Preemptive purchases to continue repair of IBM 1130 +6V supply

FOUND AND PURCHASED SOME ORIGINAL 108 TRANSISTORS FROM EBAY

Two different providers had multiple of the Motorola manufactured 108 transistors. These are high current Germanium PNP transistors that IBM used in all of their SMS and many of their SLT power supplies. One of the suppliers could deliver four in just a few days, which should be sufficient to handle any failed transistors in the +6V supply. 

CIRCUIT BREAKER ORDERED JUST IN CASE IT NEEDS REPLACEMENT

I found a small breaker that I can wire in to the supply replacing the existing switch type breaker, in case the breaker itself is the issue causing the supply to trip out. I should have it by the weekend.

Power testing on IBM 1130 System under restoration, part 2

REPLACED ORIGINAL RELAY IN VOLTAGE TEST BOARD

I had mentioned previously that the voltage testing SMS card had one of its relays removed and replaced with some IC-like substitute. I found the original reed relay in the parts box, tested it and found it worked just fine. I removed the substitute and put back the original on the board. No idea why it was changed, although it was the one to validate +6V, the same rail whose regulator had failed. 

REPAIRING THE 6V REGULATOR

I didn't have a bench supply able to delivery more than 20A of 6V power. The unit installed in the machine as a substitute, a $12 part, wasn't going to be used if I could help it. I pulled the regulator and began to trouble shoot the unit.

This supply uses a set of six power transistors in parallel to handle the +6V current. One of them was dead shorted and causing the regulator to pop its breaker. I removed that one transistor, leaving five that could handle 21A out of the 24A original capacity. This is a short term situation while I acquire a replacement transistor to mount in the supply. 

I set up the supply on the bench, fed it with 9V power and set its output to 6V. I hooked up my electronic load, which is only able to sink up to 5A of current. I don't have large resistors or other means of driving the supply beyond this level, but it did perform well and stay right on voltage through that level of load.

I had to verify all the wiring surrounding the 6V supply. The terminal block was missing a couple of metal bridges that tie pairs of terminals together, pretty critical to the operation of the regulator. With all of them in place, the other ends of all the wires verified, and the regulator back in the machine, I was ready to adjust all the rail voltages.

CHECK REGULATED DC VOLTAGES

With all the raw DC and AC power looking good, it was time to validate the controlled voltages for the SLT logic produced by three regulators. I switched on CBs for +6V, +3V and -3V one by one. After powering up, I checked the levels at the wiring going to the rail connectors on the main logic gates. I was tweaked all of them to the middle of the valid ranges. 

With this done, the next step was to connect the power to the logic circuitry and power up to begin the debugging. I still have an issue with the 7.25VAC supply for the panel lights, but the rest of the machine is good as far as power.

I powered up the system, which then lit the RUN lamp indicating it is executing some code. I flipped the rotary mode switch to Single Step (SS) which extinguished the light as it should. Once I have the lights working I can check to see why the RUN condition is active. There are two relatively innocuous reasons - a hot interrupt or a hot cycle steal (DMA access) request from a peripheral device that is not currently connected. Other defects could cause this too. 

BREAKER TRIPPED AGAIN ON THE +6V REGULATOR

The regulator tripped causing the machine to power down exactly per its design. There are two possibilities here - the breaker may be weak or the regulator may have additional power transistors failing. The breakers on similar supplies used on the IBM 1401 systems do sometimes start tripping well below their target current level, requiring replacement with a proper breaker. 

I can quickly determine whether we had a new transistor failure by resetting the breaker and powering up again. If the system comes up for a short time, then the issue is the breaker. If it trips immediately, we have more repair to do on the regulator itself.

BEGINNING TO INVESTIGATE THE LACK OF PANEL LIGHTING POWER

I put in fuse F7 for the panel lighting power supply, then checked the AC inputs to the supply which should have 115V present. The lines were not energized. I began testing continuity of the wiring back to the sequencing power box which houses the fuses. I ended for the day before I completed this, but with the problem sitting on the AC input side, I am confident I will have this back in operation shortly.

