Is the 555 still a viable IC?

[schmitt trigger]

I originally posted this question on another EE forum, but I believe that the technical expertise and the geographic distribution of members is significantly different between the two, to warrant posting it again here.
DISCLAIMER: I am not building anything with a 555, this question is only sheer curiosity..... Having cleared that out of the way:

It can be safely said that the 555 is one of the most ubiquitous and influential ICs in history. I am also sure that most if not all of the old-time forum members have used it at least once.
But the IC is almost 52 years old, and although it has been updated to CMOS versions, its age and limitations show up.
The question would be: do you think that the 555 is still a viable IC in 2024?
Let me give my personal opinion; the problem with the 555 are the required external passive components. To obtain a modicum of stability and accuracy, the cost of those components may exceed the cost of the IC itself.

[tooki]

Define “viable”.

Because clearly it’s a viable product, in that the manufacturers can manufacture them at a cost low enough that customers are still buying them in volumes sufficient to keep production going, but with a sale price that makes it worthwhile. (The sales number always thrown around is “over 1 billion per year”.)

They ceased being sensible as accurate timers long ago. If they ever were to begin with. But clearly they’re being used for something.

[MarkS]

Depends on your use case. If you need accuracy, no. If close enough is good enough, yes. What you'll find is that 99% of the time, "close enough" is perfect.

[schmitt trigger]

You are correct, "viable" is a very elastic word. Its meaning can be stretched in a number of ways.

By viable, as a narrow definition, I was thinking whether it could still find a place on low to medium volume commercial products. Nowadays, in 2024.
High volume I believe is out of the question, whatever function the 555 can do, it could be integrated somewhere else.

Yet, they still sell..... Who would be the potential customers?

As previously mentioned, this is only curiosity.

[exe]

Ha-ha, did same research ago. There are plenty of threads like this on the Internet.

As of me, I used it to generate square pulses to drive a mosfet. It worked flawlessly, can't complain. I used CMOS-version of time because it's presumably less noisy. I decided not to go with MCU as I didn't want to mess up with firmware and programming.

The only thing I didn't really like is that, in order to get less than 50% duty cycle, I had to use some "wizardy" with a diode.

[ejeffrey]

If you just need a roghly accurate square or pulse generator, they are fine. I would use them for low to medium volume.  A tiny MCU is cheaper in volume but needs to be programmed, which adds steps to manufacturing.  In addition, I'm certain I will be able to buy 555s in 10 or 20 years, so for low volume products with long life I would really consider a 555 over an MCU.

There are 555 variants at a wider range of operating voltages without a level translator.  That could be a slight bonus for the 555.

For low volume you might even choose to use a simple discrete solution like a 555 even over integrating into an MCU that's already part of the design just because it keeps the development simpler.  In many projects, the software development is the overwhelming driver of development costs.  If dropping an extra chip on the PCB can reduce software development complexity then it's often worth it.  Of course for high volume products, manufacturing costs outweigh development and it you want to minimize the BOM cost.

[coppice]

There are lots of variants of the 555, like the various CMOS versions. Between them they still sell by the billion. I wonder how close any of the currently available ones are to the original die developed at Signetics?

[selcuk]

Most of the digital circuits include a MCU nowadays since you can fit them to any budget. Then you set timers and PWM outputs of the MCU to do the job of 555 in a versatile way. It is not required as in the past anymore.

Nevertheless, I'm using it in my designs frequently (not the original one, but TLC555). If I use a watchdog chip and want to extend the output pulse, then I use it. Easy to find and cheap.

[jpanhalt]

Like a Schmitt inverter, they can be stabilized with a crystal for better timing (see attachment).

[Gyro]

Viewed as a pair of analogue comparators with bias network, an RS latch, and a couple of fairly beefy output stages, one P-P and the other O/C, with a wide supply voltage range and all in a single 8 pin package, it can still be a handy IC. It's not just a timer, and you can't directly match it with a micro (within its niche).

[SiliconWizard]

Yeah, not sure what "viable" means here. But I've used the TLC556 (dual) just a couple years ago for a simple application that would otherwise have required either more discrete ICs, or a MCU with firmware.
Sometimes simplicity just wins.

[Psi]

Yeah, It's useful, but not all that common for a production design any more due to cheaper and more flexible options, like a cheap mcu, but any MCU option requires coding and flashing so has it's own disadvantages even if it is more flexible

There's plenty of situations where you might want a 555 for a quick and dirty solution to a problem.
You could throw together a test jig for something in under an hour if the accuracy isn't all that critical.

[xrunner]

Yet, they still sell..... Who would be the potential customers?

As previously mentioned, this is only curiosity.

