SKTF -- Fall 1987

The Smith-Kettlewell Technical File

A Quarterly Publication of
The Smith-Kettlewell Eye Research Institute’s
Rehabilitation Engineering Research Center

William Gerrey, Editor

Issue: [current-page:title]

Original support provided by:
The Smith-Kettlewell Eye Research Institute
and the National Institute on Disability and Rehabilitation Research

Note: This archive is provided as a historical resource. Details regarding products, suppliers, and other contact information are original and may be outdated.

Questions about this archive can be sent to







AC Power Protection

by Bob Margolin

This article is reprinted from the September 1987 issue of Powertechnics Magazine with the kind permission of its publisher, Jeff Shepard.

Considering that fuses consist of nothing more than a piece of metal in a glass or plastic housing, they hardly qualify as sophisticated devices. Even so, they are among the most misunderstood of passive components. Consider, for example, their current rating. Ask a group of people what the current rating stamped on a fuse means and most will say that it is the current at which the fuse blows. It isn't. It isn't even the maximum continuous current the fuse can handle without blowing, or clearing as it is called in the industry. It is, in fact, just the opposite. The rated current is the minimum continuous current the fuse can handle without clearing. Within the fuse industry, this is known as the must carry current.

The obvious question is, then, at what current will the fuse clear. The answer is that it all depends on the period of time that elapsed before the fuse opens. With the exception of a few specialty fuses, the determining factor is the energy being delivered through the fuse to the load. Referred to simply as the "I squared T," it is the product of the current squared and the time that this current flows.

According to the U.L. (Underwriters Laboratories) fuse standard, a conventional fuse must clear within 60 minutes when the current equals 135 percent of its rating. Note, however, that there is no minimum clearing time. Because of this, a fuse carrying 135 percent of rating might open in 59 minutes and 59 seconds while another fuse of identical rating and manufacture but of a different type might open in less than one second. Despite their widely divergent performance, both meet the U.L. standard. To overcome the confusion that may arise from this, most fuse manufacturers publish a family of "melt" curves for each of their fuse lines. These curves graphically relate the clearing time to the current flow for each member of the family.

U.L. and CSA (Canadian Standards Association) recognize just three kinds of fuse--micro fuses and normal-sized fuses with time delay and non-time delay clearing action. Delay and non-delay fuses carry the same maximum clearing time ratings, but the time-delay standard sets the minimum must-carry time at 200 percent of rating. Differences between the U.L. and CSA standards are slight. The most notable is that the U.L. standard sets the maximum clearing time for a 200-percent current at 2 minutes for all current ratings while the CSA standard is 2 minutes for fuses rated at up to 30 amps, and 4 minutes for fuses with higher current ratings. Of course, in the event of a true short circuit, the current will be thousands of times greater than the fuse rating and its action will be nearly instantaneous.

Although not recognized by either U.L. or CSA standards as distinct types, there are a number of specialty fuses made. Included are some very-fast-acting types that limit the let-through current. Let-through is the current that flows during the period from the inception of the fault to the instant that all current flow ceases. The key to minimizing let-through current is maximizing the speed with which the link melts and the resulting arc is quenched.

Fuses carry both current and voltage ratings but the voltage specified is not a working voltage. Rather, it is the maximum voltage that can appear across the fuse when it clears. To appreciate why a fuse is rated this way, you have to remember that fuses were originally electrical products characterized by the same engineers who characterized motors and generators. Because of this, the action of the fuse was not thought of as a switch opening a circuit, but instead as an increasing counter EMF that reduced the net voltage driving current through the arc that exists between the ends of the melting fuse link. Since the voltage rating relates to a maximum withstand voltage and not a working voltage, it is not necessary to use a low-voltage fuse in a low-voltage circuit, as some people believe. In fact, a 250-volt fuse will usually work quite well in a 5-volt circuit. Most fuse makers do, however, recommend the use of low-voltage fuses when the maximum circuit voltage is just a volt or two. The voltage rating stamped on a fuse usually includes an ac or dc designation. The existence of ac and dc fuses may seem incongruous, considering we are talking about a piece of metal in a glass tube, but there really are differences between the two. An ac voltage falls to zero as it periodically reverses polarity, whereas a dc voltage does not. Because of this, it is far easier to quench an ac arc than it is a dc arc and, as a result, different link alloys or different ratings for the same link are required. On the other hand, the peak voltage across the gap in a cleared fuse is much higher for a given RMS value in an ac circuit than it is in a dc circuit.

For many years, the standard miniature fuse for electronic applications was the 3AG-style glass fuse, which measures 1/4 inch in diameter and 1-1/4 inch in length. Recently, however, the increasing value of real estate has driven designers to look for a smaller fuse and they found it in the European 5 by 20-millimeter fuse. Able to provide the same performance, it requires about 35 percent less space than the 3AG-sized fuse. As a result, it is rapidly gaining popularity in this country.

The use of European-style 5 by 20 fuses is not without its problems, the most important of which is the possibility of incorrectly-rated fuses being installed in the field. This problem stems from differences between the standards used to rate fuses in the U.S. and Europe. The U.L. fuse standard, for example, requires a conventional fuse to carry 135 percent of rated current for no more than 60 minutes while the European IEC (International Electrotechnical Commission) standard requires a fuse to carry 150 percent of rating for no less than 60 minutes. Moreover, whereas U.L. recognizes only time-delay and non-time-delay fuses, the IEC recognizes five different categories of clearing time--very fast acting Type FF, quick acting Type F, medium time lag Type M, time lag Type T, and long time lag Type TT. As a result, domestic and European fuses with the same current rating may behave quite differently.

From a logistical point of view, the existence of different kinds of fuses that are physically identical and have the same ratings can create severe headaches. With this in mind, work is now in progress under the sponsorship of IEC to develop a universal fuse link standard that meets the requirements of all Western safety agencies. This work is being conducted by Working Group 6 of the IEC's miniature fuse committee (32C). However, while progress has been made, it is unlikely that such a universal fuse will come into existence in the near future. One of the major obstacles is the not-invented-here syndrome coupled with nationalistic concerns. Even so, the need for a universal fuse link is so great that this and other problems will eventually be overcome. However, when it is, the resulting fuse will most likely be intended for use in newly-designed applications and not as a replacement for existing fuses.

Circuit Breakers

Even though fuses are inexpensive, even when the cost of a fuse holder or fuse clips is included, they present the user with the need to maintain an inventory of spare fuses. Moreover, in a case of a board-mounted fuse, replacement may also require some disassembly. True, if the fuse clears on account of a real fault condition, disassembly is required to affect repairs. However, if the fuse clears because of a transient or self-healed phenomenon of some sort, the down time resulting from disassembly or just hunting for a replacement fuse can be costly. Of greater importance, replacing a fuse in the ac power line can be hazardous if the proper precautions were not taken, especially if it is mounted in fuse clips rather than an enclosed holder. For these reasons, many designers prefer to use circuit breakers rather than fuses. Although more expensive and often less precise, they can be reset by simply pushing a button or flipping a switch. Moreover, some of the newer miniature breakers require no more space than a conventional fuse holder.

Circuit breakers for electronic applications come in two basic varieties-thermal and magnetic. Thermal breakers are nothing more than a variation of the bimetallic thermostatic switch adjacent to a small heating element through which the load current flows. Excessive current flow results in excessive temperature and that in turn results in the bimetallic switch opening the circuit. Magnetic breakers, on the other hand, are based on a solenoid-like action in which the field strength acting on the movable core is proportional to the load current. To prevent inadvertent trips due to transients, a time delay is introduced, usually through a hydraulic damper. Positive switching action is achieved by means of mechanical latching.

Thermal breakers, as might be expected, cost much less than equivalent magnetic breakers, but also perform with less precision. The two biggest shortcomings are restability and thermal sensitivity. Once opened, the breaker cannot be reset until the temperature of the heating element drops sufficiently for the bimetallic strip to close. While this may actually be an advantage if a real fault exists, it can be a pain at times. Because their operation is temperature dependent, thermal breakers are more likely to trip on transients when the ambient temperature is high than are magnetic breakers. Practical considerations also limit the maximum current thermal breakers can carry.

