stopping hi gain ocillation on breadboards?

[deadastronaut]

during my ' hi gain' testing on breadboard i get the usual oscillation/squeel like a pig stuff..

its just pi==ing me off while i'm testing stuff out...if i put my hand over it it goes quiet...but then i can't obviously whack my guitar....is there a practical and easy-ish way of stopping this?

tin can over the board?...but then i don't have access to it...

could i build it into some kind of 'open' box?....(a kind of hi gain specific breadboard)

or should i just accept a load of squeeling piggies as part of the process...

any ideas?......i have a 3pdt toggle on it thank @#$%..... thanks in advance rob.....squeeeeeeelllll  squeeeeeeeeeeeeeeeeeeeelllll....

[R.G.]

Mother Nature is telling you about gain-phase ( sometimes called "Nyquist") oscillation and parasitic capacitance.

Oscillation cannot happen without three things: (1) feedback of output to input, (2) a phase change which makes some frequency be in phase at the input after being fed back, and (3) a gain at least sufficient to make up in the forward gain path for any losses in the feedback path. These three things make up the Nyquist Criterion, more commonly stated as "it will not oscillate if the gain is less than one at all frequencies where the phase change through the amp and feedback path is less than that needed for positive feedback."

You are setting yourself up for this by having very high gain and a wiring layout that has many stray paths from the output to the input. Since capacitive paths exist for infinite distances, there is *always* a stray capacitance path from output to input of some magnitude. There is also always some frequency with a phase shift making positive feedback. All you have to have is enough gain at a frequency where the feedback goes positive and you get an oscillator.

With enough gain *everything* oscillates.

To stop it, you can do one of three things. Aw, you guessed!
1. Eliminate feedback. Well, OK, as I said this is impossible. So it can't be done in this universe.
2. Prevent the phase from going to 360 or a multiple thereof. Oh, wait. I said that can't be done. Ooops.
3. Cut the forward gain or raise the attenuation in the feedback path. Hah! Now we're getting somewhere.

Breadboards with flying wires going everywhere  have lots of capacitance between parts and wires. Guitar effects have high input impedances, which mean that they don't attenuate stray capacitive feedback much; add to that high gain and you're set up for oscillation. What you can do is the classic layout techniques for (especially) RF:
1. Keep outputs away from inputs, especially high impedance inputs. 
2. Route wires so that they don't run parallel for wires from different parts of the circuit, minimizing capacitance between different parts of the circuit.
3. Where signals simply must go out and return from the circuit, twist the send/return lines, ideally with a ground wire too, to force the capacitance between the send and receive, and capacitance to ground to be maximum. This keeps the energy from being so widely radiated.
4. Where you can't do the above, use shielded wires. This is the fundamental reason coax exists. It's self shielding. It was only after coax was invented that it's transmission line properties were discovered.
5. Bypass the power lines and use star grounding in hard cases. Wires conducting ground and power currents also carry signals and can be a feedback path.
6. Don't use breadboard for tough cases. The little clips inside the breadboard have capacitances to each other. Or ground every other one, cutting the capacitive transmission a lot.
7. Use shield plates and so on.

That is - the three most important things are layout, layout, and layout.

[therecordingart]

I was thinking put a metal mixing bowl over the breadboard, but I'd probably listen to RG first.

[deadastronaut]

cheers RG.  a great explanation.....though a lot goes way over my head....but thanks.

i think i'll put a metal bowl over my head and just breadboard a volume pot..

but seriously, i guess thats why there aren't too many ultra metal/thrashy diy projects  because of these design/testing problems and the ability to handle squeeeeellllllll.........hmmmm.....shame.

i love the sound ive got, its just a piggy...maybe i'll get my wife to hold her hand over it all day...(no jokes)

i'll persevere with it...ive had 4 days of squeeeelllll...but i'm nearly there anyway, hopefully it'll be ok on pcb and shielded/boxed etc......thanks.

@recordingart, i'll try a metal bowl anyway for a laugh..

[iccaros]

I use terminal strips instead of bread board for a lot of these things.. Less issues, harder to use a little.. but not too much


Bad picture of a design I was pulling apart

[R.G.]

but seriously, i guess thats why there aren't too many ultra metal/thrashy diy projects  because of these design/testing problems and the ability to handle

It's much worse for RF. RF experimenters can't even think of using breadboards. They do something like dots or go direct to surface mount.

See http://www.geofex.com/Article_Folders/protostyles/proto_styles.htm for some prototyping suggestions.

High input impedance, high gain and long wires are a recipe for disaster.

