Aside from the usual methods -
1. Power filtering with capacitor to ground from 9V supply,100R in series with 9v supply
2. 220p or other small cap to ground at FX output to remove high frequency buzz
3. Ground connection to chassis
4. Shielded input wire
etc etc
are there any more 'pro tips' or lesser known methods for reducing noise in stompbox builds? I've been wanting to improve the quality of my DIY builds,especially in lowering noise levels in high gain designs. Are there any particular circuit design thumb-rules I could follow to ensure quieter operation and lower noise?
Quote from: jishnudg on February 23, 2013, 02:00:03 PM
Aside from the usual methods -
1. Power filtering with capacitor to ground from 9V supply,100R in series with 9v supply
2. 220p or other small cap to ground at FX output to remove high frequency buzz
3. Ground connection to chassis
4. Shielded input wire
etc etc
are there any more 'pro tips' or lesser known methods for reducing noise in stompbox builds? I've been wanting to improve the quality of my DIY builds,especially in lowering noise levels in high gain designs. Are there any particular circuit design thumb-rules I could follow to ensure quieter operation and lower noise?
Good post, good question!! I also am on a quest to fight the hum! :icon_twisted:
I'm fairly certain that if a few intelligent circuit design practices are followed, hum, buss,noise can all be drastically reduce...I hope this thread becomes a dictionary of all available methods,tips and techniques.
Okay, to set the ball rolling...here's something I've been doing lately,and I think it has helped a little. I'm plugging in a 100R resistor in series with and a 220p capacitor to ground before and after every gain stage in the pedal in question...it filters the sound reasonably well,although I'm usually left with a little more bass in the output signal than I'd want...particularly if there are multiple gain stages. Can this method be improved upon?
there is not always a one size fits all solution. lots of funny noises can come from a guitar pedal but they are not created equal. use the right tool for the job. target the problem, understand the problem, find a solution to the problem. thats how I roll.
(http://fuzzcentral.ssguitar.com/schematics/screamingbirdschem.gif)
I built one of these and it had an unwanted high frequency tone that was barely audible. so what did I do? I added a 560pf from collector to base. that is called negative feedback. it was a good guess and it worked. I did not need to desolder anything. sometimes you need to scope it at different places to find the problem. it helps to understand all the ways oscillation can happen. is your effect input out of phase with the output? is it high gain? what is the resonant frequency of the circuit? what gain stage is creating the most noise or oscillation? where are the 90 degree phase shifts in the circuit? how clean is ground at any given place in the circuit? what outside forces are acting on the circuit? electric fields are invisible but they do exist. sometimes the electric field is created by the same pedal that is amplifieing it!
1) If there is any appreciable gain in the circuit, use some shielding between the input jack an the circuit board, unless it is a very very short distance.
2) Use op-amps that deliver lower noise in the circumstances used (bifets are quieter when used with big input and feedback resistances).
3) Never use any bandwidth you don't need, at either end of the spectrum.
Mark, could you elaborate on the third point "never use any bandwidth you don't need" - I'm sorry I'm still relatively new to this and don't fully understand how this relates to effect noise and hum. Could you explain, perhaps with an example circuit?
There's a problem with asking for hints and tips on highly technical subjects. Even the best hints and tips are highly, highly distilled nuggets from a much larger foundation of knowledge.
The real answer to your question is to go read and understand something like the book I always refer to when I have noise questions: "Noise Reduction Techniques in Electronic Systems" by Henry W. Ott. The pearls fall out of having that whole oyster.
Noise is not one thing. Defining noise is much like defining "weed". A weed is any plant you don't happen to want growing where it's growing. Similarly, noise is any signal that you don't want that is interfering with your enjoyment of the signal you DO want, even if the signal you're wanting at the moment is 0.00000000000V.
Any lucid understanding of noise starts with understanding the different kinds of noise, from flicker noise, recombination noise, thermal noise, and on to unintended-signal-pickup noise. They all come from different places, and have different cures, although some techniques overlap. Perhaps the overlap is what makes this confusing.
Mark's third point comes out of understanding the wide-spectrum nature of different noise types. One way to have a quieter system overall is to make the system deaf to signals you don't want. If your signal simply can't get over perhaps 7kHz, having an amplifier that responds all the way to FM radio and UHF radio bands is just asking for something to come in and make things noisy. Likewise, thermal noise above your signal will be more audibly obvious because there is no wanted-signal there to mask the noise.