Magnetic breakers are relatively expensive because of their complexity and the cost of the solenoid mechanism. However, they are virtually immune to the effects of ambient temperature and provide very positive switching action. Moreover, magnetic breakers can interrupt a circuit much faster than thermal breakers, thereby limiting the let-through current more effectively.

Current-Limiting Thermisters

Because the ac line is presented in switch-mode power supplies with the extremely low impedance of discharged storage capacitors, the inrush current when these supplies are first turned on is very high. Aside from the stress this places on the input components, it can blow the input fuse or open the input breaker. To prevent this, virtually all high-power and many low-power supplies have so-called soft-start inputs that limit the inrush current. In high power supplies, the current limiting is provided by SCR or TRIAC circuits, but at lower power levels negative-temperature coefficient (NTC) thermistors are often used. Made of various oxide materials, these thermistors present the ac line with an impedance high enough to limit the inrush to acceptable levels. Over time, the current flowing through them generates heat which raises their temperature and that, in turn, lowers their resistance until equilibrium is reached. The cold resistance of typical current-limiting thermistors ranges between 1 and 10 ohms with the hot resistances usually about 5 percent of the cold resistance.

With a hot resistance typically just a few tenths of an ohm, the effect of these thermistors on the circuit after the input current stabilizes is minimal. Even so, some designers would prefer even lower hot resistances and might be tempted to parallel them. Unfortunately, this approach does not work. The problem is that thermistors of the same rating are not exactly identical, even when they come from the same batch. Because of this, their resistances drop at slightly different rates and stabilize at slightly different values. Since more current flows through the thermistor with the lower resistance, the "I squared R" losses in it will be higher. This further reduces its resistance, further increasing the current flow and the "I squared R" losses, eventually leading to a form of thermal runaway. The bottom line is that one of the thermistors will burn out, forcing all the current through the survivor. Depending on its size and value, it too may very well burn out.

Some designers try to sidestep the problem by designing the circuit so that the thermistor runs hotter than normal when the nominal input current is flowing. Since the hot resistance is dependent on temperature, this hotter operation results in lower resistance in series with the input circuit. There are consequences to this approach, however. A universal rule of thumb in electronics is that higher operating temperature implies shorter operating life, and this is as true for thermistors as it is for any other component. Of equal consequence, the heat generated by the thermistor is bound to raise the temperature of nearby components. And considering how compact new designs are, there are bound to be quite a few components nearby. Although most thermistors can be operated at temperatures well above 200 degrees C, most manufacturers recommend operation be limited to lower temperatures--typically 150 degrees C.

Although the peak inrush current is a function of the thermistor's resistance, it is also a function of the ac line voltage. This is of no importance if the supply is designed to operate from just one input voltage. It can be a problem, however, for dual-input supplies because the peak inrush current will be double at 230 volts what it is at 115 volts. In low power supplies, this probably means the inrush will jump from 20 amps to 40 amps, still no big deal. But at higher power, it can mean jumping to 150 amps or more. While thermistors cannot be paralleled, they can be connected in series. It is possible, therefore, to connect a second thermistor into the circuit when the input is configured for 230 volts so that the peak inrush current is nominally the same regardless of the ac input.

EMI Filters

During the past few years, there has been a growing awareness of the need at least to minimize conducted EMI (Electromagnetic Interference). Although the FCC (Federal Communications Commission) and the IEC have developed standards for EMI emissions in residential and industrial environments, they apply only to the finished product. This means that there is no statutory requirement that a power supply or dc/dc converter meet any EMI standard. However, since switch-mode power converters generate the lion's share of the EMI conducted out of an electronic product into the ac power lines, many users now demand that the power supplies they build or buy be filtered. While this may mean nothing more than buying or building an appropriate low pass filter, choosing the right filter is not as straightforward as it seems.

The basic problem is that the electrical noise to be attenuated is not a simple signal. It is, instead, a complex spectrum of frequencies that exhibits itself as both common and differential-mode signals. Before the required filter can be characterized, the kind and amplitude of the signals to be filtered must first be determined. Unfortunately, this cannot be done through theoretical analysis. It requires careful measurement of the conducted noise under controlled conditions using instrumentation not normally available to the power supply designer. Moreover, because the wave forms generated by the switching action are squared, they are rich in harmonics that extend well into the megahertz region. This means the output noise is affected by component placement as well as circuit design. Thus, the measurements obtained apply only to the component layout of the unit tested. Even small changes in placement may change the EMI signal mix.

Once the spectrum analysis is complete, the designer can determine the amount of attenuation needed to bring the unit into compliance with the target EMI specification. However, simply thumbing through a catalog and picking an EMI filter providing the required attenuation will often as not end in figurative disaster. The reason for this is that EMI filters are characterized in a 50-ohm system. The impedances of the ac power line and the input of a switch-mode converter are neither 50 ohms nor purely resistive. The resulting impedance mismatches can significantly change the attenuation curve of the EMI filter such that the unit is no longer in compliance with the target specifications. For this reason, if no other, it makes sense to work closely with the EMI filter manufacturer, which not only has the required EMI/RFI test equipment, but also the experience needed to design the filter that best fits the unit in question.

Another important consideration is the placement of the filter. It should be at the point where the ac power line enters the shielded enclosure. In the case of products in which only the power supply generates electrical noise, an enclosed supply is highly desirable with the EMI filter as close to the ac input as possible. Most EMI filter manufacturers offer filter assemblies that combine the filter and the ac input receptacle in a single housing. Some even include on-off switches and fuse holders or circuit breakers in the assembly.

If the entire product is contained inside a shielded enclosure, an open-frame or pc-board supply can be used if the EMI filter is at the point at which the ac enters the enclosure. This means either using a separate filter or positioning the supply so that its ac input is immediately adjacent to the point of entry in the enclosure. Moreover, the filter itself should be housed inside a shielded box if any other circuit inside the enclosure emits electrical noise. If, as it sometimes happens, the power supply EMI filter is located some distance from the ac point of entry, the wire between that point and the filter can act as a very effective antenna. The energy induced in this antenna essentially bypasses the filter and is conducted out of the enclosure. The only solution other than repositioning the filter is shielding the connecting wires.



Described is a very simple method of obtaining audible output from the "Stud Sensor," Model SS-9434, made by Zircon International (also sold by Radio Shack as catalog number 64-2825). This device is extremely useful in finding internal framing members of walls, so that installation of shelf brackets and other wall-mounted furniture can be done properly.


Zircon's simple electronic Stud Sensor is actually a relative capacitance meter whose output is a column of light-emitting diodes (LED's). It is capable of detecting the change in dielectric constant caused by the density of a wall framing member--such as a vertical stud, ceiling joist, or horizontal fire stop. This is far superior to other stud finders which are metal detectors that look for nails or screws.

The Zircon device is a hand-sized plastic box whose rear surface is equipped with two strips of felt-like material to protect the surface being scanned. The back also contains the cover for the 9-volt battery. The front sports a vertical column of four LEDs in its top center just under a raised pointer line on the top of the unit. The left and right edges of the unit are slightly indented, with the on/off pushbutton being in the left side's concavity.

In operation, the Stud Sensor is moved left or right in search of the studs. As the finder approaches the edge of a big chunk of solidity inside the wall, successively higher lights come on with the detection of higher capacitance.

First, however, there is a "calibration phase" during which the instrument adjusts itself to the baseline amount of capacitance which will be seen between the studs. Therefore, the unit is held against a section of wall which is believed to be free of internal members; then the "on" button is depressed. For a period of about three seconds, the Stud Sensor "calibrates" itself to the piece of "standard" (meaning empty) wall that you've given it. During this "calibration" phase, the top-most LED stays on. When the calibration is complete, the top lamp goes out.

This stud finder does not tell you exactly where the edges of the framing members are. What you can detect is a position, either side of center, where the "capacitance meter" saturates--quits changing as you pass over the member. Using a little raised pointer on top of the instrument, you mark the place on both sides at which the highest lamp is lit; then you "bisect" the distance between those points to find the center of the member. (There are some situations where the capacitance never gets high enough to saturate the system, in which case the top-most light may not come on.) In other words, once a stable indication is reached, which is usually when the top LED is lit, the position of the little pointer on top of the unit is noted and marked. Without releasing the on-off switch, this same member is approached from the other side and the position similarly marked. When both sides of the member have been checked, the center between the two marks can be found with a ruler. One may now drill, nail, or otherwise bash away with confidence that there will be other than plasterboard and air behind the point of entry.

