Get free access to our online edition!

Tech Forum

LED Compatibility? February 2017

I replaced some outside 60W bulbs with CREE dimmable LED replacements. The lamps are  controlled and dimmed using X10 switches. When switched off, the lamps still glow at about 20% and will not shut off completely unless I use the disable feature of the switch. However, this prevents the timer from automatically controlling the lights. What causes this and is there a fix, or are LED replacements not compatible with X10?

Christoffer Mortensen
Piscataway, NJ


Conventional x10 switches require a small current to run through the load (i.e. an incandescent light) in order to work correctly. For non incandescent loads such as CFL or LED lights you need a x10 switch specifically made for them. I currently use a WS13A x10 wall switch and also an XPFM x10 fixture module to switch LED lights (and CFLs). These x10 switches are not dimmable though. In general, LED lights need a dimmer specifically made for LED lights. I have had a good success with Lutron CL digital dimmer (e.g. MACL-153MH) as a manual dimmer but I do not know of an x10 compatible dimmer designed for LED lights. Perhaps someone else knows of one that will work.

Dan Koellen
Roseville, CA

I go through this problem whenever I use x10 in a small project. x10 appliance modules need some kind of load resistor, but due to the 110VAC appearing when it turns on, I do not advise it. Instead, a batter way to solve it is to connect a 110VAC relay parallel to the LED bulb. If you want to use a resistor 33K 1W will be OK (I tested up to 42K that works), BUT be very careful about insulating the wire leads. I tested both methods (resistor turn on only a second in on state), they work perfectly.

Ankur M Bhakta
Tulsa, OK

How Much Tolerance Is Enough? January 2017

When designing circuits, is there a rule of thumb for picking voltages and tolerances of components? For example, if my power source is 12 volts, is an electrolytic capacitor with a 24V rating  “better” than one with a 16V rating? What do good designers use as a margin?

Enrico Gutiérrez
Panama City, FL


I used to design circuits for military applications and in that environment, a voltage rating double the expected maximum was considered adequate. In an automotive application, the spikes from the starter can exceed 60 volts, so you need to keep that in mind and provide isolation.

Ripple current is another capacitor parameter that needs to be addressed. It turns out that ripple current rating increases with voltage rating, so you might use an electrolytic cap with 10 times the needed voltage rating just to get the ripple current rating.

In general, the MTBF (mean time between failures) is calculated based on the stress on the component.  A component (resistor, transistor, transformer, etc.) that is rated 70 degrees C but is running at 150 degrees C will have a short life, but if it’s running at 30 degrees C the life will be normally long.

Russ Kincaid
Milford, NH

Turn Signal Signal January 2017

I have a newly restored 1971 Honda CB350 motorcycle that I ride for fun on the weekends. One problem is forgetting to turn off the turn indicators. I have found a kit that “beeps” every time the indicator lights up, but it's very annoying as I sit at a light. I would like a circuit that would alert me only if the turn indicator stays on for more than two minutes. Schematic would be welcome!

Leland Collins
Gulfport, MS


Sounds like a job for an Arduino Pro Mini and a few discrete parts.

First you need to convert the on-off 12V turn-signal levels from the lamps or LEDs to 5V logic levels. You can find level-shift circuits via a Google search. Then you’ll have two sets of pulses, one from the left signal and one from the right signal. Run the 5V logic signals to two inputs on the Arduino Pro Mini.

Second you need a control program that determines what to do with these pulses and when to turn on an indicator (visual or audible). When the software detects a pulse on either input it starts two timers, Timer1 for 2 minutes and another (Timer 2) for about 1.5 times the length of a turn-signal on-off period (1.5 times the flash duty cycle). Each incoming pulse restarts Timer2. So as long as the Arduino Pro Mini receives turn-signal pulses, Timer2 continues to run. If at the end of the 2-minute period Timer2 is still running, the indicator turns on. The indicator turns off as soon as Timer2 stops running. That indicates no more pulses from the turn signal.