My guess is learners and classes are buying many of them. One hobbyist after another, one learner after another, year after year. Each person learns using one or two, burns a few by mistake, and moves on. Look at all the tutorials for 555 timers that are out there.

I wonder what the quantity of new or good 555 timers are just sitting in parts drawers never to be used again. Nothing wrong with them. If we could only get these to new learners they'd never have to make any more ever again. 

[Psi]

There is also the LM555-MIL  for military stuff.
It's active apparently, but I've never seen it for sale at the usual places.

[EPAIII]

Consider, if you built a circuit with a 555 timer 50 years ago, which is still in use and the chip went bad, you could just get another 555 and drop it in and the circuit would still work today. About a one hour repair, including the time spent ordering the new 555.

If you built a circuit with a micro processor 20 years ago and the chip went bad, that micro processor would probably not be currently available. You would probably need to build a completely new circuit with either an "improved" version of that processor or with a completely different one and then wonder where the original code was. Oh, that's right, your IS department replaced the desktop where the code resided 10 years ago and they don't keep back-ups for over 5 years. So you would wind up re-writing the code from scratch.

So in order to save a few pennies and use the latest and greatest technology, 20 years ago, you wind up spending several days building the new circuit and writing the code for it and testing, of course. Now, tell me which would have been a better choice.

[Psi]

Consider, if you built a circuit with a 555 timer 50 years ago, which is still in use and the chip went bad, you could just get another 555 and drop it in and the circuit would still work today. About a one hour repair, including the time spent ordering the new 555.

If you built a circuit with a micro processor 20 years ago and the chip went bad, that micro processor would probably not be currently available. You would probably need to build a completely new circuit with either an "improved" version of that processor or with a completely different one and then wonder where the original code was. Oh, that's right, your IS department replaced the desktop where the code resided 10 years ago and they don't keep back-ups for over 5 years. So you would wind up re-writing the code from scratch.

So in order to save a few pennies and use the latest and greatest technology, 20 years ago, you wind up spending several days building the new circuit and writing the code for it and testing, of course. Now, tell me which would have been a better choice.

Yep, speed of repair and maintenance is very important, especially with fixing production test jigs or any production line equipment where down-time is expensive.
Not having to worry about firmware is a big time saver. Not just because of the time required to code and flash it but having to test it and any regulatory or procedural hoops required to change it.
On the other hand having a MCU makes diagnostics easier as you can easily add hardware checks and warnings in code to quickly diagnose and find faults.
So it depends on the situation and is often a trade-off

[Smokey]

I've only done one design that used a 555.  I needed an adjustable frequency and duty square wave that would be user settable with pots to drive a pneumatic valve as an air burst thing for machining.  While it "worked" it's fiddly to adjust since the duty and frequency are not fully independent.  I should have just used a micro. 

[johnboxall]

They're easy, you can find success without much effort which is great for education.

[baldurn]

As a counter argument against the 555 I have observed that nobody here at the local makerspace would use the 555. The typical hobby person is doing Arduino or ESP32 for everything. People tend to solve everything with the one tool they know, so the 555 does often not even get considered. We do have some classes where we play with old chips, and where we might do something fun to learn the 555 - but the day after everyone are back with their ESP32 project.

Funny that you say software development is the reason to avoid a MCU and use 555. These days most people think the other way around. We go with a MCU even when a 555 would have done the job. It is easier to figure out a simple program than it is to do "real" EE with discrete components.

[SiliconWizard]

Funny that you say software development is the reason to avoid a MCU and use 555. These days most people think the other way around. We go with a MCU even when a 555 would have done the job. It is easier to figure out a simple program than it is to do "real" EE with discrete components.

Yes. That is, until people actually have to make stuff that is provably reliable - that's usually when they reconsider doing everything with software.
But for tinkering, it may look easier.

[JPortici]

Funny that you say software development is the reason to avoid a MCU and use 555. These days most people think the other way around. We go with a MCU even when a 555 would have done the job. It is easier to figure out a simple program than it is to do "real" EE with discrete components.

Yes. That is, until people actually have to make stuff that is provably reliable - that's usually when they reconsider doing everything with software.
But for tinkering, it may look easier.

You don't have to prove to a committe that your independent software timer that resets everything in case of a failure actually works as intended
and it's another independent MCU you don't have to program

[baldurn]

A new YouTube video about this exact topic: https://youtu.be/SgV6_Y_sg4k?si=hC28IX6KcHHjxwQU

He proposes using ATTiny 412 instead of 555.

[jpanhalt]

Way back in 1999, High Tech Chips produced several Microchip 12F PIC-based devices that were pre-programmed to do 555 type stuff -- of course, VCC and GND were reversed.  I thought, there's a gold shovel without a handle.  The name is still found by Google, but its not clear whether they still produce those items.  I have included 2 pdf's of their datasheets from that time.