The short course on noise is:
(1) Lay out the circuit with the absolute minimum conductor lengths and with inputs separated from outputs by as much distance as possible given that you're minimizing lengths. This seems like a simple, obvious statement, but entire textbooks can and have been written about how to do this well, and the theoretical and practical consequences of not doing it well.
(2) Use a Faraday shield, this being an overall conductive enclosure around the whole mess. This minimizes AC power line hum and helps with RF pickup.
(3) Power supplies should be ideal voltage sources. Since those don't exist in the real world, lay out power and ground feeds to minimize resistance and inductance between power sources and the using devices. Since this is impossible to do really well, use local power filtering in the form of resistors and inductors for series isolation and capacitors for both noise impedance mismatch and as local buckets of charge for signal peaks.
(4) Know what currents from what signals being handled flows on every ground wire. Otherwise, you circuit can generate its own noise, both as interference and full-blown oscillation.
(5) Every active device generates noise. It may be more current-noise, more voltage-noise, or simply thermal noise from the random motion of charge carriers, and the generation/recombination of charge carriers. It may be because of the particle nature of charge carriers, like electrons hitting a tube plate like hailstones on a tin roof. Know the device you're using and adapt the biasing, impedances and amplifying techniques to minimize it per device.
(6) Inductors and capacitors do not generate noise, except in the parasitic resistances in side them.
(7) Shielded wire is a "Faraday Cage" for a wire. It helps exclude the jungle of exterior signals from a wire. If your layout is good enough, you don't need it. However, getting a layout "good enough" may be impossible in the real world even if you are experienced.
(8 ) Electricity flows in circuits - that is, around loops. You can force currents to NOT flow by breaking loops. Metal enclosures, for instance, should be "grounded" at one and only one point. Sometimes using one low frequency ground and on high frequency ground can help, if you have special circumstances.
and
(9) Amplifying signals you don't want is always a waste, so if you amplify signals outside the frequencies you want, you'll hear them as noise.
(10) Magnetic fields are very hard to shield away. There are no magnetic "short circuits" or "insulators". The only good way to avoid magnetically coupled noise is to not pick it up by being far away, or making your circuit small (small pickup loops), or orthogonal to the field, which is possible only in special cases.
It goes on, but those are the big hitters. As you may have guessed, there is a lot buried under each of those.
Quote from: jishnudg on February 23, 2013, 05:20:52 PM
Mark, could you elaborate on the third point "never use any bandwidth you don't need" - I'm sorry I'm still relatively new to this and don't fully understand how this relates to effect noise and hum. Could you explain, perhaps with an example circuit?
Probably 65% or more of what people here seem intent on making or modding are distortion circuits. These generally apply a fair amount of gain to the signal, such that any hiss on the input becomes magnified considerably. One can stick a small value cap to ground on the input to bleed off hiss to ground, but there might still be some left, or some generated internally, and once it gets amplified a few hundred times, it's not the sort of thing you want. You CAN address it with a treble control on the amp, but then you lose tonal flexibility. Best to nip it in the bud within the pedal. So, that can take the form of things like caps in the feedback loop of op-amps that trim off treble you won't be making use of.
Basically, the strategy is that if there is more bandwidth in the signal at one or more points than you are actually going to make use of, lose it.
It's not just for high gain devices, but for others as well. If you plan on using lots of regeneration with a flanger of phaser, it can produce lots of noise. Stick a small cap in the feedback loop of the last phase shift stage to keep it under control, or bleed some treble off to ground in the feedback loop of the flanger.
Analog delays tend to accumulate grit and noise over multiple repeats. Whip up a simple lowpass filter for the feedback path and trim a little more treble off it with each repeat. The treble is not likely to sound all that great on the 3rd repeat anyway, so its not bandwidth you need. Dump it.
Ah. Thanks Mark. I forgot sampling noise and beyond that quantization noise in digitally sampled systems.
if the type of effect allows (ie not reverb, delay or anything else that trails off), gating the output based on the input, or when the output goes below a certain threshold. via vactrol, fet or whatever.
Quote from: Mark Hammer on February 23, 2013, 06:05:19 PM
Quote from: jishnudg on February 23, 2013, 05:20:52 PM
Mark, could you elaborate on the third point "never use any bandwidth you don't need" - I'm sorry I'm still relatively new to this and don't fully understand how this relates to effect noise and hum. Could you explain, perhaps with an example circuit?
