1 - Beginnings
3 - Hello World
4 - More On GPIO
5 - Interrupts
7 - Timers
10 - Buttons and Bouncing
Would you like to be notified by email when Jason Sachs publishes a new blog?
When life gets rough and a circuit board is letting you down, it’s time to turn to test equipment. The obvious ones are multimeters and oscilloscopes and power supplies. But you know about those already, right?
Here are some you may not have heard of:
Non-contact current sensors. Oscilloscope probes measure voltage. When you need to measure current, you need a different approach. Especially at high voltages, where maintaining galvanic isolation is important for safety. The usual approach is a current transformer, perhaps with a Hall-effect sensor for DC-accurate current sensing. These typically have split ferrite cores that clamp around a current-carrying conductor. The current probe has a secondary winding with a large number of turns (100 or 500 or 1000), and sends current to cancel the field in the core. The current transformer is a passive component that works on its own for AC mains frequency up to the megahertz range, if the magnetics are designed properly. If there’s a Hall-effect sensor inserted into a gap in the core, active circuitry can be used to null the field in the core and produce DC-accurate current sensing.
There are a couple of different types of devices that are typically used. One is the passive clamp-on current transformer. The least-expensive of these are available for under $100 and are typically used at line frequencies only. The BK Precision CP-3 is one of them.
The second type of current sensor is a probe typically good for up to 100A, and from DC-100kHz. This is more expensive, but it’s reliable even for switching power supplies or motor drives if the switching frequency is low. There are a few models of these that all look the same:
They’re more than similar. They all have the same geometry, the same dual-range switch, the same dial for offset compensation, the same BNC connector, the same battery compartment, and the same specs. These are allegedly all made by the same company, Chauvin Arnoux and rebranded by these test equipment houses. The Chauvin Arnoux EN3 is the original part, and is available in Europe; in the USA it’s sold under the AEMC brand as the AEMC SL261. The prices from Chauvin Arnoux or AEMC are significantly less expensive (about $450 vs. $600+) than the rebranded versions, so unless you want a fancy brand painted on the side, don’t bother with the Fluke/Agilent/Tektronix versions.
The third type of current sensor is a fancy probe good up to 50MHz or more. These are expensive ($1500+) and often require their own special signal conditioning unit. Tektronix has some good ones that look sort of like Star Trek weapons. If you’re doing high-frequency switching power supplies, you’ll need something like this.
Finally the last type of non-contact current sensor is a Rogowski coil. This is a helical coil wound around a conductor that picks up its changing magnetic flux to sense current. They don’t work down to DC, but depending on construction and signal conditioning circuitry they can get close (below AC line frequency, and up to tens of megahertz). PEM UK sells one called the CWT Ultra-Mini that has a 30MHz bandwidth, and is small enough to fit around the leads of a TO-220:
I don’t know what it costs, though it’s probably not cheap. And you’d have to infer the DC current here or combine signals with a DC-accurate probe, but again, if you’re designing switching power supplies this can help you see switching transistor currents. The big clamp-on current probes won’t be able to do this unless you add a loop of wire big enough for it to fit around.
Instrumentation current shunts. If you don’t need galvanic isolation, a good old current shunt will do.
I’ve used some from Empro in the past. Murata sells similar items. They’re about 50mm x 25mm and have Kelvin connections: you connect current-carrying wires and sense wires with ring terminals. Low-tech here: it’s just a metal conductor which someone trims at the factory to be precisely 50mV or 100mV at some rated current. And by trims, we’re being literal: shave off a little metal and the resistance goes up just a smidgen.
These are typically $20 - $50 apiece and are available from 1 ampere to hundreds of amperes.
AC line current separator. When you’re measuring AC line current, you typically have a problem. The wires are in the cord right next to each other, and there’s no easy way to put a current probe around one of the wires. Enter the AC line current separator:
These are typically in the $15-$25 range, and let you plug in a standard 3-prong power cord, and they separate the wires and give you a loop to clamp on a current probe. Usually they provide a 1x loop and a 10x loop (wire loops with 10 turns): the clamp-on current probes really sense magnetic fields, which are proportional to amp-turns, so they can’t distinguish between 1 turn of 8.0A and 10 turns of 0.8A. Extech and BK Precision both sell these.
