Difficult breadboading a LM324N

[dcbrown73]

As I learn or see interesting circuits / components that I want to learn, I order them. 

Well, I ordered a pack of 10 (or maybe it was 20) LM324N ICs.   They work as an opamp fine, but it has four opamps on a single chip which seemed to me like a good thing especially since I was going to be learning while making an example function generator that required three opamps.  (do it all with a single chip with one to spare!)

Well, in practice it seems that breadboarding with one of these seems a lot more difficult than what three singles would have been.  Four different opamp pin locations being close together and all on one side of the chip makes for layout issues.  While I'm sure it's not an issue on a pcb, it makes for more of an issue on a breadboard.  Especially if you are a neat freak in that you at least like to be able to see (read understand) the circuit layout once it's finished.  (or if you need to post a picture and ask for help!)

Maybe there is a better way to lay it out, but I haven't gotten that far yet.

Anyhow, that brings up my next question.  Is there any best practices or "list" of pitfalls to avoid when you're first learning?  If so, I would love to review it!

[tggzzz]

Pitfalls to avoid?

Firstly make sure unused inputs aren't left floating?

Secondly make sure you haven good decoupling and a good ground plane.

Finally, don't use solderless breadboards; you will spend more time debugging the breadboard thank your circuits. See https://entertaininghacks.wordpress.com/2020/07/22/prototyping-circuits-easy-cheap-fast-reliable-techniques/ for good cheap fast  alternatives used by very experienced engineers.

[edavid]

Well, in practice it seems that breadboarding with one of these seems a lot more difficult than what three singles would have been.  Four different opamp pin locations being close together and all on one side of the chip makes for layout issues.

Nobody said you have to use all of the opamps.  Just stick another chip in if that's more convenient.

Firstly make sure unused inputs aren't left floating?

Doesn't really matter for the LM324.

[GerryR]

Seems like nobody uses wire-wrapping breadboard techniques along with component carriers.  Makes life of bread-boarding, much simpler.  (Guess I'm showing my age now, but I still use wire-wrapping to make / test circuits.)  Also, pretty easy to strip down and re-use if you only need to test a design, but also good for permanent circuits.

[tggzzz]

Seems like nobody uses wire-wrapping breadboard techniques along with component carriers.  Makes life of bread-boarding, much simpler.  (Guess I'm showing my age now, but I still use wire-wrapping to make / test circuits.)  Also, pretty easy to strip down and re-use if you only need to test a design, but also good for permanent circuits.

With modern components the length of the wirewrap pins become a problem. Apart from that I still like wirewrap

[floobydust]

Firstly make sure unused inputs aren't left floating?

Doesn't really matter for the LM324.

I don't agree, there is circuitry common to all four amplifiers. If one amplifier's inputs are floating and it's HF oscillating, I have seen other amplifier channels get affected.

[tggzzz]

Firstly make sure unused inputs aren't left floating?

Doesn't really matter for the LM324.

I don't agree, there is circuitry common to all four amplifiers. If one amplifier's inputs are floating and it's HF oscillating, I have seen other amplifier channels get affected.

On a breadboard there is more common circuitry than that. Wire is a 1nH inductor for every mm of its length. So a 10cm wire in the PSU rail connection plus a 10nF decoupling capacitor will resonate at what frequency?

[edavid]

An LM324 is much too slow to oscillate at HF 

[T3sl4co1l]

Yeah, the pinout is kind of suck.  I prefer duals, like LM358.

Or since I no longer give a crap about cents of component cost, I reach for TLV2372s now, for breadboarding.

I don't agree, there is circuitry common to all four amplifiers. If one amplifier's inputs are floating and it's HF oscillating, I have seen other amplifier channels get affected.

Depends who made it.  Different versions have independent biasing.  Anyway, that should only be a problem if the inputs fall too high or too low.  Which, open circuited should float too high if I'm not mistaken; unused inputs should be grounded for this reason.

On a breadboard there is more common circuitry than that. Wire is a 1nH inductor for every mm of its length. So a 10cm wire in the PSU rail connection plus a 10nF decoupling capacitor will resonate at what frequency?

