... that still (may) work.
Here is the story:
One of the builders of my kits succeeded with his build (as expected), the diagnostics page ran fine, but then he ran into weird problems with his keyboard (a converted Apple II keyboard as per Mike Willegal's instructions you can find on www.willegal.net). These conversions typically work without a fault as long as the key switches and the keyboard encoder card are good.
But, alas, the screen output tended to be garbled and multiple cursors appeared, etc., etc.
We first suspected wiring errors in the keyboard cable as it requires some permutations which really can get wrong (nope), then a bad PIA (nope) but finally the ugly truth came out: he had used a keyboard cable which is 3 foot long ! Wow ! (Why did he do that ?)
To understand the failure mechanism, we need to know that the CLR SCREEN input signal on the keyboard connector pin #12 (on the Apple-1 side) is not just a TTL input. It goes to a signal line which is also - occasionally - being driven by the cursor state machine and feeding back into it. This is done via a diode (CR4 terminal section schematic area B-1) and if there is a long keyboard cable run attached to this line, bad things can happen due to reflections, ringing, and crosstalk. The cursor state machine will get very unhappy and spurious cursors may appear, and from there it's downhill.
I suggested a solution involving cutting a trace on the motherboard and inserting a series diode into the CLR SCREEN signal path - but he did not want to cut a trace. So the diode went into the cable, close to the header that plugs into the Apple-1. On the keyboard, a pulldown resistor was added. Now, unless the CLR SCREEN button (the repurposed REPEAT key) is pressed, the diode does not conduct and isolates the cursor state machine from that long keyboard cable and anything that happens on it.
This solution works. To verify it, I made a long keyboard cable myself (52" long, a tad below 3 ft) and tested it, and it works fine.
Here is how it looks (and why I chose 52"):
Problem solved !
If anyone out there ("Hello - hello - hello - is there anybody out there ?" - Pink Floyd, The Wall) is interested in the details of this long keyboard cable solution, I will post more documentation, but because preparing this would cost me precious RQLT, I just won't do it unless there is a real demand for it. I find the long keyboard cable I made for testing the concept to be very useful for me, as for the first time it allows me to use a real Apple II keyboard on my crowded work bench (the monitor is behind the keyboard, the other one with my infamous "IC graveyard" on top of it you can see behind the dictaphone died a few months ago, maybe in sympathy with all the dead ICs on it).
Oh, and the reason for that builder to have that long keyboard cable was his desire to put his Apple-1 system in an attache case and take out a keyboard (mounted in its own small enclosure) he could put on his lap. So an improvement to the famous original Apple-1 in the attache case, morphing it into a desktop / laptop hybrid. Actually, quite clever with better ergonomics. Except for screwing up the internal signal of the cursor state machine by using a keyboard cable being just too long. But that little added diode solves that. I wonder if with this concept we could make keyboard cables that are 4, 5, 6 ft long !
Comments invited !
I have seen the diode "fix" multiple times in working on original Apple-1 computers with their original keyboards. It's usually done in the wiring of the cable near the connector.
In post #2, Corey986 wrote:
"I have seen the diode "fix" multiple times in working on original Apple-1 computers with their original keyboards."
Uncle Bernie answers:
Alas, I have no such original Apple-1 with the diode fix, and never saw one of those. So I could not know that I did re-invent this fix. Which has no inventive height anyways and took me one minute to figure out the root of the problem and its solution.
Still, your comment is welcome as it proves two things:
1) engineers all over the world think the same way and will likely arrive at much the same solution for the same problem.
2) I'm on the right track with that solution. Adopters should have no fear. It's proven !
Oh, and I wonder if the historical solutions also have that pull-down resistor on the keyboard end to make sure the diode is properly and quickly turned off once the CLEAR SCREEN key is released. Some people do only half-baked jobs.
- Uncle Bernie