Thursday, October 31, 2019

Continuing diagnosis and repair of tape load on drive A of my Telex 8020 system

VERIFYING VACUUM AT TAKE-UP REEL HUB

I pulled the hose off the differential pressure switch for the take-up reel hub and held my finger over the end. When the tape threads around the reel I should feel the vacuum suck in on my fingertip slightly. This test determines whether the fault is in the switch or in the vacuum piping.

Vacuum hose on rear of take-up reel motor

Vacuum hold-down holes in take-up reel hub

Tape above on supply reel and take-up reel below
When I powered up and pushed the Load button, the tape loaded and wound around the bottom hub. I could feel the vacuum on the hose end that I had removed from the differential switch. This suggests that the defect is in the switch itself, or the cabling, since it should activate with the vacuum.

DIVING INTO THE SWITCH TO DIAGNOSE THE PROBLEM

The switch closes to connect a signal wire to ground. Other switches on the same holder are properly connecting to ground, thus my problem has to be one of these:
  • Defective switch, won't activate at the vacuum level
  • Blocked switch due to dirt in the passage (variant of the above condition)
  • Wire lacks continuity from the switch back to J03 pin 11 where it enters the control logic
  • Vacuum level too low, sufficient for PEOT switch but not for Hub Vacuum switch
  • Switch activates but vacuum switched off as diverter solenoid changes to run mode
Four vacuum/pressure differential switches - hub vacuum switch on bottom
I put my scope on the switch line and set it to trigger once if the switch activates. After pushing Load and watching the tape try to load, I determined that the vacuum hub switch never closed. It is definite that the load sequence is stalling in Thread waiting for the Hub Vacuum Switch to close as the trigger to move onto Dump state.

In Dump, the diverter solenoid releases to Run condition, putting the vacuum in the tape columns. The feed reel will rotate clockwise to feed in tape while the take-up reel turns counter-clockwise so it too feeds tape into the columns. Once the loops of tape are in the proper spot in the columns, the tape would begin moving slowly forward until the reflective spot comes under the phototransistor. This is the Beginning of Tape (BOT) marker and then load sequence completes.

I am not getting into Dump because of the failing switch. I decided to test the switch with a stronger human powered vacuum. I yanked the vacuum hose from the reel hub end, waited during a load until the tape went around the take-up reel and then sucked in to try to activate the switch.

In spite of my inhaling strongly the switch never registered so the load operation again failed. I noticed that there is a fairly high flow when I suck on the hose which stops if I block the back end of the switch.

Remaining possibilities as I see it:
  • Differential switch is broken inside
  • Connection from switch to backplane connector J3 is broken
  • Something in circuit within the logic cage is holding the signal to +5 in spite of the switch grounding
Next step was to disconnect the plug from the differential switches and pull the board. With that done I could perform a binary search. From the pin for the switch, I should see a path to ground when I inhale on the hose; that means the switch is good and the cabling or logic cage is bad.

Differential switch board with three set for vacuum, one set for pressure

Micro Pneumatic Logic, In. MPL503 part
The switch is clearly bad. I have more resistance inhaling on the hose to other switches, plus they do activate but the suspect switch does not. I removed the board and will take the switch apart to see if it has "user serviceable components inside". If not, I can buy replacements on eBay.

FIXING THE VACUUM SWITCH

With a bad switch, I have four options:
  1. First is to grab a good switch from my drive B; that eliminates the chance I can get the second drive operational. 
  2. Second is to swap the role of the Cartridge Detect and the Hub Vacuum switches, by moving both wiring and hoses; this eliminates the ability to use autoloader cartridges and forces me to manually remove the tape seal band. Assuming the other switch is good, this will allow me to continue further on the load cycle. 
  3. Third is to buy a suitable vacuum operated switch, hook it up to the hose and wire it to the signal line bypassing the bad differential switch; this adds non-original parts to the drive but allows for both autoloading cartridges and a potential working second drive. 
  4. Fourth is to find a replacement of the exact same switch component and install it. 
I removed the switch from the board, which was easier than I initially assumed. I began by desoldering the pins that fit onto the PCB, also removing the screws for the plastic holder on the top side of the board. After removal, I realized that the soldered pins where lugs that slide into the switch body, and the plastic holder remains in place.

The side of the switch had a small screw, which I tried adjusting. I discovered that when rotated fully clockwise, the switch was closed. I then backed it out about a third of a turn from when the switch opened and tested with vacuum. It worked perfectly! Somehow this was misadjusted at the Telex factory when assembled onto the PCB.

I don't have to buy any replacement switches or disable any functions of either drive. Good news indeed. I had to reinstall the board onto the tape drive and insert the four hoses correctly. This took a bit of time because of limited access, but eventually it was done.

TESTING LOAD AFTER REPAIR OF THE VACUUM SWITCH

Pushing Load resulted in the tape starting to turn until it hit the PEOT sensor but not threading into the tape path, immediately signaling a Load Check condition. Time to check my hose connections to the differential pressure board because it seems that one of the switches is operating or failing to operate when it should.

Monitoring the switch that I repaired, I watched the voltmeter on the signal line as I pushed Load. As soon as vacuum came up, but before the tape had even began to advance into the tape path, the switch closed. My setting to was too sensitive now. I need to pop the switch out and adjust it for less sensitivity, iterating until I have it switch on only when the tape wraps round the hub of the take-up reel.

Wow, that switch setting is sensitive. There is a narrow zone where it works. Too sensitive and it triggers when the vacuum comes on. A bit less sensitive and it triggers when the PEOT sensor sees the end of tape early in the load cycle. A magic zone that is about 10 or 15 degrees of rotation of the screw where it works properly, but less sensitive than this it will never see the tape on the hub.

Once I had the Hub Vacuum Switch working properly, the load sequence moved to the 'dump' state where it lowers tape into the vacuum columns in preparation for moving forward to the reflective BOT marker. The diverter value switches off, routing vacuum to the tape columns.

The take-up reel winds counterclockwise allowing the tape to be sucked into the lower column. The supply reel winds clockwise feeding in more tape to the upper column. LED sensors in the tape columns will detect the tape level and stop rotating the reels when the tape is in the proper position.

The load goes wrong in a new way. Now, I hear a weird vacuum and scraping sound, the end of the tape on the takeup reel is completely pulled off the hub and we go into a Load Check condition. My debugging moves on to figuring out what is happening inside the covered tape path with the vacuum columns.

My suspicion at the start is that the load process should wind the tape further on the take-up reel before it begins the dump state. If too little is on the hub, it will slip off during loading of the vacuum column. Time to dig through the maintenance manual and begin scoping signals and timing events.

VACUUM AND AIR PRESSURE PARTIAL SPECIFICATIONS

In the maintenance manual, they indicate that the air blower must produce over 50 inches of pressure and the vacuum pump needs to run at around 30 inches. Test procedures at specific points in the drive, with a tape loaded, so 49+ for blower and 26 inches minimum for the vacuum.

I don't know the CFM requirement for the pumps in order to buy substitutes to power the second tape drive. I can probably estimate from air-holes and the pressure to sort out the probably flow rates. The cross section area of all the holes combined is probably less than .20 square inch.

ROUGH, INEXACT AND SKETCHY CALCULATION OF CFM REQUIREMENTS

That is .0014 square foot thus to get one CFM we need air to move around 733 feet in a minute or 12 feet per second or almost 90 miles per hour. If it flowed at anywhere near this rate we would hear some serious whistling and tooting. From this I expect that any pump or blower with just a few CFM will be adequate.

