Retro ICs that Will Never Die


From the 555 timer to the LM317 regulator, many old-time components still can be had from suppliers to help engineers, hobbyists, or even college students complete their projects.

I’m in the process of updating my basic electronics book Electronics Explained (Newnes/Elsevier, 2010) for a second edition. I’m adding some hands-on experiments and projects that have the reader build some interesting demo circuits on breadboarding sockets. I usually order the parts I need from online supplier Jameco and occasionally from All Electronics and Mouser

Looking through the catalog, it dawned on me that the company still stocks lots of the older ICs and transistors, some introduced way back in the 1970s. In fact, I see dozens of them. I discovered that distributors like these still sell millions of these throwback parts each year. Why?

Defying Obsolescence

Remember the 555 timer and the 741 op amp?  These parts were first introduced in the 1970s.  If you’re an engineer from that era, you may have used them in a design. I did. Even if you’re a recent grad, you may have come across them as parts used in a college lab experiment. And if you’re a hobbyist or experimenter, you surely know about these oldie-but-goodie parts. 

Some other popular parts still hanging around are the LM317 and 7805 regulators. You can also order the LM301 op amp, LM311 comparator, and the LM386 audio amp. Other memorable parts include the MC1458 dual op amp, the NE565 PLL, and the CA3080 transconductance amp—remember that? Also still in the mix are the MC1496 balanced modulator, NE602 mixer, and the XR2206 function generator. 

As for digital ICs, the old 7400 series continues to dominate in all its variations. The 74LS low-power TTL versions remain popular, and the 74HC CMOS equivalents are widely used.  Also still available are the 4000 series CMOS parts. Maybe you recall the popular CD4017 decade counter or the CD4046 PLL.

It’s still possible to acquire old micros such as the 6502, 6800, 8080, 8031, and Z80. Like me, many of you learned micros on one of these chips. All are packaged in a 40-pin DIP, something you don’t see today.

Your favorite tried-and-true transistors can be had, too, like the 2N2222, 2N2907, 2N3904, 2N4401, MPF102, and 2N3055. Old but still good MOSFETs are the IRF510 and the 2N7000.  And let’s not forget the 1N4001 and 1N4148 diodes.

What’s So Good About Retro?

The main question is, why are these so-called dated parts still around?  Here’s what I think.

First, these parts work just fine as is, and they’re still “good enough” for many designs. They’re proven, reliable, and forgiving.  Second, lots of references use these parts.  Thousands of articles, books, app notes, datasheets, and online sources let you come up with a design fast and easy. Third, these parts are cheap. Most can be bought for less than a dollar with a few exceptions. And since these parts can be found at multiple sources, acquisition is fast and easy.

Finally, the ICs are large by today’s standards and use the dual-inline package (DIP). They’re easy to work with and readily plug into the widely used breadboarding sockets. On the other hand, modern surface-mount parts can be tough to work with. They’re small and require special tools, manual dexterity, and great eyesight. 

Thousands of hobbyists and experimenters buy DIP parts simply because building with them is an easier process. Just look at the projects in popular hobbyist magazines like Nuts & Volts, Circuit Cellar, Elektor, and Make. Most use the older DIP chips.

What I really wonder is, does anyone really design new products with these chips? Probably not. Today almost everything is micro-based and SMT. If you need a multivibrator oscillator, you no longer go for the 555 timer IC. Instead, you program a cheap 8-bit PIC processor to do the same thing.  And it makes sense to use a single, embedded controller chip instead of multiple TTL or CMOS chips for most digital applications. But it’s more difficult to do something equivalent with linear circuits. That’s why so many old op amps and other linears are still around.

I just breadboarded a demo circuit that shows binary counting. It uses a 74HC161 4-bit binary counter driven by a 555 timer clock. The flip-flop outputs have enough current to directly drive the four blue LEDs. It’s one of the new projects in my updated book, and it only took me a few minutes to build. The circuit should be a cheap and easy build for a newcomer to electronics. SMT is too off-putting. 

Do you still design with retro parts?  What are they and how are they used?

Discuss this Blog Entry 26

on Dec 6, 2016

Not correct. The LM1496/MC1496, CA3080/LM3080, NE565/LM565 and NE602 are all obsolete and unobtainable except for old stock. So is the MPF102, and most of the other small signal JFETs.

on Dec 8, 2016

The source of your information? Just curious who is correct?

on Dec 7, 2016

I have a couple of Duracell flashlights, LEDs with sturdy aluminum cases. Nice, except for the fact that the batteries won't last. The multi-function button that turns on the LED evidently pulls a small current even when "OFF". I used to use 4000 series CMOS which could be powered from a 9V battery and, if gates were not actively switching, the battery would last indefinitely, probably for its shelf life, even while keeping voltage applied to the circuit. A DVM didn't register any quiescent current to keep power applied.
Why wasn't my flashlight able to achieve the performance we had 30 years ago?