Power tests of the IBM 1130 system under restoration - part 1

REMOVING FUSES, DISCONNECTING TRANSFORMER OUTPUT AND OPENING BREAKERS

In order to carefully step through the checkout stage by stage, I had to disconnect various parts of the power circuitry. There are seven fuses which control power to various power supplies, convenience outlets and the power sequencing transformer that is essential to turn on an 1130.

Each of the voltage regulators has a circuit breaker, which I opened to block them from energizing logic circuits throughout the machine. The +6V substitute regulator does not have a breaker; I disconnected its input instead. 

There are two transformers which were substituted by a prior restorer. One of them is connected only to the convenience outlet inside the machine and its inputs are connected through a fuse. The other, however, is directly connected to input power when the main contactor energizes when the power switch is thrown. It powers various fans throughout the machine which I don't want energized initially, so I have to pull the wires off the secondary to block that transformer. 

With all these disconnections and fuse removals, the machine should not provide power to anything besides the power sequence logic itself. I can then insert fuses, connect wires or switch on breakers selectively to test section by section.

VERIFY ABSENCE OF POWER THEN CHECK 24VAC POWER SEQUENCING VOLTAGE

With the plug inserted into the 240V outlet, no fans should run, the unit should not power on and I shouldn't find AC anywhere outside of the power sequencing box. I verified that this is true. 

Next, I inserted fuse F5 which provides power to the 24VAC transformer. I could see the 24V appear on the sequencing lines and the relay R3 energized which among other things allows AC to flow from transformer T3 to the convenience outlet. 

VERIFY SUBSTITUTE TRANSFORMERS DELIVER 120V

Inserting fuses F3 and F4, while F5 was still in place, energized transformer T3. I checked the voltage at the convenience outlet which was indeed 120V as intended. This gave me some confidence but I still will check the output of transformer T2 to ensure it too is delivering the proper voltage. To test this I had to switch on the power switch of the 1130 system, causing the contactor to energize. 

This was a good time to verify that all the fans began spinning around the machine. However, they were spinning without T2 connected, thus they are wired in some nonstandard way.

ANOMALOUS WIRING AND COMPONENTS DISCOVERED

The fans should be 115V blower motors whose power is fed from the stepdown transformer T2. This allows machines to be used with 115, 208 and 230 volts, the transformer converting the higher voltages down to 115 as a standard input to motors, fans and power supplies.

However, they are spinning when the only power present is 230V. I see they are wired in odd locations on the terminal bus, which is how they get power early, but I am waiting to hear from the prior installer to learn whether these are 230V substitutions or something else was done to make them run. 

When I was checking all the power sequence wiring I came across a replaced relay. Instead of the small relay used on the SMS card for power level checking, a large IC style component was soldered in place. As long as this switches on when the voltage rail is active and switches off when the rail drops, the substitute is okay to keep.

I did look at the specifications of the power regulator for the +6V rail, since a modern substitution was put in place. The regulator used by IBM can deliver up to 24A at 6V, while the modern part has a capacity of 10A on paper and probably less without serious cooling. Even if it provides noise free and stable power at 6V, it lacks an overvoltage protection which is an important feature of the IBM supplies. 

I am not happy with the 6V regulator in the system and will have to address this. Long term, I will repair the IBM regulator and put it back into service, but in the short term I will look for a different high current regulator with overvoltage protection. 

CHECK RAW DC AND AC POWER LEVELS

I hooked up the output of T2 to the power circuitry and once again switched on the 1130 system. The contactor engaged but with fuses F1, F2, F6 and F7 out, there should have been no power to the power supplies. I checked and this is true.

Next, I inserted the fuses one at a time, checking the output of the power supply it energized. F1 provides power for the raw voltages that produce the +3, -3 and +6V. F2 provides power for the 12V and 48V DC supplies for core and relays. F7 energizes the 7.25VAC power for lighting circuits. F6 is power to the various peripheral devices, which I verified at the power connectors for those devices. 

The fuse for the lighting supply was missing from the holder. When I put one in, I did not see the 7.25V appear at the lighting PCB. I will need to track this down and figure out what is happening, then repair it. 