Monday, October 28, 2019

Diagnosing failure in tape load operation in Telex 8020 drive cabinet A

AUTOLOADING CARTRIDGE AND TAPE CARTRIDGE OPENER

IBM introduced a feature on its tape drives, during the 360/370 era, where the operator only had to push the closed tape cartridge onto the drive and hit the Load button. The drive would open the seal belt around the outside of the tape and thread the tape automatically. My drives have that feature as well, but I am unable to test it because I don't have the right kind of tapes.

The normal seal is a band that clamps shut around the perimeter of the tape to keep out dust, but is not designed to work with the automatic openers. A different band, called either the wrap-around or autoloader cartridge, is instead placed on the perimeter of the reel of tape. With this band on the tape, it can be opened by the drive.

All the tapes I own have the standard seal on them. I have dim memories of having to take new tapes out of the box, back in the mainframe days, take off their normal band and put on the autoloader types before putting the new tapes into the tape library. I am seeking such a band but so far can't find any. Without that, I won't be able to test out that part of the loading function of my drives, instead having to remove the seal and stick the tape reel on sans band.

REVIEWING LOAD SEQUENCE

At the point that the load fails, it has powered up the motor for the vacuum pump and air blower, begun rotating the feed reel slowly counter clockwise, and energized the diverter solenoid. It then waits for the PEOT sensor to indicate that the tape is present. After a watchdog timer goes off without any PEOT signal, the logic records a tape load check condition and stops.

There are a number of possible errors that could cause this:
  • Diverter valve not operating to switch vacuum
  • Insufficient vacuum or air leak
  • Inadequate blower volume or air leak
  • PEOT sensor opening clogged
  • Component error on board responsible for sensing PEOT and advancing load state machine
DEBUGGING THE SIGNALS

My first action was to watch the diverter valve to see if it operated. My VOM showed the operating voltage applied to the coil and I felt air coming out of the loading shoe. Additionally, I could feel vacuum at the PEOT sense port. The shaft in the solenoid moved fully and promptly.

This narrows the problems down a bit. The diverter is working and I have some vacuum and air pressure. Time to move the meter to the PEOT switch and see if that is tripped. I can see this on J3 on the motherboard, pin 15.

I then realized that the vacuum front plate wasn't closed properly. It was snapped shut at the top but there is a second snap at the bottom. Without that closed firmly, the vacuum wasn't able to pull the tape where it had to go. With the door shut properly, the drive would almost complete loading.

Notice that the vacuum front plate is not on my list of possible causes. This is always a possibility, thus debugging requires flexibility and adaptability if you want to succeed.

The tape was threaded through and onto the takeup reel, but then it just stopped. This may be caused by the previously noted problem with the Load Point light dimly lighting. If the drive believes it found a load point, that would stop the seek operation.

I also noticed that it won't unload from this point. Pushing unload does nothing, while it should rewind the tape and then unload it completely.

I watched the BOT and EOT lines (beginning of tape or load point; end of tape) and they were properly low while tape was in the column. They are not detecting a reflective tape spot. These should trigger a clear logic 0 or 1 level on the motherboard, driving the indicator with a logic 0 to light the bulb.

I watched the output that drives the control panel light and it stayed at logic high, meaning the bulb should not illuminate. I don't know why it is partially lit but it is not confusing the load logic.

The load process seems to stop at the point where it should enter the 'dump' state where the two reels move to lower tape into the columns now that the vacuum has switched back to run state from threading state. I found a pin where I can watch to see if the state machine advances to 'dump'.

It does not. Vacuum drops after the tape winds on the take-up reel but before it tries to lower the tape into the vacuum columns. The state machine never advanced to this point. I will now check a vacuum switch that should indicate that the tape has wound onto the take-up reel. If that doesn't occur within the timer limit, the drive would give symptoms like I am seeing.

After hooking the VOM up to the switch at J03 pin 14 it is definitive. The switch is not closing. This is hooked with a hose to the take-up reel hub which has three holes that are hooked to the vacuum line. When the tape winds around the hub, it blocks the holes, producing the vacuum that should trip the switch on.

My action plan tomorrow is to verify that vacuum is applied to the hub, then verify that we see a vacuum when tape covers the hub. If the vacuum is felt then, the problem is in the switch itself.

Start testing logic and power amplifiers in drive A

INITIAL SHORT TEST

I did a quick check of the power supply connectors to see if any of the loads are shorted, using the VOM to measure resistance. All the results were reasonable except for the 12VAC line which appeared to be an open circuit. There may be a relay that connects this line only at certain times; investigation is needed.

BORROWING POWER SUPPLY AND VACUUM/BLOWER FROM CABINET B

Since I am still waiting for the fuse holder for the cabinet A power supply, I swapped the power cables over from the A drive to the B cabinet power supply. Too, I moved the vacuum and pressure hoses over to the A drive, giving me a complete unit to test.

The hose lines and power cables are all a bit short, so that I could barely get them connected across cabinets. My first attempt looped the hoses over the mid level frame bar, which wouldn't permit the two drives to be pushed together.

The power cables required that the cabinets be abutting, thus I had to redo the hose connections. I completed all the connections and slid the drive cabinets together. Everything fit together with no slack but not excessively taut.

FIRST POWER UP

Power up was done with no circuit breakers tripping or fuses blowing. The operator control panel had the Power, File Protect and Load Point lights illuminated. Load Point was weakly lit, unlike the others. When I pushed the Load/Rewind button, vacuum power came on, the automatic cartridge opener activated and the drive was trying to thread tape into the vacuum columns. After a short interval, Load Check illuminated which is expected.

When I push the Unload button, the power window on the front door lowers. Hitting Reset or Load/Rewind will raise it again. This is as expected.
Power window in front door lowered
CLEANING TRANSPORT PATH IN PREPARATION FOR A TAPE TEST

I opened up the vacuum chamber on the drive and cleaned it thoroughly using Isopropyl Alcohol and KimWipes. A lot of dust had coated the walls, faces and other parts including the tape head. It had to be removed before I would attempt to load tape.

Vacuum columns, head and tape transport path
LOADING THE FIRST TAPE

I couldn't figure out how to get the tape cartridge inserted with the tape seal in place, so I removed the seal and put the tape on the drive. Hitting Load/Rewind starts the vacuum and begins slowly rotating the tape reel counter-clockwise. This continued for some time until the time-out when the control logic declared a Load Check condition.

The control logic for the drive sequences through state machine steps to load the tape. The first step is to power on the vacuum and blowers as well as the +45/-45V supply. This also activates the automatic cartridge opener which should open the wrap-around seal and lift it to give access to the tape.

The second step rotates the tape counterclockwise until the end of the tape closes a vacuum switch that indicates the end of tape was located. This should advance the machine to the next steps where it will turn the tape reel clockwise and feed the tape down through the vacuum columns, tape heads and eventually onto the take-up reel.

However, the drive never advances. This will require some diagnosis to determine if the flaw is in the vacuum sensor or the control logic board. Very fortunately, I found a binder in the garage with the maintenance manual and other helpful documents. Just in time, too!

RECAP OF PROBLEMS DISCOVERED

  1. Load Point weakly lit at power up
  2. Fails to load tape

Sunday, October 27, 2019

More work on the operator control panel of drive B

REPLACING BROKEN DENSITY BUTTON

I installed a mini pushbutton switch on stiff leads into the PCB, after removing the broken switch. Simple and easy fix. This hides behind the operator control panel in normal operation thus I didn't need to find the same or cosmetically similar switch.