on Dec 8, 2016

Eveready sold a cheap LED flashlight a few years ago, using a LED bulb surrounded by a DC converter PC board, and it was physically disconnected from the battery by the sliding on-off switch, same as in legacy flashlights. I bought 4 of these. A rare find! With eveready 1.6V AA lithium batteries in them, they show no sign of quitting. I agree with you about the rear pushbutton junk. Streamlight seems to work very well with the rear pushbutton now, from my experience. The other manufacturers can't seem to get it right. I carry a Stramlight Microstream (about $20), which runs on one AAA battery and it lives in the bottom of my pants pocket. Excellent battery life and it does not power up by accident!

on Dec 8, 2016

4000 series logic cannot be beaten - if you want something taking nano-amps quiescent, no micro or gate array comes anywhere near it!
And it runs straight off a 9v battery!!!
And the silly idea that using a tiny micro is better than a 555C (or a spare 4093 schmitt gate) - not just cost, but someone has to programme it on every board!

on Dec 8, 2016

Many new and improved versions of the older ICs have better specifications but also a slightly higher price point. Unless your design is being mass produced in 10k+ quantities it's often hard to beat the price/performance of the proven ICs that will continue to be produced for at least another decade.

on Dec 8, 2016

Those parts are ideal for "Pease" style board-less breadboarding! I've taken those breadboards and put them into plastic boxes. A little hot melt keeps the components from shorting together and fastens the circuit to the box (use the glue gun tip to melt into the ABS box, then start dispensing glue). I wouldn't use any of those parts professionally, but they're perfect for "quickie" solutions around the home and lab.

on Dec 8, 2016

I build a pipe and cable locator for the Cathodic Protection Engineers and I use an LM386 audio amp to amplify the 120 cycle signal from the pickup coil, also I build an AC soil resistance bridge that uses a 555 and 741 Op Amp, these are very dependable IC's and need to be kept around.

on Dec 8, 2016

I'm misty. I strongly agree about the eyesight and dexterity observation. New parts seem to jump like fleas under my stereo microscope. I've found that some 'Little Logic' is still preferable to micros if you have the space. The logistics of maintaining a code base at my business outweigh the price difference for low qty runs.

on Dec 8, 2016

There's so much to be said about this topic. Yes a lot of these "modern" packages are a bit difficult to breadboard with, but it turns out what you really need are PCB adapters from SOP/MSOP/TSSOP/SOIC and the like to DIP or PGA, and you can get arrays of these inexpensively from China via eBay in all kinds of configurations, and where you REALLY need to have these are for all the higher pin count uCs from Microchip and other vendors where the parts JUST AREN'T AVAILABLE in any package that fits neatly on a 0.1" grid perf board. What I find myself missing are the "monolithic matched bipolar" series like CA3045/46/83 (these were NPN, I believe there were PNPs too) because these allowed you to wire up your own Gilbert cells and analog log/antilog circuits - and while I'm on the subject there's a really great tutorial on the general topic if you Google "translinear circuits tutorial" and go to the site in the UK, this version also covers MOSFETs which many of the others do not. As far as "retro" circuits one of my favorite series is the LM2907/17 "analog auto tach", it has a really clever implementation, it's actually still useful for a lot of purposes and it's still also available. On the other hand it's really hard to find function generator parts like XR-2206 or ICL8038, but now you have options, like if you have room for an 18 pin DIP you can put in a PIC16F1847 and program it with code from "Electric Druid". Also you mentioned the venerable CA4046, I still remember the day I got my first samples of the 74HC4046 - as I recall the VCO was voltage-programmable out of the box from about 50 Hz to 17 MHz IN A SINGLE RANGE, show me any MODERN chip that can do that! Finally one of my favorite "stunts" with "retro" part #s is to use parts from the 74HC4051/2/3 series for analog multiplexers/demultiplexers OR logic parts, they have fairly decent analog performance but they're priced like logic devices, so you get the best of both worlds.

on Dec 8, 2016

I still experiment with vacuum tubes in guitar amplifier design where the tone and performance of tubes is still preferred by the majority of players. So, never mind IC's and transistors. I'm hoping for the continued availability of leaded resistors, capacitors, etc.! But then, there still continue to be tubes available on the "new-old-stock" market, most types of which haven't been made in 30 years or more.

on Dec 9, 2016

All the active and passive parts you nostalgically reference are readily available at reasonalbe cost from RSR Electronics, Inc. or Electronix Express. .

