Another possibility is the Dallas Semiconductor (now Maxim Integrated Circuits) DS1065. It contains an oscillator, prescalers, and a frequency divider. Housed in a three-lead TO-92 transistor package, it’s a full oscillator and divider that can be set to frequencies between 30 kHz and 100 MHz.
Its single I/O pin provides the output. But with a pull-up resistor on this pin, the chip goes into a programmable mode that enables you to enter prescaler and divider selection options to get the frequency you want. And the really cool thing about this chip is that it uses zero external components. The frequency tolerance is ±0.5% initially but can vary ±3% over temperature and voltage variations—pretty good for noncritical clock applications.
Maxim’s DS1085 is another such oscillator/divider IC. It’s a pretty good frequency synthesizer that can be set to frequencies between 8.1 kHz and 133 MHz and comes in an eight-pin SOIC package. A two-wire serial interface lets you program the device to your desired frequency. Its frequency tolerance is ± 1%.
Also, Maxim’s MAX7387 and MAX7388 specialty clock timer ICs have a fixed-frequency oscillator in the 32-kHz to 32-MHz range. Standard frequencies of 1, 4, 8, and 16 MHz are off the shelf, but you can order any frequency you want in that range for a price. This particular part is used mainly for watchdog and power-fail applications.
Another interesting replacement for the 555, Linear Technology’s LTC6906, is a pulse oscillator with a frequency range of 10 kHz to 1 MHz—pretty much the range of the 555. It’s made from CMOS and uses only a single external resistor to set the frequency. No external capacitor is needed. And, the accuracy is 0.5%. The chip operates from a 2.25- to 5.5-V supply and consumes very little power (less than 12 µA at 100 kHz). The package is a six-pin, 3- by 3-mm SOT-23 (ThinSOT) surface-mount type (Fig. 5). Its external resistor (RSET) determines the frequency according to the simple formula:
fOUT = (1 MHz/N)(100k/RSET)
For greatest precision, it is best to use a metal film resistor with a tolerance of 0.5%. The N in the formula above is a frequency divider factor. The LTC6906 has a built-in frequency divider with ratios of 1, 3, and 10. The desired ratio is selected by voltage on the DIV pin: ground for divide by 1, open for divide by 3, and +V (the supply voltage) for divide by 10. You can get frequencies down to about 10 kHz or a little less up to 1 MHz. It’s a pretty cool part and one of the first in a long time that addresses clock and timing applications.
Linear Technology also has another similar part that can replace crystal or ceramic clock oscillators in some applications. The LTC6905 operates from 2.7 to 5.5 V and is housed in a SOT-23 package. Its frequency range is 17 to 170 MHz. Again, the frequency is selected with a single external resistor and a divider ratio. The frequency error is only about ±0.5%, and the jitter is less than 50 ps at 170 MHz. This is pretty good and can be used in many digital applications as a clock oscillator without the need for an expensive crystal.
Variations include the LTC1799, LTC6900, LTC6902, LTC6903, and LTC6904 with different frequency ranges and the single external resistor programming. The LTC6903 and LTC6904 have a frequency range of 1 kHz to 68 MHz. The LTC6903 is programmable via an I²C bus, while the LTC6904 is programmable via an SPI bus.
If you absolutely must have crystal accuracy at low cost, you might be able to use National Semiconductor’s MM5369. It uses a low-cost 3.579545-MHz crystal to make an oscillator followed by a 17-stage divider. The divider is programmable to provide a wide range of frequencies, including 60 Hz.
Probably the most popular approach to generating a clock of programmable frequency is to use a cheap embedded controller (Fig. 6). Any 8-bit microprocessor can easily duplicate the operation of a 555 and then some. All you do is write some code that puts the processor into a timing loop or two and pulse an output port.
But is it wise to replace the super-cheap and easy-to-use 555 with a more expensive and complex microprocessor that requires us to write software? Absolutely. Most products are digital anyway, and virtually every one of them contains at least one embedded controller. So, it is natural to implement timing functions with a microprocessor.
Embedded controllers like the PIC, 8051 derivatives, 68HC05/11, and others are dirt cheap in high quantities. Also, most of you are already steeped in programming, so writing the code is trivial. Besides, a crystal or ceramic resonator usually controls the microcontroller clock, so it produces really accurate timing pulses. This mitigates the problems inherent in attempting to achieve timing accuracy and temperature stability with an RC network.
With an embedded controller, you get high accuracy with ultra-reliable repeatability. Add to that the presence of pulse-width modulation (PWM) functions on many microcontollers, and you have added functionality without new hardware—just some more code. By using a long count loop or nested loops or the built-in counters and timers plus precisely counting instruction execution times, you can program the chip to provide even the oddball frequencies your application may need.
So What About The 555?
I’m not knocking the 555. I’m just trying to understand its ongoing use when so many other solutions are available. And why has it survived when other ICs of the 1970s and later have become obsolete? In most cases, microcontrollers and programmable logic devices like PROMs, PALs, GALs, and FPGAs have replaced the once-ubiquitous DTL and TTL logic chips as well as CMOS logic chips. Yes, you can still get TTL and CMOS logic, but they aren’t often used in new designs.
Older linear devices are probably still more common than older digital devices. Almost as old as the 555 timer is the 741 op amp. Like the 555, it probably will never die, especially since good linear devices never seem to fade out. These linears, for example, are still around: the 301 op amp, 78xx linear regulators, 317 regulators, the 1496 balanced modulator, and the 602 RF mixer. I still like the old RCA 3028.
My guess is that the 555’s long life comes from a combination of factors like very low cost, multiple sources of availability, extreme versatility, and widespread design knowledge. Its old-style DIP format also makes it easy to use, plus it is rugged and forgiving – that’s hard to beat.
While you won’t find the 555 in many new designs, I suspect we will continue to use it in our hobby projects, and schools will continue to teach it. Every now and then, you may find the ideal application for it. Just remember that there are alternatives that will produce smaller circuits with fewer discrete components.
For now, congrats to Hans Camenzind and the crew of designers at Signetics, wherever you are. Long live the 555.