A good reason to include a nice lossy electrolytic capacitor somewhere in the supply.  100nH and 10nF is a few ohms resonant impedance, well damped by a capacitor of the same ESR.  If there's no smaller impedance directly in parallel with the capacitor, then its ESR will dominate and this damping will apply to any series LC branching off it (give or take the total parallel equivalent of them all, obviously).  Or, place the electrolytic at the end point (i.e., in parallel with the 10nF, reasonably nearby), so there's effectively a big resistor in parallel with the tank, damping it all the same.

An LM324 doesn't have much impedance, or need much around it; just a few uF electrolytic alone is adequate.

An LM324 is much too slow to oscillate at HF 

HF starts at 3MHz or so, near the GBW of the amp.  The full loop (input to output) is pretty unlikely to run that hot, but a capacitive load on the output might induce that section to oscillate up there.

Tim

[David Hess]

Yeah, the pinout is kind of suck.  I prefer duals, like LM358.

I prefer dual parts as well and the selection is better.  When laying out printed circuit board, I find that they pack better than quads anyway.

For unused operational amplifiers, the safe thing is to connect the output to the inverting input and the non-inverting input to a fixed voltage within the common mode range.

Connecting the inputs to the positive or negative supply can be used to disable that section but it depends on the details of the amplifier type.

[Kleinstein]

The LM324 should go to a defined state with floating inputs, as it does not have bias current compensation. So with open inputs they just miss the base current for the input stage. So most of the time it's OK to leave the inputs open. The better way is to connect the inv. input to the output, making it a follower.

The LM324 pinout is not even that difficult, it's rather symmetric and easy to remember.
For decoupling on the breadboard, I some times / for a few parts I have have there have the decoupling cap directly soldered on top of the chip. For the LM324 / LM358 or similar even an electrolytic cap (e.g. 10 µF)  is "fast" enough and even shows the polarity.

The problem with a 4 fold OP is more that everything is so close together. So I prefer dual OPs.

The cross over distortion of the LM324 / 358 sometimes gives some unpleasant effects.

[tggzzz]

For decoupling on the breadboard, I some times / for a few parts I have have there have the decoupling cap directly soldered on top of the chip.

Agreed, but most people using a solderless breadboard wouldn't know that was beneficial. Plus if you are doing that, why not go straight to "decent" manhattan/deadbug/ratsnest techniques?

[magic]

The LM324 should go to a defined state with floating inputs, as it does not have bias current compensation. So with open inputs they just miss the base current for the input stage.

Input bias isn't cancelled but the base current of the second stage may be cancelled by deliberate imbalance of transistors Q19, Q2, Q5 such that a constant current is injected into Q6, roughly sufficient to sink its collector load. Then, when the loop is closed, the input stage would only supply a minor correction to get it perfectly right. I'm not saying that they bothered, though.

[Brutte]

The problems of solderless breadboarding only come from the low quality breadboards and jumpers.
If you buy a decent ones with phosphor-bronze nickel plated contacts that have indent to hold the leads in the center, unless you mechanically violate those, work just fine.

But if you spill a coffee (with sugar), push in a >-1mm² wire or TO220, use dirty components or try to bend breadboards, etc, well then even a phsphor bronze won't help and you'd better use soldered breadboard. I'd also recommend using tinned jumper wires.

The LM324N is absolutely fine as long as you use 2 - 3 boards stacked. With single board it is indeed PITA.

The breadboards that I use look like this one from Distrelec (except mine were half the price). The power line bus can be detached and the cores can be stacked so with 2 boards you will get 63 rows that have 5-2-5-2-5-2-5 tie points (5 holes - 2 voids - 5 holes - ...... ) and if you stick LM324 in the center, there is a plenty of room for connections.

[dcbrown73]

Agreed, but most people using a solderless breadboard wouldn't know that was beneficial. Plus if you are doing that, why not go straight to "decent" manhattan/deadbug/ratsnest techniques?

The only issue I have with manhattan / deadbug / ratsnest is I'm learning and am likely to need to move and change things often as  I learn.  If I'm actually soldering and desoldering to change everything.  That can take 15-30 minutes of learning and turn it into hours.  Not to mention, I'm probably sure to mangle a few components in the process.

Maybe I breadboard it and once it's functioning to some extent.  Move to a different method for further testing?

[rstofer]

Why don't you use single or dual op amps (or even just one op amp in a 4 pack) so you can spread out your work?  It makes troubleshooting a lot simpler.

There's no good coming from trying to jam together all the parts for 4 op amps in a single package. 

