Tuesday, June 30, 2015

Working on install of VM/370 R6 and DOS/VS R34 on the P390 system, plus work on the ztex board problem

Crushing day yesterday (Monday) with no free time other than work. Today I got a bit of time to go into the workshop.


I rerouted the last of the long cable, allowing the 1053 to sit properly in its cradle on the 1130. Things are buttoned up and ready to do final testing when I get the SAC Interface back online.


I found two virtual disk volumes that were corrupted, which explains why the OS/390 image doesn't boot up fully, triggering an abnormal end in the virtual disk driver code of the P390 software. Whether this is recoverable if I substitute a clean empty volume is still to be determined, since it depends on whether anything essential was on those volumes. Not a high priority.

I located and began transferring the public domain versions of DOS/VS, VM/370 and MVS 3.8, which would let me completely customize and install the systems as I want them. Some of the distributions are already in the right format for P390 (AWSTAPE format for tape reels and AWSCKD format for disk volumes), others are in formats that can be converted using utilities from the Hercules 370 simulator system on a PC.

At lunchtime I transferred over the DOS distribution tape images and began setting it up to begin generating my running dos image. I may have a problem if any of the software tries to use 2K storage protect keys, an option that went away on newer machines and is not emulated on the P390. It appears there are workarounds that have been developed by people that I can locate on the internet if necessary.

Having set up everything that should have allowed me to boot the DOS/VS distribution tape, format the virtual 3350 disk drive and then restore the disk volume from tape, I started up the P390 system and booted the tape drive but it just sat in "loading" or "stopped" status with a PSW of 000000 and no signs of life on the console. I am not sure what is wrong but will start investigating.

I brought over my VM/370 R6 distribution materials and started installing that. First step is to boot the disk dump restore program from tape, then restore the two disk volumes vmrel6 and cpr6l0. The restore job takes about two minutes on a PC running the Hercules 390 simulator, according to posts form those who have run it, but it is taking forever on my P390.

Finally, the two packs were restored and in the interim I had a clue as to why it was so slow. The x235 is a dual processor machine but OS/2 was only running with one of them online. This shortfall in capacity caused the CPU busy to pin at 99%, with the P390 processor mostly waiting for the x86 processors to do the IO emulation tasks and handle the virtual peripherals.

My CMOS battery is dead, which means it doesn't hold settings while powered down. I think the default configuration it is reaching does not activate the second processor. The new battery should be here in a couple of days, after which it won't suffer setting dementia, but I can take a bit of time to explore the configuration settings at first power-up to try to get both engines chugging away.

My attempt to boot the VM system lead to an invalid PSW loop, which I dimly remember being mentioned as a sign that VM is trying to use 2K storage protect keys, something not supported by P390. I think there is an easy fix for this - time to do some research.

It is indeed the issue of 2K keys and the solution is in how Control Register 0 is set by CP. There are source mods I can use to reassemble the CP supervisor but I have the 'bootstrap' problem right now. Either I temporarily bring over a VM system that already has the mods applied, or I create a new AWSTAPE image for the pack with the modified file included.


I am quite fortunate to have a fellow enthusiast, Richard Stofer, who also has a ztex board and has found a way to get his fpga files to load to the onboard flash and configure on powerup. I am exchanging settings and other information until we figure out what are the differences.

There seems to be some undocumented requirements where certain installs of Xilinx ISE set the parameters in a way that fails with ztex while others set parameters that allow success. Richard had one project that would fail and a different project under the same ISE that works, indicating that the conditions causing the problem are subtle.

Richard ran my project on his system and had a file that loaded to flash and booted automatically at power up time. Something different, but even more subtle than I thought. I have a different installation of the ztex software on the home PC, which I will try to see if that is able to load my fpga bitfile so that it configures at board power-on. I will have to try this tomorrow.


I found a GenRad Bug Hound on ebay for a very low price and bought it. It finally arrived and appears to be working fine. It is a tool to track down shorts, problems with wired-or buses and trace signals; it had been recommended by a reader of the blog, thus when the right price arose I added it to my arsenal of tools. 

Sunday, June 28, 2015

P390 System running several versions of MVS, OS/390 or Z/OS

I spent the day visiting with visitors from NY and Ireland - took everyone to Dim Sum at Rincon Center in SF. Travel was challenging, with a confluence of events today that overloaded transit.

The annual Gay Pride parade, a Grateful Dead concert, and an SF Giants game meant that the Caltrain was so jammed that people could only board at the first 4 or 5 stations, the next 20+ had huge frustrated crowds trying to push onto a train with zero room.

The roads were even worse,  which is why we didn't try to drive in. Originally we were going to head up to arrive 2.5 or 3 hours before our meal reservation, but we switched to mass transit plus a 25 minute walk from the train station to our restaurant.

Finally got back home in the early evening, so not much accomplished in the workshop.


I was able to fire up several versions of the OS, validating that this system is in great shape. MVS 5.2.2 came up fine, as did Z/OS 1.4, but I had a failure during IPL on the OS/390 2.10 system.

My problem with the OS/390 boot was in the OS/2 driver that handles the virtual disk drives (AWSCKD for count-key-data type disks); it failed and disabled all disk access. That is undoubtedly due to an integrity or corruption issue on one of the virtual disk packs. I expect that once I know how to use the utilities to load and examine those PC virtual disk pack files, I can fix this somehow.

I own a couple of cards that would act as a 370 channel, allowing any physical peripheral for MVS mainframes to be hooked to the P390 system. However, these cards are designed to fit into the older IBM style PCs - the Microchannel Architecture bus (MCA) - and not into the x235 or any modern PCI bus oriented machine.

If I can find a card for the PCI machines that acts as a mainframe channel, I would put in into the system. I don't have any devices that require bus and tag channel connections, which makes this a bit theoretical.

The P390 system does provide SCSI connections which might allow my IBM 9 track tape drive to be attached (as a 3420 tape drive as far as MVS is concerned). That wouldn't require the channel card, it could be connected and tested much sooner.

Saturday, June 27, 2015

P390 checks out and runs MVS, replacement op amp commits suicide, 1442 card reader news of the world


While I haven't made any progress on repairing or replacing the punch feed wheels for my card reader/punch, the National Museum of Computing in the UK, which has a restored 1130 system, has had great success in clearing up a problem they have been experiencing for months on their 1442. Congratulations to Peter Vaughan of TNMOC.

The team at the Techworks museum in Binghamton NY are engaged in battle with their 1442 reader/punch, although this one is attached to their IBM 1440 computer rather than an 1130 system. I wish them success as well.


I ran out and bought a new op amp and a few 10uf and 22uf tantalum capacitors, allowing the completion of the replacement of the failing capacitors on the servo board and an attempt to restore the regulated power supplies.

After putting everything back, swapping the suspect op amp and swapping out the 22uf tantalums, I fired up the bench test and saw an essentially perfect -4.98V level. Time to move back to using the main power supply, delivering -20, +20 and +10 raw power, plus hooking in the logic board and both heat-sink mounted sets of power transistors. If I find all four regulated voltages are good and the fuses hold, it will then be time to proceed forward.

Moving forward means that I have to re-install the three crowbar SCRs which will blow a fuse if the regulated voltages go too high. Unfortunately, the +5V crowbar SCR was bad and the replacement SCR I bought on ebay was also bad, so I left that circuit unprotected. Then, I hook up all the cables and boards and repeat the power testing to ensure I now will hold power at the proper levels and the fuse-blowing is far in the past.

Alas, when I fired it up, watching the +5V supply, I saw it soar to about 8V and saw a small wisp of smoke as my replaced op amp committed suicide. Something is still not right. This is not making me happy. I am putting far too much labor into this, given that this was an opportunistic buy and not something that is mainstream for the 1130 project.


My suspicion that the MVS wait state was due to a simple operational issue was confirmed - the wait state I received means that no console was available for the IPL to continue. I took a bit of time today to fire up the system again and try to figure out what I have to do to set up a virtual console terminal that MVS will find during IPL. This will involve starting at least one 3270 session.

When I started a TN3270 session as the P390 was booting, I  had a main console and was able to do a cold start boot and see MVS come fully up. I did some minor operator commands to poke around and convince myself I had a workable system, then quiesced for the day.

