The first job was to address the problem of the magnetic starter tripping out every few lines. I found online several documents regarding the pulsing nature of the current drawn on the input of a VFD. I think the pulsing current may be fooling the current sensor on the magnetic starter, making it signal an overload and turn off the power. In the manual for the VFD they provide recommendations for input wiring protection based on the capacity of the VFD rather than that of the motor. By following these recommendations I set the current limit on the magnetic starter to a higher value, and as a result the power did not shut off at all during today’s work.

For reference of what it is I’m trying to cast, here is the proof I took at Mono U of the caster output from my ribbon after hand corrections:

This is cast in 9 point Times Roman on a 10 point body, with 3-point leading between the lines. Although the lines were supposed to be justified, due to the keyboarding errors and manual corrections, only three of the 8 lines that should be justified actually are (“1. Select the location…”, “2. When you click…”, and “also be options for fixed…”). My corrections to the ribbon should improve that average a bit.

The third line, once corrected, was actually too long so part of the hand correction was to hyphenate the word “pointing.”

By the way, the text is a short excerpt from the documentation for a computer program for managing digital Matrix Case Arrangements (MCA’s).

Sunday afternoon, I tried casting the ribbon again. One thing I noticed about the line lengths was that the lines that were quadded out at the ends of the paragraphs seemed to have more accurate line lengths. There was one adjustment I had not done on the caster, which is to adjust the space transfer wedge so that types cast with the S punch in 3/8 justification come out the same width as the same types with no S punch. I manually punched a ribbon to set 3/8 justification and cast a series of em quads. By running this ribbon and stopping it after it got to the quads I could cast a line of justified em quads, then disconnect the air supply and cast a line of non-justified em quads for reference. The justified ones were casting wider than the non-justified ones, but by adjusting the transfer wedge I got these to cast the same width.

Later that afternoon, with a bit of futzing about, adjusting line lengths as the caster was running, I got the ribbon to cast at one go with no freezes and no motor cutouts. I did have one line break not work, and by the time I noticed there was most of the next line in the type channel. I stopped the caster and carefully removed the first line and set it aside. Later I integrated it into its proper place in the form. There were a few small blobs of metal under some of the types in the earlier lines (near the bottom of the form). Perhaps these were caused by bleeding feet on the type. In any case, here is the result, with some of the spacing adjusted to fill the lines more or less:

I made a list of all the errors, and reviewed the ribbon to see if they were keyboarding errors or casting errors. In the latter case I compared what was cast with what the ribbon called for to try to determine if the error was a hole not read, a hole read where there was none, or some other more complex error. Most were of the first type, which might be corrected by increasing the air pressure.

To address the problem of incorrect line lengths, I made up another Excel spreadsheet to calculate the proper justification codes for a line on the ribbon. You enter the set size and line length once, and for each line, you enter the number of S punches (which are adjustable-width characters), the number of 1 punches through 14 punches, and the number of rows on the ribbon with none of 1 through 14 punched (which selects row 15 in the mat case). The spreadsheet then calculates for each line the two justification codes required.

I found several of the lines were way off, usually due to some other mispunch on the line. For instance, there is one place where two letters were punched in the same row of the ribbon, because the ribbon did not advance in the keyboard. This will only cast one type (by 25% chance it is actually one of the punched letters), but the keyboard counted both of them when calculating its justification. I corrected these justification errors using tape and the hand punch.

In the evening I ran the ribbon again, with even better results. The line lengths were fairly even (though not perfect). Only one line ended up too long, due to one type casting as an em quad instead of a narrower letter. Other than that all the lines fed onto the galley with no problems. Turning up the air pressure a bit had reduced the number of miscasts.

I haven’t taken a proof of this yet. That loose “g” at the end of the fifth line actually belongs at the end of the third line on the word “pointing.” This is also the line that cast too long; you can see the rogue quad at the start of the line between the words “on” and “the.”