Temperature indicating stickers. These come in two varieties: reversible and non-reversible.
The reversible ones are usually liquid crystal strips that change colors when the temperature changes. You can take these and stick them on the outside of a motor, and they’ll tell you how hot the motor is, so that you don’t have to wonder whether the motor is going to overheat, or burn your fingertip trying to see if the motor is too hot.
The irreversible ones are paper strips with some sort of coating that changes color (I like to think they are just using lemon juice, like the secret writing in the old kids’ detective stories). These are good to put on transformers or capacitors or other components that risk overheating. They won’t tell you what the temperature is; they’ll tell you what the temperature was, so if the color doesn’t change, then you know it hasn’t overheated — at least not yet.
Both kinds are typically about $1 - $2 apiece.
Multi-range / constant-power-envelope power supplies.
Yeah, yeah, I said you know about these already. But here’s the thing. When you buy a lab power supply, you’re typically paying by the watt. The typical power supply has a rectangular operating envelope: let’s say you have a 120W power supply that can provide up to 5A and up to 24V. You could use that power supply at 12V and 5A. Or you could use the same power supply at 24V at 2A. Or you could use it at 24V and 5A. But if you need a 25V power source at 0.25A (6.25W power), or a 3V power source at 6A (18W power), you’re out of luck, and you’d need a different power supply instead, even though you don’t need anywhere near the rated 120W of the power supply you have.
Certain switching power topologies (particularly resonant and quasi-resonant power supplies) transfer a certain amount of energy per switching cycle, and it doesn’t make much difference whether you’re delivering 5V at 10A, or 10V at 5A. So these can have a wider current / voltage rating for the same maximum power level. I have on my bench the BK Precision 9110, a 100W 60V 5A power supply. I can use it at 20V 5A, or 60V 1.6A: the ratio between the ends of the constant power curve (20V 5A and 60V 1.6A) is 3:1. I like it a lot. It’s lightweight and relatively inexpensive (about $250) and barely takes any room on my bench.
A few other companies make similar power supplies with a hyperbolic output envelope, but most are expensive or heavy. BK Precision also sells a larger rack-mount power supply (the BK 9115, a 1200W 80V 60A supply, and the BK 9116, a 1200W 150V 30A supply). Kikusui’s PWR series from 400W - 1600W is another multirange supply. It’s a little more expensive (looks like around $1500 for the 400W version) but the same idea, and has an output ratio of 5:1 continuous or 8:1 peak. Here’s a graph of the output capability:
This is a really great idea and I hope more manufacturers follow suit.
Spring-tip probes. Pomona Electronics deserves a whole column just on its own. They sell lots of goodies. The obscure one that I like is the spring-tip probe:
Your typical DMM probe has a rigid tip. Here’s a question: When you are trying to press a probe tip to make good contact with a trace on a PCB, how hard do you have to press? I have no idea. It’s really hard to control the force that you exert on a sharp object. Designers of automated test equipment have known about this for years, and typically use “pogo pins” to make good contact and be insensitive to mechanical misalignment.
Pomona makes spring-tip probes along the same lines. You don’t have to worry about how hard you press; you just hold the probe still and the spring-tips take care of providing consistent contact pressure.
IC test clips. When you need to look at the voltage of an IC pin on an oscilloscope, it can be tricky to do without having to manually hold the darn thing against the IC pin. IC test clips will clamp onto the leads of an IC and bring out the signals to some pins on top of the clips. These allow you to hook a scope probe onto the pins.