This P390 system is a normal PCI based x86 system plus the PCI card for the processor. Unfortunately, that means the parallel channel card I own, which is designed for the IBM MCA bus, can't be used to allow this to hook up via bus and tag cables to physical peripherals. I would need to find a PCI version of the parallel channel board in order to achieve that added functionality. 

Friday, June 26, 2015

Doing well with Pertec power supply, plus checkout of a donated system


I swapped out all but one of the 2.7uf tantalum capacitors on the servo board. One of them is just under a thin wire jumper on the back side, looks like an engineering change, which I don't want to disturb. There are still several 10uf and 22uf tantalums of the same type that I should replace.

I put the removed components back - two transistors, a diode, and another 56 ohm resistor. I want to isolate as much as possible so that I can figure out the source of my problems without the confusion of multiple power supply levels and interconnected boards.

I set up my own -20V power supply and tried to isolate all but the negative power supplies in order to track this problem down and obliterate it. So far, still chasing voltages and validating the behavior. First thing I realized is that I didn't have the power transistors from the heatsinks plugged into the servo board, which is necessary for the board to develop the full negative voltage.

I removed both heatsinks so they could be hooked up as needed on the bench, to continue my isolated testing. With the heatsink for the -10V hooked up, I got an exact -10V as previously determined, but my -5V was down at about a third of a volt (negative). The reference voltage to the op amp is -5.01V, an excellent value, but the other input is our defective -0.373V yet the output is not swinging to a negative value near the low rail (-20V).

My current guess is that the op amp that should steer the overall regulated voltage to -5 is not working properly, based on the lack of a big negative output. However, there are some components that might be causing the symptoms, so I will methodically eliminate them before replacing the op amp. Nothing looks likely, other than the op amp, but checking will proceed carefully since there is a several-day lag for some parts orders to arrive.


I am checking out a donated P390 server - an x86 server (IBM x235) with the P390 processor card that hosts it a S/390 5 MIPS processor that will run MVS or Z/OS on the server. The x86 server has a bad CMOS battery, which is not surprising, but OS/2 came up fine and some initial x-server diagnostics gave the machine a thumbs up. I started the P390 and it did a bunch of 390 processing before it halted with a disabled wait state.

I would need to look further to see why MVS stopped. It could be a startup issue I could solve with a different kind of start, it could be lack of a virtual peripheral or console session . . . I don't know that much about the P390 system yet. However, all signs are good that this is a complete and working historical artifact.

Thursday, June 25, 2015

Bit of work after I return from biz trip - still a draw between Pertec and me

I am back from a multi day trip to the east coast on business. While there, I visited the Techworks in Binghamton (www.ctandi.org) where I met a few of the restorers and looked at their 1440 system with 1442 card reader/punch, 1311 disk drive, and 1403 N1 printer. It is mostly working, just some remaining erratic console printer behavior (yes, another 1053 mechanism with its dreadful old grease causing problems), and issues getting the card reader working.


I swapped about 15 tantalum capacitors out from the servo board, which has all the power regulation circuitry as well, plus replaced the 56 ohm precision resistor for the -5V regulator. It is a very hot day so my tolerance for working with a hot soldering iron and a hot desoldering iron under those conditions is limited - I do a burst of 6 to 10 component swaps then walk inside for a cool-down.

By late afternoon, I attempted a measurement of the voltage levels, now that I have replaced the resistor and cleaned up the capacitors. When I fired it up, the replaced 56 ohm resistor emitted quite a nice stream of smoke and charred itself up. Working hypothesis is that the resistor failure is a result of some other component problem, and shouldn't be replaced until I sort out the underlying cause.

I tried to check the components in the -5V section while mounted on the PCB - not perfect but looking for anomaly not explainable by the circuit schematic. I removed a couple of components to test them out of circuit, but they weren't defective.

Sunday, June 21, 2015

A little puttering on Father's Day before the family assembles


I moved the console printer into its final position - although I have a bit of cable to rethread into the 1131 - the cable for the 1053 is l-o-n-g compared to the distance when the printer is in place. About four or five spare feet have to be stuffed down and tied into place.


Found a bias resistor for the final power transistor for -5V that was damaged and had increased from its nominal 56 ohms (2%) to 2500+ ohms. I don't see how this could cause the +5V circuit to jump to over 8 volts, but it certainly could yield the low voltage for the -5V side. I will replace it when I return. Still looking for a cause for the problem with the +5 supply.


I found something I had bought more than a decade ago but never got around to testing or using. It is a US Navy workstation, I think it was made by Pace, that includes a vacuum pump, twin modulated AC power plugs for soldering irons, a Pace 'dual path' desoldering iron that fits the unit, plus mechanical drive motor and some other near-welding capability. I cleaned out the tip and verified the vacuum works well.

This will help a lot with all those pesky tantalum capacitors I want to swap out - the process of wicking off solder to open up the holes on the PCB was the tedious part. Now, I can take them off and use this desoldering unit to prep them for the replacement easily. Sometimes it pays to clean up back rooms a bit.

Saturday, June 20, 2015

1053 console printer assembled in covers, fully tested, not quite ready. Pertec power supply travails filled my remaining hours


The tab operation is working well now - minor adjustment. Carrier return works almost flawlessly - occasionally it won't trip as it strikes the left margin, instead reaching further left before resetting. If I could tab/space past the left margin position, it would be a pretty benign failure, only impacting left registration of certain print lines. However, the carrier has great difficult spacing past the left margin stop.

I suspect there should be a way to move past it, or to slide the margin over past the carrier, but I haven't figure it out yet. The parts diagrams for each section don't show parts that might be different with rare options - there is one page covering the high speed carrier return feature but it doesn't snow the parts on the carrier itself related to the changed margin tops. The motor power is reconnected to the 1130 itself, so that the typewriter motor comes on when I power up the 1130 computer.

I have the front panel and covers ready for assembly, once I do a software driven test to be sure my selector magnets can correctly trigger the five operational actions - space, tab, carrier return, index and backspace. I will also validate that it shifts upper/lower, shifts between red and black ribbon, and of course that it still types characters correctly.

As I prepared for the test, I decided to reverse the order of my plan, fitting the front panel and then the cover on the typewriter before I did the testing. It just felt better to see it coming together as a complete unit, even if I run the risk of having to open it up for future adjustments.

I found several ways to assemble the two parts - front and cover - in the wrong sequence. I also learned to make sure the column pointer from the carrier is free to move in the front panel before tightening it up. Since the screws go in from the bottom, a very inconvenient angle, it took a few tries. Then, the cover is a bear to get seated properly without hanging up in a position neither on nor off, but unwilling to budge further. A few tries of this and I finally had it together.

I powered up the 1130, hand entered code to make the typewriter type characters or execute various functions, then began putting it through its paces. It did type characters and do spacing operations, as commanded. However, I didn't get a good tab, backspace or carrier return from program command. The front panel buttons work fine for CR, tab and space.

I saw the nylon tape change tension when I issued a "shift ribbon color to red" function, but I didn't see any sign of the counterpart, "shift ribbon color to black" at least in visible tape movement. One of the characters I requested typed as something different - might have been my error setting up the program, or something different. In fact, it could be errors due to my SAC Interface box, which is cabled on the system but was not powered on.

Given the chance that the behavior will change when the SAC testing resumes, I am going to put this down into the recess on the 1131 and install some paper. There is one part missing right now, a spring loaded bar that holds the incoming paper forms, but I know where to look for it.


I have the power supply out of the drive and set up on the bench alongside the servo board which includes the regulators. I will check out as many components as I can in this way, but I will definitely need the switching transistors hooked in. They are installed on big heat sinks on the drive chassis itself, so connecting to them will be an added hassle.

After moving everything close to the chassis and setting various boards on insulating layers, I validated that my +10V and -10V supplies are spot on. The +5V and -5V were essentially off, which I suspect is because they have a 5VSense line back from the using boards that is used to maintain the right voltage level for the logic board. This normally is looped through the plug P210 that delivers power to the logic board, delivering the +5 supply back as +5sense, -5V back to its sense and even a ground sense line looped through the cable.