Replaced Density pushbutton switch on the operator control panel PCB
VERIFYING THE PUSHBUTTON LOGIC WORKS

First step was to beep out the connections between J1 that connects the drive motherboard and J2 which connects to the pushbuttons and lights on the panel. It was important to verify that I knew what was connected directly and what was driven by the flip flop that recorded the Online status of the drive.

I then did some beeping out of wiring on the backplane to identify places to read switch state and to command light activation. I wanted pins well enough separated for easy access with the probes and other wires.

The sequence of testing planned is:
  1. verify that the 'start state' line on J4 pin 3 was high when I powered up
  2. verify that the Reset button connected ground and J4 pin 4 when pushed
  3. verify that the Unload button connected ground and J4 pin 5 when pushed
  4. verify that the Load/Rewind button connected ground and J4 pin 7 when pushed
  5. verify that pushing the Start button changed J4 pin 3 to low
  6. verify that pushing Unload doesn't connect ground to J4 pin 5
  7. verify that pushing Load/Rewind doesn't connect ground to J4 pin 7
  8. verify that SW1 toggled to offline and back to online causes J4 pin 3 to go high again
  9. verify that the Power lamp is always illuminated
  10. verify that grounding J4 pin 17 turns on the Select light
  11. verify that grounding J4 pin 11 turns on the Load Point light
  12. verify that grounding J4 pin 9 turns on the Load Check light
  13. verify that grounding J4 pin 15 turns on the Ready light
  14. verify that grounding J4 pin 10 turns on the End of Tape light
  15. verify that grounding J4 pin 12 turns on the File Protect light
  16. verify that pushing the Density switch on the PCB will connect ground and J4 pin 1
Status of first set of tests:
  1. Failed - no signal seen on pin 3
  2. Success
  3. Failed - no connectivity
  4. Failed - no connectivity
  5. Failed - no signal on pin 3
  6. Failed - no connectivity
  7. Failed - no connectivity
  8. Failed - no connectivity
  9. Success
  10. Success
  11. Success
  12. Success
  13. Failed - no light
  14. Success
  15. Success
  16. Success

Moving a VOM directly to the flip flop I did see that the Start push-button turned it HIGH and SW1 toggled to offline toggled it LOW, so indirectly I confirmed tests 5 and 8. However, something is still wrong as the signal isn't reaching J4 on the motherboard via the cable from J1 on the operator control panel PCB.

Starting testing of logic and amplifiers in drive B

SCHEMATICS OUT OF SYNC WITH THIS DRIVE

The book of schematics I own are a match for cabinet A, both the drive and the integrated control unit, but I found noticeable differences in the cabinet B boards. If I have to debug or repair any of the boards that differ from the diagrams, it is going to force me to compare and partially reverse engineer.

One example was the operator control board, which had some additional ICs and a hand executed engineering change that involved cutting board traces and a jumper wire. That one was a simple board and not many changes but still took 30 minutes of tracing before I was sure I had found all the changes.

SHORT TESTING OF LOGIC CAGE BEFORE APPLYING POWER

After inserting all cards in the logic cage, I used the VOM to test for any short circuits before attempting to energize the power supply. With that successful I could move forward to fire up the system.

VERIFICATION OF DRIVE SPEED CONFIGURATION

Boards 7, 9, 10, and 11 have configuration differences depending on whether the drive is to operate at 75 inch per second or the full speed at 125 ips. I verified that boards 9, 10 and 11 were properly set. However, board 7 has a problem.

On board 7 the specific part values for the configuration are all encapsulated in a DIP 16 package that plugs into the U27 spot on the PCB. This provides six resistors and two capacitors to the rest of the circuitry on the board, which is the pre-amplifier that controls the capstan motor. The package is missing, thus my board can't work as it sits.

I have the proper values and wiring required, thus I can build the parts onto a DIP socket and install this onto the board to complete its configuration. I won't be able to have tape move forward or back, nor read nor write, until this is completed.

INITIAL POWER ON AND OBSERVATIONS

With the boards installed and everything wired up , I applied main power to the drive power supply. Many of the PCBs in the logic cage have small red LEDs which illuminated and the fans for the +45 and -45V supply came on. No fuses blew or breakers tripped.

The operator control panel had the Power and File Protect lights illuminated. No button presses changed anything. I believe that I will need to place a tape on the drive before the logic will attempt to load - firing up the vacuum/blower motor and rotating the reels trying to thread the tape end into the vacuum column and through to the take-up reel.

CLEANING TAPE PATH

I chose to clean the tape path at this point, using lab low-lint wipes and 99% isopropyl alcohol. There was a lot of dust picked up but finally I am ready to try to fit on a tape reel and test out the loading sequence.

Converting the Telex drives to different speeds

DOWNGRADE MODELS AND PRICING STRATEGIES FOR MAINFRAME PERIPHERALS

Mainframe makers had long pursued a strategy to build a peripheral device with multiple speeds of operation and optional features, allowing them to win business with more price sensitive customers while protecting the prices and margins for those who valued the highest performance and added features.

The device was designed for its top performance and features, then methods are developed to slow it down. This can be by changing the size of a pulley on a mechanism, or adding delay logic. An advantage of having one design with slowed models is that a field upgrade can be offered, allowing a customer to pay the market price difference and receive the benefits of the faster model.

A sizeable price difference may be implemented by a pulley or jumper, with the cost of parts changed way out of proportion to the price changed. It is most visible in processors with slowed speeds, where the CE key, used to change metering of hours used between the customer and the service meter, also increased the machine to its full performance.

With the IBM 3420 tape drives, IBM offered models with three different speeds - 75, 125 and 200 inches per second. The Telex 8020, a competitive drive that was program and plug compatible with the 3420, offered models at 75 and 125 ips.

In looking through the schematics for the tape drive, I found that the same PCB design would support both speeds, differing only in the value of some components on the board. If my drive was a lower performance model, I could easily modify the machine to give myself a field upgrade.

Of course, if my drive was already set up to run at 125 ips, there is no need for changes. I pulled a board (the capstan drive pre-amplifier) from slot 7 and looked for the parts that vary based on speed. I discovered that Telex had been a bit clever, making it harder for a customer to upgrade the drive themselves.
Board 7 schematic where parts values determine speed of the drive
The differences in speed for this board are set by the values of six resistors and two capacitors, as shown in this chart on the bottom of the schematic page. I looked over the board to find parts with numbers like R302 and C302 but quickly realized that they were all installed inside a pseudo integrated circuit U27 that had a DIP 16 mounting but was a big encapsulated block with an obscure part number.

Table of parts values to set capstan tape speed

Pseudo-IC containing the discrete parts
Using the pin numbers on the schematic, I checked the values of the embedded components and found that my drive A was the full 125 ips version. If it had not been, I would need to remove the DIP 16 block and wire up discrete components of the proper value on a DIP socket to plug into the board.

I pulled board 07 from cabinet B to check which speed it implemented. To my surprise, the pseudo-IC has been removed from the board. I can see that the board has a flush level socket into which the IC 'plugs' instead of being soldered. The good news is that I know the values and can build a DIP 16 socket with the parts.

Flush socket with removed pseudo-IC on drive B board 07
This cabinet had been destroyed by a forklift through the window of the front door and subsequent fall that was hard enough to bend the frame and break some welds. Typically in those cases the insurance carrier pays the owner or vendor and the machine is supposed to be scrapped.

The missing chip may be because the normal installation procedure at the customer site would have had the CE plug in the appropriate chip to set the speeds, but it is shipped sans IC. However, this board appears beat up and used, indicating that the chip was removed deliberately. Telex may instruct their CEs to remove these modules because they can be used to upgrade other tape drives in the field - before release the unit for scrapping.