As mentioned in the Stud Sensor's instructions, there are cases where the framing members present a lower capacitance than the empty wall. A noted example is where the building has been insulated by foil-covered fiberglass. In this case, the indication of the framing members is reversed. In fact, it is necessary in this case to experimentally "zero in on a stud," and let the instrument calibrate on top of it; this will maximize its sensitivity to that situation.

It must be remembered that the unit will find any change in capacitance. As far as the user is concerned, it cannot tell wood from metal or any other substance. Therefore, we have not lessened the danger of drilling into wires, pipes, ducts and all the other little traps hidden in the walls of our secure little castles. While testing the finder on a portion of wall, you will sometimes find things which you wish were wooden studs, but which do not fit the expected pattern of framing. These objects should be treated with caution, since running a drill bit through an electrical cable is just too enlightening. It helps to become familiar with the construction of your house--such as the typical distance between structural members--so that things that appear in odd places will alert you to these dangers.

The Zircon Stud Sensor's detector is a capacitor whose plates are pieces of foil on one of its printed boards. A large central plate is flanked on both sides by two narrower vertical strips which are jumpered together. This sensor is located behind the top-most piece of felt on the back of the cabinet. During calibration, avoid this area with your hands, or the instrument will appear to malfunction.

No attempt was made to understand the operation of the capacitance sensing circuitry or of the automatic calibration process. As no circuit diagram came with the thing, and since it uses standard components (thank you, Zircon, for not erasing the chip numbers), the circuit of the "output board" (carrying the LED drivers) was merely traced. The LED's are driven from outputs of an LM324 quad op-amp which acts as a set of comparators. One input of each of these comparators goes to its assigned point on a resistor string; the free inputs of the comparators come from a nice analog signal (off the sensor board) whose level reflects the changing "capacitor." This analog signal is an inverse function of the capacitance; in other words, when the top light is on, the analog voltage is about 2 volts, and when the bottom light just comes on, the voltage is about 6 volts.

Adapting the Stud Finder

An obvious first try is to look for the LED's with a light probe. It is indeed possible to find individual lights on the box and hear them turn on and off; however, it is very difficult to move up and down the row of LED's--following their changes--while moving the finder back and forth across the wall. You might try this before modifying the finder; if you like it well enough, you can use the rest of this article to put yourself to sleep.

Another approach would be to give the user a series of tones to represent the various LED's. This is easy enough, but it usually means a set of diodes to isolate each driver output from its neighbors, and a selection of resistors.

We chose to run a VCO with the analog signal that drives the comparators. However, this left us with another problem: Other than by counting to five after you press the button, how do you know when the "calibration" procedure is complete? The Stud Sensor lights the top LED during this phase, and when that goes out the sighted user knows when to go hunting. The answer to this turned out to be automatically taken care of by the top lamp's comparator signal. Unlike the other comparators, this one gets the analog signal through a resistor (100K), so that a transistor can also operate on the comparator's input. Thus, the collector of a transistor pulls this input down until the calibration phase is over, at which time the transistor opens and allows the comparator to look at the analog signal.

The signal on this top comparator's input, then, is an ideal take-off point for our VCO. When you first press the "on" button, this point shorts to ground until the Stud Sensor calibrates itself, after which we have a voltage which "dips," or decreases, when the instrument sees an increase in capacitance.

Thus, we need a simple voltage-controlled oscillator whose frequency is an inverse function of input voltage, assuming you want a higher pitch in accordance with higher lights in the read-out column. Since the signal to be tested is coming through a 100K resistor (most of the time), we also want a circuit with an input impedance of at least ten times this to avoid loading the visual read-out system.

The circuit chosen uses the old standby 555; oh no, not again. Its configuration is the usual free-running oscillator, except that a PNP transistor has been slipped in between the charging resistor and the trigger/threshold pins on the chip. Placing the transistor here, with the 22K charging resistor as its emitter resistor, means that the input impedance will be quite high; the impedance at the base will be the emitter resistor times the transistor's gain. (Assuming a transistor "beta" of 100, the input impedance with 22K in its emitter will be 2.2 megohms.) Through a 1meg resistor as protection in case something goes wrong, this base is taken to the aforementioned analog signal on the output board of the Stud Sensor.

Add-On VCO Circuit

The negative lead from the finder (wire No. 1 of the ribbon cable) goes to circuit ground and to pin 1 of an NE555 timer chip. The plus finder lead (wire 7) goes through 10 ohms to VCC of this VCO. VCC is bypassed to ground by 100uF (positive of the electrolytic at VCC). The 555 has pins 4 and 8 tied to VCC. Pins 2 and 6 are tied together and go through 0.027uF to ground; pins 2 and 6 also go through 10K to pin 7.

Pins 2 and 6 also go to the collector of a 2N2907 general-purpose PNP transistor. The emitter of this transistor goes through 22K to VCC. The base of the 2907 goes through 1 megohm to the signal lead from the finder (pin 3 of the ribbon cable). Pin 5 of the 555 has no connection. Pin 3, the output, goes through 47 ohms (1/2 watt), then through the speaker to VCC.

Dissecting and Modifying the Stud Finder

The added VCO was built on a small piece of perforated board and attached to the front of the cabinet just below the LED's. Behind the nameplate is a convenient hole in the case which, after boring through the plastic nameplate, can be used to bring wires out to the add-on. Some sort of small coverlet should be fashioned to enclose the modification. We made a very temporary enclosure out of fish paper which is taped to the cabinet. Our added little board is just held onto the nameplate by double-sided foam tape.

On the back, just above the battery door, there are two screws hidden under one of the pieces of felt. (Actually, this "felt" seems to be female Velcro, and it has an adhesive backing.) Carefully remove this felt and put it aside so that you can re-install it later.

Underneath this felt are found four holes. Two round ones accommodate recessed screws which keep the halves of the case together. The other holes are rectangular slits exposing test points which are of no immediate interest.

Besides the screws, a projection with a lip serves to hold the top end together. Once the screws are removed, the two halves of the cabinet should be pried apart nearest the end of the battery compartment, so as not to damage the plastic "catch" at the other end.

It took a bit of force to get ours open the first time; there may be a couple of glued points along the sides. Eventually, however, the halves popped apart without noticeable harm. The on-off actuator will immediately dislodge and fall out. This is easy to re-install, although it takes a little examination to see how it is able to operate the little PC-mounted switch that it rests on. (This actuator doesn't really push in when you press on it; it rocks in a direction that forces a projection against the switch.)

There is a board in each half of the box; these boards are connected together by a 7-conductor ribbon cable. The front of the box contains the output board to which we will make three connections. Holding the unit with this front panel on the right, the ribbon cable will be seen to dive under the bottom edge of the output board, and to have its ends brought up through holes, whereupon they are soldered in the usual way. These connections appear as closely spaced points at the very bottom edge of the output board. With the LED's facing the table and the battery compartment toward you, these ribbon connections will be numbered 1 through 7, from left to right.

Counting from the left, the first pad is ground, the third is the analog input signal we wish to monitor, while the seventh and right-hand one is plus 9 volts. You will want to "tack" (solder) wires onto all three of these pads, being very careful not to bridge across to neighboring ones. These wires can be knotted or marked with tape so they can be identified after they have been brought through the case.

[The term "tack" refers to a soldering procedure which is done by laying a well-tinned wire against a previously made connection, then heating it until the solder melts and the new wire becomes involved in the joint. Usually, no new solder is necessary; however, the wire you are adding must be generously tinned (wet with solder), since no flux is available to clean the materials and promote "wetting."]

If the thought of tacking wires onto the board intimidates you, it should be possible to simply cut the ribbon cable in half and splice it back together again, including three wires to bring out for the adapter. Don't forget to re-insulate the new connections carefully.

Just below the bottom edge of the board, about 3/4 of an inch to the right of the ribbon cable, is a nicely placed hole (obviously intended for something that became unnecessary). The nameplate is smooth thick plastic tape which has been glued to the front of the instrument: it can be cut or punched to accept the wires.

The case of the finder can now be reassembled and the felt installed over the screws. (We chose to attach this felt to the battery door, so we could get back inside when we wanted to.)