I temporarily connected those three to their supply equivalents, to permit me to verify the power regulation circuitry on the servo board without actually hooking up the logic and downstream boards yet. I still see fuses blowing and thus can't get reliable +5 and -5V on the board.

The resistance I see on the +5 and -5 outputs is about 2.8K ohms, which would only draw milliamps, so the only way I can get this dramatic suicide of the raw supply is for another of the crowbar circuits to be firing, shorting the input rails. To finish the debugging, I tagged and removed the other two crowbar SCRs, after which I was able to get the power regulators to stay up long enough to figure out what is wrong.

With no crowbars firing, I checked the voltages on TP4, which should be +5V, but was +11V. That certainly would fire the crowbar. The voltage on TP12, which should have been -5V, was hovering at less than one volt. There is definitely something wrong but at least I can now check levels against what I expect and hopefully zero in on the failed component(s). There are less than fifty to check out, since I know that +10V and -10V are correct at TP21 and TP25.

Further, there should definitely be a load on each supply output line, since the servo board uses all four voltages with other parts. Before I dive any deeper, I am going to verify that my sense lines are feeding back the voltages as they should, since these are essential to the output level of the regulators. My current method, having jumper wires pushed into the sockets, doesn't guarantee a solid connection.

The regulation itself is driven by an op amp, fed the +5V sense level on one input and a voltage divider, in parallel with a zener diode and two ordinary diodes, as the other input. The op amp drives a cascade of transistors to drive power at the selected voltage. The -5V regulator has a similar mechanism. The ground sense line is part of the voltage divider for +5V, making it as critical as the other two sense lines.

With the logic board plugged in, ensuring good sense voltage values, I again found the +5V running high at over 8V, but the -5V was near zero still. Further, I am getting fuses blowing in the main supply. How a 10A fuse can blow with a 2K to 3K resistance for a load is beyond me, unless there is another crowbar somewhere causing mischief.

The board has only one other SCR and it is part of the brush motor control circuit, not used for a crowbar. My assumption that the load is almost 3K must be wrong - it may be at the low voltage of the meter I am using, but there are tons of zener diodes in the regulators. Any that fire above the VOM voltage could produce a short in conjunction with another failed component, something I am missing at low voltages.

The main power transistors are mounted on heatsinks on the chassis, connected by cables to the servo board. This allows me to easily test their condition with a parts tester hooked to the cable. All of them appear to be functioning properly.

I will do some continuity testing of all the paths where a short could be in series with a zener diode.
Nothing found by the onset of evening. With Father's Day tomorrow and a business trip beginning early Monday, I will not be getting back to the workshop for quite a few days.

Friday, June 19, 2015

1053 carrier return working well, continued work on Pertec drive


I finished my adjustments and it seems that the carrier return is working perfectly, under motor power as well as with the hand crank. Hurray! The repaired margin rack latch is working fine and the soldered part is not getting much impact since the margin rack movement is buffered by the air cylinder.

I would have been testing with software except that I noticed my tab button is erratic - sometimes it doesn't move the carrier, other times it works fine. Better that I get this sorted out while the mechanism is still out of the covers and accessible. I am agonizingly close to finishing the restoration and putting this together atop the 1131.

This problem could be caused residual gunk on the carrier assembly stopping the tab operation from latching open, it could be a flaw in the mechanism that moves the tab bar to cause that latching, it could be a problem with inadequate spring tension on the mainspring, it could be some kind of drag or interference on the carrier itself, or . . . .? Some of these, such as the residual gunk, would clear out over time with steady use. Others require adjustments.


My 125 replacement capacitors should arrive tomorrow afternoon (Friday), meanwhile I can do some more testing/diagnosis on the boards with the first ten caps changed out. I put it together and fired it up at lunchtime. Alas, the crowbar for the 10V raw supply immediately forced a fuse to blow.

I zoomed in to look at this circuit and adjacent parts. I quickly found the SCR that implemented the crowbar, a 2N3228 device, was defective. It was removed and a replacement ordered, which I should receive on Wednesday while I am away on my business trip. I will be able to put it into place and test everything late next week.

I should be able to run the board with the crowbar SCR missing, as all it does is force a short circuit. I can see that the effective resistance at this point is high enough that it will only draw milliamps, so the only risk I face is if the voltage is considerably above 5V. I did this test quickly and observed that the voltage jumped to over 8.7 volts, also the other crowbar fired due to other voltages also exceeding spec.

Time to dig back through the voltage regulation circuits and diagnose any problems. Starting at transformer, rectification and filtering, where I should be producing +10, +20 and -20V DC, I see I am delivering more than 14V for the 10 supply and almost 23 for the 20V supply. This design requires that the transformer be connected and jumpered for each small granularity of mains voltage. This box is wired for lower voltage, while my outlet is delivering about 122-123V, thus the higher raw output.

It is quite possible this is enough to push the regulators up over their maximum output levels, without any flaw in components. I will rewire the transformer, recheck the raw power levels, and reconnect it to the regulators once I feel the raw supply is working properly. I hate designs that are so finicky they can't deal with the legitimate range of mains voltages that will be found in buildings.

The wiring is now changed. Too, the mail came and I have 125 more tantalum capacitors to swap for the dying ones on these boards. I powered up, found the raw voltages much closer to target, but I am seeing fuses 2 and 3 (+20 and -20 raw supply) blow out when everything is cabled together. Time to back off and test inch by inch through the regulator circuitry.

Wednesday, June 17, 2015

Replaced ten tantalum capacitors, 100+ to go in the Pertec disk drive, plus fighting the Ztex/Xilinx toolchains and designs


When I brought the main logic board in to check, I spotted an exploded tantalum capacitor right at the top of the board. That made my decision for me - replace them all. I have 125 new tantalum capacitors coming in a few days from Digikey, which I will swap for ALL the accursed Mallory tantalums that are bombs waiting to go off.

My earlier order of a few replacement capacitors (and an op amp) will arrive today, so I can at least replace the known exploded component on the board and recheck resistances. I prioritized capacitors in the power supply regulation circuits, particularly those involved in the 'crowbar' circuits that force a short circuit to blow fuses if the voltage levels get too high, thereby protecting other components in the system.

This first round of changes has resolved the problem with the -20V raw supply fuse, but there is still something triggering the +10V raw supply crowbar. I began checking additional components on the board. I had a total of ten replacement capacitors on hand, so that is how many I replaced, one of the logic board and nine on the servo board surrounding the power regulation circuits.

I didn't get a chance to do a power-on test, due to the late hour when I finished, but that will be the way I check to see if I still have problems after this round of replaced capacitors. The challenge is that the crowbar circuit blows the +10V raw power fuse so quickly, I can't tell what might be wrong. My storage oscilloscope might be put to the test if I still have the problem tomorrow.


I did some initial trials to use alternate means of loading the flash on the ztex board so it will boot up when it receives power. However, so far it is not doing any better. I will keep trying different variations until I can make this work as intended.

Tuesday, June 16, 2015

Repaired 1053 part, repaired Pertec drive board but more problems remain


I picked up the 1053 margin rack latch part from Dynasty Jewelers this afternoon, after they used a laser welder to put the broken tang back into place. The craftsman wasn't certain it would stand much abuse, but I think I will try to reinforce the back by epoxying a piece along the tang, spanning the joint, as an insurance policy against breakage.

Bottom view of the part with the repaired tag at top left
Side view of repaired part with the tang and weld on the left
I epoxied a support metal piece behind the repaired tang. The repaired part is in place and I am adjusting the carrier return behavior.

Reinforced tang with black metal and epoxy


The parts appear to be arriving on Thursday afternoon, later than I had expected. I stopped by Anchor Electronics, since the parts were very low cost, picked them up today on my way to the mall to retrieve my repaired console printer part.

I installed the parts on the board, tested continuity of those repairs, put the board back into the drive and replaced the blown fuses. It was time to try to power this up again and attempt a startup sequence. However, even though the direct short I found is repaired, there is something else shorted, probably one or more other capacitors on other boards.

The power plug to the main logic board (J210) drops the resistance of the +5V supply dramatically, suggesting we have a short on the logic board probably due to yet another dastardly tantalum capacitor. There is also a big drop when J203 is plugged in, which connects to positional transducers and the positioner lamp assembly. This could be the incandescent bulb whose cold resistance would be low.