Nothing else appears to be removed, although I need to look more carefully at all my boards just in case this was stripped for some parts before it was turned over to the scrapper. I definitely need to restore the drive in cabinet A first, as it is the most likely to have had a full complement of working parts.

Saturday, October 26, 2019

Start repair of operator control panel & finish repair of dump card from cabinet B of Telex 8020 tape system

CONDITION OF CONTROL PANEL

The panel has a number of wires dangling loose and I wasn't sure that everything was even still there. This drive took such major trauma I assume that when it was hit with the forklift tine to cause the front damage, it knocked the unit over backwards strongly enough to twist the frame and jar things loose.
Damaged operator control panel
RESTORATION OF WIRING AND REPAIR OF PARTS

My first step was to take the schematics and write down the wiring points for the pushbuttons that are disconnected - which plug pin they are hooked to and whether the wire is on the normally open, normally closed or pole side of the button. I also have disconnected wires that I think go to the indicator lamps.

I did continuity testing and identified all the wires and switch positions to which they belong. It was a quick task to solder on all the wires and verify their operation. The two lamp connections are more problematic. These lamp holders have a top tab that represents one side of the bulb and two side tabs for the other contact of the lamp. On the first two lamp holders, the top tabs have broken off clean at the level of the case.

I can see a touch of metal where the tab broke off on each housing; I should be able to insert a wire and solder it to the remaining metal or at least make a good friction fitting. I made a first stab but didn't get a good solder connection onto the tab that was broken off flush. With a bit more work, however, contact was restored.

I was not getting reasonable results either looking for the switches to give me continuity when pressed nor for the lamps to light (except for the Power On which is hardwired to light). I decided to plug in the main power supply to the unpopulated card cage and test with the plug on the rear that is connected by ribbon cable to the operator panel.

After some study of the card it became obvious that my schematics do not match the card. There are two additional ICs on the board, an open collector hex inverter and a dual JK flip flop. Only two of the six inverters are wired and only one of the flip-flops. Clock and D input were wired to ground, so this operated solely but using the notSet and notClear inputs. For some reason, the notQ output is sent through one of the inverters however the output of that inverter is not connected to anything.

I also found signs that this board was reworked to change its function. Two traces were deliberately cut and a jumper wire was soldered elsewhere on the back. It is clear that I would have to reverse engineer the card to figure out the schematic before I can do more debugging.

After I beeped and traced out the circuits, I realized that the one of the cut traces was done to wire the flipflop and inverter into the circuit, breaking the straight connection of Start pushbutton to output. I would have expected to have the flipflop on the main control logic board down in the card cage, but I suspect that the rework may give more flexibility in resetting the drive.

The other cut was for pin 2 - the gate to turn on the +45 and -45 power supply - which was originally directly connected to ground on this PCB. The result would have been immediate powerup of the power supply, but I believe that this rework ensured that the drive could intelligently power up and down rather than always driving the 45V supply.

I checked for continuity of the +5V supply to the VCC pins of the TTL chips, but the circuit wasn't connected! I found that the choke coil was poorly soldered to the board. That was quickly repaired. I then tested the various light driver signals, grounding each gate signal to verify the lamp illuminated.

Testing the switches will be a bit more complicated since some of the signals are now driven by the flipflop that is set by the Start button and reset by the RESET button or when the switch S1 on the PCB is turned to the offline position. That means I can't directly check the Load/Rewind or the Unload buttons, these are only active when the Start flipflop is energized. I can and did verify the basic behavior where the button wiring enters the PCB but can't check the outputs of the PCB right now.

The pushbutton on the back for Density (S2) is partly broken, but I can can wire in a new pushbutton switch to replace it. I am assuming that the drive will normally select density automatically upon reading and picks the highest density by default to write; the button must be to override this behavior.

MOUNTING THE PANEL

The reason for this damage is the failure and loss of the top bar across the frame, caused by the traumatic bending damage it sustained. That bar mounted hinges that were attached to the bottom of the operator panel. It tilts up for service and rotates down to position the buttons and lights facing forward.

I need to create a replacement mounting for the panel of some sort. The more significant decision facing me is whether to repair the frame first. If the drive had the vacuum pump/blower assembly, I could conceptually restore it to full operation, however that doesn't exist. If I gamble that I can find pumps and blowers that match the flow requirements of the drive, restoration is possible.

To repair the frame would entail removing all the covers and all the interior parts. It is not all that arduous a task. I probably would need some kind of clamps and a hydraulic piston to attempt to straighten this. Alternatively I could take the stripped down frame to a automobile body shop where they do this kind of repair all the time. They could also weld on a replacement top bar.

The downside is expense. Frankly I don't have a use for these drives so the restoration is simply a hobby effort, after which I will find a home for the working drives. I have about $150 sunk into them so far including the replacement parts I bought. I am not sure I want to step up to the expense of the substitute vacuum/blower plus the frame repair work.

DUMP CARD DIAGNOSIS AND REPAIR

The plug-in board for the cabinet B power supply, the "Dump circuit", drives the shunting of +45 and -45 power supplies through big 300W resistors. Based on a suggestion by a blog reader and after some study of the schematics, it appears that the dump circuit is used to sink energy from the reel motors when they are slowing down from a high speed rewind, a form of dynamic braking.

With my dump card installed, the +12V power supply output went to zero. It returns when the card is pulled out. The card itself is not very complicated. It uses two op-amps to compare the 45V levels against the 12V levels. Since the +12 and -12 supplies are regulated using zener diodes, they are used as a reference standard.

Appropriate resistor dividers take the +12V and the +45V  down to a common voltage when both are on spec - thus the inputs to the op amp are balanced and it does not drive any output voltage. If the +45 level increases above nominal, the difference is amplified by the op-amp, with its output further amplified by a transistor on the card. There is a second circuit comparing -45 and -12, a mirror image of the first circuit, thus the card has a +dump and -dump output plus inputs of +12, -12, +45 and -45.

Dump card with the shorted capacitor clipped out ot the circuit
I found that the +12V input contact was shorted to ground, which certainly matches the symptoms observed. Looking at the schematic, the most obvious source of the problem would be a short in the electrolytic capacitor sitting between +12 and ground. I clipped off one lead and indeed, that capacitor is a dead short. The contact itself is no longer shorted with that capacitor removed from the circuit.
Our culprit
I found a suitable replacement for the 15uf, 20V part, soldered it onto the board, and tested my cabinet B supply with the dump card installed. +12V is working exactly as intended. Whether dump is working properly really can't be tested until I get the whole drive working and can put it into high speed rewind then monitor the 45V rails.

Final repair of power supply in Telex 8020 tape drive cabinet A, except for replacement fuse holder

INSTALLING THE REPLACEMENT RECTIFIER DIODES

I ordered the diodes from Newark, early in the day on the 22nd. Although the website had indicated that it would be shipped that day, USPS Priority Mail, tracking the assigned number  proved that it wasn't handed over to the post office until 9PM on 10/23. That delays delivery from the expected Thursday until Friday. Sellers are quick to fill out electronic pre-shipment and try to claim that as the mailing date, but they can dawdle in actually handing off the packages believing this is hidden.

As a backup I ordered two more diodes from a "Fast and Free" ebay seller whose promise was Monday the 28th. I had a credit on ebay from a past purchase with a guaranteed delivery date that wasn't met, so it was cheap insurance for the Newark order. Once I saw the shipment handed in the need for backup lessened, but that didn't come until late in the day past the promised ship date from Newark.