The adapter/oscillator was constructed on a small piece of vector board measuring about 1-1/2 by 2-1/2 inches. This board also contains the speaker.

This speaker is a very thin, apparently ceramic, unit. Available from Jameco as part number TS30S, it measures about 1-1/4 inches square and about 3/16 of an inch thick. There are two nice mounting holes in diagonally opposite corners by which it may be screwed to the board. The solder points for this speaker are found near one of the corners which does not contain a mounting hole; they appear as small double dots on the edge of the round section.

As mentioned, the board is mounted on the outside of the finder with double-sided foam mounting tape; this will stick very well if you first clean the front of the finder with rubbing alcohol. We leave fashioning a cover to the creative scrounger. It is likely that half of some small plastic container or even one of those ubiquitous plastic "bubble packs"--in which so many things are shipped--will do. It would seem like overkill to install a complete project box on the front of the finder for such a small board.

Operation of the Adapted Stud Finder

Immediately when you squeeze the "on" button, the oscillator's highest pitch will be emitted. In three seconds or so, this pitch will drop, letting you know that the "relative capacitance indicator" is ready for business. If this is done in free space (away from the wall, and with your hands away from the sensor), you can test the instrument by noting that bringing your hand near the upper-most felt on the back will cause the pitch to rise sharply.

In actual use, knock on the wall to find a place that sounds unsupported, place the Stud Sensor against it, and squeeze the "on" switch. (Never let go of the switch until you are through using the finder, or you will have to go through that calibration phase again.) When the pitch drops, start moving the unit from side to side until you cross a stud that makes the pitch rise.

The spots you mark on either side will be the ones at which the pitch levels off. In other words, the framing members will look broader than they really are, with a relatively unchanging capacitance being seen when you're right over one. Once you find the edges of this "broad peak," you mark these, and the midpoint between will be the exact center of the stud. Good hunting, fillies.

Parts List

  • 1--NE555 or Jameco LM555N
  • 1--10-ohm 1/4 watt resistor
  • 1--47-ohm 1/2 watt resistor
  • 1--10K 1/4 watt resistor
  • 1--22K 1/4 watt resistor
  • 1--1megohm 1/4 watt resistor
  • 1--0.027uF disc or Mylar capacitor
  • 1--100uF 10-volt electrolytic capacitor
  • 1--speaker, Jameco TS30S
  • 1--Zircon "Stud Sensor," No. SS-9434


  • Jameco Electronics, 1355 Shoreway Road, Belmont, CA 94002; Phone: (415) 592-8097.
  • Zircon International, Inc., 1580 Dell Avenue, Campbell, CA 95008; Phone: (408) 866-8600 or 800-245-9265 (outside California)

Crystal Calibrators and Marker Generators Revived

Compiled by Members of the Smith-Kettlewell Staff


Crystal-controlled marker generators provide a cheap way for blind radiomen to find their way around the dial. In the age where consumer products have a "product life" of only two years, it is unlikely that talking frequency read-outs will be designed for many of the best-selling radios, and an independent means of determining frequency may be desirable. The devices described generate marker signals at selectable intervals; with a little detective work, you can find your way around the dial without having access to the visual display.


Back in the days of steel men and wooden radios, so-called "crystal calibrators" were very popular for everybody, sighted or blind. A standard configuration was to generate marker signals every 100kHz; the finicky radioperson could recalibrate his/her visual dial at the nearest marker, thus attaining an additional measure of read-out accuracy in the portion of the band being used.

Using crystal calibrators somewhat differently, the blind operator found them to be a way of frequency read-out. By counting markers from a known position, he/she could locate a portion of the spectrum without using the visual dial.

For the crystal-calibrators' original use, much effort went into making them accurate and drift-free. Crystals were housed in temperature-controlled ovens, and periodic recalibration to short-wave primary frequency standards was done by the meticulous radioman. Highly linear tuning dials--either optically encoded rotating discs or pushbutton entry systems--have eliminated the need for recalibrating radio dials. However, the blind operator's need for "markers" still exists, and some easy-to-build modern circuits are listed here.

Only one of these circuits is "trimmable" (precisely adjustable to radio-frequency standards such as radio station WWV, for example). Since what we're trying to accomplish is a series of "where am I?" signals, circuit simplicity is a main point of this article.

General Descriptions

Crystal oscillators can easily be fabricated using logic gates. Both TTL and CMOS circuits are listed here. An advantage of using gates, as opposed to more traditional circuits of linear devices, is that harmonics are generated which can be heard up in the 2-meter band. Thus, if you build a 1MHz oscillator, you will hear markers every 1 megacycle on your dial. (theory describes perfect squarewaves as being made up of only odd harmonics. Fortunately for us, no circuit is perfect, and even harmonics are present in all of them. It is true that the odd harmonics are stronger than even ones, however.)

What you hear, as you tune across each marker, is a continuous-wave carrier. Since there are lots of dead carriers on the band (stations about to commence, or broadcasting soft music, for example), it's nice to have a way to quickly identify the "markers" as unique. To do this, the marker generator can be pulsed on and off a few times per second; very few outside signals will be found doing that.

The circuits which pulse the markers will be called "interrupters" in this discussion. An interrupter is simply a slow-running oscillator which either interrupts the power to the crystal oscillator, or gates its output on and off.

Additional markers at various minor divisions can easily be created by putting the 1MHz crystal frequency through arrangements of flip-flops--"dividers." Nicely packaged sets of these "dividers" are available in the 7490 series chips. These allow you to operate on the 1MHz frequency by divisors of 2 and 5. Thus, you could have markers at the following dial intervals: every 1/2 megacycle (dividing by 2), every 250kHz (dividing by 4), every 200kHz (dividing by 5), every 100kHz (dividing by 2 times 5), every 50kHz (dividing by 20), every 40kHz (dividing by 5 times 5), every 20kHz (dividing by 2 times 5 times 5), and every 10kHz (dividing by 10 times 10).

The editor's favorite arrangement is to have the following divisors on a rotary switch: Position 1 is off. Position 2 gives you 1MHz markers; position 3 gets you 500kHz; position 4 is 100kHz; position 5 is 20kHz, and position 6 gets you markers every 10kHz. The "Fowle switching system" will also be described. Using four DPDT toggle switches, all combinations are available. (A nice feature for hams would be to have markers every 25kHz; these would land on the official band edges. Too bad--if you want this, you need another divide-by-2 flip-flop--dividing by 10 and then by 4. You get to design that one, "OM.")

Whichever circuits--or combinations thereof--that you choose, the eventual instrument can be housed in a small project cabinet of your liking. An antenna post must be provided so that a length of wire can be arranged near your receiver. The unit can either be battery powered, or run from AC mains; an adequate power supply circuit is given.


Like everything for blind folks, a system of this sort taxes your cleverness and detective ability. Exemplary situations will be described in this section.

You can picture using a marker generator as trying to find your way around and measure things while being placed in the middle of a Braille Ruler which is infinitely long. Long and short marks--major and minor divisions--give you precise information about relative distances, but you sort of have to ask around to find out whose neighborhood you're in.

It helps to know some details about your radio: If it has a band switch, you should write down the frequency ranges that each "band" covers. It will help you to know about "neighborhoods"; for example, there's a shortwave broadcast band that stretches from 9.5MHz to 10MHz, another from 11.5MHz to 12MHz, and another that goes from 15MHz to 16MHz. In the U.S., our Bureau of Standards broadcasts conspicuous signals at 10MHz (the top of one broadcast band), and at 15MHz (the bottom of another broadcast band). Thus, if I don't find a signal from the Bureau (station WWV) at either end of the band I'm searching through, I might suspect that I'm on the band from 11.5 to 12 megaHertz.

Dialing in a desired frequency is perhaps the easiest task. If I feel like tuning in a station at 9.530MHz, I set my marker generator to produce pulsating signals at 1MHz intervals; then, I fish around until I find the WWV signal which is at the upper end of a broadcast band. I then set the instrument to give me half-megacycle markers and tune down to the next signal below WWV; this gets me to 9.5MHz. Switching to the 20kHz position, I tune beyond the first mark above 9.500 ... you can guess the rest.