[deadastronaut]

@steve: arghhh i'd get lost with that...its hard enough for me on a breadboard..i can see why you do it though...cool.

you gave me this idea though..use terminal block for pots...i use those pots with pcb pins/lugs..they'll pop straight in....consider that nicked if they fit ok.....they always pop out of my bboard.....cheers!.

@RG. i remember reading that a while back, great article..i like the dead bug...i also remember doing a cardboard  booster circuit when i first started......it worked great..
then i bought a breadboard....then i made it true bypass, then i added pots...etc etc....great fun!..

i'll persevere today, let squeeeeeeeling commence..hopefully when i lay it out for pcb it 'll be ok...

cheers guys!.rob.

[DougH]

In addition to what RG said, keep your jumper wires as short as possible.

[deadastronaut]

just an update on this squeel like a pig lark...

1: i attached a ground to a tin box lid with a croc clip lead and rested it on top...sorted

2: i put a simple buffer in front, and squeel went away...

or...

3: just make a clean boost and forget the idea.

[Transmogrifox]

That was 2011.  Did you ever finish prototyping and build it?  It sounds like a high gain beast with potential to be a challenging layout on the real thing too.

RG's post above should be in a sticky somewhere.

On these high gain circuits I often find tight decoupling caps on each power supply pin set are helpful.  The ground and power come from the same strip pair -- it's really tempting (esp w/ IC's) to take ground from 1 side and power from the other making something like a foot-long return path) I make a little power supply zone on my BB for each gain stage so there aren't feedback paths formed through power supply rails.  This is mostly to prevent RF instability, that in turn becomes audible -- especially because RF oscillations can more easily cross-couple into other circuits creating a feedback loop where normally one would not expect inter-coupling capacitance or mutual inductance to be sufficient at audio frequencies. The RF unstable amplifier gets modulated at audio frequency, transmits across the board to an input, gets demodulated by nonlinear elements and in turn feeds into the RF oscillating amp, modulating it...  I have had an op amp unstable at 55 MHz before that operates like an op amp at audio frequencies because the average behavior during its time between the rails filters down to relatively clean sounding audio with a bit of hiss.  If that average behavior is modulated onto a 55 MHz carrier, then it only needs to be demodulated and fed back.  I just mentioned that because sometimes there will be things happening that don't make any sense if you're only considering the only instability that is audible must happen in the audio band.

For dual-package IC's, don't pair circuit gain stages that are separated by filters (like EQ or tone).  For example if the other half your output buffer op amp is used for a gain of 1000 ahead of an active EQ which has some gain of its own the cross-coupling between packages may be insufficient to prevent unintended feedback.  This is true for final layout too, but it's worsened by having long wires in close proximity further degrading the amount of rejection between signal paths.

Keep wires short and keep high gain nodes connected with tight loops between related components.  For example, you wouldn't want to have the base of a BJT gain stage on one side of the board and then take a long wire from the collector to the resistor 2 cells over across the board.

If your breadboard is like mine, it's mounted on a metal backplane.  Keep the backplane grounded.  If it doesn't have a backplane then the grounded tin box lid thing is a cheap way to make one.

Another thing I do is I use my breadboard power supply rails as a storage facility for all my large-valued capacitors.  I usually have all strips lined with caps keeping ground and +V decoupled every 2 inches or so with banks of caps between 10 uF and 1000 uF depending on the combination on-hand of 330 uF, 47 uF, 2.2 uf, ... etc.  Just a hodge-podge.  Then I usually keep some short-trimmed 0.1 uF scattered about to reject high frequency stuff.

[LightSoundGeometry]

first day in EE class today and we all got a Bread board kit..surprisingly most didnt  know what the heck was what.

So I made a rangemaster out of ll the parts we were given  and impressed everyone  ..all the kids and professor were like ..oooh whats that ?   


I was able to fix a few pedals I did that squealed using RG's advice. a few I was not able to get fixed all the way as they would squeal only if the effect was on, guitar unplugged and not chained to any other pedals...makes me wonder if the buffered pedals do something.

I have learned something new, maybe not related but the 1 meg resistor after the input cap helps quiet the pedal down to. I think , not sure!

[SISKO]

Great advices from RG and Transmogrifox!

[LightSoundGeometry]

Mother Nature is telling you about gain-phase ( sometimes called "Nyquist") oscillation and parasitic capacitance.

Oscillation cannot happen without three things: (1) feedback of output to input, (2) a phase change which makes some frequency be in phase at the input after being fed back, and (3) a gain at least sufficient to make up in the forward gain path for any losses in the feedback path. These three things make up the Nyquist Criterion, more commonly stated as "it will not oscillate if the gain is less than one at all frequencies where the phase change through the amp and feedback path is less than that needed for positive feedback."