Once I had the diodes in hand, I soldered them into the circuit and put them back on the heat sink. My first test was to fire up the full wave bridge, sans capacitors or any other load, just to verify that the rectifiers are working properly.

The diode bridge worked, then everything still okay with the capacitors in circuit. Finally I put in the fuses and checked the outputs at the J5 connector. I didn't see +45, which turned out to be a broken cartridge fuse holder!. It had cracked during the removal of the enclosure from the tape drive cabinet.

The supply only need a replacement panel mount fuse holder to be put back into service. Everything works perfectly at this point. The darned holder is over $27, since it is a 600V, 30A spec holder for FQN time delay fuse cartridges. I should install it by mid next week, then the enclosure will be put into place and testing can move forward to active electronics throughout the drive.

FURTHER INVESTIGATION OF VACUUM/BLOWER MOTOR CONTROL CIRCUIT

Having proven that the vacuum/blower motor jack appears to have 240VAC even when off, I tested this by plugging the actual motor to the jack and then toggling the control pin 12 of J4 to verify that the motor goes on and off. Everything worked well.

The triac has a high resistor and 1uf capacitor installed across the two main terminals, which causes the 240 to appear on the socket but no power is delivered until the triac fires, gated by the control signal.

CLOSE UP AND FINAL TEST OF ALL VOLTAGES

I fired this up as a closed unit, verifying the +45, -45, +12, -12, +5, +5 for operator control panel, +6.4, unregulated +8, and the 12VAC outputs were correct. It passed all tests, with the control pins 13 and 12 of J4 switching the AC for the 45 V supply and vacuum/blower motor.

Tuesday, October 22, 2019

Evaluating restoration of integrated control unit from Telex 8020 cabinet A

ROLE OF THE CONTROL UNIT

The integrated control unit in this Telex 8020 tape subsystem is a plug compatible alternative to the IBM 3803 which requires its own cabinet. The control unit front end attaches to an IBM mainframe using a pair of hefty parallel cables called Bus and Tag. This control unit causes the mainframe to believe there is a string of IBM 3420 tape drives available; it responds to read, write, control and sense commands just as the 3803 would.

This control unit back end attaches to each of the tape drives in the string, using the simple Telex protocol with the drives. The parallel cable used for the connection has lines for the 9 data bits read from the drive, 9 data bits to be written to the drive, plus status and control lines. It can command these actions and more from the drives:
  • Read a record
  • Write a record
  • Read one record backwards
  • Space forward one record
  • Write a tape mark
  • Rewind
  • Rewind and Unload
LIMITS ON TESTING

I will be testing each drive individually with the standalone Telex Tape Tester, a device which plugs into the drive and controls the signal lines that normally would come from the tape control unit. Thus I will know if the drive is correct as far as movement commands before I get the control unit working.

What I don't have is a mainframe or channel tester that I could use to drive the control unit to verify that the front end works properly. What I would need to do is send various channel commands to the controller and read both data and the sense (status) data from the control unit. For this to work, the control unit must properly respond to the signal protocol on the tag cable and transfer the correct data or sense bits on the bus cable.

I may be able to produce test sequences of pulses using my Analog Discovery and capture the resulting pulses from the control unit on my logic analyzer, for simple sequences at least. This could partially test the control unit.

CONTROL UNIT POWER SUPPLY

The wiring diagrams show that the control unit is fed 240V from two legs of the 3 phase master input, just like the power supplies in cabinets A and B. However, I couldn't find any large socket on the control unit for power. It appeared to be completely featureless on the outside, only having ribbon cables coming out to the bus/tag connectors and up to the two tape drive card cages.

I finally spotted a tiny US style male socket recessed up on the rear of the control unit,which must be the power input. Totally unlike the connectors for the supplies, so it must draw far, far less current in order to use the widespread 15A connectors.

Begin restore of power supply in Telex 8020 drive A

DIFFERENCES NOTED IN POWER SUPPLIES

Visually it is different from the other unit, having no space on the front for plug-in boards, using fuses instead of circuit breakers for some power levels, and employing a bat handle toggle switch along with a circuit breaker for main power control.

Front of new power supply
It does have the same sockets as the other enclosure - J4, J5, J6, J7 and J8 - and undoubtedly does the same things just with a different cabinet arrangement. AND - what a difference internally. This is packed with almost no spare room, with all the PCBs designed and built by Datapro/Telex. The first one I did had two power supply units they sourced externally and simply integrated into the enclosure.

View inside from the top

I can see that the regulated power supply is a printed circuit board just behind the front cover. The AC switch control logic is a small PCB on the bottom, but with no card cage. Similarly, the dump function is implemented internally without a plug in card. The triacs are mounted on a mid level block, while power transistors are spread across blocks on the right and on the front PCB.

Low voltage supplies board, behind front panel
The two 300W resistors are mounted vertically along the right edge and the huge transformer is stuffed in behind them. With the compactness of this arrangement, I will take more time to isolate sections and test them, with key parts of the wiring buried beneath all the other parts.

Top cover
Fortunately, I believe I have the schematic pages that fit these units. The major downside of this enclosure is that there is no labeling on the various terminal blocks, unlike with the first unit. I have to study how they seem to be wired and match what I see to schematics until I can definitively label them TB1, etc.

REMOVAL AND INSPECTION

This enclosure is mounted down at the very bottom of the cabinet, in front of the cage where the IBM bus and tag cables are connected. This is unlike the supply in the B cabinet which sits on a shelf about 18-20" above the bottom level.

I unscrewed the socket head bolts and then fought to slide it out the side of the cabinet and then to haul it up to the table top level. I believe each of these weights about 125 pounds and is unwieldy to boot.

Nothing appears damaged or out of place. I had to disassemble layers to figure out what parts are where and how the wiring runs, in order to isolate each section for independent testing. It is much more tightly integrated - no separate PS1 and PS2 components, instead one board that produces +12, -12, +5, +6.4, 12VAC and +8 unregulated.

LOW VOLTAGES SUPPLY TESTING

I fired this up and found that all low voltages are working properly. This part is good to go. Proper voltages delivered for +12, -12, +5, +6.4, +8 unregulated (about 11.5) and 12VAC.

HV SUPPLY TESTING

This is controlled by grounding pin 13 of J4, just like the other power supply, but there is also a bat handle switch here that blocks the power to the high voltage supply transformer. When I switch on the toggle switch, but don't ground pin 13, no 45V power is available.

When I use the jumper to pin 13, but leave the toggle switch, there is still no 45V. This is the expected behavior. When I flip on the switch the main circuit breaker instantly pops! Something is shorting the high voltage supply, probably a bad filter capacitor or other failed component. This is going to take some work to debug.

The circuit is pretty straightforward. The power flows through the transformer, through a full wave rectifier bridge and then into two huge filter capacitors. Large fuses sit between the filter capacitors and the rest of the circuitry - mainly the shunt regulation since nothing is plugged into the power supply.

My next test was removal of the fuses to isolate the transformer, rectifier and capacitors. I also removed the capacitors from the circuit by taking the wire off one lug. Triggering power with pin 13 and then flipping the switch, I will either see 45V unfiltered or the breaker will still pop.

With the fuses out, it still popped. I then removed the capacitors and it still popped the CB. That leaves just four rectifiers and the transformer. Removing the transformer leads from the diodes, the circuit breaker was happy, no tripping.

I removed all the wiring from the rectifier terminal block (also has the two transistors that shunt current to the 300W resistors), then quickly ascertained that I had a dead short across at least one of the 1N1186A diodes. This is a 40A, 200V stud mount rectifier. I began unwiring and unmounting them for the final check.