Initial checkout of the second IBM 1130 system

QUICK INSPECTION OF THE SYSTEM

Generally the system looks in good shape. There is a missing blank keycap on the keyboard, the console light panel needs a bit of work to secure the lamp pcb, and some substitutions have been made in the power system of the system. A superficial look at the 1053 console printer highlighted a missing rotate tape and the typeball rotates easily by hand. Otherwise I don't see any damage or missing bits.

New system in front of my working 1130

DEEP CHECK ON CONFIGURING THE POWER

IBM 1130 systems can be set up to run on three voltages - 115V, 230V or 208V. My shop is able to support 230V (240V) so that is how I have to power it. One of the people who dropped it off in my workshop said that it was set up for 208V.

Switching between the levels involves moving jumpers and connections at several points inside the machine. Both 208 and 230V use stepdown transformers to produce 115V as a source for fans, motors and power supplies throughout the machine. IBM made use of big ferroresonant transformers, which have the property that they produce almost flat voltage output in spite of shifts in the input voltage. They are large and require capacitors to be attached.

In this system, someone had replaced the two big transformers with smaller modern transformers. Initially I was worried that these might be 208V only transformers but upon inspection I found they had multiple windings to accept a range of input voltages including 230V. 

Substitutes for the large ferroresonant transformers

I went through all the logic diagrams and other documentation I had to determine how the machine was set up. It was a pleasant surprise to find it was already configured for 230V in all the proper locations. That means I don't have to rewire it for the sake of the restoration and won't have to reset it afterwards to match its permanent home.

I did see that one of the voltage regulators, the one which takes raw 7.8V and produces 6V output for the SLT logic, had been replaced with a modern small switching power supply. The regulator and the overvoltage protection cards were removed and the unit was unwired. Likely this was due to a bad transistor or bad regulator card component. These supplies are very similar to the linear supplies in the IBM 1401 and of course the 360 line of mainframes. I may restore the original and put it back in service, but will start off with the substitute unit just to get through other restoration issues first. 

NEXT STEP - POWER VERIFICATION

I always check the power very carefully before subjecting the logic or core circuits to power. In this case, with modifications having been made, I will check even more thoroughly to be sure that everything is good before I unleash fresh electrons on the SLT cards. 

I worked up a step by step testing plan. The first step in IBM 1130 power is a small transformer that produces 24 VAC power for sequencing. This runs through various emergency power off switches as well as the main power switch on the 1130 console. Once I know that power is good, but everything else is isolated, I can begin to apply power stage by stage.

Next up is the replacement transformers, one of which only feeds a 115V convenience outlet inside the machine and the other which feeds all the power supplies. I will first test the voltage on the convenience outlet, to be sure that the transformers are doing what they promise by stepping my 240V down to 120V. 

The regulators get switched off so that the raw DC and AC power from the supplies does not go anywhere in the machine. I can power each section to verify the +3, -3, +6, +12, +48 and 7.25VAC levels, at least the raw levels, are good. 

When that is done, I can unhook the logic circuits from the regulators, power them up and verify they produce good clean voltage at the proper level. I will focus extra hard on the substitute regulator. 

The IBM 1130 has a series of relays and a main contactor to power up everything. This logic verifies that the main logic voltages (+3, -3 and +6) are good before it brings up the higher voltages. If any of the voltage rails drop it brings down the system in the proper sequence too. A time delay relay keeps the CPU in reset for five seconds, so that power can stabilize before the system is released to begin logical operation. 

Power sequencing circuitry

Second 1130 system arrives, to be restored and returned to its museum

IBM 1130 SYSTEM NEEDING RESTORATION FOR MUSEUM

A museum from the northeast has an IBM 1130 system they acquired that needs restoration, since they want to show it running. I offered to do this in my workshop where I have all the tools, space, parts, manuals and even a second system for comparisons. 

SIDE BY SIDE WITH MY RUNNING 1130 SYSTEM, READY TO START WORK

It was delivered today and rolled into place near my working system and close to where I have the 240V power outlet. I will lay out the ALDs (Automated Logic Diagrams) and other manuals that will be important for this process, using a spare 6 foot folding table. 