Probably 65% or more of what people here seem intent on making or modding are distortion circuits. These generally apply a fair amount of gain to the signal, such that any hiss on the input becomes magnified considerably. One can stick a small value cap to ground on the input to bleed off hiss to ground, but there might still be some left, or some generated internally, and once it gets amplified a few hundred times, it's not the sort of thing you want. You CAN address it with a treble control on the amp, but then you lose tonal flexibility. Best to nip it in the bud within the pedal. So, that can take the form of things like caps in the feedback loop of op-amps that trim off treble you won't be making use of.
Basically, the strategy is that if there is more bandwidth in the signal at one or more points than you are actually going to make use of, lose it.
It's not just for high gain devices, but for others as well. If you plan on using lots of regeneration with a flanger of phaser, it can produce lots of noise. Stick a small cap in the feedback loop of the last phase shift stage to keep it under control, or bleed some treble off to ground in the feedback loop of the flanger.
Analog delays tend to accumulate grit and noise over multiple repeats. Whip up a simple lowpass filter for the feedback path and trim a little more treble off it with each repeat. The treble is not likely to sound all that great on the 3rd repeat anyway, so its not bandwidth you need. Dump it.
so by bandwidth you mean hertz ranges?
Quote from: Thecomedian on February 24, 2013, 03:54:16 AMso by bandwidth you mean hertz ranges?
yes. you dont need anything below 100hz or above 10khz. I sometimes change those numbers to be a little more wideband if I make something for bass or if i make a treble booster. I'm sure 10khz is too high for most people. I see someone mentioned 7khz already so they will probably say something.
While mixers and many similar devices can certainly handle content above 7khz, a great many guitar amp speakers roll off pretty seriously above that. Heck, a Celestion Vintage 30 rolls off big time above 5khz. They can still reproduce higher frequency content, but it is waaaaayyyyyyyy down in amplitude, not contributing enough to the tone that you'd miss it. Of course, that means such a speaker will also downplay any noise in that range, but why the heck amplify stuff in the 6khz range x300 in a distortion if its only going to make the noise more noticeable?
Quote from: EATyourGuitar on February 24, 2013, 10:23:46 AM
Quote from: Thecomedian on February 24, 2013, 03:54:16 AMso by bandwidth you mean hertz ranges?
yes. you dont need anything below 100hz or above 10khz. I sometimes change those numbers to be a little more wideband if I make something for bass or if i make a treble booster. I'm sure 10khz is too high for most people. I see someone mentioned 7khz already so they will probably say something.
but..
http://www.psbspeakers.com/articles/The-Frequencies-of-Music
http://www.seventhstring.com/resources/notefrequencies.html
you still want below 100hz with low D,E or a bass. If you cut out those frequencies, you won't have them when playing, no matter how much you fiddle bass boost/cut on a guitar or bass.
I *think* he might have been talking about a distortion box for guitar.
That said, filtering before and after clipping is a whole other art - point is don't use what you don't need. Psycho-acoustics are also into play - there are preciously few bass cabinets that produce a low-B fundamental (31Hz), yet they still sound good. Remember a plucked string isn't just a sine, it carries a *lot* of overtones, which together make the sound you hear as one note.
BTW: yes, a guitar E is 82Hz, using full bass *into* a clipper will result in a horrible farty sound. Boosting 100Hz with medium -wide Q *after* clipping tends to move toward that Nu-metal chunk.
Hi, I'm looking for some help, already read many forums. Built an orange squezer from tonepad.com.
When pedal is turned off, sound is clear, no grounding problems, but when I turn it on, there is a loud, high frequency, continuous sound (like scream). However the pedal is actually working cause I hear the guitar sound, and its compressed.
So does anyone know what can cause this crazy sound.
I have the same high pitched squeal/osc with my rat clone. awesome thread
It's happend the same thing to me, for the first time, I think.
I never had too much problem of noise or hum, even with high gain pedal: BSIAB, Rat, Triple Wreck, Tight Metal... But yesterday I got this high pitch squeal with a clean booster, and just with the PSU. It was perfect with the battery.