As an alternate, an Arduino Pro Mini could simply count the number of pulses that occur for your vehicle in a 2-minute period and when the count equals that number, it turns the indicator on. You might include a “kill” switch for the indicator in case you need to keep the emergency flasher on for more than 2 minutes. For more information about the Arduino Pro Mini, visit:

Jon Titus
Herriman, UT

Trainsformer Needed January 2017

I found a bargain at a local thrift store and now have a 1959 Marklin HO Scale train, cars, and track. The set didn’t come with a transformer, so I thought it would be a fun project to build from scratch. Does anyone have a schematic or suggestion for a DIY train transformer they can share?

Alfred Thompson
Kingsport, TN


Boy are you lucky!!! I started collecting Marklin HO scale back in 1962, so a 1959 vintage should be fabulous.

The old equipment had AC motors in them, and the original transformers ran 16VAC to 20VAC. In other words, they were variacs which controlled the speed by changing the voltage to the tracks. This did pretty well, except at the very lowest speeds.

So, there are a number of ways to control the locomotive. The easiest would be to use a transformer output through a potentiometer, driving a 30 or 40 watt amplifier, (direct coupled). This output would directly couple to the track. The center spikes are the Hot, with the rails being the neutral.

A more complicated, (but fun way depending on your skill level), would be to use a PIC microprocessor to PWM a sine wave to directly drive the center rail of the track. Then you would have the most control.

Mark Lampkin
Grand Rapids, MI

Uno-Known Device January 2017

I’ve used the Arduino IDE with a genuine Arduino Uno board for experiments for a few months without trouble. Recently, I wanted to permanently put an Uno in a desktop project, so I purchased some budget Arduino Uno compatible boards. The budget boards seem fine, but the Arduino IDE doesn’t recognize any of them. When connected, the boards show up as an “Unknown Device.” I am using Windows 7 32-bit. Does anyone have any pointers on how to make this board work?

Michael Allison
Camden, NJ


Some of the cheaper Uno and Nano board spinoffs (among other off shore versions) use the CH340G series USB serial interface chip rather than the FTDI USB chip. All that is necessary is to download the CH341 driver and install it in your Windows/Mac system. You can get the driver from or Google search “”. Just make sure your comm port settings in the software match the USB driver in Windows. You can leave both drivers in Windows without uninstalling the other.

Rick Choy
Winchester, VA

On the bottom of the Nano board, are a few chips. One of them is much larger than the rest. That is the USB chip. You need to create a solder bridge between pins 25 and 26. One of those pins is GROUND and the other is a manufacturer’s TEST input... which they left floating. It needs to be grounded. Once you solder those pins together, your Nano will work. I’ve done this fix twice, now, and it solved my issues 100%.

William Barnett
West Haven, CT

Tube Tech November 2016

Is there any "technical" difference between tube amp distortion and solid-state amp distortion? I have heard tube amps described as “warm” sounding but I can’t find any info as to why. Isn’t “clipping” just “clipping” no matter the device that is performing that function?

Alison English
Tampa, FL


The soft or “warm” sound of tubes relates to four properties.

  1. Rolled off frequency response of the amplifier related to the inductance of the output transformers.
  2. Slower transient response again due to the high inductance in the output transformers.
  3. Soft clipping for very large amplitude signals due to the transfer characteristics related to the amplifier tubes.
  4. Magnetic hysteresis in the transformer itself which alters the output signal of the tubes.

Transistor amplifiers have much higher slew rates, higher frequency response, and hard clipping when the large signals cause the output transistors to hit the rail voltage of the power supply.

Solid state amplifiers are directly coupled to the speaker load, have a very high damping factor due to the negative feedback, and therefore produce much more accurate output than a tube type, amp. This is the reason for the harsher sound which is really more accurate than the output from a tube amplifier.

If you compare the FFT (Fast Fourier Transform) frequency domain traces of the two types of amplifiers, the differences will become readily apparent. A solid state amp will have many more harmonic components than a tube amp. The direct coupling of the solid state amp to the speaker, also eliminates the hysteresis from the transformer core.

