Atkelar, from Austria, posed the following to John Minck, the self-appointed guardian of the HP Archives first complied by Marc Mislanghe thirty years ago.
Atkelar wrote: "I try to make it short, but my idea is a bit hard to explain without abit of backstory. I recently got my hands on a µLab, HP 5036A, including the manual. Since I already have some other period correct test equipment (almost exclusively HP too) I wanted to make a little "documentary" style presentation of the set of tools and how the early digital days had ideas for tools that not all worked out. I found some info on how these tools were intended to be used, but I couldn't find out a "real reason" why some of these were not working out. It would be nice to have someone's input who "has been there". I am only almost 50 years old, so the 1970s didn't see me with much interest in the inside of electronics :)
I'd especially be interested in contacting somebody who had any hands on contact with "Signature Analysis", "Map Mode of the HP 1600A" - the two diagnostic aids that I assume ended up too cumbersome to keep around. The logic probe set that includes 545A, 546A, 547A and 548A are also part of the video idea. My videos have a slightly "different" presentation style, here's the µLab one for reference: https://youtu.be/wwUFyNjO8xA. Greetings from Austria,
The actual primary question in my mind is: was there ever an official reasoning behind getting rid of "Signature Analysis" and "Map Mode"? My guess is that both of these would have led to quite intense documentation updates with every revision of a circuit and that probably wasn't worth the gain for diagnostics. Logic probes and test clips have most likely fallen victim to the speed of signals, so digital scopes pretty much took over as more versatile options.
I found this question interesting. I penned, probably too quickly, the following answer:
Atkelar, thanks for your questions re some very old and almost unknown HP logic instrumentation history. I haven’t thought about those issues in years to be honest.
You are quite right that these features had a high degree of documentation requirement, and it was ever-changing if the circuits were modified in even very simple ways. But that was not the primary reason the features disappeared, in my view.
I don’t know about any “official reasoning” for getting rid of Signature Analysis (SA) or the HP 1600A map mode, but neither sold very well, and usually that determined the fate for some of HP’s instrument features disappearing. With respect to Signature Analysis, this was an end-run by Santa Clara division when the Colorado Springs Logic State Analzyers proved more popular than Santa Clara’s Logic Analyzer. The CS machines, aside from the HP 1645A, were focused on parallel logic, while Santa Clara’s were either node-only or serial data stream analysis. This at a time that 8 and 16-bit microcontrollers and then microprocessors were becoming popular, which put a premium on a designer understanding parallel data streams. In fact, although the Santa Clara team would not admit it probably, the Signature Analysis technology was pioneered in my Colorado Springs logic lab by Dan Kolody, whose first product for us was the HP 1620A (see my memoir for a picture of that box, which also sold in very modest quantity). At the time, Santa Clara’s logic probes and clip were selling somewhat better, and the two divisions were in a bit of a skirmish for ‘charter’. The prevailing history was that C Springs was an oscilloscope division, getting its brains beaten by Tektronix on a daily basis for nearly two decades, and HP top management wanted them (us, meaning me and my team) to stay focused on Tektronix, never mind the digital world. And Al Bagley, division manager for Santa Clara, was vocal about his division being the “RIGHT” place for all things digital, but he had already lost the argument for hosting the first Desktop Calculator, and he was adamant that he should have the Logic instrument business.
The “charter fight” answer by Bill Terry, HP’s VP for instrumentation, was to get HP Labs involved in two aspects—one, choosing the correct letters to portray in the Signature Analysis probe (so if any element of an LED seven-segment display failed, the character would not be mis-recognized), and two, choosing which division should build it. HPL chose FUCHPL, although they never wrote it that way until we pointed out the clever arrangement of the letters. And we graciously gave the Signature Analyzer charter to Bagley and team because we concluded it was almost irrelevant to the way test requirements were going to unfold after we modeled the Desktop Calculator ’signatures’ and found out how much work it took to get them node-by-node when no one had time to do that any longer. So, I don’t think SA ever sold more than a handful of units. This was at a time that Gate Minimization (Kline-McClusky techniques) were still being taught, and algorithmic design was quite novel and hardly taught anywhere).
The map was invented for the HP1600A, and carried along for several years. It was a great tool for mapping the issues around memory chip utilization, which for a brief period was an issue when solid-state memory chips were first available, and relatively expensive (i.e. a penny per memory cell, or $10 for a 1,024 bit memory chip). But the chip memory wars soon produced “free memory” and designers had little need to optimize chip utilization cell-by-cell. The other thing the map could easily do was portray system execution pathways, which was sometimes valuable when a machine crashed, and you didn’t know what error sent it in the weeds. The map could indicate, if you had captured it during run-time, the point at which the execution path deviated. But break-point triggering, especially multiple-depth triggering that soon appeared on the HP 1610A and HP 1615A was a much more effective way to find such issues, so the map was also rendered irrelevant.
I always loved the map, but somewhat as with SA, few people used it, and the “maps” produced were so arcane and obscure in their specificity that very hew designers ever attempted to use them. We did produce one or two Application notes about their use, but almost never got questions from customers about the mode and its value.
I hope this helps answer the questions you posed.
I dont’ know who could answer this better for you, but I will suggest that Sam Lee, Dan Kolody, or Tom Saponas at HP Colorado Springs would have some information. All three are still alive, but I don’t have current contact information. Gary Gordon at HP Santa Clara, one of the most brilliant engineers I've ever known, could also shed light on this, as might Webb McKinney. I am copying Webb on this, and he may have contact info for some of the others.
No comments:
Post a Comment