The plug J210 goes to the logic board but also passes through to other PCBs through J107. Yanking the passthru plug out still gives me a low resistance, so I will check out each board at a time. If I find more bad tantalum caps, I will probably buy a bag full of new ones and replace all the originals.


Richard Stofer helped me with a review of the Ztex documentation and site; we thought about how to get a loadable fpga bitstream onto the board so that it powers up with my logic design in place.  I have some steps to take, when I get a chance.


After a reader of the blog had mentioned the GenRad Bug Hound, I found one on eBay although not well described. This is often an opportunity to get a very good price, since potential buyers won't find it with searches. Indeed, this worked for me as I bought it for $5.50.

Monday, June 15, 2015

Jeweler working on 1053 part, plus miscellaneous work on the Nixie/Dekatron clock and the Pertec drive


David Cortesi offered an interesting idea to make the repair on my 1053 part - find a jewelry repair or manufacturer who is comfortable dealing with small parts. I stopped by a local shop to see what they think. One referral to the next, then sent to the Great Mall to find a kiosk in the aisles of the mall, no name given, which has a "$7,500 machine" that will do the weld.

After stopping at the two kiosks that were remotely related to jewelry repair, I was referred to Dynasty Jewelers in the same mall. They believed that their shop could use their laser welder to fix the part using a silver solder. It wouldn't be smooth at the joint but should be strong enough with a bulge of solder around the repaired break. Seems like a good solution to me - I will pick it up tomorrow afternoon.


My supply of 10A fuses have arrived, so that once I replace the capacitor and op amp chip on the board, I can power the drive up and continue diagnosing its operation. I will probably get the replacement parts on Wednesday.


My sockets for Nixie IN-12 tubes have arrived. I began to wire them up, sixty-six wires to and between the sockets. By the time the sun had set, I had four sockets completely wired out of six. I need to do some mechanical work, fastening the nixie sockets onto a carrier board, but the main task remaining is the wiring of the last two sockets.

Sunday, June 14, 2015

Failed component found in Pertec drive, plus spent some time with 1053 and SAC Interface box issues


My thanks once again to Peter and Johannes who offered spare parts from 1053 printers. I am no less appreciative even though my model is different and their machines didn't have the same part. Unless I am extraordinarily lucky and find yet another owner of 1053 devices or someone with stock of this rare part, I have to abandon the hope that I can find a ready-made part as a replacement.

I feel that finding a way to 'fix' or modify the current broken part is still the easiest path. I might be able to drill a hole and fasten something to the bracket, although space is limited. The smaller pivoting piece doesn't impinge on the wee space where I might be able to insert a teeny screw to hold on a bit of metal that will curve down in place of the snapped off piece.

Potential area for a drilled hole to affix a new part
The piece that snapped off was at the right side of the part inside the red oval, and it extended downward out of the picture plane for about 3/16". I may be able to create an L shaped piece that is bolted through a hole I drill in the middle of the red oval, the piece underneath what we can see and the 90 degree bend taking it downward where the original end projected.


Jack Rubin posted a suggestion to use a Genrad Bug Hound, a tool that helps find tiny variations in current flow to pinpoint where a short exists in a situation such as mine. While I don't have that device, I do have an HP Current Probe and Logic Pulser, which together accomplish the same thing. The pulser injects a series of voltage pulses to the shorted circuit, while the current probe lights with intensity related to the amount of current flowing.

There are over sixty components that sit between the +10V rail and ground, any of which could have been the cause of the short. Using the probe, I followed the current and was able to rule out sets of parts when no current flowed in those branches. Further, I found a spot where there was heavy current flow in the ground side, as well as heavy current flow on the +10V side. They triangulated to a 2.7uf filter capacitor (C36), which I then desoldered from the circuit.

The capacitor was a dead short when tested out of circuit. Further, the board itself now had reasonable resistance, over 2K, between +10 and ground. I will order the 2.7uf capacitor and a replacement for the one LM741 op amp that I cut off the board earlier in my debugging.

The capacitor was a tantalum polarized capacitor - these are notorious for failing with a short circuit and often with dramatic results. in this case, the case of the capacitor was barely affected, too subtle to notice on visual inspection. Tektronix oscilloscopes and plug-ins are notoriously prone to tantalum capacitor failures. Unfortunately, IBM uses these on almost all of their SLT cards and Pertec used them for the PCBs in the disk drive.

Once I have the new parts installed, I will put the drive together and go back to testing its startup. Should be just a few days for parts to arrive and I have other things to do, which is why I don't just drive down to Anchor Electronics and pick up the replacements right away. I am still waiting on replacement fuses anyway, which come early next week.


I am trying to load the fpga configuration bitstream in the flash chip onboard, which appears to complete normally. However, when I drop power and reapply it, the fpga fails to load that bitstream. At this point I am going to look carefully at the bitstream I produce, which loads just fine as a direct operation from the USB but fails from flash. This has to be some subtlety which impacts self-load by the fpga or a configuration issue I don't understand.

I read through the entire Ztex site and its associated wiki, but didn't find any help. Several similar problems occured when the USB firmware didn't have support for loading fgpa from flash, but both firmware images do. Other problems occured if the fpga chip wasn't correctly described, but my parameters are correct. Time for more digging and testing, I guess.

Saturday, June 13, 2015

No luck with two different sources offering parts from 1053 consoles; some work on Dekatron/Nixie clock and on Pertec drive


This was a morning of doors slamming shut on the various alphabet of alternate plans for the broken margin release latch for the 1053. I have had extremely generous offers from two separate restorers of 1130/1800 systems, but in each case, the mechanism they had available did not include the high speed carrier return feature, so the part I need was not present.

In the usual Selectric mechanism, the margin rack (which has the movable left and right margin components that determine the begin and end of a print line) is spring loaded to the right by a coiled spring installed at the left side of the margin rack. There is a bracket on the right side which trips a link when the margin rack is pushed leftward into the spring. This bracket is black, short and accommodates only a small movement of the margin rack.

On the high speed return equipped Selectrics, the margin rack moves further. There is an air cylinder low on the left side, which connects to a lever which pivots left and right at the top. It has a spring pulling the lever to the right. The lever fits into a space on the margin rack. The vertical lever from the air cylinder is what pushes the margin rack to the right. The rack moves leftward when the carrier strikes the left margin component on the rack during a return.

The margin rack locks in the leftward position by the latch I am trying to replace. When a carrier return operation is initiated, a rod pulls on the part, thus its name of margin release latch, and that unlatches the margin rack so the spring from the air cylinder lever can push the rack to the right.

Normal Selectrics don't latch the margin rack to the left and therefore don't have any mechanism to release it to the right at the start of a carrier return. I have another IO selectric already, but it has only the normal margin rack mechanisms, no high speed carrier return. The two printers that my peers checked both use the normal rack mechanism.

With several doors shut, there are only a few options remaining:

  1. Find a way to repair my current latch
  2. Find a way to modify the latch or other parts of my machine to function properly
  3. Find a collector with a 1053 having the feature who also is willing to swap my broken part for the good one they own (yes, this is not a very realistic option but included for completeness)
  4. Get a machine shop to build a replacement part, after building an accurate plan
  5. 3D print this in stainless steel after capturing the 3D shape using cameras and software

The margin release latch is really two parts riveted together. The bigger part mounts to the margin rack and has the broken end that trips the CR unlatch pivot. The smaller part pivots on a pin pop-riveted to the bigger part. This smaller part is spring loaded so that it causes the margin rack to latch at its leftmost position, but a rod in the other end of the small part will pull it away and release the rack to move rightward.

I have to make the bigger part, drill out the pivot pin to remove the smaller part, then find a way to rivet or secure a new pivot pin that will hold the small part on the big part. In some ways, options 1 or 2 are much easier because of the complicated dual part nature (option 3 the absolute easiest). Option 4 could include the drilling and riveting of a new pin, whereas option 5 requires me to accomplish those last steps.

I am looking to find microwelders in the area, who might have equipment fine enough to repair the tab. As it is a weekend here, I can't proceed directly on this but I did some research for numbers to call on Monday.