FIRST STEPS IN THE RESTORATION

Before I begin debugging any of the logic itself, I think it very important to verify all the power supplies and sequencing logic. Severe overvoltage or incorrect sequence of connecting power can damage the components inside the machine. 

The minus-1 step is to wire up the machine for the voltage levels in my shop. An 1130 can be set up for 120, 240 or 208 volt operation, generally involving moving wires on terminal blocks to choose the primary windings of power transformers. There are quite a few scattered through the machine.

1130 Expander closeup and static validation

FINISHING THE CONSTRUCTION AND EARLY CHECKOUT OF 1130 EXPANDER 

I finished trimming the rubber sleeve and mounting the 160 pin connector assembly on the rear of the chassis of my 1130 Expander box. Mechanically, all is ready to close up but I want to verify all the wiring connections from the Storage Access Channel cable (160 pin connector) to the various screw terminals on my board. 

I went through the signals on the SAC connector, looking at my documentation for which FPGA circuit each should wire to, then beeping continuity to assure it was properly connected. I found one pin whose wire was not connected to the screw terminal, thus I had to hunt a while to find the end of the wire in order to fit it back where it belonged.

Eventually I had all the signal traced and proven. I then checked my auxiliary cable, the one that controls interrupt levels 0 and 1 as well as managing the automated program load sequencer. These five wires were tested through the cable to the remote end where they did connect to the appropriate chip inside the IBM 1130. 

The next steps will be to hook this up to the 1130 system, power up everything and drive it with the PC based GUI program I wrote for the previous implementation of this box. That should allow me to boot up diagnostics, verify the signals, then boot DMS2 with virtual disk drives. 

Meanwhile guests visiting are sharply limiting my time in shop. 

Understand the flash chip process better, restructuring the program

FLASH CHIP MORE COMPLEX THAN IT SEEMED

What I discovered was enough to explain the odd behavior of the chip I was working on. Working from the tables and timing diagrams led me to an oversimplified view of its operation, and thus my code in the Arduino wasn't sufficient to properly control the chip. 

The chip provides automatic programmed operations which includes the ability to chain multiple requests together while the chip stays in its mode, for example sector erase or program (write). In order to get it out of those modes I needed to issue a reset signal which tells the microcontroller in the chip to go to sleep so that it reverts to acting like a read only SRAM. 

Secondly, I thought I could just monitor the RD/BY# signal which shows when it is busy doing some operation like sector erase or program, but in fact I have to do a read of the data lines to get the status of the operation. 

Third, if an operation such as a program (write) fails, the chip stays in write mode and won't respond until it gets a reset sequence.

The automatic operation mode means that within a period of time the chip will treat new input as a continuation request. For example, when I send the codes, five writes of specific patterns to request a sector erase, after I send the sixth that gives the sector address, it will watch for 50 microseconds for additional requests. Those are queued up and once nothing new arrives for 50 us, it processes the entire batch. It appears that writing data bytes (a program operation) can similarly be stacked.

RECODING THE ARDUINO TO HANDLE THE CHIP SUBTLETIES

This means that I had to turn the data lines on PORT E to input while monitoring status, then back to output when sending commands. I also had to flip the OE# to see the results on the data lines. Specific bits have assigned meanings, some for showing an operation is in progress by flipping the value of that bit or inverting the value of the data bit that was written. Others are set when there is an error in sector erase or program (write).

The good point is that I don't have to depend on the RD/BY# line any more, other than checking for consistency when the other status bits indicate that an operation is complete. This is the trailing part of each operation and naturally paces the rate of commands to the chip to when it is ready to act upon them. 

For simplicity sake, I will NOT use the stacked capabilities of erase or program, instead issuing a rese after the status shows completion of the prior request. I will be cycling the chip between modes for every byte programmed or sector erased, slower than otherwise possible but safer. 

WORKING ON ASR-33 TELETYPES FOR A FEW DAYS

I will be diverted to the work on the teletypes for the next few days, although I brought the flash chip programming setup home so that I could do work if I have any idle time.