The perils of.. using breadboards (was.. of using any old schematic off the web)

[smjcuk]

Blowing up stuff is half the fun

Ok so after a couple of hours of playing with this I've come to the conclusion: yes it was the sodding breadboard. I found that the moment you apply pressure on one part, two of the rows shorted out. They happen to be the ones that would short the output to ground meaning it sinks as a short, latches down and goes phut. The output voltage still stays high due to the size of the capacitors. I went round the entire board with a multimeter and two pins to found this out.

So the thread title should be, The perils of using a breadboard. If any moderators fancy changing it, please do.

So I played around with the component values and found a suitable operating point. I built the entire thing ugly style this time as well:



Running that into a 470 ohm load which sinks ~20mA of current shows the following outputs which is fine. I'm drawing 10mA from this so I'm going to stick a small linear reg on the end of it as well to float the voltage at a reasonable level. Meters show source current and output voltage.



And a hat tip to Bob Widlar, the breadboard has been dealt with appropriately:



Thanks to paulie for the education which was very valuable here

[paulie]

LOL. Yes, THAT's the ticket. Are those UT33? If so very good meter. And that is proof it's better to buy a couple low cost instead of taking out a loan for one bulky overpriced "real" meter.

Moral of the story, RTFDS.

Datasheets are for GIRLS!!! Real men use the wire 'n fire method. The trick to that approach is never buy just one component but get a whole bag of the damn things.

[c4757p]

And a hat tip to Bob Widlar, the breadboard has been dealt with appropriately:

It ain't Widlarized until it's powder!

[IanB]

So the thread title should be, The perils of using a breadboard. If any moderators fancy changing it, please do.

No need for a mod. Just edit the first post and change the title.

[smjcuk]

LOL. Yes, THAT's the ticket. Are those UT33? If so very good meter. And that is proof it's better to buy a couple low cost instead of taking out a loan for one bulky overpriced "real" meter.

Yes they are indeed UT33As. Lovely meters - cheap and reliable.

Datasheets are for GIRLS!!! Real men use the wire 'n fire method. The trick to that approach is never buy just one component but get a whole bag of the damn things.

Exactly this.  If there isn't the burning acrid smell of semiconductor or resistor in the atmosphere you haven't learned anything.

As for powdering the breadboard, my piddly little hammer didn't have enough watts.

[dom0]

As for powdering the breadboard, my piddly little hammer didn't have enough watts.

More like you not having enough watts

[atferrari]

I have also used a similar circuit many times with no hint of failure. Besides 555 a variety of other drivers from gates to MCU pins. Assuming no wiring error then almost certainly caused by intermittent plug-board contact. There are several points in this circuit that will cause self destruct if opened even briefly.

I stopped using those years ago. IMO strictly for the beginner/tarduino crowd who need the education (generally what NOT to do). Contrary to urban myth, solder/deadbug is actually quicker and more convenient in addition to being a cheaper and infinitely more reliable method.

Yes.  Time "lost" in planning the distribution even in perfboard and then soldering ends being less than chasing random "failures" in a protoboard. These have also an awful distributed capacitance.

A Month ago, a transistor, refused to switch in about 2 usec until properly soldered in perfboard. Little by little, my basic proof of concept are all soldered.

[smjcuk]

I really can't do perfboard. Just doesn't work for me. I'm probably doing it wrong. Some people seem to use component leads for tracks but I can't get on with that approach as I switch out components a lot. I've settled on (as of yesterday), Jim Williams style prototypes then Eagle+OSHpark. I've got three perfboards here; one burned to death and two unused.

[tggzzz]

I really can't do perfboard. Just doesn't work for me. I'm probably doing it wrong. Some people seem to use component leads for tracks but I can't get on with that approach as I switch out components a lot.

Why don't you design the circuit?

On second thoughts, if you use a breadboard with its unpredictable parasitic capactiance, inductance, and resistance, you probably can't design a circuit.

Throw the damn things out, save yourself time and money, and get a sense of accomplishment when your circuit works according to theory and your design.