The regulated +10V supply is nearly a dead short with nothing attached to the servo board. I am reading about .6 ohms which is close to the resistance of the probe contact and traces. The +10V supply is delivered to all the op amps on the board, which I suspect is where the problem is occurring.

While I had earlier thought that the markings on U6, a 741 op amp, where the result of a manual soldering, I switched to a working assumption that this chip is indeed dead as a doornail, shorting the +10 supply to ground. To test this, I cut this chip out of the circuit and remeasured the resistance.
Alas, no change.

It will be very tedious to have to find and lift two dozen or so connections of the +10V line until I find the place that is shorting it to ground. Other than the op-amp supply, I don't see any places where single component failures would short the rail down to ground.

I have built a comprehensive list of all the components touched directly by the +10V line on the servo board, which I will use to very closely examine them in the search for the shorting component.


I wired up the Dekatron tube and did a temporary wiring of one of the nixie tubes to allow me to check the operation of the newly assembled control board. The Dekatron is a tube with ten 'dot' positions in a circle around the top of the tube, which were used to display a decimal digit on the tube by lighting the particular dot on that circle. It is a gas discharge tube, like a more complex version of a neon lamp. Nixie tubes have cathodes formed in the shape of each of the ten digits, plus an anode, allowing the tube to light up with the numeral visible.

With some sophisticated control, a Dekatron tube can display a glow on each of the thirty positions around the circle. As originally used, each dot that would be lit had two steering grids on each side, used to steer or attract the gas discharge to make it move as the digit value increased. Thus, twenty of the spots were used for very short durations as the gas discharge moved from one of the ten main positions to the next.

With clever logic and the right sequence of voltage changes, any and all of the thirty positions can be lit, not just the original ten main positions. This clock uses this to display a 'pendulum' pattern as the clock ticks away seconds. Six nixie tubes display the digits for hours, minutes and seconds. Finally, four simple neon bulbs are used to display the colon (:) between hours/minutes and minutes/seconds.

The Dekatron does its pendulum dance, the one nixie tube lights its numeral, and the neon 'colons' dance on and off. My pushbuttons for hour and minute put the clock into setting mode and advance the digits. The one function that is not known to be working correctly is the sound from the speaker, which should be issuing chime sounds and acting as an alarm. Since I can't see all six clock digits until I wire up all the Nixie tubes, I don't know if we got to a chiming interval nor can I see what time I have set for the alarm.

I should design a nice retro look cabinet for the clock, whcn I get some spare time, and after I permanently wire up the six Nixie tubes and finish the testing. This is a nice diversion and will be a cool conversation piece.

Friday, June 12, 2015

Possible spare part source, digging into Pertec drive problem


I received a kind offer from another restorer of old computers, who has a spare 1053 printer that is deemed too far gone to restore to operation. He will check it to see of the margin rack latch is there and has offered to send it to me in that eventuality. There is a definite chance the part won't be on the machine, either because it was already removed, it is broken too, or the 1053 does not have the high speed carrier return feature.

 I will be waiting to hear from him, like a child laying in bed Christmas morning eager to unwrap presents and see what delights await. We shall see.


I saw marks on the board at the pins of IC U6 and at all four points of two electrolytic filtering capacitors C75 and C76. The IC is responsible for detecting and smoothing the head retracted switch signal, but the two capacitors are completely unrelated to the IC. Further, I saw signs of solder spatter on the pins of the IC and along one of the leads of C76.

I had assumed these were the points that emitted smoke, but the lack of correlation was suspicious. Even more so, the two capacitors seem to be a late design addition - not on either of the sets of schematics I have access to, yet all the rest matches up well. The capacitors help the positive and negative power rails so they don't sag much when the draw is instantaneously high.

After quite a bit of thinking and inspection, I know think that these two capacitors were replaced (or changed as part of an engineering change). Further, IC U6 is pin identical to the other signal smoothing op amps yet it is a different part number. I believe, on close inspection, that the hold chip was snipped off and this one was soldered to the leads, again probably part of an EC done in the field. Thus, solder resin left on the board causing the marks and bits of solder spatter due to hand soldering.

I am going to ignore those, treating them as false clues for now, and continue looking for the cause of my smoking. Because the initial smoke production was associated with movement of the emergency unload relay, my first suspect is the rather large electrolytic 'emergency unload' capacitor. If that has failed, it well could have caused the symptoms. However, it doesn't appear to be bad.

Next steps are going through each circuit hanging off the board, validating reasonable levels and testing that transistors, etc are working. I will put a bit of time in on this but at some point soon, it isn't worth that much mucking around compared to getting the main parts of the 1130 system and SAC box working.

Thursday, June 11, 2015

Welder can't fix 1053 part but made some progress investigating problem with Pertec disk drive


I visited the most likely looking welding shop in the area to see if they could repair the margin rack latch part. They felt it was not possible due to the very small size of the broken off part. Will keep looking at my plans B, C and D.


I found that the 25V raw power fuse had blown, a 10A 3AG type, which I picked up and installed in order to diagnose the state of the drive. I discovered that one of the motor plugs was not replaced when I reinstalled the power supply assembly. Perhaps this was the cause of the chain of events leading to a puff of smoke and the fuse blowing.

Installing a new fuse, I again saw smoke but once again couldn't localize it to the exact component(s) before it stopped. The fuses are popping too quickly to spot the smoke source, so I began a more rigorous checkout through the circuit.

Unhooking the logic board doesn't stop the fuse blowing behavior. Since I saw the smoke on the servo board (which also houses the power supply regulators), there is probably something bad on the board. However, it could have been due to a problem in some attached circuit, such as a short circuit that overloaded the parts on the servo board.

If I could see any obviously damaged parts, it would be easier, but everything is seemingly as good as new. I will try to isolate all the circuits that plug into this board, checking them for reasonableness, hoping to locate a primary cause for the smoke.

Wednesday, June 10, 2015

Reverse engineered the disk interface and dug further into the restoration of the drive.


I located a welding shop that I think might be able to fix the part. I will drive by there and see what they can do, but I might not get to their location today. This is plan A, repair of the bracket.

Plan B is to get a replacement part from a fellow restorer, but it has to travel from Finnland. Further, it renders one of his printers incomplete.

Plan C is to pay a prototyping machine shop to make a copy of the part, once it is scanned, measured and converted to a very precise diagram. A pivot for a secondary lever mounted to my latch is pressed on somehow, which introduces extra manufacturing steps in order to get an exact substitute.

Plan D is to drill and tap into parts on the printer in order to create a vertical tab in place of the one that broke off - this is the least aesthetically desirable of the four choices.


The four motor windings are good - no shorts or opens. However, I did notice that the pin in one of the plugs wasn't pushed into the connector all the way, thus backed out when I stuck in an ohmmeter probe. This goes to start winding number one, which if it wasn't securely connected would cause the symptoms I saw.

I have some work to do reinstalling the start capacitors and putting the power section back into the bottom of the drive, before I can test again with the reinserted pin in the motor connector. Meanwhile, I did more reverse-engineering of the interface terminator card to build out an accurate diagram for the two 50 conductor cables that form the external interface to the disk drive.

As with the RK-05 and 2310 drives, these require an external controller to generate the serial data stream that is written on the disk, and that decodes the incoming stream to read the data under the heads. The 2310 logic is embodied in the disk drive adapter in the 1130 system, while DEC schematics describe the controller used with RK-05 drives on PDP systems.

Pertec D3422 drive
These are so similar to my Pertec drive that the only variations are caused by the increased linear bit density and rotational rate. The 2310 spins at 1500 RPM but this drive operates at 2400. The 2310 implements 203 cylinders at 100 tracks per inch (cylinders per inch), but this drive is built at 200 TPI and thus has 406 cylinders. The 2310 records at 1100 bits per inch while the Pertec is packing 2200 bits per linear inch. Finally, the 2310 drive has cartridges built with eight physical sector pulses per rotation, while the Pertec drive delivers 24 physical sector pulses.

The logic for reading streams from the head is the same and the drive delivers a clocking pulse along with the data value, making it easy to retrieve the disk contents as long as the circuit can operate at the higher bit rate, 3.2 times what arrives on a 2310 and 1.6 times what arrives from an RK-05.