The lowest distortion figures will always be obtained from a solid state amp, in the less than .1% range. Tube amps, conversely are in the 1% to 5% range depending on their design.

Ron Hoffman
Solon, OH

Your looking in the wrong place: Maybe not definitive, but look here: Even harmonics are more pleasant to the ear.

Ron Dozier
Wilmington, DE

Yes, there are technical differences. Both tubes and transistors are nonlinear devices, and the transfer curve for each is unique. The transfer curve defines how the output should respond to the input. Within a narrowly defined range of input values, the output values change in “mostly” linear fashion — in math terms we would say the function is monotonic. When your input values start to go beyond the linear operating region the output is no longer a simple (linear or monotonic) function of the input.

Every device, triode, pentode, JFET, MOSFET, BJT has a unique transfer curve. Imagine that the transfer curve for a triode is not a straight line, but more of a “lazy-S” shape — the middle section is pretty close to straight, but the top and bottom of the curve rolls over. As mentioned above this curve “maps” your input signal to the output signal; any given value of input is a point on the curve that defines the output. But when the signal is near the very top of the curve, the output signal change for a given change of input is diminished (like a demon turning down your volume knob). This results in a soft clipping effect if the top of the transfer curve is relatively smooth. Gentle excursions into nonlinear behavior in a triode tends to produce a nice mix of even and odd harmonics; and, if I recall correctly even harmonics lend warmth to the sound.

In the case of a BJT (bipolar junction transistor), the “lazy-S” curve looks more like a “Z” drawn backwards, where the extremes of the transfer curve don’t bend gently. Instead, they have sharp “corners” and tend toward a “flatter” transfer function at the extremes. When the input signal gets into this nonlinear region a large change of input signal results in almost no change of output signal, but can produce lots of harmonics — predominantly odd-harmonics.

The clipping is more aggressive at the extremes of signal input; almost an “all or nothing” affair. Contrast that to the tube clipping, which is more like “diminishing returns.” I have heard of an amplifier circuit that adds even-harmonic components of the signal. I haven’t built it, but the idea is that it would create a warmer sound.

Having said all that, clipping and harmonic content (warm versus cool) should not be thought of as synonymous. Clipping occurs at the extreme limits of signal input. Nonlinear transfer curves can create harmonics at any value of input signal. When you look in the mirror in the morning, you are seeing a “linear” reflection of yourself. At the carnival or fairground when you stand in front of curved mirrors that distort your reflection, you are seeing an exaggerated “nonlinear” reflection of yourself. While seeing such exaggerated nonlinearity is humorous, in the audio realm it would be intolerable to listen to... maybe.

Guitar effects pedals intentionally distort the signal, sometimes to an extreme that is almost unrecognizable  — but, that’s an article for another day.

Doug Manchester
Rocklin, CA

The answer is simple, really. Tubes don’t so much clip, as go soft, rounding the peaks off the waveform. Solid state, however, works all the way up until it hits the head; at which point it cuts it off sharply. The resulting distortion can be either modeled as odd harmonics for solid state, or even harmonics for tube amps. And it’s all in how they are perceived by the ear; the softer clipping that tube amps do causes it to be more perceived as being not quite as loud, whereas the hard clip of solid state tends to be rougher sounding. And to conclude - if your system routinely clips, you need to Tim Allen (MORE POWER!) it.

Ralph Phillips
Bossier City, LA

In this case, all clipping isn't the same. A transistor circuit is fine up to the power supply voltage, where it mows the peaks off square and flat. This produces a harsh distortion similar to the fuzz pedal for a guitar. A tube circuit starts to round off the peaks before they actually run into the power supply voltage. The rounder peaks account for the "warmer" sound.

