There are many places where the skill bar is rising, in some places in ways which are clearly not for the benefit of anyone who wants to mess with the technology. This in itself can be a whole discussion

I would not lump SMD as a whole in there, and definitely not lump all of SMD into one "thing". Indeed, it is not through hole, but in 99% of the components i might use in a design, i get to choose the component size to my liking within what's available

Check out the following component sizes and see if thy are ok for you :

 - ICs : SO and SOIC packages, 0.05" pin pitch. For 16+ pin ICs a "wide body" option is often available. DIP adapters are available to plug those into breadboards as through hole components

 - Transistors : TO-261 (SOT223), TO-252 (DPak), TO-263 (D2Pak). Those are generally packages of higher power transistors, but in many cases they can be used for small signal stuff without issues (just being overkill in size and cost), at least as long as you have 5V+ supply available to drive larger MOSFETs sufficiently for small signal work

 - Small transistors : TO-243 (SOT89), SOT23

 - Diodes : DO214 (3 sizes : SMA SMB SMC), SOD123

 - Resistors and capacitors : 1206 and larger (2010, 2512)

My own preference for components : TSSOP/MSOP generally ok, (SO/SOIC for 8 pin chips, SO wide-body for 16+ pin chips if i want to trade space for making it easier to assemble), SOT23 (SOT323 and SOT363 in a pinch), SOT23-5 (some small chips), SOD323, 0805 (0603 if i want to save space, which i usually do)

That is what i can do with just an iron, without much effort, and without needing the magnifier (though i do have a few circline magnifiers and a microscope). I can do some higher level stuff, but dont want to, and i avoid it from things i design. I consider my SMD skill as medium-low to medium



The type of iron you use will absolutely determine your soldering experience

The entry level for anything more than TH components on single sided PCBs is a closed loop temperature controlled iron, such as most entry level "stations" (including the many Chinese clones of old Hakkos)

(Admitting, i have successfully got away with 25W "just soldering iron" and 30/120W "two level handgun" irons through most of my life as a younger kid, but keep that to "in a pinch" situations nowadays)

The next level, good for easy SMD jobs, is an iron with improved heat conduction. Those are the newer ones with element built into the tip, rather than a ceramic element with screw-on tip as most older "stations"

The further next level is an iron with heat generation at the tip itself. Those are induction irons. Here i am now with an old Metcal SP200 and 1.5mm flat tip



Nothing beats Through Hole, and many things can be done with Through Hole at the level you are now exploring. This circuit is no exception, the CMOS 555's and the suggested transistors i provided are all Through Hole. (You may find better ones in SMD, but those will work)

And if you have to use some component which is SMD, you still can switch only this one component to SMD in a Through Hole circuit



I have not suggested the bipolar-based entire circuit route. My suggestion remains with CMOS 555 and a suitable output transistor, with the following options :

 - Use a MOSFET which will work at the 3.x V drive current and handle your load. Not all MOSFETs will have low RDS at such voltage, but the good part is, the MOSFET's resistance will just behave as another series resistor with the LED. It will self regulate to a point where the LED can still work with the provided voltage, even if a little dimmer. At such small loads and with the MOSFET being otherwise adequate for the current and power dissipation, you are unlikely to run into SOA problems

 - Use any MOSFET + a charge pump driving circuit (requires extra 1 capacitor, 1 resistor and 1 diode). This will open your choice to a wide range of MOSFETs with good RDS and current ratings, but it will be absolutely essential that the pump keeps "pumping" (ie. that the 555 keeps flashing the LED, and not just stay on), or else your circuit will drop into the twilight zone of insufficient MOSFET driving

 - Use a bipolar transistor. This can be easily driven by the 555, but requires non-zero driving current (generally not an issue) and drops a fairly fixed voltage even at low currents, limiting the voltage headroom for the LED

@Medved

A symmetrical multivibrator you pictured does not have problems starting even using silicon transistors. In practice, RC constants of base networks will always be slightly asymmetrical, enabling the circuit to start.

It is not that the circuit would stubborny refuse to start when powered ON, it is just not reliable, it requires sufficiently fast turn ON on the supply. And the reliability gets worse once the capacitors ESR increases or if the transistors have higher internal series base resistance (like most high beta transistors do, mainly important at higher currents) and/or the supply has no other source of noise/disturbance on it (sufficient fast ripple on the supply can restart the circuit from the deadlock, but then speeds up the oscillation)
It has nothing to do with symmetry. Try to ramp up the supply slowly, mainly with LEDs as the loads, and you get both transistors steady ON, whatever assymetry you have.
This circuit worked reliable with tubes, because even at the "both ON state" (with grids forward biased and anode voltages low) they had some gain greater than unity, so the circuit started to oscillate. But with silicon transistors featuring hard saturation the transistors lose all voltage gain, so no oscillation start.
Normally when trying it, you start with both capacitors at zero voltage, so when the power supply appears suddenly (rampup faster than time constants) the discharged capacitors do make a kind of bistable flip-flop with it which does select one state (which one depends on symmetry or random noise), where once one transistor becomes off for at least a brief moment kick starts the circuit.
But some brownout, or the supply not rising fast enough and the circuit stays in that "both ON" deadlock.