Suppose, of an afternoon, you hear some interesting talk about the 1988 Summer Olympics, and note that the station you are listening to is KBS (the Korean Broadcasting System in Seoul, South Korea). You reckon that you would like to re-find that station, rather than just leave it to a chance meeting. Other than being on short-wave, you don't know where you are; the only thing you can surmise is that, since it's still daylight, you must be somewhere above 9MHz (at least that's the way band conditions work out here in California).

Turning on the marker generator and throwing in successive dividers, you find that you hear a beat note--somewhere above the top of the piano, but not that of a dog whistle--when you have made the instrument place markers every 20kHz. (It helps to be able to recognize the difference between a 5kHz beat note and one of 10kHz, but you can do without this talent.) On this 20kHz position, you note that you hear this beat note on the high side of the station and not on the low-frequency side. Also, you note that you are not "centered" between 20kHz marks; you are closer to the upper one (and, since these stations are spaced at 5kHz intervals, you surmise that you are 5kHz below the nearest 20kHz mark).

Your next job is to count 20kHz markers over to a more significant one. Counting carefully, testing each one by switching to the 100kHz position, you note how many 20kHz marks you pass before getting to a 100kHz mark. (In this example, you will find a 100kHz mark 25kHz above the station, and another at 75kHz below.) By the same procedure, you count how many 100kHz markers the station is from the nearest 500kHz one. (In this case, you will find that this major half-megaHertz mark coincides with the 100kHz one below the station, and that you'll have to count four 100kHz marks to find the next highest). Whichever 500kHz mark you find, test it to see if it coincides with an even megaHertz.

In reviewing what we've done: No matter which fork in the road we took, we found out that the station was just short of 80kHz up from a 100kHz marker--one which turned out to coincide with a 1/2 megaHertz marker--with this major point not coinciding with a 1 megaHertz mark. Thus, we know that the frequency is "sum'm point 575" MHz." If we know something about our radio's band switch, we'll know that we're in the area of 15MHz. However, if we know nothing, we can count 1MHz markers until we find WWV (which, in this example, will be below our station). Through detective work, we arrive at the conclusion that Radio Korea has this daily broadcast on 15.575 megaHertz. (Of course if we'd listen, instead of playing with the radio equipment, we'd hear them announce that fact.)

All that sounds pretty easy. Doesn't it? Well, the catch is that it doesn't always work quite that well. When conditions are good, there will be stations that are so strong that they obliterate our markers. Sometimes this can be compensated for by giving the marker generator a longer antenna, or bringing it closer to the radio, but this has its limitations. Depending on the lack of sophistication of the radio receiver, more marker signal strength may clutter the band with false signals; called "birdies," these come from beating of the radio's local oscillator with markers that have no relation to what you're doing.

You can identify a birdie by oddities in how it is received. Often, it creates a heterodyne that moves across other signals as the dial is tuned. (Being a product of the radio's local oscillator means that it looks like a carrier that tunes with the radio. The fact that it moves in relation to other stations, and makes tweeting noises as it does so, is the reason for calling this aberration a "birdie.")

I have a little pocket short-wave radio which prohibits me from using any marker spacing closer than 100kHz because of false images and birdies. A high degree of coupling between the generator and this radio causes more trouble than I can tolerate. All this means is that I must refine my detective work. For example, if I tune for a fair distance without hearing a 100kHz marker, I just assume that I've missed hearing it.

Caution! If you lose patience because the markers are weak, resist the temptation to directly couple the marker generator to the receiver's input. Those markers may be weak at the frequency you're listening on, but be assured that the output of the generator has a 5-volt or 9-volt swing at low frequencies. That can cook your radio for breakfast!

Harmonic Analysis a la Al Alden

Analysis, in the "frequency domain" of any periodic waveform, shows that it is comprised of harmonically related sineusoids. The amplitudes of these components come together to create the shape of the original signal.

It helps to know something of the expected patterns of harmonics; this will explain the difference in signal strength between various markers of these generator circuits.

Where periodic pulses of a specified duty cycle exist, the harmonics' amplitudes are described by the expression: the sin of the angle N times pi times D, this divided by N times pi times D, where "N" is the "number of the harmonic" (first, second, etc.), and "D" is the "duty factor" (the fraction of the period). Two "duty factors" are of interest in this article: Where the output is a squarewave--having a 50 percent duty cycle--D equals 1/2. If the output is taken off one of the divide-by-five counters, D will equal 1/5.

Ignoring the denominator for a moment, we see that the numerator is a sineusoid. Whenever the "angle" (N pi D) is an integer times pi, the sin goes to zero. At the output of a divide-by-five counter, the multiplier of pi will be an integer whenever "N" is a multiple of 5; considering the fifth harmonic, 5 times 1/5 is the integer 1, and 10 times 1/5 gives us an integer of 2. Therefore, we can see that the amplitude of every fifth harmonic will theoretically be zero. In the case of the squarewave, which comes out of a divide-by-two section, every even "N," when multiplied by 1/2, will be an integer; a squarewave is theoretically made up of only odd harmonics.

Looking at the denominator (N times pi times D), we note that the sinewave envelope gets weaker and weaker with increasing N. The implication of this is that, as the divider system is set to generate finer divisions, the harmonics heard at a given frequency will be significantly weaker.

It's lucky for us that circuits misbehave, and are not "ideal" in their performance. The harmonics at supposed "nulls" are still there--although weaker than their neighbors--and the outputs of these various circuits are still hearable at 150MHz.


Each of the various approaches here have their advantages and disadvantages. For example, you can buy packaged crystal oscillators whose connections are trivial; however, they draw lots of current, and their frequency settings are not "trimmable." A lovely trimmable crystal oscillator circuit, built around a CMOS inverting gate, is listed here; however, it requires a variable capacitor, a crystal which costs more than one of the prepackaged oscillators, and is therefore expensive and hard to build.

Two fully designed implementations will be described here. The advanced builder will see how different approaches can be taken, and is invited to take bits and pieces from the sample circuits to design his favorite instrument.

The SaRonix Crystal Oscillator Modules

Contained in hermetically sealed metal cans, these 4-pin modules are intended to fit into the space of the standard 14-pin DIP IC. (The four pins are at the corners, and occupy the positions of pins 1, 7, 8 and 14.)

They are available in either CMOS or TTL forms. No distinct advantage is gained by using the CMOS versions; they still can drive ten standard TTL loads, and their current drain is 10mA (as opposed to 20mA for the TTL ones). From Jameco, the TTL 1MHz module is under $4 (cheaper than the $5 naked crystal), while the CMOS 1MHz module costs about $5.

Their disadvantage is that the frequency is not intended to be adjustable. By acts of violence, I was able to pull my TTL one's frequency just a little. By denting the top of the module near pins 1 and 14 (which is apparently near the crystal), the frequency can be lowered slightly. Oddly enough, a similar dent near the pin 7 and 8 end seems to raise the frequency slightly. The case is rather thin metal, so there's no need to use a ball-peen hammer. Also, if you go too far, the case shorts stuff out in there, and the oscillator will stop (at least as long as you press on it).

The modules that Jameco has have a "frequency stability tolerance" (whatever that means) of 100 parts per million. Thus, at 15MHz you might be off as much as 1.5kHz. SaRonix lists models with other tolerances; for example, the TTL version can be gotten with a "frequency stability tolerance" of 25 parts per million, and the CMOS ones can be gotten with a tolerance of 50 parts per million.

The pin arrangement is the same for any of the units. Pin 1 is the metal can, and is usually grounded. Pin 7 is the negative supply pin, usually ground. Pin 14 is VCC--5 volts for the TTL units, and 4.5 to 7 volts for the CMOS ones. Pin 8 is the buffered output, and this can drive a 16mA load.

In Jameco, the numbering scheme is simple: The TTL 1MHz unit is called an OSC1. The CMOS 1MHz oscillator is called the CMS1. The SaRonix numbering scheme is as follows: The TTL units bear the prefix of NCT. Available frequency ranges are classified by a three-digit number: The 040 pertains to units between 250kHz to 3.999MHz; 050 is for 4.000MHz to 20MHz; 070 can be gotten from 20MHz to 60MHz (these being termed "overtone" in the data sheet). Six "frequency stability tolerances" can be specified by letters A through F: In parts per million (PPM), A stands for 25 PPM, B for 50 PPM, C for 100 PPM, D for 500 PPM, and F is for 1000 parts per million. Finally, preceded by a hyphen, the specific frequency is tacked on. Thus, for the Jameco OSC1, the SaRonix number is NCT040C-1.000.