You are setting yourself up for this by having very high gain and a wiring layout that has many stray paths from the output to the input. Since capacitive paths exist for infinite distances, there is *always* a stray capacitance path from output to input of some magnitude. There is also always some frequency with a phase shift making positive feedback. All you have to have is enough gain at a frequency where the feedback goes positive and you get an oscillator.

With enough gain *everything* oscillates.

To stop it, you can do one of three things. Aw, you guessed!
1. Eliminate feedback. Well, OK, as I said this is impossible. So it can't be done in this universe.
2. Prevent the phase from going to 360 or a multiple thereof. Oh, wait. I said that can't be done. Ooops.
3. Cut the forward gain or raise the attenuation in the feedback path. Hah! Now we're getting somewhere.

Breadboards with flying wires going everywhere  have lots of capacitance between parts and wires. Guitar effects have high input impedances, which mean that they don't attenuate stray capacitive feedback much; add to that high gain and you're set up for oscillation. What you can do is the classic layout techniques for (especially) RF:
1. Keep outputs away from inputs, especially high impedance inputs. 
2. Route wires so that they don't run parallel for wires from different parts of the circuit, minimizing capacitance between different parts of the circuit.
3. Where signals simply must go out and return from the circuit, twist the send/return lines, ideally with a ground wire too, to force the capacitance between the send and receive, and capacitance to ground to be maximum. This keeps the energy from being so widely radiated.
4. Where you can't do the above, use shielded wires. This is the fundamental reason coax exists. It's self shielding. It was only after coax was invented that it's transmission line properties were discovered.
5. Bypass the power lines and use star grounding in hard cases. Wires conducting ground and power currents also carry signals and can be a feedback path.
6. Don't use breadboard for tough cases. The little clips inside the breadboard have capacitances to each other. Or ground every other one, cutting the capacitive transmission a lot.
7. Use shield plates and so on.

That is - the three most important things are layout, layout, and layout.

RG, is this why, in the chorus pedal I just built, it had one ground isolated from everything else and going directly from dc input to circuit ?

[smallbearelec]

attached a ground to a tin box lid with a croc clip lead and rested it on top...sorted

Rob--

At the time the original thread was started, I had not yet put my furry brain to this problem. And I was thinking of it only as a way to enable stage use of a prototype. That said, An IdeaBox

http://diy.smallbearelec.com/HowTos/IdeaBoxAndBoard/IdeaBox.htm

costs more than a tin box lid, but it is waaay more quiet in operation than an open breadboard. You can probably fashion it from locally available bits.

Regards
SD

[Transmogrifox]

... That said, An IdeaBox

http://diy.smallbearelec.com/HowTos/IdeaBoxAndBoard/IdeaBox.htm

I think you had this before Zvex publicized the Inventobox   Too bad you didn't run to the patent office.  He beat you to that one:
http://www.google.com.au/patents/USD630616

Really nice howto either way.  It really puts the idea of the inventobox within reach of the average DIYer.

[smallbearelec]

I think you had this before Zvex publicized the Inventobox   Too bad you didn't run to the patent office.  Really nice howto either way.  It really puts the idea of the inventobox within reach of the average DIYer.

Thanks for the props! Bearish Ruminations:

I think Zach had the Invento first, and it did not take off because too expensive. Dano had the same problem with the BeavisBoard. Both very clever ideas that depended, IMO, on too many made-to-order pieces--the DIY market is very price- sensitive, as I know better than most! I don't know if Carlos Snyder:

http://killertone.com/

is still doing the Tone Crafter, but I believe he sold a bunch of these. While it's not cheap, he limited the number of custom-made pieces and so left room to offer a whole package including parts for multiple effects.

Where am I going with this? Without false modesty, I note that I had a good deal of prior art to suggest possibilities. Also, New Sensor had released that very reasonably-priced 1790NS enclosure; that was most important to me, because it offers a lot more real estate than you get in a BB. And I did not have to pay for the tooling! Often, other people's new products are enablers for the rest of us.

So I'm a happy camper: Based on sales of the brackets, there are more than a few of my ins-bear-ations in use out there. Even in the absence of prior art, I don't see it as the kind of breakthrough that would justify the expense of getting a patent. Also, as R. G. has noted, the patent only gets you the right to sue for infringement. I'm comfortable with profiting on the parts and being able to use the idea as a come-on-to-visit my site.

Regards
SD