I've yet to meet an up-and-coming technician or engineer who places much value on experience when it comes to manually adjusting the components in an RCL discrete circuit for best performance — or one who can figure out how to carry out the many forgotten shortcuts in tasks that most engineers just don't need to think about anymore. After all, with better designs and today's broadband systems, how often do you need to call on these skills?
But if you had to, you'd likely think you could easily work it out. In truth, you might never get the job done. I first saw an example of this in tech school, where an instructor (with sleight-of-hand skills but, regrettably, no desire to provide us any formal instruction on his technique) quickly constructed the “perfect” S-curve response for an FM discriminator using just a general-purpose signal generator and a simple voltmeter — no wobbulator with frequency-marker system, no visual scope methods. Every hot shot in class tried to duplicate what he did, but failed.
My memory of that performance isn't so clear, but one action stood out that I'd never seen before: Each time he adjusted the generator's frequency for somewhere I assume in the FM band, the instructor would momentarily set at least one variable element (i.e., IF input transformer, output transformer, RF stage, or oscillator trimmer capacitor) a little bit away (i.e., in the opposite direction) from the DC meter reading he ultimately wanted to see at the output. Then he might adjust another one or two components, also moving them away just the “right” amount before returning to readjust all the components to get closer to the desired meter reading. He'd repeat that operation for every step in the process until he was satisfied.
In the usually difficult world of FM discriminator alignment, this iterative procedure was seemingly accurate and pretty quick — just five or ten minutes. Yet the approach appeared counter-intuitive, akin to telling the novice golfer to keep the clubhead away from the golf ball as long as possible (as teaching pro Phil Ritson once said) to achieve maximum distance off the tee. But as often happens in science, some “laws” seem to have universal application.
As I gained experience, I came upon my way for understanding this class of largely RF network problems, such as antenna tuners — although I have yet to precisely define what class or topology. It seems to work for me. So when I think about tweaking circuits today, I imagine how I'll step back before stepping forward to increase the chances of quickly solving a circuit's defining equations.
I envision a circuit with x degrees of freedom corresponding to y number of tuning components. In the initial phase, I naturally do what I expect most engineers would: Set as many of the variable components away from their endpoints, somewhere at 30-70 percent of their tuning ranges. To do otherwise cuts off my chances of a quick solution by a lot — almost as if the various components have no degree of freedom at all — i.e., they're not really in the circuit, and not a meaningful part of the equation set.
In part 2 is this blog, we'll continue looking at this counter-intuitive method of tuning and calibration with the antenna tuner example. This should lead to some insight into why the method works.
- Integrating the Trimming & Satisfying the Customer’s Needs, Part 3
- ELF Receiver Design Avoids Inductors
- Analog or Digital Filters: Oh Yeah?!
- Digital Tuning for RF Analog Front End
- Build Your Own Software-Defined Radio
- More Thoughts From RF Integration Chat: Wakeup Radio
- RF Integration: Can We Get It All?
- Filter Techniques in Circuit Integration
- Reclaiming Lost Knowledge