Google link: http://www.hightechips.com/   Yes, it is spelled differently.

[Warhawk]

A new YouTube video about this exact topic: https://youtu.be/SgV6_Y_sg4k?si=hC28IX6KcHHjxwQU

He proposes using ATTiny 412 instead of 555.

Maybe for hobby but not for production. Go and find the second source for ATTiny. Or just pre-programming ICs for production is pain.
I know about automotive projects using 555. It is not stupid as long as it does the job.

[SeanB]

Yes as a relatively robust driver nothing to beat it, especially as you can get level shifting, from almost any voltage logic level above 3V, to 12v for almost free, just add in a single resistor or a zener diode, plus you get the ability to drive an indicator LED to show output low almost free as well, and it can also drive another load active low as well. 2 of them, and a single small signal transistor as inverter, and a few resistors, and you get a robust full bridge driver, and add in another small signal transistor and a pair of resistors and you also get the ability to turn the bridge off as well, using the reset inputs. Level shifting from logic to 15V included free as well, plus well defined switching thresholds for slow edges included as a bonus. Most of the non CMOS bipolar ones will do a half amp as well no problem, the CMOS at least 100mA which is a good drive for a lot of things. Just need a 1000uF 16V capacitor close by for power decoupling.

[ejeffrey]

Funny that you say software development is the reason to avoid a MCU and use 555. These days most people think the other way around. We go with a MCU even when a 555 would have done the job. It is easier to figure out a simple program than it is to do "real" EE with discrete components.

I don't think programming an MCU is any more or less "real" than configuring a 555.  I don't think trying to make a value judgement there is helpful.  But there is a big difference in what makes sense for a hobby project, prototype, or one-off than for something that is going to need to be manufactured and supported in production.

An MCU is great because it's super flexible, you can drop it in your schematic and start manufacturing without even being sure what you need it to do. It also can easily be hooked up to a computer for automation.  But in a manufacturing setting, even for low volume it adds steps to the process and additional configuration to manage.  Unless you as the designer wants to manually program hundreds or thousands of units before they go to end users you need to document the flash process and make sure your manufacturer is doing it correctly, and possibly deal with a situation where the process was messed up and you have a bunch of defective units in inventory that need reflashing.  You can order pre programmed chips for a reasonable setup cost and that's a decent option but still comes with process overhead.

Again, in high volume you can sort this out and it's worth it to save a few cents on the BOM.  Or more likely you integrate the functionality into a part that already exists on the design and ask the software team to add some timers to their existing code base.  But for low volume products I have found that the fewer programmable devices  the better.

[Zero999]

Another example of a digial replacement is the 74HC5555, but it's no longer made.
https://www.mouser.com/datasheet/2/302/74HC5555-1597181.pdf

The original NE555 is not the most beginner friendly of ICs. It draws nasty spikes, when the output changes state, which can cause interference and spurious triggering.

Newer CMOS variants such as the TLC555 and ICM7555 don't have the same issue of current spikes, but they have much weaker output pins, especially when sourcing.

[ejeffrey]

Also to be clear , I'm not suggesting nobody should use a MCU when a 555 will do the job.  Just that there are factors beyond per unit cost, and that the programmability of a MCU is not always in the + column.  As always it is situational but I don't think the 555 is obsolete even when a small microcontroller can do many (but not all) of the same jobs for lower cost.

[Bud]

I am wondering why noone yet said "Just use an FPGA!", as it typically happens. 

[Ian.M]

Hmm.  FPGA board with three ADCs, and some level shifters and output drivers implementing a soft core CPU running a SPICE simulation of  a 555 in realtime . . .     

The only flies in the ointment are the quiescent current, power dissipation and board area used!      Lets not even mention the price. 

[Zero999]

I find myself being drawn more towards other jellybean ICs, for simple timing/oscillator functions, when it's not critical: the LM393, LM311, 74HC123 and CD4060 spring to mind.

[tooki]

Yet, they still sell..... Who would be the potential customers?

As previously mentioned, this is only curiosity.

My guess is learners and classes are buying many of them. One hobbyist after another, one learner after another, year after year. Each person learns using one or two, burns a few by mistake, and moves on. Look at all the tutorials for 555 timers that are out there.

I wonder what the quantity of new or good 555 timers are just sitting in parts drawers never to be used again. Nothing wrong with them. If we could only get these to new learners they'd never have to make any more ever again. 

