30 dB in 30 Minutes...

A couple of years ago: the phone rang, and it was another potential client.  A referral from a referral... in any event, they thought they had an antenna problem.  They were right.

It seems they were building an interesting on-body medical sensor which included a data transceiver for a clinical application. (If this sounds familiar, it should.  These days, I seem to do many projects with "on-body" and "medical" in their descriptions.)  This transceiver was at 2.4 GHz, and the size of the device was bigger than a wristwatch, but smaller than a deck of cards; it was powered by a coin cell.  It was located on the body of a patient in a hospital ward, and it needed to communicate with a central monitoring station about 25-50 feet away. Presently, it was doing a lousy job of it.  Sometimes, they had to lay the receiver on the patient's chest to get the data.  This was not only not good, it was not wireless. 

I needed to know where the client wanted me to take them. "What is your goal?"

"We need these 25 systems to work for the clinical trials.  The hardware is all built, and time is short.  It doesn't have to be pretty, but it has to work."

I scheduled a one-day session for them at my lab.  At worst, I thought, we'd get to the bottom of the problem and at least know what was going on.  At best, we'd solve the problem and know how to modify the built units.  I made sure the benches were (sufficiently) cleared off, the spectrum- and network-analyzers were warmed up, and the 'fridge was stocked.

They had built up 25 of these devices for a clinical trial; a sort of beta-test, but with lots of interested parties paying close attention.  The issue of way-too-short-range of the data link was not the death knell of their product, because they had really cool sensor technology which was the primary focus.  But, it wasn't going to impress anyone either.  They wanted the problem solved.  Now.

The Senior Engineer on the project showed up at my lab at the appointed time of 10:00am, and we got down to business; we'll call him Joe - not his real name.  He brought several of the sensors, complete with extra batteries and spare units I could play with.  He also brought their prototype monitoring unit, which had a simple off-the-shelf sleeve-dipole antenna on it.  We fired up one of the sensors, and it began transmitting packets of data about once every ten seconds.  

Sure enough, the receiver had to be within about two or three feet in order to hear the signal.  Not good.  I asked Joe lots of questions about how the unit gets deployed in the field, where the receiver would be set up, what the design of the receiver was, and so on.  I set up our spectrum analyzer and a calibrated dipole to observe the signal.  Man, that was one weak signal.  With Joe's permission, I cracked open the case of the unit.

Joe explained that they were using a Chipcon transceiver chip (now TI), and they followed the data sheet's schematic exactly.  Further, they made a "loop" antenna following some design equations one of their engineers found in the literature.  Hmmm.  I looked at the loop of wire, and something was bugging me.

"What was the calculated size of the loop?," I patiently asked.

"I believe it was one-half wavelength," Joe patiently explained.

"Did you say ONE-HALF wavelength?"

"Yessir."

If you see where this is going, congratulations.  Joe didn't, and his folks back at the office sure didn't, and that's why he was in my office.  I carefully measured the dimensions of the wire antenna they had made and noted it in my lab notebook. It was sort of a fat folded-dipole shape.  I took pictures to document everything (a habit of mine that often pays off in 1000-word increments).  Then, I opened my toolbox and removed the most powerful weapon I had.

I readied my high-quality wire cutters.

"Joe, watch the spectrum analyzer carefully."

With both his eyes on the screen of the analyzer, and mine split between that screen and the device, I cut the wire right at its midpoint.  The formally puny peak on the screen jumped up.  A bunch.

"Um, that just went up 30 dB."

"Yessir."  It was my turn to savor my favorite moment as a consultant.  If Joe smiled any broader, his face was gonna break.

"OK... what's going on??!", Joe asked.  It was 10:30am.

I explained that somebody's slide rule slipped.  If you make a loop of wire, and I don't care what shape, with one half-wavelength of wire, it's going to look like a shorted quarterwave transmission line, and very close to an open-circuit.  If it was going to be a loop, it should have been smaller and resonated with a capacitor, or longer and about one-wavelength long.  It was about as perfectly wrong as it could have been.  Snipping the wire at the center point essentially made it a dipole, with each leg folded back on itself.  Probably not very efficient, and probably not well impedance-matched, but... 30 dB better than what they had.

I showed him how to open each unit, snip the wire, tape the ends in place and get on with their clinical trial.  He left a happy engineer.  The clinical trial was a success, their product worked, and life was good.

After the clinical trials, I was invited to redesign their PC board, and put a custom antenna on it.  We ended up designing a cool embedded antenna which used the PC board copper and ZERO matching components.  If memory serves, we eliminated somewhere between four and six components.  The new antenna was designed and optimized using our CST Microwave Studio tools.  We optimized the antenna in situ, taking into account the battery, PC board, housing and the human body.  Performance was excellent; it worked just as envisioned in the clinical environment.

With much gratitude, I can report that we are still working with this happy customer.

Sometimes, you win the game by making base hits and hustling in the outfield.  You fight for every inch.  But, sometimes, just occasionally, with the bases loaded and the wind blowing in the right direction... you get 30 dB in 30 minutes.

I wish I could do that all the time.