datasheets.com EBN.com EDN.com EETimes.com Embedded.com PlanetAnalog.com TechOnline.com   UBM Tech
UBM Tech

A dark story

-November 29, 2012

At the beginning of the '80s, I worked at a radio communications research institute. I was a very young engineer, but I was team manager of the magnetic-recording-equipment team. One of the teams in our department had designed a control unit for our newest receiver. Its function was very simple; it would switch one of five preset FM channels. The simple unit had been designed using what for us were new CMOS ICs, similar to the 4000 series.

When the control unit was implemented in the receiver, the engineer overseeing the production process called our department to say that many of the receivers were not working properly; they exhibited a culprit frequency of an unacceptably high level. We had never seen this problem before.

An engineer from the team that had designed the receiver went over to the assembly shop and verified that many, though not all, of the stock-produced receivers had the reported problem. He ran trials to find the cause but could not pinpoint it. He began to swap out units in the receiver—which, thank God, had a modular design—changing RF, IF, and even audio units to see if that made a difference. He even took the RF, IF, and audio units from the prototype, which had not exhibited the defect, but to no avail. Production was halted. It was a state of emergency.

Our chief engineer was furious. He ordered the entire receiver team to devote its time exclusively to this problem and to report results in a day or two. The team checked everything—the workplace, the measurement instrumentation, the shielded room. All seemed OK. Our engineers placed new bypass capacitors into circuits, but the receivers still presented the high-level culprit frequency. Two days passed with no solution found.

I had never designed a receiver, but I was a creative thinker, so the chief engineers asked me to help the receiver team. When I reviewed what had been done so far, I noticed that all the parts had been changed except the control unit. “Why did you change almost all of the units but not the control unit? It’s the only new part in the receiver. Let’s try inserting an old control unit,” I said to the team.

“Vladimir,” they answered, “you must know that this part consists of only a few D-triggers. It cannot generate the culprit frequency, because it does not have any high-frequency generators.” I persisted, however, and the team went along with my request because no other new ideas had been presented. The engineers put the old control unit in place, removing the new one—and the culprit frequency was taken off too! Then the new control unit was placed again, and the culprit frequency clearly appeared on the spectrum-analyzer screen, as if by magic.

This unit really did not have any generators; it had a few D-triggers only, which were used as simple RS triggers with decoding. Still, the assembly shop resumed work using the old control units. We had time for analysis. A problem had presented itself, and we had to find its cause; it was a matter of honor.

We placed bypass capacitors and rechecked all the circuits. The result was bad.

At one point, I stood alone in the doorway of the shielded room. It was lunchtime, and the worker operating the test gear wanted to go have his lunch. He never turned off the oscilloscope or the spectrum analyzer, but this time he did not turn off the PSU, either.

As I stood there, looking at the damned culprit frequency on the spectrum-analyzer screen, the worker turned off the lights in the shielded room—and the culprit frequency disappeared from the screen! I turned the lights back on, and the culprit frequency reappeared. Mystified, I repeated my actions, and the effect was presented. The culprit frequency would appear only when I turned on the lighting.

I made a shadow with my hand and found the place on the PCB of the control unit that “generated” the culprit frequency: one of the D-triggers.  

My first thought was that light was coming through IC’s plastic package. When I placed my finger on it, I was happy to find that the culprit frequency was removed. But when I covered the IC with an opaque screen and did not get the same satisfactory result, I felt grief; I was very close to the solution but had not yet found it.

I checked the PCB layout and found that the R, C, D, and S inputs of a second part of one of D-triggers were not connected to GND or Vcc. The surface of PCB had some contamination—I think it was photoresist—that provided some leakage-current influences between the opened inputs and outputs of the D-trigger (it was a positive feedback), because CMOS ICs have very high-impedance inputs. That is why the D-trigger sometimes worked as a high-frequency generator. Unfortunately, this part was not shown on the schematic diagram.

The point is that the engineer who had designed the unit did not have experience using CMOS ICs; previously he had worked only with TTL logic. At the time, CMOS ICs were new and unknown parts for old-school engineers.

After I had found the source of the culprit frequency, we connected the R, C, D, and S inputs to GND, and our troubles disappeared. We had saved face, and I was the hero of the day.


Vladimir Rentyuk is a development engineer at Modul-98, a developer of embedded electronic systems, robotic equipment, image recognition systems, and other products for European and world markets. He holds a master's degree in radio engineering from Zaporizhzhya Machine Construction Institute (now Zaporizhzhya National Technical University) in Ukraine.

Loading comments...

Write a Comment

To comment please Log In

FEATURED RESOURCES