There is no way the “newbies” market can consume a billion 555s a year. Heck, I doubt it could consume 1% of that. Nor are people squirreling them away at that pace, there’s no logic in that, now that basically everyone has access to distributors that can get us common parts in a day or two. They have to be going into products. I know we occasionally see them pop up in teardown videos.

[tszaboo]

Viable? Did I miss it, when did it become NRND?

It has quite a bit up it's sleeves, works from 12V, it can drive a relay without any external transistor, and you don't need a firmware engineer (big plus).

I am wondering why noone yet said "Just use an FPGA!", as it typically happens. 

No, use an Arduino.

[coppice]

That "1 billion" number may be outdated. It seems to go back to a 2004 interview with the designer, Hans Camenzind. He mentions that number, and that the 555 was mostly produced by Samsung in Korea back then -- surprising to me, and maybe a hint that the information actually predates 2004?

http://semiconductormuseum.com/Transistors/LectureHall/Camenzind/Camenzind_Index.htm
Click the link to "page 2" for the relevant section of the interview.

No idea whether production numbers have gone down due to obsolescence since then, or even up due to various lower cost manufacturers?

He might be referring to the original 555. There are huge volumes in the various derivatives made by a variety of companies in the small linear and logic IC business.

[ftg]

Echoing what others have said, good availability, predictable and well known behaviour.
No production programming needed is definitely a plus for me.
And this "no production progamming, no FW development" was the reason why some 555's recently got designed into a product by me. Managment not wanting any firmware in the loop also helped. 

[xrunner]

Yet, they still sell..... Who would be the potential customers?

As previously mentioned, this is only curiosity.

My guess is learners and classes are buying many of them. One hobbyist after another, one learner after another, year after year. Each person learns using one or two, burns a few by mistake, and moves on. Look at all the tutorials for 555 timers that are out there.

I wonder what the quantity of new or good 555 timers are just sitting in parts drawers never to be used again. Nothing wrong with them. If we could only get these to new learners they'd never have to make any more ever again. 

There is no way the “newbies” market can consume a billion 555s a year. Heck, I doubt it could consume 1% of that. Nor are people squirreling them away at that pace, there’s no logic in that, now that basically everyone has access to distributors that can get us common parts in a day or two. They have to be going into products. I know we occasionally see them pop up in teardown videos.

Well I did say hobbyists et al. buy many of them, not that they bought all of them. I'm very careful how I say things on this forum after so many years here, I know what kind of trouble you can get into if you don't carefully word things. 

However, I did some research just now and i can't find any objective evidence of the total sales number for "555 timer". As others have stated, the One Billion number seems to have stemmed from some interview with this person Hans Camenzind. We simply do not know it's correct now. We also do not know how many new learners and existing hobbyists are out there (do we?).

I just checked the price for 555 timer DIP package on Digikey. It's $1.31 ea., or $1.17 ea, for quantity 10.

So this is like filling in numbers for the Drake equation. Many of the numbers come out of my a$$. If you have 1 billion being sold ( it's probably less but ...) the hobbyists / learners / educational institutions are buying some. But that doesn't mean each individual entity each buys ONE 555 timer. It means the parts go to far less entities because they buy more than one ea.

How about the individual learner / hobbyist? How many are there? 10 million worldwide? Maybe they buy five ea. on average because the price is very low. Now we're talking 10 million entities buying 50 million 555 timers per year.

The teaching institutions buy even more. Maybe the institutions that use them buy 5000 555 timers ea. (on average) and the number of institutions are 10,000. (we're talking worldwide now). So they all buy 5000 * 10,000 = 50 million 555 timers.

Now we are up to 100 million 555 timers per year just for the hobbyists / learners / teaching institutions. That's 1/10 of the mysterious One Billlion number. The rest is industry buying. If we want to agree the 1 billion number is really less now, I can easily support my claim that hobbyists / learners / teaching institutions buy many of them. But I have no way of knowing how many they buy, but like the Drake equation I can get the answer I want if you give me a little time to play with the numbers. 

Note: And each person can get the answer they want too because we do not have enough objective evidence.

[PlainName]

I just checked the price for 555 timer DIP package on Digikey. It's $1.31 ea., or $1.17 ea, for quantity 10.

Seems a bit high to me. Add a cap or two and resistors and a tiny 8-bit micro is looking quite attractive in comparison for some uses.

[Zero999]

I just checked the price for 555 timer DIP package on Digikey. It's $1.31 ea., or $1.17 ea, for quantity 10.

Seems a bit high to me. Add a cap or two and resistors and a tiny 8-bit micro is looking quite attractive in comparison for some uses.

It's always more expensive in small volumes and through hole.

The 555, can run at up to 15V, some versions go as high 18V, others will work down to 1V, it doesn't have firmware development and maintenance costs and can output 200mA, which is enough to drive a decent relay directly or switch a MOSFET, at a reasonable speed.