I can make this a compatible 2310 drive as far as the software is concerned by restricting the use of the tracks to the first 203 cylinders and by instituting a divide by three function for the sector pulses. I will be wasting a lot of the disk capacity, not only by ignoring half the cylinders, but also by storing only four sectors of 321 words per rotation whereas the Pertec can store more than double the bits per track.

My drive is described in the right-hand column, plus the 2400 rpm option
I waste more than half of the track capacity and half of all the tracks. Net, I will use less than a fourth of the cartridge capacity if I operate this in 2310 compatible mode. If I were to access the native capacity, plus make use of the fixed platter that doubles overall capacity yet again, I would have to write my own software and modify DMS to handle the device.

Continuing on the reverse engineering of the terminator board, I now have the disk to controller cable fully worked out, but it took hours to work through this and be sure it all made sense. By evening, the controller to disk cable was also documented and I felt good about my understanding of the interface overall. Nothing at all on bitsavers or the internet to document the Pertec disk interface. The manuals refer to twin 36 conductor flat cables but the terminator board has two kinds of fifty pin connector sets - one for IDC flat cable connectors and the other for wide circuit board sockets. No 36 conductor choice and the darned manual doesn't show how those work or are connected.

I spent the early evening cleaning and checking out the power circuitry, securing the two motor start capacitors in place and inspecting all the wiring and parts. The goal is to complete reassembly of the lower power section and then test again to see whether the motor is starting or I have to dig into the motor control and other logic.

At the end of the evening, all is reassembled and the motor is not starting. Time to figure out some diagnostic signals to watch in order to track down the reason that startup does not occur. I got to the point where I had to monitor relay K1 - the emergency retract relay - to be sure it was engaging. It didn't seem to pick on the first try, flickered at the end of the start cycle, then when I tried again I saw a puff of smoke and a burned component smell from roughly the area where the relay sits.

Good news and bad news. This will certainly help me narrow in on the component(s) that are bad, but I have to replace the parts whose funeral pyre produced the white smoke.

Tuesday, June 9, 2015

Digging into the Pertec disk drive while sorting out the repair/replacement part for the 1053


No luck finding spare parts to replace the margin rack latch part which I broke by snapping off a tab that ends a carrier return as the carrier reaches the left margin. I do have an acquaintance who has several 1053 printers and has offered the option of getting critical spare parts if I need them. I might need this.

The other option is fixing the part I have. I continue to look at possibilities to repair or replace it. It may be possible to make a 'daughter' part that sits under the existing broken part and provides the vertical tab I need to trip the carrier return unlatch arm.


The documentation is extremely incomplete for this drive. The terminator card does not match any of the documents I have, although the remaining cards seem pretty close. That means I can't figure out the signal assignments of the two 50 conductor flat cables that are used with the drive, nor the method of powering the terminator pull-up resistors.

There is not even a list of the pins on the 100 conductor connector that the terminator boards all plug into - thus I know there are 47 or 48 signals on the interface but no idea which wire corresponds to them. I had to begin reverse engineering the terminator card, but before I could do that I had to know which of the 100 locations on the main connector were each of the interface signals. That information was strewn across dozens of schematics.

Once I had the main connector locations of the input and output connectors, I could begin continuity tracing to find where they are connected on the two 50 conductor IDC connectors on the terminator board. This is quite slow, involving testing 5000 pairs of pins to find the 50 connections.

I traced six of the output signals, found them all to be on the IDC connector J102, then checked a couple of input signals such as the remote start/stop line, which were both on IDC connector J101. I have a working hypothesis that my inputs are on J101 and the outputs are carried on J102. That could reduce the search space in half, if I can eliminate 50 pins on the 'wrong' connector for every search.

I also traced the power to the terminator resistor packs, which pull up with a 133 ohm resistor and pull down to ground with a 266 ohm resistor for each signal line. I found that the power for the pullup comes in pin 1 of J101. The terminator board does receive the +5V power from the main connector, but it only runs to pin 49 of J102, to let the other end of the interface to do termination of the signals we are outputing to that end over J102.

My solution was pretty direct - jumper pin 1 of J101 to pin 49 of J102 which provides the main board +5V to the pullup resistors of the terminator board. That did allow me to initiate a startup sequence when I pushed the Start/Stop button. I am now looking at the next problem to diagnose. I hear a groaning sound but don't see any rotation of the blower/spindle motor. It hums for a few seconds then times out and resets the drives state.

It was time to investigate the mechanical condition of the motor, belts and spindles, to ensure they could turn freely, which they did. I then started to look into why the motor isn't starting. It may be a failure of the starter capacitor, due to age. Time to dig into the schematics and begin collecting data to diagnose and then repair.

I am pleased because this is definite progress. While I am looking mechanically and checkout out the starter capacitor, it would be a good time to clean the fixed disk platter, the platter in my removable cartridge and the four heads. The motor and mechanical items turn easily, so my attention turns to the motor and the control circuitry. Could be starter capacitor, could be starting windings, could be the centrifugal switch in the motor, or could be the logic in the drive malfunctioning. For example, the motor is used as a brake when slowing down to stop, but if the brake logic is active it could stop the motor from starting.

My check of the pair of start capacitors ruled them out. It is dark out now, but tomorrow or the next day, I can check the four windings and the internal thermal switch to see if the motor itself is good.

Monday, June 8, 2015

1053 margin rack disassembled and old lube removed, diagnosed why Pertec drive won't load and go ready


I put out feelers for a spare part to replace the margin rack latch I broke yesterday, meanwhile I will think about how I might repair the one I have if that is my only option. I may be able to epoxy a tube over the remnant end where it broke off - assuming there is enough to affix to. Otherwise, I might have to drill a hole and attach a threaded rod in place of the missing vertical bar.

The 'good news' is that, since I have to remove the latch itself to fix or replace it, I might as well remove the entire margin rack assembly. That let me totally remove the stale old lubricants that might have been impeding its movement. With the parts disassembled, I have everything clean and ready to put back together with only new lubrication.

What a difference in crisp return of the margin rack when released. I also see enough of a stub that I should be able to recreate the missing part of the vertical bar which trips the latch release to end a carrier return.  No guarantee it will work, but worth a try. If I can buy a good replacement for the margin rack latch, that would be ideal.

Another option, since I have the end that broke off, is to see if it can be welded in place. The material is not very ductile, sort of a cast aluminum or pot metal, and the inside where it broke is granular in appearance. Not sure whether a repair can be made by welding but will look into it.

PERTEC DISK DRIVE (5440 style single 14" platter cartridge)

I began checking out the condition of the drive, using my multimeter to check various switches and the power levels. My cartridge presence switch is working, as is the switch that signals the platter cleaning brushes are in their parked position. The power supplies are delivering the four DC voltages needed, +10V, -10V, +5V and -5V, thus I am digging into the startup logic to see why I am not getting a start attempt.

The problem was quickly found - the start/stop signal from the interface is not pulled up by the terminators to a logic high level. None of the input signals were pulled up, they were only pulled down to ground. The + voltage side of the resistor dividers were only connected to the ribbon cables, not to the +5V from the Pertec drive.

Apparently the termination power is supplied by the host system - or in this case by unconnected cables. When I verify that this is what is happening, I will temporarily bridge the high side of the dividers to the +5V of the drive, which should pop the interface signals to their default state.

With the label on the drive, I can tell this D3422-E024-NSC drive is a 200 track per inch, 2400 RPM, top loading (5440 format cartridge) with 24 sectors, using the normal Pertec interface, but I only have the special Diablo compatible interface described. The ebay listing had the number mixed up, which is why my earlier posts listed a 4xxx model number.

All the docs on bitsavers describe the interface cables as 36 conductor flat ribbon cable, yet the cable and connectors are 50 pin. The interface from the terminator board onto the logic board is 100 conductor, which may mean a simple one to one assignment of the 100 lines to 50 lines on each of two cables. I will do continuity testing to figure this out.

When I know the interface and how power is applied to pull up signals, I can move forward to work the machine towards loading the heads. I will go cautiously, ensuring the drive spins, blowers work, and the disk brush sweeps the platters before I let it go far enough to put the heads onto the live surface. In fact, I need to clean the heads, the fixed platter and the platters of the three removable cartridges that I own before any heads are loaded.