Chip Veres
Miami, FL

It's not about clipping. Tubes and FETs have greater inherent 2nd harmonic distortion which gives them the warm sound. Look up the books and articles by Douglas Self to truly understand why modern bipolar output amplifiers are very hard to beat for sonic clarity. Some of the new Class D amplifiers (TI, others) are quite amazingly clear, too. I still use FET input op amps (LF412) when I want some of that tube warmth. But when I need absolute clarity, modern bipolar devices (LM833 and newer) for crossovers, and bipolar output stages, are (in my opinion) best. Read what Self has to say.

Jim Lacenski
Bellevue, WA

It has to do with the type of distortion between the two architectures. Tubes tend to have more odd-order harmonic (3rd, 5th, etc.) distortion; solid-state amps are relatively distortion-free, and any distortion they have is generally with even-order harmonic (2nd, 4th, etc.) distortion usually generated by their feedback circuitry. With clever signal filtering schemes, solid-state amps can mimic the warmth, etc. of tube amplifiers, without the power-wasting (i.e., heat) that tube amps have.

Ken Simmons
Auburn, MI

Simple answer: To the first order, they are pretty similar, however, typically tube circuits are operating at a much higher voltage — a transistor circuit operating at lower voltage, will tend to have higher harmonic distortion than a tube operating at a higher voltage. If you use a high voltage transistor then you can get harmonic distortion from a transistor which is comparable to (or better than) a tube, but you usually have a higher noise floor.

Mark Sauerwald
Tacoma, WA

Getting Started With SDR December 2016

I want to start experimenting with Software Defined Radio (SDR) There seems to be a fair amount of info, but I’m still not sure where to start. What are the minimum requirements to get started on a small budget that the wife will live with? Is there a best “starter kit?”

Amaranto Melgar
Wetmore, KS


I use SDRSharp software with a DVB-T dongle with a Realtek RTL2832u control chip. These dongles are quite reasonable, and some are specially designed for SDR.


Hit By A Speeding Card December 2016

My Raspberry Pi Model III uses a microSD card. The cards are available in different “speeds.” Is there any significance to how “fast” a card I get?

Rick Holtz
Grand Rapids, MI


The answer isn’t really clear cut.

Class-10 SD cards are rated on the basis of streaming a single file to a clean (recently formatted) card. Essentially, what would happen with a video camera.
Class-2/4/6 SD cards are rated on multiple small files written to a fragmented card. Still image cameras where one has erased a few images.

One would hope that a Class-10 used for multiple small files would not be slower than a Class-6 card, but this is not always the case. Especially when used with a journaling file system — the ratings are based on FAT — where for single file writes one basically has the data file, and the File Allocation Table; journaling file systems, especially for something like EXT3 or EXT4 (common for Linux OS), will have things like the data file, inodes, journals, and bitmaps.

Flash memory needs to be erased to all “1” before it can be written — writes can only convert a “1” to a “0” but not the other way around. Erasure is done in large chunks. When doing small writes, the card allocates a cleaned chunk, then if needed, copies previously written data to that chunk before adding the new data.

Higher quality cards have the ability to “hold” multiple allocations “open” at the same time. These cards will function better on a journaling file system then a card that only holds two open allocations a time (which is all the Class-10 streaming model requires). Everytime a “2 allocation” card changes files, it has to close/flush any data changes to the card, then prepare a new “allocation” if changes are expected for the next file to be opened.

For FAT file systems, that means just closing the data chunk and opening another. For EXT3, that could mean closing a data chunk, opening an inode chunk and updating it, then closing it to open the next file chunk. Cards supporting 6 open allocations can minimize the number of times they have to perform the flush/close operation. They can flush a data chunk while still holding the inode/bitmap chunks active.

This may not be a concern if the device (RPI, Beaglebone Black) is used in an embedded mode, where the only data being written is a log file from some running application. But if the device is being used as an interactive computer, where one is compiling files, building applications, etc., a high end Class 6 card may be better suited than a low end Class 10, even if Class 10 implies nearly twice the transfer rate of Class 6. The RPI “NOOBS” OS starts out on a FAT system (it allows one to use native Windows to create a new card), but I believe the first step it performs is to repartition the card with an EXT3 file system, and copy the running OS into that partition. The Beaglebone Black OS images are currently EXT3 — and need special tools to create a card image.