[themadhippy]

I just checked the price for 555 timer DIP package on Digikey. It's $1.31 ea., or $1.17 ea, for quantity 10.

Time to get a new supplier,cpc  over here have them at  32p ea  or 25p ea for 10,including  the tax.

[coppice]

I just checked the price for 555 timer DIP package on Digikey. It's $1.31 ea., or $1.17 ea, for quantity 10.

Seems a bit high to me. Add a cap or two and resistors and a tiny 8-bit micro is looking quite attractive in comparison for some uses.

In any significant quantity they are a few cents. Looking up 1 off prices for components is just silly.

[ejeffrey]

However, I did some research just now and i can't find any objective evidence of the total sales number for "555 timer". As others have stated, the One Billion number seems to have stemmed from some interview with this person Hans Camenzind. We simply do not know it's correct now.

It easily could have dropped but TI alone has about 14 million in stock.  That's not counting stock at distributors, much less all the other manufacturers. 

It's still a huge volume mass production part not significantly dependent on hobbyists and education markets.

I just checked the price for 555 timer DIP package on Digikey. It's $1.31 ea., or $1.17 ea, for quantity 10.

But a reel of SOIC parts is $.07 each.  The single part price is irrelevant.

[baldurn]

The ATtiny 212 is about $.4 in the same 8 pin package as the 555 chip. It might be more compared to 555 if you buy the later in quantities, but you save on other components. It won't accept the same voltage or current ranges as 555, but for the saved components you could pair it with a mosfet.

In circuit programming is really easy. Just make three pads (vcc, gnd and program). Touch it with a pogo pin device for a few seconds and it is programmed. Can easily be automated.

[factory]

Bigclive found a new product using the 555 only 2 years ago.



David

[Zero999]

Bigclive found a new product using the 555 only 2 years ago.

David

It uses the 555 as a Schmitt trigger oscillator to generate a sawtooth, which compared against a potentiometer, using a comparator to generate PWM. I would have used a quad op-amp/comparator IC: one channel for the Schmitt trigger oscillator and the rest as comparators for the PWM. The LM339 will directly run off 24V, so no voltage regulator is required.

[SiliconWizard]

Definitely consider the TLC variant which is under $1 per 1 at Digikey, $0.37 per 1, $0.23 per 100 at LCSC. It's also better on almost all counts than the original.

[baldurn]

Bigclive found a new product using the 555 only 2 years ago.

David

It uses the 555 as a Schmitt trigger oscillator to generate a sawtooth, which compared against a potentiometer, using a comparator to generate PWM. I would have used a quad op-amp/comparator IC: one channel for the Schmitt trigger oscillator and the rest as comparators for the PWM. The LM339 will directly run off 24V, so no voltage regulator is required.

The 8 pin ATTiny212 would have done the job with one chip instead of three chips and a much simpler schematic. You would configure three pins as ADC to sample the potentiometers and three pins as PWM output. The remaining two pins are VCC and GND.

[julian1]

The pattern of having analog comparators, controlling the inputs of a S/R flip-flop for stateful behavior, is really common in things like smps etc. 
I feel like the Camenzing's 555 would be more general purpose/have higher utility, without the 5k/5k/5k biasing divider, which needs to be worked-around when a circuit that is not a simple timer/oscillator is called for. Although it's possible there is also a pin count constraint. 
Perhaps someone made a variant?

[Gyro]

There is the Control pin (pin 5) which gives direct access to the other input of the Threshold comparator, but this also interacts with the Trig comparator other input via the 5k/5k divider formed by the bottom two resistors in the internal chain. There was never any variant without the resistor chain that I'm aware of though.

[schmitt trigger]

A classic book, copyright 1976. Although this particular edition is 1978.

[tooki]

How about the individual learner / hobbyist? How many are there? 10 million worldwide? Maybe they buy five ea. on average because the price is very low. Now we're talking 10 million entities buying 50 million 555 timers per year.

The teaching institutions buy even more. Maybe the institutions that use them buy 5000 555 timers ea. (on average) and the number of institutions are 10,000. (we're talking worldwide now). So they all buy 5000 * 10,000 = 50 million 555 timers.

Now we are up to 100 million 555 timers per year just for the hobbyists / learners / teaching institutions. That's 1/10 of the mysterious One Billlion number. The rest is industry buying. If we want to agree the 1 billion number is really less now, I can easily support my claim that hobbyists / learners / teaching institutions buy many of them. But I have no way of knowing how many they buy, but like the Drake equation I can get the answer I want if you give me a little time to play with the numbers. 

