Memory is a good thing

Note to my loyal readers: I am on vacation this month, so I’ve asked Gregory Davis, a technical marketing manager here at Tektronix, to take over and share a few of his thoughts on signal integrity. Don’t worry. You’re in good hands.

In a previous life I worked for an aerospace company responsible for designing and building the main mission electronics for an orbiting satellite. My job was to support the testing of the power systems that supplied a variety of voltages and currents throughout the satellite. For this particular satellite, all of the power systems ran off of a single DC bus; each power supply would convert the main bus power to the voltages and currents that were needed by the electronic subsystem that the power supply supported.

Sometime late in the development process we discovered a problem with one of the power supplies. When we turned the unit “on,” we would occasionally see an “in rush” current that not only exceeded the fuse ratings but also violated the satellite specification. It was a problem. We discovered the anomaly during environmental testing of the third unit in a series of five, serial #3. Unfortunately, we had already delivered serial #1 to the customer, and it was sitting on top of a booster rocket at the launch site. T minus 40-plus days and counting. Did I mention it was a problem? It was actually a big problem. If a fuse blows while in space, not only is the power supply inoperable but the system that it powers is useless, as well. And, it’s hard to send a repairman out to fix it.

The first order of business was to determine if serial #1 displayed the same problem. So I was sent to the land of gators, the second Disneyland, and “the right stuff” to test it. As you can imagine this was not a simple task. The unit was 30 stories high on top of a rocket, connected to the satellite, and all of the command lines that I needed access to were not available at the power supply; they could be accessed only at the control center, which was about a half mile away. I then had to create and submit a special test plan that outlined what was needed: test personnel, panel removal, connecting a current probe, etc. Once approved, I was given a schedule (2 a.m. to 4 a.m.) to run my test. Apparently I wasn’t the only one who wanted access to the satellite.

At the appointed time, I took the elevator up to the appropriate level and walked along a gangway where only a flimsy metal bar was in place to keep me from falling to the tarmac below. I entered the covered compartment, and positioned in front of the satellite to run my test was the instrument I had arranged to use. It was an analog oscilloscope with a persistence mode (Did I mention I was only 12 years old when I was running these tests?) connected to a current probe attached to the power supply’s main input power lead. I set up the scope to trigger on the in-rush current and set the persistence to infinite so I could take a picture of the resulting waveform with a scope camera. I was in communication with the control center, so I would tell them to send the “on” and “off” commands as required. The sequence went something like this:

1. Clear scope trace
2. Set-up scope to trigger
3. Me: “Send command number XXX” (to turn on the P/S)
4. Control center: “Sending command number XXX”
5. Me: Look for anomaly
6. Clear the oscilloscope trace
7. Me: “Send command number YYY” (to turn off the P/S)
8. Control center: “Sending command YYY”
9. Set up the scope to trigger
10. Repeat sequence 3–9

We were on about our 75th cycle when we finally did see a failure, but my muscle memory was in overdrive and I cleared the scope before I realized I had an actual failure. D’oh! We had to continue the testing to see if we could force another failure. Luckily, it only took about 15 more attempts before I was able to capture the anomalous event on film.

Occasionally, I take a look at the oscilloscopes that are available today and I think about how much easier this job would have been using a digital oscilloscope with memory. Just the process of saving data would have been much easier because instead of having to use a camera to record the data I could have saved every in-rush current waveform, allowing us to compare and analyze not only the obvious problems but also marginal data that might have been considered good when looking at the initial trace. With the computational ability of today’s oscilloscopes, we could also have created a statistical analysis of the failures to see if there was an event at the beginning of the turn-on that indicated that a failure was imminent. I most certainly would have avoided my faux pas of clearing the screen before taking a picture—even with my boneheaded move of clearing the screen the data would have been saved.

The point here is that digital oscilloscopes with memory have made an electronic engineer’s life a lot easier. We save, analyze, and share data much easier than it was done 30 years ago. (I was 12, remember.) If you remember, the original complaint about digital oscilloscopes was that they did not have a persistence mode, so it was difficult to capture intermittent events. But with breakthroughs such as digital phosphor displays, users are presented with an experience that is very similar to an analog scope.

So the truth is I was older than 12 when I had to run those tests. (Would you believe 15?) I didn’t have to walk four miles to school in my bare feet, and I didn’t have to do my homework by fire light. But I did have to use an analog scope and a clunky film camera to capture images of waveforms. A scope with memory is a good thing, especially when you’re dangling 30 floors up, trying to debug a satellite power supply.