[smjcuk]

I do design the circuits...

In this case I was purely being lazy. TBH breadboards have been little trouble for me under 4MHz or so over the last 20 years. This is the first identifiable failure caused by one.

I'm using neither perfboard or breadboards now. Perfboard is inflexible and requires too much effort. Breadboard is unreliable. So dead bug/Manhattan. Seems to be the best compromise. When it works, over to PCB.

For ref this entire design is scrapped now. I've designed something far more efficient on paper using discretes. 64KHz multivibrator, buffer, inverter, two 10v +/- regulators. No ICs. Just transistors, zeners, diodes, caps and resistors. 22 components for the entire solution.

[Zero999]

I really can't do perfboard. Just doesn't work for me. I'm probably doing it wrong. Some people seem to use component leads for tracks but I can't get on with that approach as I switch out components a lot. I've settled on (as of yesterday), Jim Williams style prototypes then Eagle+OSHpark. I've got three perfboards here; one burned to death and two unused.

I'm not a fan of perfboard either.

My advice is when you do use perfboard, don't use the component leads to make connections. Use thin lacquer coated wire, the kind you can just solder directly to. It's much easier. Any components you think you may change can be socketed.

I prefer stripboard for prototypes. I'm not a fan of breadboard but use it occasional for small things.

[tggzzz]

Perfboard is inflexible and requires too much effort. Breadboard is unreliable. So dead bug/Manhattan. Seems to be the best compromise. When it works, over to PCB.

I agree, except that I use perfboard for up to 4 components or where I need a little mechanical strength.

[smjcuk]

I'm using offcuts of stripboard glued to the PCB blank for that sort of stuff. I assume there is no mechanical stability to start with after having a toddler attempt to kill an entire box of my stuff

[tggzzz]

I'm using offcuts of stripboard glued to the PCB blank for that sort of stuff. I assume there is no mechanical stability to start with after having a toddler attempt to kill an entire box of my stuff

You used to be able to buy that kind of thing, whit varying track width for different characteristic impedances. Very useful for RF stuff. The radio hams still use such techniques as a matter of course.

[smjcuk]

Yeah I've got the ARRL handbook here and read about that. I'm trying to stay under 20MHz as it's a bugger above that. The company I used to work for was all microwave RF stuff. Waveguides, cavity resonators and the like. Ugh couldn't be bothered with all that.

[Howardlong]

FWIW I do every construction method under the sun bar wire wrapping.

It depends on the situation. For almost every MCU project that has a DIL part I'll use a breadboard or perfboard, and if there's some nastiness like SMPS or high speed mixed signal then it's either Manhattan/dead bug or I design a board. Sometimes it's a hybrid with the MCU still on breadboard. For larger prototypes with large SMD devices, I tend to use Schmartboards or other breakout boards together with the odd mini custom board as appropriate. Schmartboards are pretty good because they have ground planes making decoupling and ground connections easy with 0603 caps and 0R resistors.

For HF and below with SMD parts, it's either perfboard or I make up a board. VHF and above I always make up a board, too much time wasted trying to make any other way work IMHO.

I realise that there's a tendency to complain about breadboards, but I rarely have a problem with them and I do use them pretty much every day: I can have a basic initial rubber stamp DIL PIC project up and debugging in 10 minutes, but I am geared up for exactly that scenario as it happens so frequently. The trick is knowing when to move off the breadboard... and I'd say a 20mA charge pump really oughta work on a breadboard, so I don't know what to make of what happened. Indeed I made a charge pump last week from an oscillating PIC pin to aid bootstrapping a mosfet gate, no problems at all, although it was only a few hundred uA rather than 20mA.

Irrespective, it's time to get out that current limiting PSU of yours!

[kolbep]

I had a 555 and 4017 blow their lids into the ceiling and across the room.
Hint: if you do not know enough about capacitive power supplies, you DO need the Zener diode....

Come to think of it, a pic also suffered the same fate

That is why it is only linear power supplies for me