I’d guesstimate those numbers at something like 10% of that. Me, I haven’t bought a 555 in years — and that time included my 3 years of electronics technician training. In that, we did use the 555,  but because we breadboarded them, almost all of them were kept for reuse, so very few being purchased. These days, for better or worse, in education you end up simulating more than testing actual hardware.

And I don’t anticipate buying any more 555s because I think there’s a better solution for almost every application. At least that’s been the case so far.

[Zero999]

Bigclive found a new product using the 555 only 2 years ago.

David

It uses the 555 as a Schmitt trigger oscillator to generate a sawtooth, which compared against a potentiometer, using a comparator to generate PWM. I would have used a quad op-amp/comparator IC: one channel for the Schmitt trigger oscillator and the rest as comparators for the PWM. The LM339 will directly run off 24V, so no voltage regulator is required.

The 8 pin ATTiny212 would have done the job with one chip instead of three chips and a much simpler schematic. You would configure three pins as ADC to sample the potentiometers and three pins as PWM output. The remaining two pins are VCC and GND.

That would be two ICs, since the MCU requires a 5V regulator.

Unless it's just a hobby project, I would be hesitant at using an MCU for such an application because there's only one supplier and the firmware cost. The LM339, is just one IC, which runs directly off 24V, is cheaper than any MCU and can be bought from multiple suppliers. The only downside is more passive components are required, but they're very cheap.

Definitely consider the TLC variant which is under $1 per 1 at Digikey, $0.37 per 1, $0.23 per 100 at LCSC. It's also better on almost all counts than the original.

The downside of the TLC variant is the output is much weaker, than the NE555. It can only source 10mA, which isn't enough to drive a relay and will considerably slow down the turn on time of a MOSFET. It can sink 100mA, but sometimes sourcing is required because the relay or driver transistor needs to be on, when the output is high.
https://www.ti.com/lit/ds/symlink/tlc555.pdf

[AndyBeez]

I recently encountered a 556 (dual 555) on a plug-in mouse scarer. One stage provided a low frequency pulse that gated the second stage; which ran 40KHz into a transducer. The DIL packaged chip was dated c.2010, so there were a few analogue diehards out there. But they possibly have long retired, leaving mouse scaring to the Arduino kids.

[baldurn]

The 8 pin ATTiny212 would have done the job with one chip instead of three chips and a much simpler schematic. You would configure three pins as ADC to sample the potentiometers and three pins as PWM output. The remaining two pins are VCC and GND.

That would be two ICs, since the MCU requires a 5V regulator.

Unless it's just a hobby project, I would be hesitant at using an MCU for such an application because there's only one supplier and the firmware cost. The LM339, is just one IC, which runs directly off 24V, is cheaper than any MCU and can be bought from multiple suppliers. The only downside is more passive components are required, but they're very cheap.

For this extremely simple purpose, you could probably get away with just a resistor and a diode for a voltage regulator accepting anything between 6 to 24 volts. The total BOM could be 1 resistor, 1 diode, 1 MCU, 3 MOSFET, 3 pots. But I am not convinced this is a real product - frankly it looks like a beginner project for DIY :-)

Anyway, using an op amp for saw tooth is not an example of the usefulness of the 555 chip.

[MT]

5:23 Big Clive about the 555 quote: "its ancient, one of the earliest "micro controllers", goes back long long way. 
https://youtu.be/0WHFwqzyu0U?t=322

Interview with Hans Camenzind inwentor of 555.
http://www.semiconductormuseum.com/Transistors/LectureHall/Camenzind/Camenzind_Page2.htm


https://www.planetanalog.com/at-50-years-the-555-timer-ic-defies-conventional-wisdom/

Bigclive found a new product using the 555 only 2 years ago.



David

[PlainName]

I recently encountered a 556 (dual 555) on a plug-in mouse scarer. One stage provided a low frequency pulse that gated the second stage; which ran 40KHz into a transducer.

Sounds like a perfect use for one (albeit dual). Just does what it's meant to and you don't have to worry about battery voltage or anything.

[EPAIII]

Perhaps a bit late in this thread so you may need to read the first 20 posts which this follows.

I just want to say that I am fully aware of the value of programmable devices. I just looked in my personal stock which goes back for many years and I did find 10 555s and 3 556s. They were in a plastic drawer with the paper napkin packing on top of them from my last move back in 2011. So I haven't touched them since then.

OTOH, on the workbench to my immediate right and out in plain sight, there are several different types of single chip processors that I have been playing with. Arduino, Raspberry Pi, etc.

In this wonderful world of electronics that we live in, THERE IS ROOM FOR BOTH TYPES OF CIRCUITS!

