Well I found some time and pulled the DRAM bus termination resistors from one of my rev 0 replicas, to see what would happen. I expected to find the machine still worked but with some erratic behavior. I planned to take before and after O’scope shots of the DRAM address bus to be able to demonstrate the difference.

What I expected and what I got were quite different. What I got, was a machine that wouldn’t boot and give me a monitor prompt at all. I couldn’t really do before and after O’scope images, because I couldn’t generate an apples to apples comparison without putting the processor into the same tight loop for both test cases. However the display on the video was stable, so it shows that DRAM access was at least mostly working without the termination resistors.

I tried 4K and 36K DRAM configurations, and found no difference in behavior.

While my reproduction isn’t an original Apple preproduction board, I don’t think behavior would have been significantly different on a preproduction Apple II.

Just to speculate a little. There may have been enough board to board variation that some pre-production units worked better than others, but I imagine that on the whole, things didn’t look very promising when the first prototype Apple IIs were built. Imagine the struggle that the early Apple employees had, finding a way to stabilize the computer enough to show at the West Coast Computer Faire, and more importantly get it in shape for revenue shipments.

There are stories floating around, about how early prototypes of the Apple II were not very reliable and Rod Holt made some fixes to make the design more reliable. A recent example is this Mike Scott interview.

Just what exactly was wrong and how it was fixed is not mentioned in any of these stories. I’ve always had my suspicions about what one of the problems was. However, without confirmation, I’ve been reluctant to speculate publicly in my blog or elsewhere. Last spring at VCF east, I had a chance to chat with Dan Kottke. Dan was very involved at the technical level with both the Apple 1 and Apple II. More importantly, he seems to have retained in his memory a great many details about those early days at Apple. Dan confirmed my suspicions about what one the of problems was.

Since I first heard the story of the flakey Apple II prototypes, I was very suspicious of the two SIP resistor modules shoehorned at the end of two of the three rows of ram. Here is an crop of an image of Geoff Harrision’s rev 0 board, #97 showing two blue SIP packages.

SIP Modules Shoehorned in between chips

These resistors are used to terminate the DRAM address bus. Termination is often added to a bus to improve signal quality in cases where noise and ringing cause performance issues. That was the first clue, but there are others. Nowhere else on the board are parts jammed so close together with so little spacing between traces. Also note the silk screen set at an angle – no other parts on the Apple II are labelled that way.

This image of the copper layer on my rev 0 replica shows how tight this area of the layout is.

Copper layer around SIPs

Dan confirmed for me that these SIP packages were added after prototypes exhibited problems with a DRAM address bus that was, in Dan’s words, “all over the place”.

Satisfied that I have found and had confirmed at least one of the problems with the preproduction Apple IIs, I still have two questions left to answer.

Well, I think I have the means to get some idea of the answer to the first question. I will be doing some experiments in the future and reporting results in this blog.

The answer to the second question will only be answered if someone comes forth with a pre-production Apple II, without the termination resistors. Knowing how engineers save prototype hardware, I’m thinking that some early Apple employee will come up with one, at some point in time. Who knows, maybe Rod still has the one in which he grafted the resistors onto, in order to test his fix.