Others are individual micro-hooks:
Agilent also has their “wedge” which jams contacts between the pins of an IC:
Tweezer multimeters. I sooooo want one of these Extech RC200 tweezers:
When you have a surface-mount resistor or capacitor that’s small and unmarked and you need to check the value, it can be a real pain to do so reliably. Even with spring-tip probes it’s hard to get DMM probes to touch both sides of the component and keep it from dropping or slipping out of contact. These tweezers are much better.
Data acquisition systems
I could write another article just on data acquisition. I often use pieces of test equipment to record data for further analysis. If you have an oscilloscope, you already have something that does data acquisition. And you might think, “Hey, great!” Except that when you read the fine print on oscilloscopes, usually the vertical axis gain accuracy is only 1%. Oscilloscopes are designed to show you qualitative waveforms in real-time. If you really want precise digitized data, and you don’t care about the real-time part, there are many other options to choose from, and it depends on the feature set you want. Here are a few typical use cases:
Remote monitoring. The HOBO data loggers are pretty well-renowned. These are small battery-powered gadgets that acquire data periodically for weeks or months, and you connect to them via USB or WiFi to download the data.
Industrial data recording. You want something that records data at 100Hz or 10kHz or 1MHz, maybe it’s 10 bit, maybe it’s 16 bit, maybe it’s 24 bit. I’ve used a number of different systems.
LabJack is a series of low-cost data acquisition systems which are pretty versatile for basic tasks. They’re inexpensive enough that it’s worth getting one as a general tool in your toolbox.
I’ve also used Agilent’s 34970A Data Acquisition System, which is a standalone datalogger, more like a high-end DMM with data recording capability: it is high accuracy (22-bit) but with a low acquisition rate. You purchase plug-in modules for signal conditioning, like the 34902A multiplexer which can measure currents, voltages, and thermocouples.
Lately I’ve been using a Data Translation DT9836 for high-speed 16-bit simultaneous sampling of voltages and quadrature encoder signals. They’re kind of pricey, but the signal conditioning is really high-quality and the results are low-noise waveforms. To give you an idea, I’m using them with a Honeywell CSNX25 current transducer at 50A peak current capability, but I can see sub-50mA current levels. That kind of dynamic range (1000:1 and greater) is hard to get without careful design.
Digital signal recording / logic analyzer. The Saleae Logic and Logic16 are inexpensive ways to record high-speed digital signals and decode protocols like I2C/SPI. I have a Logic16 at work and I’d get it again in a heartbeat. The datalog capability of these devices is pretty nice: they only save data on a bit change, so you can log data with 100MHz time granularity and as long as the signals themselves don’t change very often, or are only occasionally “bursty” (like I2C traffic that has only a few transactions once a minute), your log files don’t get very large.
You also can’t forget target-side datalogging. When you’re using a microcontroller or DSP and you want to record the data in your program variables, usually the approach is to spit this data out on the UART to record on a PC. Here you don’t need a data acqusition system, just a way to get serial data into the PC, and nowadays that means UART to USB conversion. I really like the FTDI2232H chip; it’s fast (up to 12Mbaud!) and will go directly from a microcontroller’s UART pins to a USB connector; all you need is a crystal and a handful of resistors and capacitors, and you can put a USB connector on your circuit board. For prototyping, I have a DLP USB1232H board on hand: it’s inexpensive and I can stick it into a small solderless breadboard and connect short wires to my microcontroller’s UART signals.
Really! I’m serious. Go buy a cheap stethoscope — there are some on Amazon for under $10 — and pull off or cut off the metal piece at the end, so you have a rubber tube for listening to things. (If you’re too cheap to do that, buy a 2-foot length of rubber tubing from the hardware store.) Why? Because occasionally there are things on your circuit board which may hum or whine or vibrate, and you’ll want to figure out what those things are, and you’ll want to do that safely. If you have a 500V switching power supply, this is not something you want to stick your ear next to. Use a stethoscope; enhance your hearing capability and directionality and make sure there’s insulating rubber between you and dangerous voltages.
That’s all for today – happy testing!
© 2014 Jason M. Sachs, all rights reserved.
Add a Comment