Sunday, June 7, 2015

Extremely close, but now I have to fix a part that I broke before this endless restoration of the 1053 is done


I added a knot at the end of the carrier return cord, installed it, put on the mainspring and tensioned it. Next, I installed the escapement cord. Now, with the slight shortening of the CR cord, there is a bit more tension on the cords although the spring loaded tensioner on the side is not pulled in to the desired line yet.

The changes was enough so that even when the carrier slammed to the extreme left due to failure to unlatch the CR mechanism, the cord stayed on both drums. I could unlatch the mechanism manually then resume spacing or tabbing to get the carrier over to the right again.

I can see that the problem is a failure of the carrier to push the margin rack leftward, partly due to the air cylinder keeping the rack from absorbing all the energy. I think the rack isn't moving fast enough due to residual good on the margin rack pins which slide left and right.

Margin rack, right side, top pivot cocks the rack to the right

Left side of margin rack, vertical lever connects to air piston to handle high speed carrier return

Full pivot for the left side of margin rack, piston rod at bottom goes to air cylinder offscreen to right

Air cylinder to handle the deceleration and cushioning of the high speed carrier return option
If it would move easier, it would complete the release of the CR latch. However, if it moves too slowly, the carrier bounces past the rack and continues to the left, without enough margin rack movement to release the CR latch. I will work on further freeing of the rack, plus work on the release so that it triggers even with a partial movement of the margin rack.

I am very, very close at this point, I think. I hope. Things will move quickly when I get the CR to unlatch properly, so this is the top priority. First I worked on clearing as much lube and resistance out of the rack and air cylinder as I could.

It still doesn't pop back when released, instead kind of oozing. This is typical behavior of IBM mechanisms once the lubricants of the sixties became the tar pits of this century. All I can do, unless I totally disassemble the affected parts, dip them in mineral spirits and reassemble, is to work some clean new light lubricant (I use a very light Mobil One synthetic) into all the sliding and pivoting spots, hoping to displace the old gunk through repeated cycling.

Next, I saw that the unlatch isn't triggered until the margin rack is moved to almost its full leftward extreme, something that happens with manual operation using the handcrank wheel, but that doesn't work right when it is returning rapidly under power.

white plastic is pivot for unlatching the CR mechanism, when pushed by the vertical bar to its right

The bracket at the right of the margin rack with its vertical bar
There is an adjustment to move the bracket on the margin rack end to be closer to the fulcrum that releases the CR latch. Even with it at its full extreme, there is a substantial gap between the fulcrum and the margin rack bracket arm that trips it. Since it is a arm bent at a 90 degree angle to the main part of the bracket, I thought I would try to bend it to lessen the gap. Instead, I snapped the arm off! ARRRGGGGHHHH.

Saturday, June 6, 2015

Still working to get the escapement and CR cords on the 1053 exactly right, plus checkout of the Pertec disk drive


I removed the spring on the carrier return torque limiter and replaced it with the correct spare parts spring, since the torque limiter must be adjusted so that it doesn't produce excessive pull, which could have caused the problem I had.

I have begun the tedious job of installing the escapement and carrier return cords on their respective drums and pulley systems. I need to have this finish up with the carrier at the far right, since that is the position to put the initial tension of five turns on the mainspring.

With everything hooked up, I did some hand cranking of CRs and tabs and it all seemed good. Power was applied to the motor, tabs again good but the CR zoomed all the way to the left, popped the escapement cord off the drum again, and hadn't unlatched the CR clutch as it should have.

I only need to reattach the escapement cord to try some more, but it is going to be necessary to figure out what is malfunctioning before I do the next power return or I will be in an infinite loop of escapement cord reattachment.

I did a few rounds of carefully setting up the cords, drums, mainsprings, pulleys and have come to the conclusion that I need to put more tension on the cords by shortening the carrier return cord. Normally, the tension is adjusted by loosening two screws that hold the CR drum in position on the mainspring shaft, rotating the drum in relation to the escapement drum, but I have a problem.

The mainspring shaft is a bit scraped up so that the setscrews and CR drum are no longer adjustable. I will have to shorten the cord instead of rotating the drum to move the attachment point further. The daylight is fading, so this will have to wait until another day.


The drive won't start - no spinning of the motor, blowers etc. I need to get in and debug this, using documentation from a very similar Pertec drive. I think the documentation I have is from the successor version of my drive, very very similar but some differences that are easy to spot.

The drive, opened up and cartridge removed
This may be simple issues, such as the start/stop signal line on the interface appearing to be active since I don't have any terminators on the cable. It could be that one of the microswitches that signal the presence of a drive is not working. I can be a logic board failure as well. At least all the superficial stuff looks good - fuses are intact, boards in place, connectors inserted properly, no wiring out of place.

The removable disk cartridge (5440 type for top loading) but similar to the front loading 2315 used with the 1130
Two unused cartridges, in box, that were inside the cabinet holding my drive

Unused cartridge in the box

Friday, June 5, 2015

Studying how to set up escapement and carrier return cords on 1053, plus newly acquired Pertec cartridge disk drive


I need to think about and research how the escapement cord is affixed to the escapement drum - the knot wedged in a slot just popped free when the carrier slammed into the far left end. This could be caused by the spring I stuck on the torque limiter, it may be caused by the way I installed and wound the cords on the drum, or it could be a misadjusted or malfunctioning part in the air piston that decelerates the carrier at the end of a return.


I found a recycler offering a Pertec 4246 unit which had a single platter top loading cartridge (5440 style) which is an IBM repackaging of similar technology to the 2315 cartridges used in the 1130. There are two 14" platters, one fixed inside the bottom of the spindle and the other contained in the cartridge.

I figured that the drive may have disk heads I could use to restore the 2310 compatible Diablo unit I own. When I got it home, I found it had two new-old-stock 5440 cartridges in boxes inside the cabinet.

I hadn't opened up the unit yet to check the condition of the heads, but with a fixed plus a removable platter, there are four heads and I only need 1-2 good ones. Late in the early evening, I did open it all up. I found I had a good removable cartridge inside, plus the two NOS onces. The disk surface on the removable pack and the fixed platter underneath were scratch free, although they had a coat of dust that must be removed.

Inside, I looked at the disk heads in situ, which doesn't give me quite the access I might have if I removed them, but decent enough to see that there are no visible signs of scapes or smears of brown oxide. This is very promising.

The logic comes up enough to indicate the write protect status (using independent rocker switches for the lower and upper platters) and that it was in 'safe' mode - absence of recognized error conditions. However, no fans were spinning, no blower going and no spindle rotation. This may be a safety feature when the interface is not properly terminated; my two ribbon cables were hanging loose.

The biggest issue is that I have no documentation on the drives, no schematics, and no docs on the interface. A quick peek at bitsavers and a few google searches came up with nothing. 

Thursday, June 4, 2015

Just about done with 1053 console printer

Extremely busy at work and with other obligations, keeping me out of the workshop until midday today. I was able to sneak in a few times and get things moving forward again.


Lunchtime allowed me to use the FE maintenance manual method of installing the cords onto the drums and setting up the mainspring. I did that and installed the motor starting capacitor into place. I knew that the backspace interposer to operational latch spring had to be installed, but expected everything else to work properly.

I got the cords installed, apparently correctly and with the proper spring tension. I tried a series of tab operations which moved across to the right, then a carrier return operation which brought the carrier back to the left. However, something went wrong as it got to the back, because it would no longer tab or space. In fact, I saw that the cords had become unwound somehow.

When I get a chance to go back to work on it, I will try to figure out what went wrong and correct it. I should be right at the end of this restoration, ready to verify the operation, put covers back on and stick it back on the machine

In the late afternoon I snuck out to the garage and saw that the knotted end of the escapement cord had come loose from the escapement drum. I reinstalled the cord, making sure it was firmly in the notch with the knot to the side. I had to get one turn of cord on the drum, then snake the rest around a pulley and a tensioner before hooking the clip onto the side of the carrier.