[PlainName]

Clearly, there is room for only one on your bench with the loser being relegated to the drawer 

[Smokey]

Clearly, there is room for only one on your bench with the loser being relegated to the drawer 

Hey you chips punks!  Get off my lawn!

[tszaboo]

The ATtiny 212 is about $.4 in the same 8 pin package as the 555 chip. It might be more compared to 555 if you buy the later in quantities, but you save on other components. It won't accept the same voltage or current ranges as 555, but for the saved components you could pair it with a mosfet.

In circuit programming is really easy. Just make three pads (vcc, gnd and program). Touch it with a pogo pin device for a few seconds and it is programmed. Can easily be automated.

The other components are literally 0.1 Cents each in quantity, and you need like 10 of them?
Programming a board is expensive on the other hand. I looked at production programmers for microchip, the SOFTLOG ICP2 is about 500 EUR, so below ~10K boards the 555 wins hands down. Above 10K boards the time spent on programming is significant. A worker at 20 EUR/hour which is almost minimal wage, cannot program more than say 120 boards an hour, which is 16 cents. Microchip's programming fee for pre-programmed part is 5 cents for this part for 5K pieces.
And I can buy NE555 compatible parts for 2 cents in China.

[baldurn]

That board also needs to be mounted in some product. The same worker or robot can do the programming. You will have a jig where you drop the board for a few seconds until a LED turns green. The worker will be mounting a board while the jig is programming the next board. This way the worker will never wait on the programmer and the extra working time per item is a few seconds at worst.

And really? Expensive to make a firmware that reads three ADCs and programs three PWM outputs accordingly? Anyone can do that in less than an hour, even a total beginner.

This is a strange product by the way. All popular multicolored LED strips these days are WS2812 based, which you could not possibly program without using a MCU. This simple MCU could do it however and with even less components (no MOSFETs).

Also the price for the programmer is really nothing as you can make your own using nothing but an old Arduino or Raspberry Pi Pico.

The Raspberry Pi Pico debug probe at €13: https://raspberrypi.dk/en/product/raspberry-pi-debug-probe/

If that is too much, you can get the Raspberry Pi Pico at €4 https://raspberrypi.dk/en/product/raspberry-pi-pico/ and download the software yourself. Takes less than 5 minutes to convert that thing to a debug probe.

Then all you need is a cable with three pogo pins and a small jig. I would 3D print it. I could easily design and print such a thing in about an hour.

[SiliconWizard]

I recently encountered a 556 (dual 555) on a plug-in mouse scarer. One stage provided a low frequency pulse that gated the second stage; which ran 40KHz into a transducer. The DIL packaged chip was dated c.2010, so there were a few analogue diehards out there. But they possibly have long retired, leaving mouse scaring to the Arduino kids.

Arduino? Nah! You need at least a RPi 5 for this and the WiringPi library. Probably Python to run on top of this, all in all requiring nearly 1GB of RAM.
But, then you'll be tickling true progress. You can even connect to some cloud provider to log the pin toggling action that happened locally and see that on a nice graph on your mobile phone.

55... what? You said?

[tszaboo]

That board also needs to be mounted in some product. The same worker or robot can do the programming. You will have a jig where you drop the board for a few seconds until a LED turns green. The worker will be mounting a board while the jig is programming the next board. This way the worker will never wait on the programmer and the extra working time per item is a few seconds at worst.

And really? Expensive to make a firmware that reads three ADCs and programs three PWM outputs accordingly? Anyone can do that in less than an hour, even a total beginner.

This is a strange product by the way. All popular multicolored LED strips these days are WS2812 based, which you could not possibly program without using a MCU. This simple MCU could do it however and with even less components (no MOSFETs).

Also the price for the programmer is really nothing as you can make your own using nothing but an old Arduino or Raspberry Pi Pico.

The Raspberry Pi Pico debug probe at €13: https://raspberrypi.dk/en/product/raspberry-pi-debug-probe/

If that is too much, you can get the Raspberry Pi Pico at €4 https://raspberrypi.dk/en/product/raspberry-pi-pico/ and download the software yourself. Takes less than 5 minutes to convert that thing to a debug probe.

Then all you need is a cable with three pogo pins and a small jig. I would 3D print it. I could easily design and print such a thing in about an hour.

I take it you never had to setup a real production line?
I've sit behind production workers with stopwatch to optimize speed. Even if you as an engineer can demonstrate it very quickly what to do, that's not sustainable on the long term. It's not possible to make 15 second tack times 8 hours straight doing the same boring task without mistakes. Also programming is often times a separate step at a separate station, because they will simply skip the test if there isn't something blocking them to do so. And then you might need to rework hours of production.
You also don't use "Arduino or something like that" to program in a real production facility. Each testing JIG is valued at ~5K EUR just because engineers had to design it, order it put it together. Also production programmers are a LOT faster. And you didn't address the bottom line, that I can make the circuit for ~3 cents in components.