For the CMOS ones, the prefix is NCC. The number 030 stands for the frequency range of 250kHz to 3.999MHz; the 060 is for the range of 4.000MHz to 24MHz. Frequency tolerances are: B for 50 parts per million, C for 100, E for 500, and F for 1000 parts per million. The frequency, preceded by a hyphen, is tacked on at the end. Thus, the Jameco CMS1 bares the SaRonix number of NCC030C-1.000.

Discrete TTL Crystal Oscillator Circuit

(This circuit is thrown in here for the sake of completeness.) Two inverting gates are used; the inputs of each gate are tied together. Each gate has 1.5K connected across it. From the output of the first to the inputs of the second is 0.0047uF. The crystal goes from the output of the second back to the inputs of the first.

"Interrupter" for Non-Gatable Oscillators

In this system, the negative supply terminal of the crystal oscillator is operated by the output of a 555. Pin 1 of the 555 is grounded; pin 8 goes to VCC. Pins 2 and 6 are tied together and go through 0.1uF to ground. Pins 2 and 6 also go through 1 megohm to the 555 output, pin 3. The VCC pin of the crystal oscillator goes to VCC; the negative supply pin of the oscillator goes to pin 3 of the 555. Pin 4 of the 555 can be used to disable the "interrupter"; when pin 4 is high, the interrupter is working, but when pin 4 is tied low, power to the oscillator is constant.

Frequency Divider Systems

The 7490 series "BCD counter" is ideally suited for our purposes--dividing by 2, 5, or 10. In the 7490's intended use, the cascaded flip-flops count from 0 through 9, triggered by negative transitions of the clock pin. However, this "string" can be broken in two--an initial division by 2, followed by a counter which counts from 0 through 4. (In other words, followed by the 0 through 1 section, an arrangement of three more bits takes care of 2, 4, 6 and 8; the second section resets before the binary version of 10 can be reached.)

The two sections, seen separately, divide a clock by 2 and 5, respectively. Thus, we can use this chip as a whole (to divide by 10), or we can switch in the sections separately (to divide by 2 or 5). The most straightforward switching arrangement is to provide a switch for each divider section as follows:

The Fowle Divisor-Selection System

A DPDT 2-position switch (without center-off) is used for each divisor. (More than the four shown could be included.) The output of the crystal oscillator goes to the arm of S1 section A. The arm of S1 section B goes to the arm of S2 section A. The arm of S2 section B goes to the arm of S3 section A. The arm of S3 section B goes to the arm of S4 section A, with the arm of S4 section B going through 0.05uF to the antenna.

Each switch has an "out" position and an "in" position (the latter signifying that its divider is switched "in"). On each switch, the "out" positions of sections A and B are jumpered together; thus, if all the switches are in the "out" position, the signal from the crystal oscillator will go to the antenna. As for the "in" position, section A goes to the clock input of its divider, while section B goes to the output of the divider.

All "Set" and "Reset" pins should be grounded. On the 7490, 74LS90, and 74C90, there are two "set" and two "reset" pins. Oddly enough, however, these are not parceled out as one pin for each section; rather, they are combined in an AND gate, and each pair is common to both sections. In the dual version of the 7490, the 74390, each half has only one "reset."

The above Fowle system will work with TTL or CMOS. However, as with all CMOS implementations, inputs must never be left floating. Therefore, besides going to contacts of respective switch sections A, the clock inputs should also go through 100K resistors to VCC.

The Gerrey Marker-Selection Scheme

The editor wanted a simple rotary switch to select his favorite markers. This was accomplished by arranging all the dividers in a string--with the output of this string being coupled to the antenna--and then just "stepping" the 1 megacycle driving signal along the string with the switch. If 2-input AND (or NAND) gates are used to couple the various sections of the 7490's together, you create a free input at each junction which can go to its appropriate position on the switch; the output of the crystal oscillator goes to the swinger.

One problem has to be conquered: What if, just by chance, the output of the previous flip-flop--on the other input of your desired AND gate--happens to rest at zero? In that case, the driving signal will be gated off. A simple solution to this is to "set" the flip-flops frequently and periodically; this assures that interruption of the signal won't be for long. The interrupter can provide this "set" signal. The only sacrifice is that it is not practical to turn the interrupter off any more, or that random flaw in the system will be retained.

Since the eventual design has very few components other than the IC's, it lends itself to wire-wrapping; a wire-wrap table to follow will serve in place of a detailed circuit. For now, the circuit can be summarized as follows:

The device is powered from a 5V supply, capable of supplying 200mA. One of the crystal oscillator modules is used; its negative supply is operated by the 555 interrupter listed above. The output of this interrupter also goes to the "set" terminals of two 7490's. The "reset terminals are grounded.

A 7400 quad NAND gate is used for coupling. The "final" gate in the series has its output going through 0.05uF to the antenna. The Q0 output of the second 7490 goes to one input of this NAND gate, while the other input goes to position 2 on the rotary switch. (Position 1 is "off," while position 2 is for 1MHz markers.) The clock of this Q0 flip-flop (the section that divides by 2) goes to the output of a previous NAND gate. One input of this gate goes to the Q3 output of the 7490, while the free input goes to position 3 on the rotary switch. (Position 3 divides 1MHz by 2, getting 500kHz markers.) Preceding this 7490 (on the remaining clock terminal) is another NAND gate driven by the Q3 output of another 7490, and so on.

Using a 2-pole 6-position switch, one pole can turn the system on for positions 2 through 6. The other pole gives you markers of 1MHz, 500kHz, 100kHz, 20kHz, and 10kHz, respectively.

Wire-Wrap Table for the TTL Marker Generator

  • NE555 Interrupter:
    • 1 to ground.
    • 2 to 6: also through 0.1uF to ground.
    • 3 through 1 megohm to 2.
    • 4 and 8 to VCC.
  • Jameco OSC1 or SaRonix NCT040C-1.000 Crystal Oscillator: (Between pins 7 and 14 is 0.05uF.) 1 to ground.
    • 7 to 555-3.
    • 14 to VCC.
  • Both 7490's, BCD Counters used as dividers: (Between pins 5 and 10 is 0.05uF.)
    • 2, 3 and 10 to ground.
    • 5 to VCC.
    • 6 and 7 to 555-3.
  • 7400 NAND Gates: (Between pins 7 and 14 is 0.05uF.)
    • 2 to first 7490-12.
    • 3 to first 7490-1.
    • 5 to first 7490-11. 6 to second 7490-1 7 to ground.
    • 8 to second 7490-14.
    • 9 to 7490-11.
    • 11 through 0.05uF to antenna.
    • 12 to second 7490-12.
    • 14 to VCC.
  • Rotary Switch (2-pole 6-position), Section 1:
    • Position 1 is open.
    • Position 2 to 7400-13.
    • Position 3 to 7400-10.
    • Position 4 to 7400-4.
    • Position 5 to 7400-1
    • Position 6 to first 7490-14.
    • Arm of Section 1 to Crystal Oscillator's pin 8.

Finally, positions 2 through 6 of section 2 go to the input of a 7805 5volt regulator. The input of the regulator is bypassed to ground by the parallel combination of 0.05uF and 1000uF(negative of the electrolytic at ground). The output of the regulator goes to the VCC line of the project. This line is bypassed to ground by the parallel combination of 0.05uF and 470uF (negative of the electrolytic at ground). The "Common" terminal of the regulator is grounded. The arm of switch Section 2 goes to the positive output of a bridge rectifier, with the negative bridge output being grounded. The bridge, in turn, is being driven by a 12-volt transformer whose primary is fused and runs off the mains.

As a final note on these TTL versions, you should know that the editor has noticed false signals--parasitic oscillations, internal modulation products, or whatever--which have not been eliminated. I suppose there's nothing to stop you from trying ferrite beads on power lines to the chip, or other such fanciness. However, I could usually tell which signals were real and which were not. False signals often sounded "hummy" (producing rough beat notes), and they were generally weaker than the intentional markers. You know that something is wrong when, on the 500kHz position, you get randomly spaced rough-sounding phantoms which are rather weak and too frequent. It helps to know that, when you select markers which are widely spaced, they will be nice and strong (since you are orders of magnitude closer to the fundamental).

