Paul Gilmartin wrote: > I'm surprised at a factor of 5. Wouldn't doubling the > gamut of characters result in half as many copies of > each on the train, and no worse than halve the throughput.
You have made an unwarranted assumption. A TN print train did not merely add 26 lower case alphabetic characters, nor halve the number of copies of alphabetic characters. A TN train layout also added symbols (such as the so-called box printing characters, and many others). A "genuine" IBM TN layout had two sets of everything, while a basic, standard 48-character train had five sets of everything that it had. But in order to calculate the speed of a printer that uses a print train you must know the distribution of characters on both the train itself and in the data to be printed. I think you are already aware of this fact, but it might not be obvious to most folks on this mail list, who might have never even heard of, much less seen, an actual print train. The speed of the printer has as much to do with how often the graphic characters to be printed "come around" on the print train as it does with how fast the mechanism can advance the paper to the next line. A 1403-N1 printer can advance the paper to the next line faster than the print train completes a full revolution (that is, when it is able to get back to the same, so-called "index" print slug at any given position [column] on the print line [paper]). If there were one set of characters on the train (that is, only one slug on the train for each graphic to be printed) then the printer would print exactly as fast as the train completed one full rotation. Two identical sets of slugs would enable it to print up to twice as fast. But it could not print lines any faster than the paper itself was able to be advanced by the printer, which would set a limit on the speed of the printer: for the N1 model of the 1403 this was ~1100 lines per minute. A 1416 print train had space for 240 slugs. They could all be the same, or they could all be different. A vanilla TN print train from IBM had exactly two sets of 120 different slugs. A vanilla PN print train had four sets of 60 slugs. The more commonly-used 1403-N1 GN print train had five sets of 48 different slugs (I think, I have not been able to put my hands on a layout for this train, as I have the others). If a GN print train could achieve 1100 LPM, then a TN print train should be able to run at 2/5th of that (or 440 LPM). The math should be simple at this point. But ... it isn't. Why? The difference between even a PN print train and a TN layout caused production printed output to print too slowly, and changing a print train on a 1403 was a very dirty, time- consuming process. Few sites that needed to print characters that were on a TN train wanted to print very many of them at one time, or in even a significant minority of their output. They just needed to do this now and then, and were willing to slow down regular printing just a little bit in order to avoid the time-consuming job of changing the print train to a TN layout. Another consideration was that 1416 print trains were expensive. You rented them and could not purchase them, and management did not think that renting another print train (for the same, full monthly rate as the ones that were in use all of the time), just to sit around on the shelf most of the time, was a good idea. Thus, what most sites did to address the problem of printing "text" output was to order a customized 1416 print train that had only a single slug for most of the really rare and obscure characters (such as the superscript numbers, the box drawing graphics, the round and square bullets, the underscore [then used only in PL/I programs], the quotation mark ['"'], etc.), but which probably did not include the really rare graphics (such as the lozenge, which was very rarely used at all, even in "text" oriented output). It was common for an IBM customer to order such a customized 1416 print train whose layout was identical to one ordered by a customer across the street or in the same city, or whom they met at GUIDE or SHARE. This was because IBM charged a fee to make a new, distinct, different 1416 train layout. If you could determine the RPQ number for a layout that you could live with, then you could order that, and IBM would build another one and not bill you for the one- time RPQ fee. A common, so-called "TN" print train was really nothing more than a basic, 48-character GN layout, with one of the five sets of what would otherwise have been the same 48 slugs as in the first four sets replaced by the most interesting 48 characters chosen from the 60 additional ones that were on a real TN print train (that were NOT on the standard PN layout) plus the 12 additional ones that were on a PN print layout. This usually meant that you got the 12 characters that were on the PN train, plus 26 lower case letters, plus 10 others, which meant that you did not get the superscript numbers, the plus-or-minus sign, or the broken vertical bar, etc. That was what most "TN" print trains actually were. When "text" was printed (i.e., something with lower case characters) on such a train, it would run exactly as fast as the 1416 train itself rotated: ~200 times per minute. Thus, most so-called "TN" print trains actually ran at about 200 LPM in round numbers when printing text, but at (nearly) full speed (880 LPM) when printing ordinary upper case data. -- WB ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
