SparkFun Electronics Commentsurn:uuid:214d0e4e-f1b1-d287-ce26-ac5b4c9f82492024-03-28T15:18:22-06:00SparkFun Electronicsdksmall on Thirty Years of Test Controller Design Progressiondksmallurn:uuid:384c81ec-b0ec-3b0e-bc8b-2c27f9da99b32019-03-15T11:49:54-06:00<p>The board was designed starting in 1985. We probably built at least 10 of these full wire-wrap versions before going to copper, so some of them were built later. And over the decades, things break or burn up, and need to be replaced. I'm actually debugging with that exact controller now, and I'm afraid to touch it or move it around.</p>
saccade on Thirty Years of Test Controller Design Progressionsaccadeurn:uuid:0fabed85-ee82-1b04-5fbf-cd58149dfb9a2019-03-11T16:23:42-06:00<p>I couldn't help but notice the "From 1986..." board photo has parts with '87 and '88 date codes.</p>
dksmall on Thirty Years of Test Controller Design Progressiondksmallurn:uuid:f0404dfc-ba7c-f904-aad4-06cca27a6cd32019-03-06T09:44:09-07:00<p>Nothing wrong with wire-wrapping, if you do it correctly., i.e. square posts, not round. I have my trusty wire-wrap pencil I bought back in the 80's, probably from Radio Shack. I see SparkFun still sells the exact same tool, except it's blue. This article referenced the treasure chest I did 9 years ago in the bio. There are several pictures there that show the wire-wrapping that was used as well.</p>
wsanders on Thirty Years of Test Controller Design Progressionwsandersurn:uuid:45392645-7069-9f53-0454-20ec2352cfe12019-03-05T20:22:09-07:00<p>Still have this horrible thing: https://photos.app.goo.gl/8qXrdaXFajVaFRSAA. I made a 4K static memory board for my ZX-81 with it. It worked for a year or then started glitching out. Typical of the technology.</p>
dksmall on Thirty Years of Test Controller Design Progressiondksmallurn:uuid:d58842da-397b-07f0-655d-ca32f6701d8a2019-03-05T15:21:47-07:00<p>Yes there are many times I must remind management that our functional testers are to verify the board was built correctly, not verify the design. Although there is certainly an overlap. I worked on the Teradyne systems when I was doing ICT and some functional on an L210. Later we started using the Z1860 (Zhentel, then Teradyne) for ICT. I know a few guys that work for MicroChip down the road and they do the IC test developments that you describe.</p>
Customer #134773 on Thirty Years of Test Controller Design ProgressionCustomer #134773urn:uuid:cc79a610-0a60-b483-2273-be5196d69d312019-03-05T14:18:23-07:00<p>Thanks for the nice "walk down memory lane", Kelly!<p>Much of my career was also in test engineering. I was mostly involved in testing at the IC level. I have a few things you're probably aware of, Kelly (and Pete), but some readers might not be aware of and will at least find interesting.</p><p>First, you mention that your boards involve a lot of wire wrapping. There were (and probably still are, though I haven't needed it this millennium) "NC" (Numerically Controlled) machines that automatically do the wire wrap connections. Also going into this discussion is that CAD (Computer Aded Design) systems for PCB (Printed Circuit Board) design, until the last couple of decades, have required much more computer power than could be provided by the typical PC of the day, so it was often not practical to do a custom PCB design when you only needed a few copies. (Today there are open-source PCB design packages that output the industry standard "Gerber files", and lots of vendors that will turn those files into physical PCBs at reasonable cost in a relatively short time.)</p><p>My next point is that the goals of PCB-level testing are somewhat different from the goals of IC testing. For PCBs, the goal is to verify (either directly or indirectly) that the UUT has no shorts or opens, that all of the correct parts are installed corectly, and that they are all functional. For IC testing, our ideal goal is to verify that for each UUT, or as we refer to it, DUT (Device Under Test) every single transistor on the device is good, that it will function over the entire range of speeds and voltages that the spec sheet says, and that it lives up to the specified output drive, input load, and power consumption specs. The testers are also used to do a whole bunch of "characterization" testing, used to establish things like temperature ranges, and all of those bizzare charts at the end of the typical data sheet. (High reliability parts, often referred to as "MIL-SPEC" parts, are given much more extensive testing on a per-DUT basis than are "commercial" or "automotive" grade parts.)</p><p>IC testing has traditionally fallen into 3 separate catagories: logic (or "random logic"), memory, and analog (sometimes referred to as "linear"). One of the sayings is that "you use logic to test memory and memory to test logic". Logic testers (where I spent most of my time) have huge arrays of memory that store "stimulus/response" test vectors which will be presented to the DUT sequentially (usually at, or just above, the DUT's guaranteed maximum speed), to verify its functionality. Memory testers (I was involved in designing the architecture for the memory test "options" for multimillion-dollar IC testers) use logic to put patterns into RAM and verify that they can correctly store them. (Pattern sensitivity is the big bug-a-boo in the memory world, especially for DRAMs.). Analog testing often times involves external instruments (lovingly referred to as "rack and stack"), though things like "arbitrary waveform generators" and high-speed high-precision DACs are common.</p><p>One last item for this comment: Board-level testers (especially "bed of nails" testers) often have much lower speed capabilities than do the IC (or "chip") testers, partly due to the differing goals mentioned above.</p></p>