A full wave rectifier is built from four diodes, usually drawn in a diamond shape. At two opposite apexes, the AC input from the transformer is connected. At the other two opposite apexes, the + and - rectified current appears. The two bad diodes were both connected to one of the AC inputs, but failed differently. One of them failed open circuit, the other in a short.

I ordered the replacement diodes and expect them to arrive later this week, at which time I will install them and continue with the power supply checkout. There may have been a sizeable short, well more than 40A, to cause one of them to short out. Interestingly, the lines from the rectifier and capacitor are fused with 30A cartridges, leading me to suspect the filter capacitor as the short that took out the diodes.

Using my ohm-meter and then an ESR meter, I saw reasonable readings from both of the large filter capacitors. The ohmmeter charged up the capacitor slowly and then I could read the charge in millivolts, when the capacitor is good. My ESR meter measured less than .01 which is fine for the usage here where the power dissipated across that resistance would be about 3 or 4 watts at full draw of the supply and the voltage drop about a quarter of a volt.

The big decision I had was whether to wait for the original equipment rectifier diodes to arrive at the end of the week or whether to convert this to use of a modern rectifier bridge. I have an HBPC5010 component on hand that supports higher current and much higher peak inverse voltage, but in a small package. However, I chose authenticity since it will only be a few days and there is plenty of other work to do on this drive.

AC SWITCH PCB TESTING

Given the breaker trip I get when I jumper the first triac, I know this part of the circuit is working. It is delivering 240V to the toggle switch and through that to the large transformer for the high voltage supply.
Triacs and related components
The vacuum/blower motor jack J6 appears to have 240 across it, just like on the other supply, regardless of whether the control line is grounded or not. At this point I have to wonder if the issue I had is due to a misunderstanding. Either the control line needs to be pulled up to 5V to turn off the outlet, or perhaps something else is going on that I am missing.

AC switch control board, below triacs
Nothing in the schematic differentiates the vacuum/blower circuit from the hi voltage supply circuit, they use the same design and thus should be switched on by grounding the appropriate pin. I decided to plug in the blower motor and see what happens. I also should look carefully through the power supply board to be sure it isn't pulling the line down on its own.

I find it highly suspicious that both cabinets have the same apparent failure, leaving the vacuum/blower motor powered continuously. I reviewed the circuitry that switches the HV and motor power, assuring myself that they are absolutely symmetrical in design and should operate the same.

The AC Control Switch board generates +5V to power the operator control panel but it uses it internally through opto-isolated triacs. The LED in those triacs on the board is driven by the +5V supply and routed out as control signals. When the control signal is pulled down to ground, the LED lights and triggers the triac.

The board also generates 12VAC separately for each of the two triacs. That 12V is switched by the opto-isolated triac to provide an AC drive signal off the board and out to the main triacs that actually switch the AC power. The main triac gate is fed the 12V AC only if the control signal was pulled to ground, the LED lit and fired the on-board triac, and then that triac feeds 12V.

I finally decided that my meter might be sensing 240V due to the capacitor network across the triac but if it isn't on, there wouldn't be any power to drive the motor. I hooked up the blower to my repaired power supply in cabinet B, provided 240 to the power supply and turned on the circuit breaker. No motor running. I think grounded J4 pin 12, the control pin, and the motor turned right on. In fact, this was working correctly on the cabinet B supply and probably works well on the cabinet A one too.

Connecting vacuum/blower in cabinet B to drive in cabinet A

REASON FOR THIS WORK

My cabinet A is in good physical condition, at least it hasn't suffered the major damage found in cabinet B. However, only cabinet B has a vacuum pump and air blower. My intent is to wire up the motor from cabinet B to the power supply in A, then route the hoses over to the fittings on cabinet A.

This should give me a fully functioning drive in cabinet A. If I could resolve the challenge of sourcing or building an equivalent vacuum pump and air blower, then it is theoretically possible to also get the drive in B working.

Damaged wiring in rear of operator control panel

Forklift through glass door and into the upper right part of drive, broken cartridge autoloader

Frame quite bent from forklift accident

However, there are substantial challenges in cabinet B - bent frame, bent parts, cracked parts for the tape cartridge autoloader, and a very damaged operator control panel. I believe I can live with the bent frame and can fix the operator control panel. It is still not clear whether critical mechanical transport parts have been bent or damaged. The autoloader could possibly be bypassed coupled with simple repairs.

Properly wired operator control panel from cabinet A

Undamaged cartridge autoloader and panel from cabinet A
THREE HOSES CONNECT TO MANIFOLD AND REEL MOTOR

The air blower provides cooling air to the take up and feed reel motors, air to the manifold and vacuum to the manifold. These three hoses were connected on the cabinet B drive unit, but all I need to do is swap them over to the cabinet A drive unit. Then, with the long power cable from the blower motor plugged into the power supply in cabinet A, it will complete the drive in that cabinet.

Good drive A, with fittings for three hoses

Vacuum pump and blower in cabinet B with hoses that reach to drive in A

Final repair of Telex 8020 drive B power supply

MISSING +12V SUPPLY

I was in error believing that the -12V supply was not working - I apparently checked the wrong pin. I couldn't see any common parts other than the AC transformer and it was working properly. However, the +12V supply is the only one malfunctioning.

The supply uses an LM723 regulator IC with support circuitry to control current, set the regulation point, and to protect against overvoltage with a crowbar circuit using an SCR. The two obvious candidates for the failure are the regulator chip or the SCR. I removed the supply from the enclosure to let me bench test it more easily.

The only complication was the current configuration which was set for 208V input. It was just easier to take it out to the garage and do my debugging there.

Lo and behold, with the supply disconnected, all voltages were working. I connected it and again the +12 disappeared. I finally traced this to the Dump card and transistor circuit that will shunt off power through the huge 300W resistors whenever the +45 and/or -45 went too high.

With my reconfiguration of the main supply power transformer, I was getting essentially spot on voltages with my household line levels. I don't expect them to surge and therefore I pulled the Dump card and will run this way will I look into the cause of the problem, It is either on the card itself or one of the two transistors which are mounted on the rear of the card cage. Both of these can be serviced with the supply enclosure bolted back in the drive.

TRIAC SWITCH FOR VACUUM/BLOWER MOTOR STUCK ON

Using the diode mode of my multimeter, I see that the suspect bad triac has no diode action between the gate and MT1 terminal, unlike the working triac with has diode action both from g-MT1 and g-MT2. A replacement triac is already on order.

ALTERNATIVE TO REPAIRING TRIAC SWITCH

However, I may not have to repair this part of the power supply, because I have a second such triac switch in the other power supply, the one mounted in the base of drive cabinet A. If that one works, I can simply use it and ignore this one. After all, I only have a single vacuum/blower assembly and will only be using one of the two drives.

REINSTALLATION INTO DRIVE B

With the +12V working, sans Dump card, and postponing the motor triac function decision until I test the other cabinet's power supply,  and everything else working perfectly, I closed it up and installed it back into the drive B cabinet.

Sunday, October 20, 2019

Power supply testing results for Telex 8020 tape drive system - almost ready

SCHEMATICS DON'T MATCH TERMINAL BLOCK WIRING

The schematics show the wiring in the power supply across four terminal blocks - TB1, TB2, TB3 and TB4. Some of the wiring does match, for example TB2 which is behind the PCB cage next to the triacs is hooked up the same way, but others seemed wide of the mark.

For example, I checked some points on the terminal blocks that should be connected to the neutral and to the hot lines, but they aren't hooked there. The wiring on TB4 doesn't seem to be connected to the 45V power supply transformer at all, but it should be. That is where the connections are swapped to set the input voltage from 110 up to 230 volts.

