Several years ago, probably more than I really want to recall, I was working as a lab technician for a company developing an ultrasonic liquid level control. It operated by oscillating a pair of piezo crystals when liquid contacted their surface in a sensor probe. The challenge at that time was to design a circuit that would have high enough gain at about 1 MHz to oscillate the crystals and then amplify this signal to saturation at the output so the signal could be reliably detected. In concept, it was a simple switch but needed to operate at this frequency with a 4- to 20-mA current loop. At the time, “high-speed” op amps were power hogs.
So to accomplish this, we decided to use a transistor array and direct-couple several stages to provide near rail-to-rail output oscillation. Since each was direct coupled, the succeeding stages’ bias point was determined by the previous stage output. Everything appeared to work well until the temperature started to dip to –20°C or so, then all oscillation would cease and the circuit would fail to operate.
After some head scratching over this, I remembered that VBE changes with temperature. Assuming a –2-mV/°C shift, I was able to calculate that the bias points were shifting and driving the amplifier stages into saturation. I re-calculated the resistors required to hold the bias points over the temperature range, replaced the old ones with new values, and the circuit worked great from –40°C to 70°C. This was by no means rocket science, but at the same time, I didn’t think it was too bad for a dumb ol’ technician!
Hey, that is pretty good engineering! Keep up the good work. And the good thinking! I had an old Heathkit radio with six (germanium) transistors that I built around 1958. It ran well for years. Then I took it out while I was shovelling snow. At 0°C, it sounded pretty bad. I replaced the transistors with decent silicon ones, and I biassed up the old class-AB stage with a diode, for temperature compensation, and it worked fine at –20°C.
I noted your design piece in the special issue of Electronic Design (“What’s All This Noise-Rejection Stuff, Anyhow?”). Of course, using an instrumentation amplifier to create a balanced input that rejects common-mode noise is a common practice in industrial audio circuit design. Unfortunately, it isn’t the case for consumer electronics. (I’m suggesting that it might be, if they had the big noise problem we’ve seen. /rap)
Yet we in the broadcast industry are often connecting consumer gear (CD players come to mind) to professional balanced audio inputs with mixed results. Your circuit is a clever way to connect unbalanced to balanced without resorting to magnetics. Audio isolation transformers are the quick and dirty way to fix ground loop noise issues, but high-quality magnetics are expensive and introduce their own distortion and bandwidth limitations on the low end.
Active balancing is considered by most audiophiles to be a superior solution. I tend to favor transformers when the longitudinal noise voltages can be extreme, i.e., telephone lines or long runs between buildings not served by the same power source. (Well, yeah. /rap)
Now, for more fun. Your piece on driving single-handed (“What’s All This Driving One-Handed Stuff, Anyhow?”), while it’s well and good to develop good coordination and teach both hands to be comfortable with each task that driving entails, I fear it would only encourage folks to multi-task beyond their abilities. I’ve always operated two-way radio gear while driving, microphone in one hand, sandwich in the other, steering wheel on its own. (Kidding!) (You do have two knees, right? /rap)
But most folks have real issues with using their cell phones or applying lipstick while driving. (Most folks are not very good drivers. Most folks are not very good at thinking. /rap) Both anecdotal and experimental data have proven a significant reduction in driving ability and safety when engaged in these multi-tasking activities. And most of those folks are driving automatics, not stick shifts!
I understand the danger of crashing is increased N:1, where N may be 4. But on many roads, where the road is nearly empty, 4 × 0 is still 0.
Ehem... N:1 on an empty road is not zero! Plenty of single-car accidents happen if the road is a bit curvy and the driver isn’t intimately familiar with it. It just happened a on the “East West Road,” as it is called in Dummerston, Vt., over the shoulder and into a tree at a curve. Granted, the tree was probably more aware of its surroundings than the driver.
So the tree was drunk? Or was the tree using a cell phone?