A 100uF filter cap in the power section replaced the 47uF of the original project, but this just lowered the pitch, not the volume of the squeal. A 47R resistor in line to the 9v solved the issue.
hi all,
love this topic, made my day, big time :)
they key help was Elijah-Baley's post, I just did what he suggested, and .. ta-taam.. silence... :)
__________________________________
my case:
I had just built Jack Orman's booster (AMZ Mosfet Booster), link here: http://www.muzique.com/schem/mosfet.htm (http://www.muzique.com/schem/mosfet.htm), but y'all know it, better than me :)
I had built it, and loved it...
BUT then I tried it with my brand new XX-cheap wall wart...
and there was this whining... at the pitch of B on the high E string, B4, so to speak...
and it was really frustrating...
cause, you know...
the first g searches returned results suggesting a small cap to the ground at the input... like 47p....
which Jack Orman's circuit has...
which made the situation appear as if prevention had been already made...
the first thing I did was put a 47R to the +9v, it lowered the whining... at least it seemed...
but it hadn't gone...
at the same time, I read the same about a 47uF electrolytic to ground: lowered it, but it wouldn't go away...
so, it all looked hopeless...
this was the point at which Elijah-Baley's post came as big help...
what I did then:
the same...
100 ohms, ( 100R, just for the searches' sake) and then a 100uF electrolytic...
I didn't even cut the legs of either of them to normal size, cause I really thought of it as a "give this one a goo, too" experiment..
and the whining was gone... totally... at full boost, too...
so, from now, I'll put this small set of whine-elimination before any circuit :)
thanks a lot to you all :) & special thanks to you, Elijah-Baley :)
Many simple pedals, especially those with only a few discrete stages, take very small current. This means you can actually raise the power filter input resistor up to 1k. This allows a smaller/cheaper filter capacitor and 100uF would be as big as you'll ever need. Extra filtering from RF noise can be had by adding a ceramic capacitor 10nF to 100nF. I habitually add a 100nF disc cap on every power input.
When you have an indicator LED it doesn't need any power filtering and its current draw will only cause a voltage drop across the filter resistance. Take the LED power connections from before the filter resistance and after any polarity protection diode because LED's don't like excessive reverse voltage.
An extra point is that the filter resistance doesn't need to all be in the positive feed. You can have the resistor in the negative feed just as well - or split it with some resistance in the positive and the negative inputs. With daisy-chained DC power cables, this is a powerful way to stop ground loop interference due to the common power and signal grounds.
- Star grounding on seperate components.
- Ground fills on the second layer of the PCB. With cuts in the fill to "split" noisy parts of the circuit (like clock signals, power, digital, etc) with clean. The "cuts" can be linked through a few traces.
Quote from: j-pee on September 16, 2019, 03:39:01 PM
[...]
they key help was Elijah-Baley's post, I just did what he suggested, and .. ta-taam.. silence... :)
__________________________________
my case:
I had just built Jack Orman's booster (AMZ Mosfet Booster), link here: http://www.muzique.com/schem/mosfet.htm (http://www.muzique.com/schem/mosfet.htm), but y'all know it, better than me :)
I had built it, and loved it...
BUT then I tried it with my brand new XX-cheap wall wart...
and there was this whining... at the pitch of B on the high E string, B4, so to speak...
and it was really frustrating...
cause, you know...
[...]
what I did then:
the same...
100 ohms, ( 100R, just for the searches' sake) and then a 100uF electrolytic...
I didn't even cut the legs of either of them to normal size, cause I really thought of it as a "give this one a goo, too" experiment..
and the whining was gone... totally... at full boost, too...
so, from now, I'll put this small set of whine-elimination before any circuit :)
thanks a lot to you all :) & special thanks to you, Elijah-Baley :)
I'm glad you solved your problem! ;)
Ironically, I built a SHO circuit board (with a volume pot) and I got a whine when I plug the PSU. It doesn't matter if I put in line to the 9v a 47R or a 100R or a diode, or both. I added a 100nF film across the ground, but the whine is stille there. The last thing I should do is to replace the 47uF with a 100uF, then I guess have to change vero layout. Because the weird thing is that I built even a Box of Rock, that is three SHO toghether, plus another one. It works fine, even the last sho section stand alone works fine.
more crap:
- if you can, avoid attenuating then re-amplifying your signal like the plague. like the plague took ebola out to dinner, got married and had kids.
- avoid using stupid-high-value or carbon-comp resistors for your input bias.
- 100R/100u power filter gives you a low-pass at 15Hz. if you don't like the voltage drop from such a high resistor, and have the space, 47R and 220uF will also give you 15Hz.