The Battery-Operated Fully CMOS Crystal Calibrator

This device is more along the lines of a proper "crystal calibrator"; its frequency can be adjusted to a primary standard (such as one of the various short-wave "time and frequency" radio stations). Its frequency does change somewhat with sagging battery voltage, however. In addition, the isolation of the oscillator is not as good as you would hope on the 1MHz position. This is probably due to the fact that only one gate is used as a buffer, and this is on the same chip as the crystal oscillator. It is as stable as quartz when the battery is alive and when you switch in at least one divider.)

The 1MHz crystal used is sold by Jameco under the catalog number of CRY1. It is said in their blurb to be cut with 13pF "shunt capacitance" in mind. You can see from the circuit that across the crystal is a 33pF fixed capacitor in series with a 35pF trimmer; thus, the range of the combined shunt capacitance is perhaps from 4 to 18 picofarads. If your trimmer ends up being near one end when adjusted, you can change the value of the fixed unit to compensate.

Because you might use them elsewhere, the various portions of the circuit bear individual titles so you can easily find them.

Trimmable CMOS Crystal Oscillator Circuit

A CD4011 quad NAND gate is used. (A quad NOR gate could also be used, but the switch to turn off the "interrupter" would have to be re-arranged.) Pin 7 of this gate is grounded; pin 14 goes to VCC (from 4.5 to 15 volts). The gate inputs, pins 1 and 2, are tied together.

Across this gate, between the input pins (1 and 2) and the output (pin 3) is a 22meg resistor. These inputs go through 33pF to ground. These inputs also go to one side of the crystal; the other side of the crystal goes through a 35pF variable capacitor to ground (the rotor, or the adjustment screw, at ground). The output of the gate, pin 3, goes through 22K to the junction of the crystal and the trimmer.

Pin 3, the output of the oscillator gate, goes to one input of another gate (pin 5 on the 4011). Pin 6 awaits a signal from the interrupter; if tied high, the output is constant, while operating it with the interrupter makes the signal pulse on and off. Pin 4 of the 4011, the output of this second gate, is the 1MHz buffered output of the oscillator.

"Interrupter" for the Trimmable CMOS Oscillator

Two unused NAND gates of the 4011 are cascaded to make an RC slow-running oscillator. NOR gates would do just as well, but the switching to disable the "interrupter" would have to be re-arranged.)

The output of one gate, pin 11 of the 4011, goes to both inputs of its neighbor, pins 8 and 9. Across the first gate, from pins 12 and 13 (which are jumpered together) to pin 11, is a 1meg resistor. Pins 12 and 13 also go through 0.1uF to the output of the second gate, pin 10.

Pin 10, the output of the interrupter goes through 100K to pin 6, the gating input of the buffer in the crystal oscillator. This pin 6 also goes through an SPST switch to VCC. When this switch is closed, the instrument produces an uninterrupted signal.

Selectable Divider System

Two 74C90's were used in our instrument, although a single 74C390 would do. Assuming the use of 74C90's: On both chips, pin 10 (the negative supply pin) is grounded; all the "Set" and "Reset" pins, 2, 3, 6 and 7, are grounded. Pin 5 (the plus supply pin) goes to VCC.

Pin 14 of the first 74C90, the clock of the single flip-flop, goes through 100K to VCC. The output of this flip-flop, pin 12 of this same 74C90, goes through 100K to pin 1, the second section. The output of the latter section of flip-flops, pin 11, goes through 100K to pin 1 of the second 74C90. Pin 11 of this second 74C90 goes through 100K to its own pin 14. The output of the last divider is pin 12.

A 2-pole rotary switch is used for the selection. This could just as well have only five positions, but six positions were close at hand. One pole connects the crystal oscillator to clock terminals along the string. The second pole is needed to switch the antenna from the output of the divider string to the crystal oscillator's output in the 1MHz position.

The arm of Section 1 goes to pin 4 of the 4011, the buffered output of the crystal oscillator. Positions 1 and 2 of Section 1 are left open. Position 3 goes to pin 14 of the second 74C90, while position 4 goes to pin 1 of this same chip. Position 5 goes to pin 1 of the first 74C90, while position 6 goes to pin 14 of this chip.

On Section 2 of the switch, position 1 is grounded. Position 2 goes to the output of the crystal oscillator (which is conveniently available now on the arm of Section 1). Positions 3, 4, 5 and 6 are jumpered together and go to pin 12 of the second 74C90, the output of the divider string. The arm of Section 2 goes through 0.05uF to the antenna.

The negative side of a 9-volt battery is grounded. The positive side of the battery goes through an on-off switch to the VCC line. This VCC line is bypassed to ground by the parallel combination of 0.05uF and 100uF (negative of the electrolytic at ground).

Pin Assignments

  • 7490 series:
  • Note: The pin arrangement for the 7490A (not given here) is completely different from that of the 7490, 74LS90, or 74C90. Also note that the pairs of "Set" and "Reset" pins are just inputs of AND gates; they are not divided up between the "sections" of flip-flops as you might expect.
  • Pin 5--VCC
  • Pin 10--Ground
  • Pins 2 AND 3--Reset
  • Pins 6 AND 7--Set
  • Pin 14--Clock for FF0
  • Pin 12--Q0, the output of FF0
  • Pin 1--Clock for FF1
  • Pin 9--Q1, the output of FF1
  • Pin 8--Q2, the output of FF2
  • Pin 11--Q3, the output of FF3

74390 Dual Versions: To avoid confusion of numbers, the two identical sets have been listed under the headings "Sections I" and Section II." Also note that only one "Reset" terminal is listed for each set.

  • Pin 8--Ground
  • Pin 16--VCC
  • Section I:
    • Pin 1--Clock for FF0
    • Pin 2--Reset
    • Pin 3--Q0, output of FF0
    • Pin 4--Clock for FF1
    • Pin 5--Q1, output of FF1
    • Pin 6--Q2, output of FF2
    • Pin 7--Q3, output of FF3
  • Section II:
    • Pin 15--Clock for FF0
    • Pin 14--Reset
    • Pin 13--Q0
    • Pin 12--Clock for FF1
    • Pin 11--Q1
    • Pin 10--Q2
    • Pin 9--Q3

Address List

• Jameco Electronics: 1355 Shoreway Rd., Belmont, CA 94002; Phone: (415) 592-8097.

• SaRonix: 4010 Transport St., Palo Alto, CA 94303: Phone: (415) 856-6900, or, (800) 422-3355.


By Bernie Vinther

Schematic Drawings

We blind electronics techs and engineers often have to discuss circuit diagrams with our sighted counterparts--as if we were on the phone or the old "2-way" radio with them--in order to describe things. Most of the time this is not too much of a problem, but when it comes to talking about, and remembering, complex circuits, I find that it becomes a complex and frustrating problem. Also, to my mind, raised-line drawing kits simply aren't good enough.

Your editor suggested that I use a "magnetic board. This involves cutting schematic symbols out of cardboard and mounting them on magnetic strips. Then, long slender pieces of various lengths could be used to represent the connecting wires from symbol to symbol. But I quickly found out that the connection pieces were seldom the right length, and it was impossible to tell the difference between cases where two wires merely cross, or are electrically connected.

I had already experimented with symbols made from cardboard which I pinned to a large cardboard sheet. Here I used elastic string of various lengths which I could stretch and pin as needed in order to make connections. I used pins with large heads to represent where wires electrically connect, rather than just cross. However, I soon found that the cardboard pieces were difficult to braille for their function or value, and in time, they and the backing they were pinned to became ragged with pin holes.

A friend of mine showed me a felt-covered board you could use with Velcro or pins. But at the price, over $100 for a 3-foot by 6-foot piece, it was a product only for people who had money to burn.

Finally, I believe I found the solution which we can all benefit from. I found some 3-foot by 4-foot cork bulletin boards which I cut into 2-foot by 3-foot pieces. For the schematic symbols, I went to Radio Shack and bought some 6- by 8-inch plain unplated pieces of perforated board (Radio Shack stock #276-1396A); a friend of mine then used a band saw to cut this material into the shape of electronic symbols.