By the way there is a remedy for it: Do not tie the top end of the base resistors to the supply, but to a separate point, fed from collectors by two diodes. In that configuration when both transistors tend to go on at the same time, the top side of the base resistors loses power, so keeps transistors away from saturation. In that state they have plenty of gain, so the positive feedback via the capacitors amplifies the noise so much it starts to oscillate.
And once running, the top of those resistors get fed from the supply voltage present at the collector of the transistor that is at that moment off (neglecting the voltage drop across the liad). Because they alternate, the bases get supply all the time, so the thing keeps running normally.
But with these diodes we are not talking about the basic circuit anymore...

I am interested in one more question :

Assuming the interest is to make it last on a battery as long as possible, changing the flashing from 50% duty cycle to 5..10% is obvious way to extend the battery life. (With 555 this requires additional resistor and diode, as the duty cycle is reverse to what is achievable with the discharge pin. I would drive the capacitor from the 555 main output. With the 2 bipolar multivibrator i guess it can be done just by using assymetrical capacitors ?)

Is there any additional visual brightness benefit (that could be turned into energy saving), if within the on period, the LED strobes at some visible (few Hz) frequency instead of steady on, like using a 2nd 555 ? Instinctively i'd think yes, if only for the facts that it allows to still have some length to the "on" period while actually further cutting down the duty cycle, and that the extra flicker may help catch the eye more for the same average brightness

There isn't any "one size fits all" for beacons, it isn't just about the brightness when it technically becomes visible, there is very strong psychological component to when it becomes noticed on one hand to how well it could be tracked after being noticed.
And all that  in a compromise against to not being too obnoxious or steal attention from somewhere else where it is needed as well or even more.

If we are talking about notifying someone about some obstacle normally not expected there, some form of flashing pattern would grab attention better, so needs less power.

On the other hands if you want to mark thins to aid alignment (like maneuvring ship under a bridhe,...) a steady nonflashing light is the best.

You may look for some standards for various uses, minimizing power need for such things is quite common requirements so most of them are really optimized to do the job the best with minimum power.


For a small device you also need to factor in the controller consumption, mainly with low duty ratio strobes it becomes very significant power consumer of the thing. There the two transistor multivibratou (keeping its reliability problem aside) is really performing quite bad.
On the other hand if the pattern would be alternating lights, it will be performing good (with the reliability fix, of course)...

@HomeBrewLamps If you insist on through-hole, I'd try something like this. Still not much components, easy to solder. Adjust R2-R3 for desired repetition rate and duty. At least the circuit does it not in 1960's style, but in 1980's style. Hey, Italo! ))  Yes, I love 74HC14 of different breeds, never let me down...

This circuit will drive many kinds of bipolar and mosfet transistors of your choice.

For production level reliability, probably add 100-Ohm resistor is series with pin 1 of the IC.

@RRK I actually have some 74HC14's laying around too. I like how that schematic reads. Very simple.

Thr electronics I tend to gravitate towards are 80s-mid 20s usually. Probably because they are "easier" to understand. Seeing a billion micro parts on a board or a black blob with a bunch if traces leading into it is not as intuitive. I am no professional though obviously. Even the 60s stuff throws me for a loop sometimes.


I actually have made an astable multivibrator once with DM160 tubes... out of curiousity (not for any practical reasons whatsoever) is it even reasonably possible to make a tube vibrator flash an LED? Probably with a more generous supply voltage like 9V I know in the past there was pocket radios with micro tubes in them. The DM160 and DM54 were supposedly used as "tune indicators". I never have seen one personally. You'd think I'd have found atleast one of those pocket radios by now... but it seems stuff like boom boxes and pocket radios don't have as much permanence as a rotary phone or fridge.

There isn't any "one size fits all" for beacons, it isn't just about the brightness when it technically becomes visible, there is very strong psychological component to when it becomes noticed on one hand to how well it could be tracked after being noticed.
And all that  in a compromise against to not being too obnoxious or steal attention from somewhere else where it is needed as well or even more.

If we are talking about notifying someone about some obstacle normally not expected there, some form of flashing pattern would grab attention better, so needs less power.