- avoid active filters with extremely-narrow Q-factors.
- op-amps have better power supply rejection than discrete transistors.
- if you can afford to kill anything above 5KHz or so, do it. bypass gain stages with small caps.
- For your high-impedance input stage, TLC227* is a fairly-cheap CMOS-input op-amp with 8nV/Hz noise rating. Bear in mind its maximum recommended voltage of +/-7.5V (ie. no 18V supplies without a voltage drop). For all low-impedance parts of the circuit, you can get away with a cheaper BJT op-amp like an NE5532 or something, which have similar noise ratings.
- Don't use the CA3080 for OTA projects - use the LM13700.
sorry if any of that has already been mentioned. here's a related question i've often pondered about:
does op-amp noise rating have any bearing on noise injected into a buffered Vref?
Thanks MrStab!
About my SHO, I tried yesterday to replace the 47uF with a 100uF, and it tamed the whine, so I can finish my first 1590A pedal.
I didn't include any resistors in line. I'm wondering if I should use a 47R or 100R anyway.
what kinda PSU are you using, Elijah? i'm unsure if that's the main issue, as you said your BSIAB works fine, though increasing the cap did tame the whine so... maybe.
you're not using this layout by any chance, are you? the input and output are right next to each other. as others mentioned, that's a no-no, and i would honestly expect that to oscillate.
https://hotbottles.files.wordpress.com/2012/03/zvex-sho-updated-from-zach-schematic-762360.png (https://hotbottles.files.wordpress.com/2012/03/zvex-sho-updated-from-zach-schematic-762360.png)
(It was a Box of Rock, not a BSIAB)
My PSU is an old Boss. I think it's ok, I don't have problem, usually. I built more than 40 pedals, I got this problem with some of the last circuits, but I always solved, escept in one case.
I used this layout: http://tagboardeffects.blogspot.com/2013/02/zvex-super-hard-on-compact-layout.html.
Now, as I mentioned, I have to solve this problem with a circuit out of box of a Woolly Mammoth, that was ok initailly, later it starts to squeal, then it stopped, than it start to squeal again. No matter if I use resistors or diode in line ot the 9v, or bigger cap across the ground. I have to work on it again.
I have a problem quite often with ground loops from the safety ground of the power supply that feeds the effects and the safety ground of the amplifier. Quite often there are capacitors to this ground from both the power supply and the amplifier and the ground is taken to different points on the safety ground bus, meaning any current from the capacitors will flow through the impedance of the ground and modulate the other ground. The idea here may be to remove the capacitance to the safety ground or to provide an isolation transformer for the effects so a safety ground id not needed. Sometimes batteries are the best answer.
Once the input signal goes through the 100nF input cap on that layout, Elijah, it's on the trace right next to the output. i wouldn't feel comfortable with that.
i meant your post that said "I never had too much problem of noise or hum, even with high gain pedal: BSIAB...". I just thought i'd refer to it as BSIAB is a notorious troublemaker!
sorry, i don't understand quite what you mean in your last post - i'm a bit of an idiot. do you mean a Wooly Mammoth clone is having squealing problems?
Don't worry MrStab. :)
About the layout I used for the SHO, I got what you mean, but I solved the problem, fortunately.
Perhaps I found always goo dlayout, but casually the last I used it was a bit tricky, it wasn't the schematic, the components valus, but right the layout. I don't know.
My last unsolved squealing problem is the Woolly Mammoth, this layout: https://www.sabrotone.com/?attachment_id=1026
I worked on it without problem changing some parts with sockets, and one day it starts to squeal, then it stopped, and then never stopped to squeal. In a few of days. I tried bigger cap, I guess 330uF, in the power root. As well diodes and resistors.
I don't touch it for about two months.
I should to try again. But that was really tricky. I thought to build it again, or at least try it on the breadboard, at least.
The problem with transistorized designs is the designer has to accommodate the wide variation in hfe of transistors and if germanium is involved, the variation in leakage currents as well. You can build a design and test it and have it seemingly be stable and doing what you want but a change in temperature may change the gain sufficiently to make the circuit oscillate. Add in biasing variations due to part tolerances and you have something that often works but you have to do a lot of calculations to prove it's always going to work.
This is one reason I prefer op amp designs to transistorized designs - if one works, a complete production run will work and all units will behave exactly the same.