Dennis Bieber
Kentwood, MI

Speaker Sharing December 2016

Does anyone have a simple circuit that will allow me to mix my iPod audio out with my computer’s audio to play through the same speakers? I want to avoid un-plugging/re-plugging just to hear some music.

Jonnie Vanalstyne
Manchester, NH


I use a simple DPDT toggle switch for something like this. I switch my computer output between speaker and headphones without wearing out the headphone jack.

In your case, the audio to the speaker would be switched between the iPod or the computer. The speakers must be amplified, and the computer’s output volume adjusted down to be close to the level coming out of the iPod headphone jack. You can mount the jack and switch on the speaker.


Fridge Alert December 2016

My young son frequently leaves the fridge door open. I’m an electronics beginner and I’d like to build a simple circuit to alert me when that happens. It should be simple enough that I could build it with the help of my son, so he learns two lessons in one sitting. Schematic and design appreciated!

Michael Calhoun
Odessa, MO


Combining utility and electronics education suggests use of a simple temperature sensor that produces a linear (straight-line) voltage output as temperature changes. The TMP36 sensor, for example, produces about 0.3V at -25°C to about 1.5V at 100°C (boiling water). I recommend the TO-92 3-pin package that will unobtrusively go in a refrigerator.

The three connections are power (pin 1), temperature-voltage output (pin 2), and ground (pin 3). The sensor provides a learning opportunity because you can use a simple voltmeter to measure the temperature in your refrigerator and experiment with ice water, hot tap water, and so on. Just solder three wires to the sensor and apply power. Use heat-shrink tubing to insulate the bare leads on the LM36 and the solder connections. A few dips into clear nail polish help with waterproofing.

The slope of the TMP36 output voltage amounts to 10 mV/°C, with a 500 mV offset. Thus at 1.5°C, the temperature of an "average refrigerator," output should read 615 mV, or 500 mV + 15 mV. At 100°C, you should measure 1.500 volts. Just remember to subtract the 0.500V offset if you calculate a temperature from the sensor's voltage output. Don't worry about the exact voltage: The sensor has a typical tolerance of ±1°C.

To turn on an alarm when temperature increases beyond a specific point, an analog comparator does the job. The LM393 offers a good example. You get two comparators in an 8-pin dual inline package (DIP) but use only one for the fridge monitor. The second comparator stays unconnected. The LM393 has two inputs, Vplus (pin 3) and Vminus (pin 2), and one output (pin 1). The IC takes power at pin 8 and connects to ground with pin 4. When Vplus > Vminus, the output becomes a logic-1. When Vplus < Vminus, the output switches to a logic 0, basically a connection to ground. This latter condition acts like a switch to ground and it lets the comparator control a small piezoelectric buzzer connected between the power supply and the comparator output. (Or you could use an LED as a visual indicator.)

To use the comparator, connect the Vminus input to the LM36 sensor output. Connect the Vplus input to a variable resistor--10K ohms will work well. This resistor lets you set the "trip" voltage between ground and the power supply voltage. Adjust the resistor to the point where the buzzer just turns on or just turns off. Then make a slight adjustment in the direction that turns the buzzer off. This change gives the comparator a reference voltage at the comparator's Vplus input. When the sensor voltage exceeds this "set point" voltage, the buzzer turns on.

PUI Audio and Mallory Sonalert manufacture loud piezoelectric buzzers. Choose one in your hearing range. AQll Electronics has a loud siren (ES-25) that operates from 6V and will alert everyone in your house to the open fridge door! Can create a circuit on a piece of solder breadboard. Try the SB300 Solderable PC Breadboard available via Amazon. For power use four D dry cells. A diode in the circuit drops the voltage from 6 to about 5.3 volts. That's within the range of recommended voltages for the LM36 sensor.

Jon Titus
Herriman, UT

Page 1 of 24 pages  1 2 3 >  Last ›