I resorted to a time consuming process of disconnecting various wires and beeping out continuity in order to build up a map of where each of the terminal block lugs actually connected. This took abouit an hour to complete. It was sped up by the realization that most of the wiring did indeed match the schematics.

The big error was on a schematic showing now the large transformer was wired up - the supply for +45, -45V - which was erroneously drawn as hooked to TB1. Once I realized that this was false and that TB1 was wired for an entirely different purpose, the output of the hi voltage supply (ground, +45 and -45), I could move on to sort out the actual transformer wiring.

No terminal block is involved in wiring the input of the high voltage transformer. Instead, six wires are plugged onto lugs on the transformer body, those wires running through harnesses to other locations. These were hooked to the cooling fan, the neutral power bus, the output of the triac which switches on power to this supply, and formed a hidden jumper function.

Two wires are to the cooling fan - terminals 2 and 5 - which is a 115V unit. Those terminals are part of an autotransformer that drops the input voltage (220 in my case) down to 115. Terminals 4 and 8 had separate wires but they are shorted together somewhere remotely, thus they form the jumper that is needed to configure these windings for 208, 220 or 230V.

Terminal 2 is connected to the neutral leg - actually just one of the legs of the 3 phase when this is connected to the 3 phase distribution panel, but as I am wiring it to household single phase 220, it is just one hot that I chose to call neutral because it is not switched.

The switched power from the triac was wired to terminal 10, which is the proper winding for 208V operation. I moved it over to terminal 9, thus converting the transformer to 220V input voltage.

BRING UP AFTER INITIAL EXAM OF ELECTROLYTIC CAPACITORS

I performed some quick tests of the electrolytic capacitors that I could isolate from the circuits, both for capacitance and for ESR values. I was trying to rule out obvious problems before I energized the circuit. The two capacitors appeared to be in excellent condition.

Next I considered hooking up a variac which will allow me to vary the voltage from extremely low levels up to the 220V target input. By beginning softly, I would limit the current inrush to capacitors, giving the surface time to 'heal' if there are any tiny shorts due to long periods of disuse. An alternative method would be to put limiting resistors in line with the capacitors and gradually lower the resistance, but a variac would be the quickest way to deal with the bringup.

However, the variac I own is a 110V unit, not 220V. In order to use this variac I would have to temporarily rewire each supply from 220 to 115V configuration then use the variac. Once satisfied that the capacitors are not going to die, I would rewire and run them at full voltage. Not worth the effort given how clean everything looks, so I will just watch carefully as I test each unit.

Each of the separate supplies - 45V, PS1, and PS2 - were powered up and the output voltages checked under partial load. I used some power resistors to load down the different supplies while I checked the resulting voltage levels and ripple with a scope.

TESTING PS2 - +6.4 AND unregulated 8V SUPPLY

I set up a plug for the 220V (actually 240 as I measure it) of my home supply, which normally has attached the fast charger for the Sonata Plug In Hybrid. I had a spare plug which I wired to the connector from the tape drive. That connector was inserted into the rear of the power supply, my plug was inserted in the extension cord to the home 240 supply.

I had isolated the main circuit breaker allowing only the PS2 supply to be energized. When I switched it on, I measured the unloaded output levels as 6.55 on J8 and 11.2V on J7, which is close enough particularly for the unregulated output.

Last step was to install resistors to put the supply under load, within its specs of course, and measure the output voltages again. With a small load (100 ohm resistor) I only had 65ma draw and under a half watt of consumption. I was still over voltage by about 2.3% but I don't think that is critical at this point so I moved on to test the next supply.

TESTING PS1 - +12V, -12V, +5V AND 12VAC SUPPLY

I removed PS2 from the circuit breaker and reattached the rest of the supply circuitry. Both plug in boards were moved from the card cage. The Dump card will short the 45V supply if the 12V sensor supply levels drop, not something I want to have happen. The AC Control card will switch on the triacs that energize the 45V supply and the vacuum/blower motor, which I am not yet ready to do.

My first test was to leave PS1 disconnected and turn on the circuit breaker, verifying that the fan doesn't run and all is otherwise good. Once that was verified, I could insert PS1 in the circuit and get ready to test its levels.

This supply delivers four different levels but all is accessible from the front panel plug J4. I was quite satisfied with this supply. The +12, -12 and +5 circuits were only a trace over the exact value, something I attribute to the operation of the sense lines. The 12VAC was solid as well.

Supplies with sense circuitry allow attachment of a second thin wire to the load, paralleling the heavy distribution wire that will carry all the load current. There is some voltage drop on the distribution line, varying as the square of the current draw, which makes the voltages different between the output and the sense pin at the supply.

The circuitry then boosts the voltage of the output slightly, compensating for the line loss, attempting to get the sense wire at the target voltage by delivering enough more on the output pin to cover the wire loss. Since I didn't have the sense wire added, I believe that the minor voltage boost was being applied, so that my 5V was reading at 5.03V.

TESTING +45 AND -45V MAIN SUPPLY

I bypassed the triac to directly power the high voltage supply. The fan came right on and I measured +52 and -52V at the outputs. Since this isn't a regulated supply, those didn't seem bad. The "Dump" PCB drives a shunt of the excess voltage across huge 300W resistors, a very wasteful power supply design. Since I don't have the dump card in place, the voltages will be high.

Two 300W resistors running left to right near the top of the pitcure
I rewired the transformer to its highest input setting (230V) in an attempt to minimize the raw voltage of the power supply, since this will lower the wasted power consumption across the resistors. The result is just a hair over 45V, just what I want to see. Ah, the days before switching power supplies, when power waste was the norm.

TESTING AC SUPPLY CONTROL CIRCUIT

Once I had the power supplies operating, it was time to verify that the AC supply control circuit and triacs were working. This plug in card has two control inputs. Grounding the input will fire the associated triac and provide 220V AC to the circuit it is switching. One switches the vacuum/blower AC motor on, the other feeds 220V to the 45V power supply..

What I found was that the high voltage supply was switched off, but the vacuum/blower motor had full power applied. I need to investigate this, as the blower should also be off. The card is supposed to deliver a  low capacity +5v supply, which is also used internally with the opto-isolated drivers for the triacs. I don't measure any supply, which is a problem.

With no supply, neither triac should be firing but I definitely have supply going to the vacuum/blower circuit. The internal 5V supply runs through a photo-diode and a resistor to the control signal line. These are not wired to anything but need to sink current to ground in order to fire the photodiode and turn on the triac.

Turns out I forgot to plug in two wires that on the side of the card, inside the card cage, to deliver power. With that installed, the +5V was working great and when I grounded the control line for the triac switching on the 45 volt supply, it fired right up.

The problem remains that my other triac is always firing, thus the vacuum/blower motor will be running continuously but should only run when the control line is grounded. Even when I remove the AC control plug in board, the triac is firing. It appears there something wrong on the back of the card cage.

Based on my diagnosis, but without actually testing the failing part, the only part that could be bad is the triac itself. I bought a new one and should have it on hand in about a week, after which I will finish the repair of the power supply. The leads have to be desoldered and the new one soldered in place.

REASSEMBLY AND FULL INTEGRATION TEST

Finally it was time to completely reassemble the power supply and all its parts, then feed power to the main input. Each and every power level was again verified to be present and correct. Lastly,  I cycled the 45V which depended on external control signals, by grounding the related pins on the connectors of the power supply. I had the 'dump' card (actually the voltage regulator card) installed and checked that the +45 and -45 seemed goo.