After checking the errors against the ribbon again, I tried to cast the form a second time, but started getting nozzle freezes. The clutch that drives my caster may be slipping a bit as it warms up, and as a result the RPMs drop and freezes can occur easier. I was also getting late so I quit for the evening.

This morning I checked the angular alignment of the nozzle on my caster and found that it was off a tiny bit, so I adjusted that. I also reset the nozzle to its proper center position on the assumption that the faulty alignment was causing the leakage.

I also took apart my mould again, including removing the two side blocks on the crossblock and the two type blocks on the main body of the mould. I cleaned off some accumulations of well-adhered type metal, and on reassembly I took great care to ensure that the mating surfaces were absolutely clean and free of oil.

I put everything back on the caster, and tried it out. It was casting beautiful quads with no sign of nozzle freeze. But looking at the nozzle area revealed why: The nozzle was leaking plenty of type metal, and all this extra flow was keeping the nozzle hot.

I lowered the pot and swung it out, and gave the outside of the nozzle a good cleaning and examination.

A close look revealed that there was something off-center or out of round, perhaps as a result of being used too much with the centering position out of adjustment.

Unfortunately the only spare I had (aside from the display casting nozzle) was in even worse shape. It looked like someone had pinched its tip with pliers!

I have a factory drawing of the dimensions of several nozzles so I knew what shape the tip should have. I didn’t have much to lose in trying to reshape these nozzles, so I started with the pinched one. I mounted it in my lathe, carefully centered it, and turned a thin cut off the conical end until the pinched area was gone. Then I cut the slight step in the shape and drilled out the hole to 0.070″ per the diagram. Finally I used a file to cut the tip down until there was a flat spot 0.093″ in diameter.

I put this on the caster and readjusted the nozzle height. This nozzle did not seem to screw in quite all the way so despite me shortening it, I actually had to lower the nozzle height so it would seat at the right time.

After casting a couple of lines of quads the mould was warmed up and ran with no freezes, and, even better, no nozzle leakage! Furthermore it was clear the mould was running tighter because the cast of the air vent on the jet had almost no flash on it, unlike previous times.

One other problem I had encountered last weekend was the casting of letters directly above the correct ones in the matcase (and also the line length being wrong as a result). I though it might be due to excessive air leakage between the pins, but reducing the air pressure had not helped. However, during the week, while I was working on one of the other problems I noticed that the front pin jaws were not closing fully around the air pin. In fact they stayed far enough open that the rack might be more than half a tooth out of position, so when the locking bar seats it actually pushes the rack to the next position from where it should have been. I traced the open jaws to the rod that connects the tongs spring box to the bell crank. A nut had come loose on this rod so the rod had some end play. When the rod was pushed by the spring box (which closes the pin jaws) it would not push far enough. I tightened the nut and thing were running properly again. Quieter too because the rack was not making any “clack” noises as it was snapped into position by the locking bar.

I tried casting the ribbon again. At first the galley was not operating when it should. It seemed that the air pins for the two justification lines had become a bit lazy so I turned up the air pressure a bit and they started working properly. I got through a run of the ribbon with only a few events: The line lengths are still all over the map so I had to stop to clear over-long lines and spend time dropping quads into the too-short lines. I had a couple of lines where the column pusher did not push far enough (instead compressing its spring box). There were also some miscasts too, often involving a position on the ribbon reading holes from adjacent positions. Perhaps I turned the pressure up too much in my effort to get the line breaks working.

Finally I still have the problem of the magnetic starter that supplies the VFD tripping its overload protection after casting 4 lines or so. The average current draw isn’t actually that high, but because the VFD is converting the supply to DC using a simple bridge rectifier and capacitors, the peak current is actually quite high but only occurs for a small fraction of each AC cycle. This spike current might be spoofing the overload sensing on the starter. Because the VFD itself provides overload protection for the motor, all I really need is a switch and over current protection (fuses or a circuit breaker) rather than the magnetic starter. For now I’ve turned up the overload limit on the starter so I hope this will stop these false trips.