on Dec 8, 2016

Still using SA602 (NE602) but in the 8 SOIC. Yes, dips are inactive. Still big enough to work with easily, I think. I have looked over time for an improved mixer (with gain) but everything new is in the 500mhz ++++ region and power hungry to boot. Probably why the 8 SOIC version of this mixer is still considered "active" by NXP and available stock at DKEY and MOUSER if not others also. Philips designed a good product that has stood the test of time.

on Dec 8, 2016

We just put into production a board containing a 555 timer as a very low freq clock. Given the low price, the lack of maintaining code, and all the associated issues with version control, it makes a lot of sense.

on Dec 8, 2016

I do miss the 3045/3046/3086,which despite may drawbacks like modest beta and high popcorn noise, really did have decent matching of four transistors (the fifth one with the substrate-connected emitter was generally best left alone). At one point ST made a part that was vastly improved and merely duplicated the old RCA datasheet for documentation. It had drastically lower popcorn noise and much higher beta, but evidently no one wanted it and it passed into oblivion. We can buy arrays now from THAT Corp., at a rather higher tariff.

Many years ago I encountered a guy who was convinced that microcontrollers should be in everything---this when the 6502 was really big stuff. He wasn't a very good designer otherwise. If he is still around I wonder if he is beating the big drum for the Internet of Things?

on Dec 8, 2016

Yeah don'tcha love incidents like the DDOS attack the other day that brought down the dyn servers? Word is the protagonists used (among other things) IoT video cameras to make the attack, we're talking cameras that were supposedly put in to ENHANCE security - problem is their firmware is full of security flaws (we know they're there because the bad guys already used them) and the vendor either doesn't provide updates or maybe even a mechanism to load them (or both)! So much for IoT fulfilling clear objectives by well-implemented means...!

on Dec 8, 2016

I miss the CA3096 also. It had three NPN and two PNP in a 14-pin package so you cobble together all kind of things,

Moog, in their iconic '70s synthesizers, made a log amp with a CA3046 and used the fifth transistor as a heater to "oven control" the temperature for stability, probably the best use to which it's ever been put.

on Dec 8, 2016

Wonderful article. Many older chips are still in use and better in many ways to their modern counterparts in specific designs due to their robustness, derating capabilities, or a specific mechanical need for through-hole design. Also, their particular parasitics can be useful for isolating modern power networks. Moreover, even modern ICs sometimes take a step or two back in process technology for economic reasons (higher yield), and because an older process still meets design requirements for certain customers. Pushing life out of older processes is sometimes critical for semiconductor companies with fabs, because it delays the need for extremely expensive retooling in the fab, and allows companies to show greater profits to their shareholders. Most engineers don't consider Cost when designing solutions to problems, but it should be considered in addition to all other design parameters. Keeping program/project cost down can be a life-saver for a company.

on Dec 8, 2016

Voltage is one thing that makes old parts attractive. Old parts work on "real people voltage," meaning 5V, 9V, 12V, etc. With too many new parts, 3.3V is considered the "high voltage" version. 4000-series CMOS will work up to 18V and down to about 3V. If you've got a 9V or 12V battery or a stack of four AA cells---CD4xxx-whatevers just work. An LM324 works from a single supply with inputs and common-mode voltages that include ground, negating the need for all kinds of level shifting tricks. You can do entirely unnatural things with an LM3900 if you know how and don't need any bandwidth to speak of. I did a heart rate monitor in the '70s using LM3900s and 4000-series logic. When the bandwidth of interest is less than 10Hz, they can't be beat.

And, yes, miracle of miracles, when a 4000-series logic chip (or even a 74HC-series one) is not working, they are OFF; they don't don't draw any power. And, don't tell anyone, but 4000-series chips can even be powered through their inputs. The idiot protection diodes on each input pin will route enough power to the voltage bus inside the chip to allow it to function in many cases.

Another miracle from the '70s is the LM10: A high quality voltage reference and an entirely decent op-amp on a chip that will work from 1.1 volt to 40V---no s#!t. I've run them off of single NiCad cells at 1.2V; a 1.5V AA cell is luxurious.

And finally, this stuff is fun! Like Lou said, you can build a working circuit, grunt simple or mind-bendingly clever, on your kitchen table WITHOUT a design team and with pocket change or less. Cool.

on Dec 8, 2016

True words. We used 4000 series CMOS in satellites since it had such good noise margin and was more rad-hard than 74HC. I watched a tech at ESL troubleshoot a rack-mount ground unit for two days before he realized the power was off and the box was kinda working by power from the inputs, as you noted.

I had a professor at GMI/Kettering that taught a whole course on the LM3900 Norton Amp. The auto business loved them since they were a dirt cheap quad with some unique applications. That professor would be flabbergasted to know that I ended up in Silicon Valley and went to parties where the designer Tom Frederiksen was in attendance. Frederiksen was right up there with Widlar for genius, just not as drunk and crazy.