Suddenly while completing the final tests of the voltages, I found that both +12 and -12 had dropped out. All others are working fine, but this adds a second fault to be addressed once I have the triac in hand.

Saturday, October 19, 2019

Beginning restoration of Telex 8020-2X 9 track tape drives (IBM 3420 compatible)

TELEX 8020-2X DRIVE AND IBM 3420 COMPETITOR

Some time ago I bought two cabinets worth of a 3420 compatible tape system for about $100. The first cabinet looked clean and unused, plus it had an integrated controller for attachment to IBM bus and tag channels. The second cabinet had suffered a forklift or other transit accident, with the frame bent into a parallelogram outline rather than a rectangle, but most of the gear inside appeared undamaged and unused.
Telex 8020-2X A cabinet
I believe this was an insurance write-off, after which it was to be scrapped but somehow wound up stored untouched in a overpacked warehouse until I bought it. It is clear, after the fact, that there was at least one other drive. I dimly remember the seller mentioning having sold some of the drives to other people.

The IBM equivalent consists of one cabinet per tape drive plus a cabinet solely to house the controller electronics (3803). The Telex series made use of a simpler controller that was fit into the first drive of a string, obviating the floorspace need for a standalone controller such as the 3803.

These tape drives have long vacuum columns to buffer tape movement, supporting operation at up to 125 inch per second in stop-start mode. To support the columns, the drive has both a vacuum pump and an air blower, since during loading and other operations it is necessary to blow on one end of a tape while sucking on the other side.

When I first looked at the drives, I believed that I would need to have 3 phase AC power since the main power cable required the customer to provide 3 phase 208V AC. However, once I began to look through the schematics, it became clear that no part of the machine needed three phase. They simply divided the loads of two tape drives and a controller across single phase 230V by wiring different pairs of the three input phases to each load. A+B, A+C, B+C.

I COULD REPAIR AND USE THIS ON REGULAR HOME 220V SINGLE PHASE POWER!

Power supply with two plug in boards removed
In the Telex design, a single AC motor is connected by a belt to both the vacuum and blower components. This takes up about a foot of height inside a cabinet and runs the full width. The space required to hold the integrated controller in the first cabinet (drive A) doesn't leave enough space for the vacuum/blower assembly. It also doesn't allow room for the 3 phase main power cable and distribution box.

The solution was to make up a special second cabinet (B) which had two vacuum/blower assemblies one atop the other, plus an AC power distribution panel that would in turn supply power to each drive. Additional drives beyond the first two contained only a single vacuum/blower assembly, also without the main power distribution box.

Sadly, what I have is an A cabinet and an ordinary B cabinet. That means I have no AC distribution and only a single vacuum/blower but two complete tape drives. It appears the special B cabinet was sold to another person.

I CAN ONLY USE ONE OF THE TWO TAPE DRIVES!

Unless I am very lucky and can find a spare vacuum/blower assembly, or can create my own sources of vacuum and blown air at adequate flow rates, only one of the two tape drives can be operated. The other will serve as spare parts. I still need the second cabinet simply to hold the vacuum/blower in place but it will not operate.

Blower and vacuum pump with motor barely visible in between
The main AC distribution panel sends the single phase power (pairs of the three 3-phase incoming leads) to each drive, which has a power supply inside. The panel also delivers a single phase pair to the integrated controller that is inside my A cabinet. Its sole other contributions are hosting a 110V convenience outlet via a step down transformer, and hosting the 3 phase circuit breaker for the entire tape string. I don't have this panel, it was in the missing special B cabinet, but I don't need it. I can provide single phase power to both the working tape drive power supply and the integrated controller.

I have removed a power supply from the cabinet and will begin to bench test this. It switches the vacuum/blower motor on, delivers 5V for the digital logic, delivers _+45V and -45V for the DC reel and capstan motors, and other voltages such as 12V for miscellaneous purposes. Once I have this working properly, I can begin powering up and testing different parts of the machine.

Power supply box

Inside the box is an enormous transformer and impressive filter capacitors. Step one is to verify the condition of the capacitors before applying power to the first section of this power supply. The output of the biggest transformer, rectified and filtered, is the +45 and -45V needed for the various DC motors. This large transformer is switched on by a triac in the AC switching board.

Transformer for +45, -45V supply
Mains input and rest of 45V supply
AC switching controls the vacuum/blower motor and power to the 45V supply. It has a triac to act as electrical switches for those motors. Its other triac delivers power to the "hi voltage" supply which exists to deliver +45 and -45 for use by the servo logic and drive motors.

Triacs to switch vacuum/blower motor and power to 45V supply

The small wide PCB (PS1) is the lower voltage board, producing +12, -12, 12VAC and 5V for distribution to the rest of the drive. That lower voltage board has several filtering capacitors that also will be validated before I apply power. It powers up immediately with the drive, unlike the 45V supply that is switched by a logic signal.

PS1 delivering +5, +12, -12 and 12VAC
A final supply exists to deliver +8V and -8V, hidden in an add-on enclosure on the main power supply box. This box (PS2), like the PS1 board, is on whenever the drive is switched on.

Misc power levels provided by this supply
Fortunately, the various circuits can be isolated and tested individually. The control logic for the AC switching is a plug-in card to the face of the power supply, with the triacs themselves mounted on heat sinks behind the card sockets. Adjoining that plug-in board is another, which monitors +12, -12, -45 and +45 volt supplies to generate a 'dump' command to unload the drive if power has dropped.

Mismatch between schematics and actual circuitry

I was examining the various components to locate everything relative to the schematics. I did this so I could decouple the different power supply subunits and test them one by one. I quickly found that the plug-in board for the AC power supply control was substantially different than the one in my schematics.

The schematic has only a single four way diode bridge in place and one filter capacitor to feed the +5V output, making use of photo-coupled triacs fed by a CA3079 chip that detects zero-crossings of the AC input and times pulses to fire triacs.

Schematic of plug in board
The actual board has many more diode bridges, two more filter capacitors, and a welter of additional components including transistors. It replaced the photo-coupled traics with optically isolated switches driven by the additional circuitry.

Actual board I have in the drive
The good news is that it still switches two power sources - AC vacuum motor and 45V servo power supply - using the same pins. I have to assume it is electrically identical at the interface and test it that way. The output of the board fires heavy duty triacs elsewhere in the tape drive to accomplish the actual switching.

The schematic lists the date of the drawing as Oct 1980 while my board has a testing date sticker of July 1980. It seems that I have an older board which was replaced by a new simpler and lower cost design in late 1980 or early 1981.

Remainder of circuitry matches my schematics!

The dump plug-in card is identical to schematics, as is the PS1 and PS2 supplies that provide 12, 8 and 5V levels. The high voltage (45V) circuits and all the other wiring also matches the schematics perfectly.

The missing 3 phase distribution panel that would have been in the special B cabinet provides sockets for cables to deliver 3 phase power to every other cabinet as well as to the power supplies in the special B cabinet. It is a squarish connector, although the only such connector I have had the cable cut off.
3 Phase AC input cable to each cabinet
This plugs into a socket on the drive cabinet, where only two of the phases are selected and the third pin is removed entirely. Thus, the cabinet is set up to run on one of the three possible load positions by its wiring and the installation would attempt to distribute the cabinets on each load position to balance out the current draw on each of the three phase wires.


The wiring inside the cabinet goes to a more common 220V connector that plugs into the rear of the power supply enclosure. This makes it easy for me to provide power to the tape drive, bypassing the 3 phase distribution box and obviating the need for the huge 3 phase connector normally used to power the entire string of tape cabinets.

Common 220V connector that plugs into power supply box

xx