Sometimes it's the little things that wreak havoc on an otherwise perfectly good design. This happened to me while deploying tunable RF bandpass filters ahead of a high-level receiver mixer. The filter design looked good on paper and proved O.K. on my
Tonne Software simulator, so I went ahead and ordered samples of miniature shielded adjustable inductors that looked like they would fill the bill electrically and mechanically.
I might add that the well-known inductor company I ordered these coils from was quite accommodating. It actually sent the samples all the way from China by way of Federal Express. Gratis. Now that's my idea of service.
With the bandpass filter board etched, drilled, screened and ready for soldering, I mounted my samples, soldered down the associated capacitors, and proceeded to sweep the assembly to determine its actual passband characteristics. The results were encouraging, but it seemed a bit of tweaking here and there might be necessary to achieve best response across the band of interest, and to reduce passband ripple.
Tweaking TroubleGrabbing my handy Nylon alignment tool, I started adjusting the ferrite slugs on the appropriate coils. Each coil slug was equipped with a hex hole to accommodate the standard-sized hex-tuning wand.
Imagine my alarm when I found that the slug on the first coil I touched jammed when it reached the bottom extent of its travel. Ditto for the second coil.
Trying to turn the coil slugs out proved impossible, so I de-soldered the cans from the assembly and opened the shields to see what could be done. To my dismay, the jammed slugs had both split in two. The slugs were no longer adjustable. A trip to the Boonton 260-A Q-meter indicated coil Q was seriously degraded, too.
A few experiments on the remaining coils indicated this jamming situation was a mechanical problem for sure, even if great care was exercised to avoid over-torque. It was obvious there was no way these inductors could be safely adjusted. If the slugs found their way to the bottom end-of-travel, they jammed. Imagine what a problem this would be on the manufacturing line. I could envision tons of rework and scrap. Not a pretty picture.
This isn't the first time I've experienced a hidden gotcha that cropped up with an otherwise perfectly suitable component. A project I worked on a few years ago (more years than I care to recall) used a certain discrete transistor as a high-voltage display driver. On the face of it, the part was suited for the task. It was electrically and mechanically ideal, with suitable price and availability specs.
However, we soon discovered transistors were turning up defective after they were soldered into place on prototype display boards. The culprit turned out to be the way we were preparing the leads. Our technicians were snipping the leads with a pair of dikes (pliers) and, apparently, the shock wave transmitted up the leads and into the package was somehow weakening the wire bonds inside. After a fair bit of head scratching and lots of consultation with the vendor, we discovered that trimming the transistor leads gently with a pair of scissors cured the problem.
In the case of my little passive inductors it was a matter of poor design. Maybe the manufacturer expected its tunable coils would never be adjusted! In the case of the shock-sensitive transistor leads, the semiconductor maker ultimately made the necessary improvements, and the problem never recurred in subsequent lots.
Sometimes there's more to a basic device than what's necessarily printed on an engineering data sheet. I'm not talking about transmission line reflections, software bugs, race conditions, or aperiodic nanosecond glitches. Those sorts of sophisticated problems are more or less expected in complex device applications based on microcontrollers and FPGAs. But jammed slugs in low cost inductors are another matter altogether.
Have you ever encountered unexpected problems with otherwise perfectly acceptable passive components or simple discrete devices? For example, have you ever had parts that wouldn't accept solder? How about packages with unreadable markings? How about resistors that changed value when the leads were flexed or bent?
What are your experiences? Write me at
amm at en-genius dot net, or post your comments below on our blog.