And if you like the LM10, check out my article about Pease that just posted today "What's all this LM10 stuff, anyway?" Pease taught me what a cool chip it was. He told me we could go look up Widlar's schematic, but we never got around to it, I sure regret not making time for that. Only minor niggle is your calling the ESD diodes idiot protection diodes. I asked Paul Grohe if National ever tried to make a chip without them. Grohe, who was Pease's hand-picked protege told me, "Yeah, they blow up if you even look at them wrong." The input bias current on an LMC660 is really just the mismatch between the high-side and low-side ESD diodes. the input MOSFETS don't even draw attoamperes.

And agreed about a kitchen table circuit that works and you can always get parts for. I remember designing in an Atmel AVR and they discontinued it. When I complained the rep said the AT Mega would do everything the old chip did. He did not seem to get I would have spin the PCB to mount the new chip, as well as fiddle with the code to get the right pins wiggling.

on Dec 8, 2016

"Just using one of those little microprocessor chips for the function" may be OK if you are building a one-off thing that is intended to be disposable. But if you are building a production run of equipment that needs to be usable for 20 years it is a foolhardy choice. First, as already stated, every one of those little micros must be programmed, and, somebody must write that code with no bugs ever! AND, if my customer in the far off boondocks needs to fix the system at 4PM on a Friday, it is a lot better to need a 4000 series CMOS device than a short-life-until-obsolete micro, which still needs to have the program loaded in, and if they have the equipment to load the program, even after it is obtained, if it is still available six mmonths later. IFF the maker of the chip has not been bought up by some other comapany and renamed, or just plain discontinued.
A major selling point of our product is that all the parts are readily available from multiple manufacturers and many suppliers. That is a valuable feature for capitol equipment to posess. Plant managers and development group managers will remember suppliers of equipment that must be scrapped after a year because parts are no longer available.
BUT, if one is in the business of selling fad products that will be obsolete before the 90 day warranty expires, that is a different market. Throw-away toys and junk should definitely use the programmable processors, since serviceability is not a concern.

on Dec 8, 2016

One more thought is that the original IC devices were designed to be universally applicable, most of them are intended for a broad range of applications. SO how do you improve on a bunch of devices like that? PLUS, not needing such a tightly regulated supply voltage makes them a lot simpler to use. One other point is that logic running on 15 or 12 volts has a much bigger noise marging, in absolute voltage, than those packages running on 1.8 volts. And often noise resistance is important.

on Dec 8, 2016

I love those parts as well. Lots are obsolete as the first poster said but those in his post are still available at Jameco as Lou said as his favorite supplier.
I have been employed in the electronics industry for over two decades but still use these parts at home for ham projects and at work as needed when prototyping. Great article.

on Dec 8, 2016

I fondly remember these devices and the comprehensive manuals supplied by manufacturers. I did and still use prototyping using 'dead bug' fabrication. Mostly ham projects.
I worked for a Jameco type supplier in the early 70's, International Electronics Unlimited. Got to work with most ICs in the article. I recall a 16 bit processor built as a multi chip configuration, CROMs and RALUs etc; this as INTEL's processor was the 4004. Also RTL and DTL logic devices. A great article. TNX

on Dec 10, 2016

Regarding the comments about processors vs. these oldie but goodie parts, I did a controller pre-Y2K using HC logic, the LM1458, LM301 and LM310 follower first as DIP then went to SOIC for production. A few months before Y2K the end user contacted me with concerns about code updates...I could reassure him then that there never would be any and the last time I heard the controllers were still working. And, as pointed out by William K, Paul Rako and others/many, if the parts did fail all they had to do was buy a replacement and solder it in. I don't think these parts will be unavailable for many years to come because of all the stuff still working with them.

on Dec 26, 2016

it makes a lot of sense.

Please or Register to post comments.

What's Communiqué?

Blogs on topics such as wired and wireless networking.


Lou Frenzel

Lou Frenzel writes articles and blogs on the wireless, communications and networking sectors for Electronic Design. Formerly, Lou was professor and department head at Austin Community College...
Commentaries and Blogs
Guest Blogs
Jan 26, 2017

An Amateur’s View on the P2 (Part 2): Slew Rate and the Oscillator 3

Justin Mamaradlo takes a further look into the P2 op amp and how it functions, analyzing the oscillation and slew-rate characteristics of the venerable component....More
Jul 15, 2016

Simple Yet Effective ESD Testing Methods for Higher Reliability 11

There are multiple ways to test for electrostatic discharge, ranging from implementing a human-body or machine model to....using a balloon and a comb?...More
Apr 8, 2016

Confabbing on the Fabless Fad 5

High capital and maintenance costs, and EDA advances along with abstractions to deal with chip complexity, have been leading contributors to the fabless migration....More

Sponsored Introduction Continue on to (or wait seconds) ×