Ever so often, a design or test project gets bogged down because some aspect of the lab work hits a brick wall. The method for getting past such a problem is not always so easy or obvious, so it helps to have people pass on their favorite magic tricks.
New technicians and engineers often don’t have a lot of lab experience. As a result, they are prone to make certain mistakes and assumptions regularly, until they get some practical insight from their more experienced co-workers. Here are a few important, general guidelines to keep in mind before diving into the real world of wires and components:
• All measurements disturb the system under test. Sometimes the disturbance solves a problem, sometimes it creates one. But it never leaves the system completely untouched.
• Stray and parasitic components (capacitance, leakage, resistance, inductance, coupling, etc.) are just as real as deliberate ones. This is the number one reason why simulations often do not work in practice, and why everything suddenly quits working when you replace the cover on the radio frequency (RF) unit being tested or aligned.
• Test equipment inputs usually are grounded on the low side. This means there will be some ground loop interference, and unwanted noise as a result, when two grounded systems that consume current get hooked together. It also means the low signal just got returned to the neutral alternating current (AC) power line, so watch out where it’s connected.
• Bandwidth is a function of interconnect. To achieve full bandwidth on scopes, generators and other wide-band instruments, correct connections, matched impedances and suitable cabling are all essential. The higher up you go in frequency, the more serious the problems of losses and mismatches become. Beyond 1 GHz, you really need to be focused on every aspect of the interconnect, or all of your signal can disappear in the cables.
• Static damage is a real, not imaginary issue. Ignore it at your personal peril. Many systems now use complementary metal-oxide semiconductor (CMOS) and low voltage metal-oxide semiconductor circuitry of one type or another, including operational amplifiers (op-amps), which are quickly and fatally damaged by just the wrong touch. Learn how to work in a "static-free" manner. It will save endless grief and your co-workers’ menacing looks.
• Many complex pieces of test gear have serious "artifacts" or false displays that are a result of certain control and signal combinations, and interconnect quirks. These include spectrum analyzers, digital scopes, and most low-level measuring devices. Think very carefully about what you see before you accept the readings as true.
From an experimental viewpoint, it is difficult to really simulate an aircraft electrical system on the workbench. Small, regulated power supplies simply do not behave like the real thing, and may mask many in-rush or transient related design problems due to their internal limiting characteristics.
To really make a bench "aircraft" supply, you need a source that has very high short-circuit current capability, and a certain degree of "compliance," so that input ripple and other generated artifacts of your system can be clearly seen by examining the input.
This actually is pretty easy to do. A pair of series connected 12-volt high capacity car batteries with an ordinary low quality charger applied to them does a very credible job, especially when connected through at least 10 to 20 feet (3 to 6 meters) of wire (the same gauge you call out in your installation drawings).
Your equipment will now see the very same environment it sees in real life. The more simultaneous runs you have to other test devices, the more lifelike your simulation becomes. Many interactions and mysterious noises that you swear are patently impossible suddenly spring to life on your personal workbench.
This test should be done on any system intended for installation in an aircraft, as the direct current (DC) source simulations given both in Mil-Std-461 and DO-160 do not take these effects properly into account. They attempt to simulate the source in a very inaccurate, but repeatable manner. Now, however, your tests can be accurate, repeatable and realistic.
You can also do a fair job of replicating this situation by feeding your reasonably sized, regulated supply through a 50-ampere high current blocking diode and into to a very large capacitor bank (100,000 uF minimum)–and remember a bleeder resistor, so you don’t get killed later when working on the wiring.
The capacitors initially take a while to charge, but then can supply huge peak currents and even exhibit some credible droop and other lifelike characteristics under high currents. If you can trip a 25-amp breaker with your test source, you are usually ready to work.
One measuring trick worth knowing is the evil effect of hooking counters to the sources you also want to measure in some other analog way. Attempting to attach both a counter and scope to a circuit can be frustrating. The counter’s clamping inputs can introduce considerable waveform distortion and also are quite a load.
You can be incredibly clever and use your scope’s vertical output to drive other instruments–like a counter–and bypass many of these problems. Even old Tek 465s can perform this stunt, and you may find it quite convenient in some situations.
Some newer scopes, with digital processing, can also display the frequency, but they tend be quite inaccurate–even on high-end scopes–compared to a real counter. Try it and see for yourself.
The true workshop wizard will also become familiar with the shop’s spectrum analyzer and make up a set of small probes to track down interference sources and unwanted emitters in "finished" products. Useful sizes are two turns of No. 22 insulated wire hooked to a 3-foot (0.91 meter) 50-ohm piece of coax, and a 1-inch (2.54-cm)-long insulated center conductor exposed on a 3-foot piece of 50-ohm coax.
These two probes can search everywhere for emissions without damaging the analyzer. And because of their small sensing aperture, they can pinpoint exact components and shield leaks with incredible accuracy, plus display the exact offending frequencies.
No spectrum analyzer handy? You can do a fair job with a high bandwidth scope (good to at least 200 MHz) with the same probes, providing you set the vertical amp input quite high and leave the sweep in Auto mode, but untriggered. Just watch the band thickness on the scope as you probe around for stray emissions.
The high bandwidth scope is not very discriminating, but it can save the day if it’s all you have available. If you use a scope’s probes, remember to terminate the scope input in 50 ohms to keep full bandwidth intact. The analyzer will have an internal 50-ohm termination, so you don’t have to worry, but your scope has a 1-meg ohm input impedance.
Speaking of scopes, having problems connecting to low-current high-speed CMOS logic signals? This can be a real nightmare, as the scope can present too much capacitance and shunt resistance to make usable measurements possible.
There are a few tricks that can help, though. First, get an active field-effect transistor (FET) probe, which can reduce your input capacitance to 2 picofarads (pF), or one-millionth of a millionth, and help control some measurement artifacts. Second, use inventor Ed Meitner’s trick of adding a small capacitor (1 to 2 pF) in series with your 10X scope probe tip. This is usable only at higher frequencies and is not hugely accurate voltage-wise. But it does allow useful examination of the previously unseeable, which can be a real help.
The last useful bit of magic is a very simple one. You can save days of frustration over "impossible measurements" with a quick calibration check of your gear at the start of each day. Also, perform a quick cross-check whenever you encounter an "impossible" reading.
Any simple reference (most scopes have one built in) will do, and it saves wasted hours because your scope probe is misadjusted, your leads are intermittent, especially broken coax cables, or your instrument has quietly failed the night before and currently is fooling you. Probably the best spent 10 minutes of your life, barring topics we can’t cover here.
Walter Shawlee 2 welcomes reader comments and may be reached at [email protected]