On the other hands if you want to mark thins to aid alignment (like maneuvring ship under a bridhe,...) a steady nonflashing light is the best.

You may look for some standards for various uses, minimizing power need for such things is quite common requirements so most of them are really optimized to do the job the best with minimum power.


For a small device you also need to factor in the controller consumption, mainly with low duty ratio strobes it becomes very significant power consumer of the thing. There the two transistor multivibratou (keeping its reliability problem aside) is really performing quite bad.
On the other hand if the pattern would be alternating lights, it will be performing good (with the reliability fix, of course)...

I am sure with a micro controller accomplishing a decent attention grabbing flash pattern is possible. I own cone lights I use for roadside service on tractor trailers that strobe and go off for around a second. Pretty sure when I took those lights apart I found a microcontroller inside them. But thr company that makes them put black epoxy over the chip so I cannot see what kindof chip it actually is.


I bet that a decent track-able pattern would be a rapid strobe and then constant on for a few seconds repeating. But then battery life issues may start happening. Especially in the winter when temperatures hit extreme lows. I dont have a microcontroller burner or coding skills at the moment to mess with all that.


Someday perhaps.

Here is my circuit. It will do 0.8sec 7Hz flicker & 3.2sec darkness, i.e. 10% real duty cycle on the battery and ~20% visible duty cycle. It should be fairly efficient, but the pauses are long

All through hole

Experiment with the frequency and duty cycle to get your preferred pattern

The 7555's can be substituted with ordinary 555's if supply voltage is increased to 5V (ordinary 555's will drain the battery more, although still not too bad compared to the energy used by the LED)

Be careful with the MOSFET gate at higfher supply voltages. The IRLU024N is 16V Vgs max, many other MOSFETs have similar ratings. If you supply higher voltage, and the charge pump further doubles it, you may end up exceeding the MOSFET ratings. The charge pump is not needed and can be disabled if you use high supply voltage

Pattern with all 3 modes (dark, flicker, steady) can be done but requires additional components

Like your creative use of DIS pin )

The question however is if DIS is a true open drain and pushing it above Vcc is really allowed. Datasheet is not clear on this, even more, a blanket statement of "Due to the SCR structure inherent in the CMOS process used to fabricate these devices, connecting any terminal to a volt-
age greater than V+ + 0.3V or less than V- - 0.3V may cause destructive latchup. " is present. TLC555 explicitly allows for this, btw.


Using a double device of 556 may save a bit of board space.

Depending on the actual battery type and load current / internal resistance, 3.15V threshold may be too high. Definitely too high for single use lithium batteries. For rechargeableies, it may be economical just to use a battery with built-in protection, though discharging them to the point of internal protection set-off every time may be a bit rough.

Indeed. TLC555C works as well (note some other variants of TLC555 may have too high min operating voltage). LMC555 appears to also allow Vdis > Vdd and have lower current draw (though both are insignificant vs the LED)

The MCP112 have few versions with similar thresholds. The next ones are 3.0, 2.9, 2.7V. (All a little lower actual tripping point than the name suggests - see datasheet). I assume a rechargeable battery, for single use there is no much point in the shutdown circuit anyway

556 (all types) is a less common device, so somewhat higher chance that a specific chip required won't be in stock

Additional design bug corrected which i spotted now - The supply to the charge pump is now from before the decoupling, to prevent the Vdd of all chips dropping when the charge pump is charging

Like your creative use of DIS pin )

The question however is if DIS is a true open drain and pushing it above Vcc is really allowed. Datasheet is not clear on this, even more, a blanket statement of "Due to the SCR structure inherent in the CMOS process used to fabricate these devices, connecting any terminal to a volt-
age greater than V+ + 0.3V or less than V- - 0.3V may cause destructive latchup. " is present. TLC555 explicitly allows for this, btw.
ouble device of 556 may save a bit of board space.

Chapter 5.1 Absolute maximum ratings clearly shows no limitation to the DIS voltage vs VDD, it just limits it to 18V (and 0.3V below GND). That means the output transistor (or an extra GGNMOS parallel to it) is all what is protecting this pin against ESD, forward body diode for negative and intrinsic bipolar Bvces for positive voltages. No diode towards VDD.
The previous "any input" line really limits to the VDD, so that tells to me there are just diodes towards VDD and GND as SD protections.

 In fact you may always check this using a multimeter (if you have some clone IC where this is really not 100% clear how much identical the design is) - if you see a diode towards a supply, biasing that in forward (even less than 1mA) may trigger erratic behavior and when excessive (above the 50..100mA the normal CMOS are usually designed for), even a latchup.
And this is by far not limited to CMOS processes, it is common to any junction isnsulated monolithic process, whenever there is something looking like a diode biased in forward, you could be pretty damn sure it will be part of a parasitic bipolar. The only thing you do not know is where are all the collectors (there are more than one) and how much of the current flows there and how that part is sensitive.
But to activate the parasitic bipolar transistors (to e.g. trip the latchup), you need to forward bias their base-emitter (aka the diode your multimeter sees) first.
But if there is no such "diode" biased in forward, there won't be any parasitic collector currents...