Most of the pieces, such as op-amps, buffers and gates, I cut to a 1- by 1.25-inch size. For transistors, LEDs, SCR's, etc., I cut discs by using a 1-3/8-inch (1.375-inch) hole saw on a drill press with the pilot drill recessed. Symbols for things like 555 and 556 timers and other IC's were made rather large, but in modules of only 6 and 8 pins. In this way, I can make up IC's with 6, 8, 14, 16, 18, and 22 pins or more by joining them together. The shaping of the pins was made trapezoidal, that is, shaped like pyramids with flat tops. I made the space from one pin to the next to be 0.8 inch so that two 8-pin chips could be cut out of each board; this left remaining scrap pieces which could be used for smaller parts such as resistors, diodes, etc.

The resistors, by the way, I cut into something more like their physical shape, rather than their zig-zag schematic symbol, so that they could be brailled and written on for blind and sighted colleagues.

Since the schematic symbols are made from perforated board, they are perfect for brailling and writing on. Sticking them in appropriate holes, I use straight pins with small round plastic heads to make Braille symbols. Inputs, outputs, and power supply connections can all be clearly marked with crayon or grease pen for both sighted and low-vision users, and changed as needed. I also use the pins to make bends in the "elastic wires."

To show where wires join rather than merely cross, I use pins with large heads. I also use these large-headed pins to represent bubbles for inverter symbols on the inputs and outputs of logic gates.

Oh yes, I almost forgot to mention that I have to cut the pins off on the back side, otherwise they become a real "pain" at times.

I keep all the pieces in a 24-compartment flat box with the walls of some of the compartments cut out to make room for the big IC pieces. I keep the pins in a 35mm film container inside the box. Thus, all the pieces, including the board, are easy to carry, set up and change.

I found that drawing symbols for transistors, LED's, SCR's, etc. with silicon rubber on the round pieces is a lot of help in identifying the symbol and knowing which lead is which. Things like JK flip-flops I made from rectangular pieces about 1.5 by 2 inches. I also cut some slender pieces with an arrow on one end (about 2 inches long) that I can use for the "swinger" on toggle and rotary switches. Switch contacts are made from 1/2 inch diameter pieces which I also cut by using a smaller hole saw. I can pin these in place as needed; then, by tying several knots along the elastic string and running it to another switch arm, I can show that the two sections are part of the same switch.

The point of this technique is to get something "quick and easy" so that everyone can "read" the circuit. Where possible, I tried to make parts look much the same as they do in a schematic diagram. Even when I deviate from this, as in the case of the resistors, I try to make the symbol obvious, and I make sure that it is well marked.

Timing Diagrams

When it comes to timing charts or diagrams for digital circuits, I believe I found a unique solution to the problem. I went to the hardware store and bought some 30-inch long "piano hinges"; I cut these into two equal lengths. I then took a saw and made several cuts from the edge of the hinge to the place where the hinge is joined together. I repeated this step down the entire length of one side of each hinge. Thus I had a hinge with a single piece on one side and several movable pieces (flaps) on the other side.

Using the contiguous side of the hinge for mounting, I simply pin these hinges down on my cork board, orienting as many as I need horizontally, one above the other. I then can use each hinge "flap" to indicate high and low conditions of clocks, timers, counters, gates, flip-flops and so on. by flipping the hinge pieces up or down for highs and lows (or ones and zeroes), I can keep track of timing pulses. Each length of hinge is used to represent a different part of the circuit, and with one hinge placed in line above the other, you can see what the timing relationship is from one to another. One flap represents a narrow pulse, perhaps at the master clock frequency; longer pulses are depicted by bringing up adjacent flaps, as many as you need for a given length of pulse.

Because flipping these hinges is quicker than drawing, and because they can be instantly changed, the sighted students, my instructor, and my electronics friends generally like this better than pencil and paper.

I'm sure that there is room for improvement to make my madness diagrams. If so, please submit your ideas to Bill Gerrey or to me:


Bernie Vinther

Phone: (509) 586-8060.

Or, send tape or letter (no Braille) to:

915 West Grande Ronde

Kennewick, WA 99336.


First, I'd like to salute the southern latitudes of this world for having the good sense to schedule their Fall season in accordance with that world-famous organ of science, The Smith-Kettlewell Technical File. I personally think this is a great time for Fall; all that pesky rain and sleet is behind me, and I can celebrate Christmas and Easter at the same time. (Christmas trees are cheaper this time of year, and colored eggs make grand ornaments, as long as you don't knock them off.)

I'd like to thank both my colleagues and my readers for contributing articles (and ideas for more articles) to help me get back on track. Not only do old editors get writer's block, but old engineers get "road block," and I was in need of inspiration for a season or two. The track we're on has tickled my fancy (I don't know my fancy what), and we're rolling again. I apologize for my slow caboose.

I think this thing on "Visual/Nonvisual Aids" that Bernie Vinther kicked off will amount to something. As you may remember, I had dreams of writing up material about how to communicate with readers, and studious stuff like that. I probably have things to say on that subject--try and stop me. However, if this topic--functioning as an integrated part of the mainstream--is to be treated fairly, it will benefit the most by opening all of our tool boxes in a forum. Bernie suggested that his treatise be put in a section called "Cross Talk," a forum section where ideas would be exchanged. I didn't follow his wishes this time because I felt that the things he had to offer could stand by themselves; no fair putting them in a subsection that made them look like my brain children.

But, on Bernie's suggestion, such a forum shall be instated. I think "Cross Talk" is being used some place, so how about the "Tech-File Forum"? More than in "Hints and Kinks," the "Tech-File Forum" can harbor lively debates, and be a place where phone numbers can be exchanged.

Back on the subject of working with sighted colleagues, isn't it true that, in the age of a computer revolution, old procedures and methods need a second look. As I review my kickoff paper, "Making Technical School Accessible to the Blind," SKTF, Fall 1984, I cannot ignore that aspects of it are dated, and thank the modern age for that.

Come on, you Kats and Katrinas, how are you doing it? How are you getting your reading done? What are your study procedures? How is your office organized? Do you get anything on computer diskette, and how is that working out? What kind of "hard stuff" are you able to down-load from data bases?

* * * Turning to the subject of instrumentation, this will take three parallel directions. All of these belong in a technical magazine, and they are listed as follows:

The fact that we can now build our own limited-vocabulary digitized speech, and thus steer clear of short-lived commercial speech products, will make filling these pages with digital meters worth the trouble. A new National voltmeter chip (the ADC3511) is now being interfaced with the "Ram Talker" in this laboratory. In addition, Tom Fowle is working on systems of connecting the Ram Talker to instruments with visual digital displays.

There is a ready-made technology that's waiting to come down in price; analog-to-digital attachments for computers will soon be cheap enough so you can make your talking or Morse Code computers do the work of your test instruments. If you are using such equipment, please send me particulars.

Because talking--or otherwise readable--voltmeters are something we are all likely to get, stuff to feed into them will be of interest. Transducers for measuring most any physical quantity can be had, and the circuitry for driving these transducers is now appearing in chip form. Whether monitoring the conditions in your swimming pool, or measuring finely machined parts, something for you will come of age and will be printed here, so keep subscribing.

* * * Consistently, now, this fine magazine exists on a diskette. It is probably time to offer it in that medium. The question now arises, in what formats? (If Apple is to be issued, I'll have to do something clever to arrange for that, since CP/M and IBM are what I have direct access to.) Which format speaks to your equipment, and what do you think of the idea? Write me two lines on a post card: First, would you want SKTF on disk? Second, even if you don't, what breed of files can you play (just so we know what kinds of adapted computers are out there)?

* * * Above all, see that this magazine finds its audience and/or haptic readership. Boy! wouldn't my financial concerns evaporate if the subscribership were large enough.

We cannot afford to be an obscure minority. Every blind person trying to enter technical professions is haunted by obstacles like, "Do a little research and find out how many other 'people like you' are able to make it." If we slip from view or lose track of each other, we'll trample the field of newcomers by way of our obscurity. From the selfish point of view, the more identifiable technically minded blind people there are, the more likely we are to get equipment designed and made for us.

You who read this are loyal supporters and good soldiers, present and accounted for. Thank you. The real need is for more of you. Go get 'em and bring 'em back alive.

Whichever hemisphere you're in, this is a new year, and I hope that it brings good doings and smooth workings for you. Radio telegraphers will note that this is the year of mutual affection (88 means "love and kisses"). A fine 19"88" to you.