Vintage controller runs environmental tests
Steve Lindberg, Tektronix - October 27, 2011
While I spend my days designing and testing analog circuits for Tektronix, I spend much of my off-work hours in my personal lab, which has about 350 instruments. Most of the equipment comes from my company's recycling/salvage practice and other equipment comes from surplus electronics stores and ham radio festivals. When I needed to automate an environmental test at Tektronix for differential probes, I used my surplus HP 9000 series controller, which I find easier to program and use than today's computers.
The tests required ten different interconnection setups and numerous oscilloscope settings on a Tektronix TDS4000 series oscilloscope, a Wavetek 9100 calibrator, a Picosecond Pulse Labs pulse generator, and a Racal relay multiplexer. That required me to navigate through three menu structures. I had to remember, for example, when does the 50-Ω terminator go in the signal path? To minimize labor and maximize repeatability, I used the multiplexer, which I found through the Tektronix recycling/salvage practice. Fortunately, the mux met my bandwidth spec.
My home lab included a 1989 HP 9000 R/332 controller, shown in Figure 1 (see the product at www.hpmuseum.net/pdf/HPChannels_1989_3_25pages_Mar89_OCR.pdf). I purchased it from R5-D3 Surplus Electronics in Portland, OR for $35.00. It was kind of wobbly because the bottom had protrusions that were metal arcs. I think the arcs discouraged users from standing it up.
I first learned of the HP 9000 series in 1983. I found them irresistible, especially the Model 9845C, which had a high-resolution color CRT and considerable computer power for its day. In the early days of personal computers, you could easily develop your own software because the hardware, operating systems, and programming languages were relatively simple--no structured programming and easy software access to the I/O ports. The HP text editor is a bit odd, but usable. The HP9000 line is really a calculator on steroids. Typing an equation at the prompt will usually work.
The HP9000 designers were masters at their craft. Its ease of use and turnkey operation were there way before the Macs touted it. Only its high price prevented the HP9000 series from being more popular.
When building a home lab, the real value is instruments that stand the test of time. I have oscilloscopes, function generators, counters, and other instruments that are just as useful as the day they were built 20 or 30 years ago. Their low cost makes for easy acquisitions. It takes a while to accumulate a bench full of equipment because you're at the mercy of whoever discards the the equipment as surplus.
The real function of the R/332 was it's instrument control. It has GPIB and serial ports, which was how you connected test equipment before computers had USB, Firewire, and LAN ports. In fact, my instruments were designed before those standards had been developed.
I brought the R/332 home and turned it on. It booted HP-UX (HP's version of UNIX) and then Basic 5.1. I had no documentation and manuals weren't available at the time so I decided to play around. I remembered the CAT (CATalog) command, which is similar to the DIR command in DOS. I found a huge program stored by the previous owner. Its 20,000 lines of code were overwhelming, but it gave the computer some unexpected value.
After running the HP9000 for a few hours, I heard a beep alarm and an over-temp LED on front panel came on. The R332 shut itself off. The fan was working and nothing smelled like barbequed electronics so I opened the unit. Tracing the high-temp LED wiring back to its source, I found that it went into a power supply. The power supply's enclosure had security Torx screws. I opened it after finding a set of really weird bits. The power supply was a clam-shell design with circuits on both halves, joined by pin connectors. I was idly complaining about the pickle of how to look at it powered when my 14-year-old daughter said "build an extender cable dad." I thought that was a great idea and cobbled one together.
The power supply had a thermistor mounted to a switching FET's heat sink. Nearby was a potentiometer. Having no idea about the circuit topology, I marked the pot's position and then tweaked it just a bit. The over-temp LED went out and stayed out. The unit has worked fine ever since. I wrote a GPIB interrogator program to find the commands of instruments whose manuals I had yet to obtain.
Eventually, I found the programming set. It's really a course in programming. The chapters on I/O are fantastic. Nothing is left to chance. The HP Computer Museum has a large collection for free. You can download at www.hpmuseum.net/document.php?catfile=256.
At this point, I brought the HP 9000 R/332 to work and received a few chuckles from my colleagues. "Basic?" they said. HP Basic is, though, quite powerful for controlling instruments. I wrote simple programs on it without any documentation.
I used the R/332 to drive a Racal 500 MHz BNC relay mux, a Tektronix 4000 series oscilloscope, and a Wavetek 9100 calibrator. Figures 2 and 3 show the equipment. The calibrator provides 200 Hz sine waves for gain measurements and square waves for CMRR (common-mode rejection ratio) measurements. A PicoSecond Pulse Labs fast-rise generator lets me measure the probe's rise time, which is related to its bandwidth. Figure 4 shows the test equipment connected to the HP 9000 through GPIB cables.
The initial effort took about three hours. I have ten test steps and entering 1 to 10 while my program displays its menu sets the instruments to the correct configuration. Then I just look at the oscilloscope screen and log the reading.
The test no longer takes three people sharing the front panel button push buttons, entering data, and swapping cables. That was very time consuming and inefficient. We all kind of felt under utilized once the system was running.
I also have National instruments GPIB cards in 3 PC's in my lab. But, I find the HP so easy to use that I've never written any code on PC's for GPIB. For serial-port control, an old modem program named Telix was outstanding and very useful because of its embedded script language. I've used it to automate hardware at five different companies.
Steve Lindberg has been in the electronics industry for 35 years as a hardware engineering tech and an electronics hobby enthusiast. By day, he's a New Product Development Technician in the oscilloscope product line at Tektronix. E-mail: Steve.Lindberg@tektronix.com.