Chapter 5.1 looks like you are looking in a TI datasheet. The TI parts indeed show no limits on Vdis > Vdd

The ICM7555 i meant initially is a Renesas part (a similar one is made by Analog Devices)

I dont expect to see anything other than an open drain FET there (unless there is some circuit closing through the substrate to the other transistors in the chip ?), there sure isn't anything drawn in the schematic (Renesas page 5, Analog page 9)



If there is an ESD diode from Dis to Vdd, this would just clamp Vdis to Vdd

In this circuit this will just clamp the charge pump, returning to 3.x V driving of the MOSFET

In some other circuit, with lower impedance source on Dis, i can see how this can lead to blowing up the ESD diode (and chip), if this results in lifting Vdd to Vdis and result in significant current draw through the diode to power other things on the same Vdd

I still fail to see how this can lead specifically to a latch up ?



If i would build this circuit for myself, i would quickly find out that something is amiss after building it on a breadboard and powering up (or maybe not, if it works fine despite the wrong use)

However, here i am fairly blindly drawing a circuit for somebody else to assemble, so if there really is something odd going on with this pin, replacing to a chip without such limitation makes sense

Yes, I was looking into TI datasheet.

And no, ESD diode won't be "just clamping". On a monolithic piece of semiconductor you will practically never get a plain "diode". It will always act as a bipolar transistor, that "diode" acting as its base-emitter junction. But a collector is always there, most often even more than a single one. And unless that "diode" is explicitely meant to carry currents during normal operation (which an ESD protection diode is definitely not),that collector may be some part of some other internal circuit. Or it could be not that greatly connected resistive substrate, allowing it to reach high potentials which in turn may activate another bipolar (and that is how the latch up starts), or at least cause internal local ground bounces, disturbing other circuitry as well.
Only if such "diode" is designed to carry current in forward during normal operation, either it is created as a transistor with the collector connected to the base, or arranged so the collector current actually helps to carry the output current (like a series "diode" with a low side driver actually being a PNP emitter follower) or at least the collector current taking care of so it won't interfere (good substrate contacts around the diode, auxiliary collector rings connected to the other supply,...), or it won't be a real diode at all, but e.g. cleverly connected bunch of MOSFET that carry all the current via their channels as MOSFETs, without activating any PN junction in forward at all. But all this means extra complexity and silicon area, so it is done only if that diode is supposed to work as a diode during normal operation. Protection diodes are not that, there the added measures are meant to just prevent destruction (latchup), but not unwanted errorneous operation of the device.


With the substrate there is one catch: Nat always it is of a P-type and connected to ground (as it is the most common way for most bipolar, as well as CMOS processes).
Best example is the CMOS4000 (~3um 16V gate oxide CMOS), which uses N type substrate, connected to the positive supply (VDD). If the thing is manufactured in that process, it becomes almost impossible to have any pin wthout a "diode" towards VDD (with parasitic collectors all over the chip).

But such diode is visible by the simple multimeter test: It reads 0.5..0.8V forward as any normal silicon diode. If there is no current (and I mean really no current, not even uA, even the highest MOhm range should show OpenLoad), there can not be any parasitic bipolar.
Still it is a good idea to bypass the potential protection or parasitic diode by anexternal Schottky. That way you ensure even when some diode is there, it wont get any current...

From practice, I am under impression that modern branded logic families like TI or NXP or so almost completely solved latch-up problem and survive quite significant abuse! Though some lesser crap sometimes still does...

Anyway, this circuit is not at risk of latch-up, but maximum gate voltage will be clamped at Vcc + 0.6-0.7V

Indeed, the latch up itself is really hard to trigger, common standards require latchup to not occure till at least 100mA at maximum design junction temperature (where the thing uses to be most sensitive), which is pretty high.

What you will meet is functional disturbance coupling via the parasitic elements on the chip.
Like pulling some inputs below ground activate response completely against normal functional logic, like pulling one noninverting double-opamp IC input negative causes both outputs to go high. This behavior is hard to predict, even when dealing with an IC from one single manufacturer. Because it may seem to be the same IC part number and specification since it was introduced 40 years ago, but usually the chip itself was transfered and redesigned for more modern (and cheaper) processes half dozen times, so you can meet easily 6 completely different designs, each with its different weak and strong points, mainly regarding out of spec condition response.