gain access, I had to detach some wires that were too short to let the side tilt out fully. I marked each and finally could look at everything. Lots of wires with gnawed off insulation, one of them barely able to conduct electricity. 

The solution is to replace all the ruined wires and reassemble the power supply box. I had a good supply of heavier wire but had to go to the local hardware store to buy 22 feet of 14 gauge for the medium current wires. I began rewiring but didn't finish before I had to leave for a dinner with friends. 

TESTING SECTIONS FOR SHORTS

I did safety checks on all portions of the power box and the printer to identify any short circuits, such as a bad transformer primary, prior to doing any testing under power. Another important test was a set of dedicated wires in the cables from the 1130 that delivery 115VAC to spin the fans whenever the 1130 system is powered on. It was important that the blower motor not be shorted. This is on the wires attached to TB1-7, TB1-13 and TB1-14. 

Input to power supply

Blower circuit

To accomplish some of the short tests, the meter is attached to the wire harness that was disconnected, since items like the blower motor are outside the power supply box. I did validate as much of the external circuitry as feasible during this exercise. 

VERIFYING UTILITY OUTLET CONNECTIVITY

A dedicated set of wires in the power cable coming into the 1132 delivers unswitched 115VAC power to feed a utility wall socket for customer engineer use. I checked that we had proper neutral, hot and ground lines with no shorts. The test was on the wires hooked to TB1-16 and TB1-17.

Once again, this is mostly done outside of the power supply box, since the incoming wires from the cable and the outlet box were both external, attached to the terminal block. 

Everything was good with this verification. 

CHECKING CE POWER SWITCH CONDITION

The wires on the power cable delivering 208/230V to the printer pass through a toggle switch on the power supply.  When switched on, as it normally would be, the power then passes through the contacts on the contactor K1 before energizing the primary of the main power supply transformer. Simultaneously it drives the main motor for the printer and the usage meter power supply in its separate box. 

The switch was working perfectly.

CHECKING MAIN AND CARRIAGE MOTORS FOR SHORT OR OPEN CIRCUITS

I couldn't spin the main motor since the 403 print mechanism is gummed and frozen with congealed old grease and oil. However I wanted to ensure that the windings all had continuity without any short circuits. This done with the wires that attach to TB1-3, TB1-4 and TB1-12 since the motor is outside of the box we are testing. 

Similarly, the carriage control motor which advances the paper was not moving freely due to old lubricants, but its windings could be checked at this point. It is supplied by 48V while the main motor runs on 208/230V. It is checked via the wires that attach to TB2-1, TB2-4 and TB2-9. 

Both motors had appropriate DC resistance and no shorts to ground. 

SHORT TESTING THE LOGIC GATE

I tested the power rails for shorts and found none. The +6, +3, -3, +12 and +48 were all good on the gate. I tested the magnet driver circuits superficially and found no shorts there.