Once the cables were in place, I had to test the tension on the mainspring while the carrier was at the extreme right of its travel. The pesky spring between the backspace operational latch and interposer popped off the spring hook and is currently hiding inside the mechanism. It is the last part to install before doing my round of testing.

Of course, it is finicky, troublesome and prone to springing off to the abyss of the mechanism, returning to the outside world at some later point after lots of shaking and inversion of the printer (or just randomly an hour later). Finally, I got it installed.

I began testing and adjusting. First up, I checked the operation of the Tab, Space and Return buttons, which work reliably now. However, I had some adjusting to do for the carrier return function. It worked fine when hand cranking, but not as well under motor power.

The high speed carrier return feature moves the carrier very rapidly leftward, with an air piston cushioning the carrier as it strikes the left margin. The mechanism to release the CR latch wasn't adjusted right, but with a bit of work I got it to reliably unlatch once the carrier strikes the left side.

The movement was tepid during CR, until I realized that the spring was not attached to the torque limiter which allows the operational shaft to slip if there is too much resistance to any action. I put on a fairly heavy spring and did get snappy return of the carrier, but my cords are still not set up right. When it moves that rapidly to the left margin, the escapement cord pops off the drum and that lets the mainspring unwind.

I will check to see whether I need a different spring with less tension for the torque limiter, then work on cord installation in a way that is going to be less prone to popping off at the end of a carrier return. A bit of tweaking until that part works right and I can then remove the temporary power cord, rewire it into the 1130 and do some testing under program control.

I was very happy with the performance of the space and tab functions, CR is probably working fine as long as the darned cords stay on the drums. It is feeling very close to a satisfactory conclusion of the restoration. It will be nice to have the console printer in its covers, with the console switches in place, ready to be used.

Tuesday, June 2, 2015

Day of unexpected teeny setbacks lost me a few hours


I have the spring attached between the backspace operational latch and its interposer. The mainspring was wound up but won't fit cleanly on its bracket. It turns out that the mainspring shaft needs to be further forward, threatening the placement of the CR and escapement drums, the cords . . . . ARRRRGH

Once again, removed parts to reassemble the mainspring and drums

CR cord loose on right, spot where mainspring, plate, drum and backspace linkage goes

Loose end of escapement cord in middle, escapement drum lying in bottom in the shadows
I thought that I had tried all the possible combinations of ways these parts could be assembled, but indeed there is at least one more way and I have not yet stumbled on the proper order. I guess doing this over and over builds skills. It definitely is going together faster on the tenth try.

As of the end of the day, I had the two drums, backspace linkage, mainspring plate and some related brackets installed. I have the infamous backspace latch to interposer spring to attach once again, the fun task of winding the two cords onto their drums and a few other tasks yet to do. At least this time, the mainspring will fit properly.

There is a procedure in the FE maintenance manual for replacing the two cords, escapement and carrier return, which pull the carrier and typeball left and right, which I will follow. It sounds like it will be easy, but then again it always sounds easy until the finicky parts, ridiculously small clearances and Rube Goldberg nature of the Selectric slap some reality into me.


The new board works fine, except I haven't gotten it to load the bitstream into flash and successfully power up using it yet. I didn't spend much time because I am obsessively focused on wrapping up the accursed console printer. Still, I can resume my interface testing now.

Monday, June 1, 2015

Getting very close to done with 1053 printer, meanwhile replacement fpga board arrived


The particular area I am trying to reassemble involves about a dozen parts that must intermesh and be assembled in a particular order, not documented anywhere other than in the school of hard knocks as I try, back out, try a different way, and continue ad infinitum.

This is a Rubik's cube of parts including the backspace levers, springs, mainspring plate, mainspring shaft, CR pulley, tab pulley, midspan antisagging bracket, CR and escapement cords, several setscrews, and a number of nearby items that can go awry as the above are manipulated.

I have dislodged the space operational latch spring and had to give up on the search and install a new part from my stock. I am now searching for the backspace latch spring which slipped off my springhook and sailed into oblivion. I do have more stock but I don't want a printer full of loose springs, so I spend a bit of time carefully searching after each spring does its dash for freedom.

I am becoming far more intimately familiar with all the power frame side operational mechanisms that I ever want to. After a bit of recovery time, I went back to my 1053 and resumed reassembly. By mid-afternoon, the carrier return pulley was tightened in place on the mainspring shaft, the operational latches and springs were in place, the mainspring plate was loosely in place, and it was time to get the carrier return and tab cords tightened before I locked the escapement pulley into place.

Getting the two cords tight, the first running over three small pulleys where it could pop out of place, then wound, the second running around the tensioner and then over a small pulley to wind on the escapement drum, while holding the escapement drum from rotating relative to the mainspring shaft and CR drum . . . takes about a dozen hands to manage all of this or just saintly patience, near infinite time and luck.

I do NOT have saintly patience, nor especially good luck, thus all I can do is throw serious chunks of time replaying the ordeal of Sisyphus. Ignore the bursts of blue language wafting out of my workshop.

After dinner, I have the escapement drum and its cord correctly wound, with a wedge holding the tensioner until everything is fully installed. However, the carrier return drum and cord are impossibly balky, since I have to hold the escapement drum so it doesn't turn while I turn the CR drum, but also hold and feed the CR cord and keep it from popping off its three small pulleys. Grrrr.

By the end of the evening I had the drums, cords and almost everything else back together, but the backspace operational latch spring had disappeared. I am actually not upset at all about this, because I have good access, amazingly, to where I need to hook up the spring or its replacement.

What remains to do:

  1. Install the backspace operational latch spring (latch to interposer)
  2. Install the mainspring and wind it to appropriate tension
  3. Install the motor starter capacitor mount and the capacitor
Once I successfully test the various button and programmatic actions of the console printer - tab, backspace, space, index, return, shift, type a character plus the tab, space and return buttons - I can remove my temporary power cord, put the covers back on the console printer and install it into place on the 1131.


To my amazement, my replacement fpga board from Germany arrived today. Quite a bit faster than the first time, although exact same class of service. Tomorrow I can return to testing the SAC Interface Box and its virtual IO devices.


In my reading of the DMS load software, it appears that any peripherals not configured in the load deck will have their 'driver' code skipped during the loading of the system area on the disk. Therefore, my disk which is configured for 1442 reader/punch and 1132 printer will not have the interrupt routines or device handlers for the 2501 and 1403 loaded on disk.

Even if I switch the flags in the main communications area (COMMA), the code itself is missing. That blocks my hope that I could boot up the 1442/1132 disk image, then patch a few areas on disk and memory in order to have it read cards from a 2501 and print lines to a 1403 - both of which would be virtual devices through the SAC Interface Box. 

Spring 20 - me 1, but 1 is all I needed. Continuing with 1053 printer restoration


With some excellent advice from members of the Golfballtypewriter forum, I was able to fit my spring onto the interposer arm. However, I found it was not an exact replacement. It was too long, so that it didn't apply the necessary tension to pull the interposer rearward.

I am working with other springs to see if I find one I have enough confidence in. I also posted a request to the forum to see if anyone has a new-old-stock replacement spring. I can additionally salvage one from another Selectric typewriter, I think. They are relatively easy to buy since most have stopped working due to solidified lubricants.

I made use of one spring, not quite identical but seemed to do the job. I tested it with hand cycling and it now latches each and every time the tab button is pressed. The process of reassembly has begun. The backspace operational latch has a spring that has to fit to the top ear of the backspace interposer - it involves pushing a spring away and looping it over an ear, rather than pulling it toward me.

Fortunately, there is an IBM service tool for this, called a captive hook spring tool, which grips onto the far end of a spring (the circular hook), which allows you to maneuver it onto its holder, after which it can be released from around the spring hook. Thank heavens I bought the specialized tools needed for Selectric work - some of these tasks would be impossible without them.

That spring attached, I could reinstall the mainspring plate, the bracket that holds up the escapement rack at midspan, and set up the various linkages. I have just three things that remain to do before this is reassembled and ready for validation testing.

  1. Attach the carrier return cord pulley on the mainspring shaft with the cord taut
  2. Attach the escapement pulley on the far end of the mainstream shaft with its cord tensioned properly by the self-adjuster
  3. Install and wind the mainspring to its proper tension