[PlainName]

All popular multicolored LED strips these days are WS2812 based, which you could not possibly program without using a MCU.

Hope I've joined the right pedant's thread...

You could use WS2812 sans MCU with some memory and a shift register, perhaps a 74HC165N. Of course, you would need have a clock, and a 555 could be great for that

[jonovid]

by the bucketload from Shenzhen Yes.

[TimFox]

I believe the 2N2646 unijunction transistor is still available, but more expensive than a 555.

[Zero999]

The 8 pin ATTiny212 would have done the job with one chip instead of three chips and a much simpler schematic. You would configure three pins as ADC to sample the potentiometers and three pins as PWM output. The remaining two pins are VCC and GND.

That would be two ICs, since the MCU requires a 5V regulator.

Unless it's just a hobby project, I would be hesitant at using an MCU for such an application because there's only one supplier and the firmware cost. The LM339, is just one IC, which runs directly off 24V, is cheaper than any MCU and can be bought from multiple suppliers. The only downside is more passive components are required, but they're very cheap.

For this extremely simple purpose, you could probably get away with just a resistor and a diode for a voltage regulator accepting anything between 6 to 24 volts. The total BOM could be 1 resistor, 1 diode, 1 MCU, 3 MOSFET, 3 pots. But I am not convinced this is a real product - frankly it looks like a beginner project for DIY :-)

Anyway, using an op amp for saw tooth is not an example of the usefulness of the 555 chip.

It's an example of the uselessness off a 555 and that there is a more suitable IC for said application.

Here's an LM393 PWM circuit I designed awhile ago. Just use the LM339 and repeat the comparator section another two times. If it's going to be run off 24V then make the pull-ups higher, say 4k7 or 5k6, instead of 1k, and add a pull-down, say 12k or 18k, to the outputs driving the MOSFETs, to limit the gate voltage to below 20V.

5:23 Big Clive about the 555 quote: "its ancient, one of the earliest "micro controllers", goes back long long way. 
https://youtu.be/0WHFwqzyu0U?t=322

I don't know whether it was tongue in cheek or a slip of the tongue and he meant to say microchip.

[SiliconWizard]

All popular multicolored LED strips these days are WS2812 based, which you could not possibly program without using a MCU.

Hope I've joined the right pedant's thread...

You could use WS2812 sans MCU with some memory and a shift register, perhaps a 74HC165N. Of course, you would need have a clock, and a 555 could be great for that

Well, yes. Implementing the required encoding for the 0 and 1 could be directly stored in the memory.
And for the clock, just a single gate inverter with Schmitt trigger input would do.
Possibly a fun exercise, rather than very practical.
But while this would work with only "discrete" logic, with some programmed memory, technically it would still be "programmable logic". You would have effectively designed a very simple and dedicated processor.

Anyway, the initial question was weird from the start. There are thousands of references of BJTs these days, so is the 2N3904 still a viable transistor?
PIck whatever fits your requirements and (if longevity matters) is reasonably (/or guaranted) likely to be available in a few years from now. Maybe that's called engineering, or something. Not sure anymore.

[coppice]

I believe the 2N2646 unijunction transistor is still available, but more expensive than a 555.

That takes me back. I used quite a few of those in my youth. They seemed like such a cool device when they were launched.

[schmitt trigger]

I believe the 2N2646 unijunction transistor is still available, but more expensive than a 555.

That takes me back. I used quite a few of those in my youth. They seemed like such a cool device when they were launched.

Even cooler were the Programmable Unijunction Transistor, PUT.
Which I called the triple-lie semiconductor:
-It wasn’t “programable”, it was adjustable.
-It wasn’t a single junction, but three junctions.
-It wasn’t a transistor but a thyristor.

Nevertheless a very useful device.

[tooki]

I believe the 2N2646 unijunction transistor is still available, but more expensive than a 555.

That takes me back. I used quite a few of those in my youth. They seemed like such a cool device when they were launched.

Even cooler were the Programmable Unijunction Transistor, PUT.
Which I called the triple-lie semiconductor:
-It wasn’t “programable”, it was adjustable.
-It wasn’t a single junction, but three junctions.
-It wasn’t a transistor but a thyristor.

Nevertheless a very useful device.

In analog electronics, “programmable” is sometimes used as a synonym for “adjustable” or “settable”. It means nothing more than that there is some property that is not fixed at the factory, but is instead designed to be user-configured, especially in cases where non-configurable is the default for that type of product.