Wednesday January 26, 2011

Attendees: Bill Newman,  Bob Erickson,  Ed Thelen, Frank King, George Ahearn, Glenn Lea, Joe Preston, Ron Williams, Stan Paddock and Judith Haemmerle

Joe Preston has made the third IBM 727 tape drive on the German machine wake up and almost load tape. Getting close to trying to read and write tape.

The IBM 1403 printer on the Connecticut system will pause the ribbon while printing and thereby cut a row into the ribbon. The ribbon supply and take up on the IBM 1403 is simple so that very few IBM CEs have ever had to work on it. That is what Frank King said when he started to work on the IBM 1403 of the Connecticut system

Well if I take it apart, maybe it will look like an IBM 026 keypunch

Well if I can't see what is going on, maybe I need more people on my team.

Well maybe I can get the whole team to see what is going wrong.

OK, maybe I should look at the mechanical drawing to see if I have missed anything.

By the end of the day we think it may be fixed. Since it only fails every 3 to 6 weeks, we should know if it is fixed when we pay our taxes.

Bill Newman and most of the rest of the team have been working on the Connecticut IBM 1402 almost since it came in. It will generate random read checks with a good deck.

We narrowed it down to cards with a large number of punches in the same row. Like 60 or more '9' punches in one card.
At Ron Williams suggestion, a jumper was moved up to the next voltage level. This helped a lot but was not the final solution.

The problem was further traced to the first read station. When there were over 60 '9' punches in a row, the roller was not being drawn all the way to ground.

We examined the brush that is used to ground the roller and found it was worn and not in the proper place. Corrected that and things were better.


We suspected the transistor that is suppose to ground the roller. While we did not have a ready source for the transistor, we bought a silicon transistor and wired it in place of the original.
When we ran the test program, we had no errors.
We wired the original transistor back in place. 
Failures again.
Bill Newman wanted to make sure it was not the final transistor that was the problem.
He removed the first transistor in the pair and one of the spade lugs fell off in his hand. (See picture above.)
While trying to add a new spade the second and third wires fell off the transistor.
Luckily, we had a spare for this transistor.
We tried it again with the original power transistor and still had errors.
We put the silicon transistor back in the circuit and have ordered a true replacement for the original power transistor.
We believe the random read problem with the Connecticut system has been identified and will be soon corrected. Until the transistor is replaced, the system is usable.

That is if the IBM 1403 ribbon transport unit is fixed.

Stan