After you get the thing working, get a solder type breadboard.  Here's a solder type breadboard that exactly matches a solderless:

https://www.jameco.com/z/SB300-Busboard-Prototype-Systems-SB300-Solderable-PC-Breadboard-1-Sided-PCB-Matches-300-Tie-Point-Breadboards_2125042.html

There are others.

[tggzzz]

Agreed, but most people using a solderless breadboard wouldn't know that was beneficial. Plus if you are doing that, why not go straight to "decent" manhattan/deadbug/ratsnest techniques?

The only issue I have with manhattan / deadbug / ratsnest is I'm learning and am likely to need to move and change things often as  I learn.  If I'm actually soldering and desoldering to change everything.  That can take 15-30 minutes of learning and turn it into hours.  Not to mention, I'm probably sure to mangle a few components in the process.

You will mangle a few components anyway

Soldering is fast and once it works it stays put. Debugging poor/intermittent connections, stray capacitance and inductance is slow. Sneeze and something will have moved, and you'll start again.

Have one plain PCB as a groundplane. Chop up some bits of piain PCB into 5mm pieces. When you want to join two components, superglue one bit to your groundplane whereever happens to be convenient, and then solder your components to that.

But if you enjoy debugging a connection where you've removed a 1mm diameter lead and inserted a 0.8mm lead, go ahead and use a solderless breadboard

[KL27x]

I became familiar with the quadpack, first. It makes perfect sense to me. It's symmetrical in all directions, other than the ground and power pins. No matter if you use the left, right, top, or bottom, it goes the same. There's no way to have dyslexia. The pin on the end is the output, the next one is In-, then In+. It's the easiest and most sensible package to me.

I have cut a quad pack in half before to make a smaller dual pack. Scored and snapped off half the opamp IC just to the side of the power and ground pins. There's no pulling down the unused inputs, anymore. Worked fine. I figure the die is a dot in the center of the chip, safe from harm from this procedure, and all I'm snapping are some teeny bond wires.

[Kleinstein]

The breadboard is nice as one does not have to heat up the iron for changes. With decent quality and not trying to force in too large parts like TO220, the reliability is not that bad.  There is some coupling capacitance, but slow parts like the LM324 are OK.
A good source for bread board jumper wires is solid telephone wires (4 x 0.6 mm).

Rats nests (air-wired balls, possibly over a ground plane) is also OK, but depending on the parts used there can also be cold solder joints and taking parts out can be a mess. If the iron is hot changes are fast, the bread board is faster.
For larger circuit I still prefer the bread-board. However I would be careful with circuits that get damages if a connection is lost.

[Zero999]

Yes, the inputs to the unused op-amps shouldn't be just left open circuit. As a general rule of thumb, the best thing to do is configure the unused sections as unity gain buffers and connect the input to a steady voltage, such as 0V.

[Mechatrommer]

While I'm sure it's not an issue on a pcb, it makes for more of an issue on a breadboard.  Especially if you are a neat freak in that you at least like to be able to see (read understand) the circuit layout once it's finished. 

no issue at all for me. if i cant stick my jumper in because there are jumpers on the left and right, i'll just use the hole diagonal to them, so my jumpers on breadboard look like zig zag arrangement. otoh pcb is a nightmare to understand just by looking, esp if traces go under the IC or the bottom layer, worse if more than 2 layers pcb.

[ledtester]

Check the input common mode range to make sure you're supplying a valid voltage when you wire it  as a voltage follower. For the LM324 this includes GND (in a single supply configuration), so this is ok. In general, though, I like to use 1/2 V+ as pictured in Fig. 2 here:

http://www.ti.com/lit/an/sboa204a/sboa204a.pdf

[Zero999]

Check the input common mode range to make sure you're supplying a valid voltage when you wire it  as a voltage follower. For the LM324 this includes GND (in a single supply configuration), so this is ok. In general, though, I like to use 1/2 V+ as pictured in Fig. 2 here:

http://www.ti.com/lit/an/sboa204a/sboa204a.pdf

Yes, that's good advice and the LM324's common mode range does include ground so it's fine in this case. Generally and extra potential divider to get half the supply voltage is unnecessary. If it's a dual supply, then use 0V or if it's a single supply and the op-amp's common mode range doesn't include 0V, then there's generally a ¹/₂V node it can be connected to. Another option is the output of another op-amp.