Circuit Design

[upspoon12]

Hey guys,

some more noobish questions for you. I don't have a GREAT understanding of electrical engineering etc. I know the rudimentary stuff, etc.

There are some things that still when i look at them in a schematic or layout etc that make me go HRM.

One thing i (after having searched many times to no avail) am having trouble with is little 'subgroups' of components that acheive a certain goal.

ie: a voltage divider. i know what this looks like and that its done for voltage referencing/biasing etc. this i can understand.

but other things.

for example you have a boost pedal. - you have input filters. you have voltage dividing resistors. - can these be anywhere in the circuit or is the fact that they are by the input what makes them the input filter and voltage divider? can you put them at the end?

I understand driving the FET/Opamp

What i'm trying to figure out is this: is there a set group of components that put together in a certain way that would form say volume control, tone control. How do you know where in the circuit to place these things? after the boost stage obviously makes sense.

I've tried to find resources on this stuff specifically as it applies to fx pedals but its difficult to come across. I've found a few which were helpful.

I'm trying to look at it like math.

theres an equation that makes up one pedal or another (in terms of parts not numbers and letterS) you can have different equations to do the same task, you can also change the equation to do the same task a different way. theres an equation of components for boost, distortion, flange, chorus etc. but are there specific combinations that perform specific modifications to the signal (obviously there is) but i'm trying to determine what it is.

at this point i'm just blathering my mind off because i get it i understand the fundamentals behind what is going on in the pedal (to a slightly above basic degree) but i don't understand how the components in a certain fashion perform certain tasks.

any light that could be shed would be appreciated

[GibsonGM]

Hey Justin - good questions, and it's nice to see your interest!  

In a way, the DIY electronic thing is about read, read, read, read, ask, read, build, read....you're noticing stuff that we all did....that there really isn't any 'volume of magical secrets', but a LOT of info in small bits.   And that a few very basic things are used, over and over, to do the same thing but in ways that LOOK different!  

The faster you look up "R-C cutoff frequency", the better off you'll be!       Then RC LPF (low pass filter), and RC HPF (high pass filter).
AC is a bit different than DC, and can be frustrating at first...

Ok, a couple of your questions:

Yes, an input filter "can go anywhere" - but would not be an "input" filter, of course.   You can shape frequencies anywhere at all, and many circuits do.   BUT - there are some 'old school rules' that you see out there.    We tend to cut bass at the input, which gives us a cleaner signal with less likelihood of (badly) overdriving the following circuitry with 'farty' effects that we DON'T want.    We tend to bring it back at the output (big output cap and such...).
Voltage division - well, that is done where it's needed!  Maybe so as to not overdrive the next gain stage - there are reasons these are where they are (that being one of them).   

A tone stack/control tends to come after a driving stage, and after clipping (which allows more harmonic content), and before a recovery stage that replaces the INSERTION LOSS of the tone control/filter.  This is because you want a low output impedance to feed the tone control, and a high input impedance to follow it.   This is called IMPEDANCE MATCHING (something to look up) - makes for the best transfer of energy between the elements, with predictable results.  < this goes with RC LPF/HPF.... All rules are made to be broken, tho, so don't be surprised to see these 'violated', too!

Much of this IS the way you are picturing: think of "little black boxes".   A transistor gain stage...opamp gain stage....opamp buffer....clipping section (gain stage plus diodes, usually)...buffer>tone control>recovery stage.....each of these things can be read about/studied/built and experimented with!    

[upspoon12]

Thanks a lot for thr response Mike.  A lot of what I K of and what you clarified for me makes much sense.  I suppose one other thing that you mentioned is where I need more knowledge.  Thr different 'sections'  and what is typically done in each section.  For example you mentioned cutting thr bass and adding it back in.  This makes complete sense and I have an recording engineer background so I understand the concepts.  But theres a lot more reading and trying I need to do so I can learn the conventional what goes where in what section for a given application.  It's certainly ly great to Know there's guys like you on this board who are looking to help how they can and enlighten.  It's definitely items appreciated.

[GibsonGM]

Hey, no problem man! Most everyone on here likes trying to help others, and thinking about all the stuff reinforces it for us too!  I was totally lost when I started trying to make DIY pedals, and know what it's like.

Check out the site Geofex linked at top of this page...R.G. on here has great articles there that will tell you much more about how this stuff works!  And if you ask questions, he's often available to help.

runoffgroove.com  has projects with descriptions - I always found the description really helped to make things make sense.

AMZ at top, GREAT stuff there!
Among tons of other places.

Everyone LOVES this one when new, too!  The concepts at work with the Big Muff are universal (that tone/recovery stuff I mentioned), so study this one, too!    http://www.kitrae.net/music/big_muff_guts.html

[PRR]

> look at it like math.

It is all calculable.

But in many practical audio systems, the tractable linear equations blow-up into non-linear equations.

Say you want a gain variable 0.1 to 10. While it is possible to design an amp to do this, in practice it is often simpler to apply a fixed active gain of 10 and a passive variable divider (pot) 1:1 to 100:1.

Which comes first? The gain or the loss? In the ivory tower, they both come out the same.

But in some Real World.....

If we apply a 1V signal to a gain of 10 we get 10V out. But if the amp only goes to 9V, the signal must get distorted (non-linear), and that is usually bad. So when we might have a strong signal, we usually put the pot first and then the gain.

OTOH, if the signal is very weak, potting it down makes it so weak it may get lost in universal hiss.

Also, as Mike is saying: Say you have 0.1V of guitar and 1V of the local radio station at 1MHz. You can't hear 1MHz, so what is the problem? Problem is that 1V of AM radio into "any" diode will "detect" (non-linear) the audio. Amplifiers are full of hidden diodes. Now you have news or sports-talk or oldies along with your guitar. If you just want to stop the 1MHz from getting to your main amplifier, the high-cut could go "anywhere". But since you want to cut the 1MHz *before* any diode in your box, you generally put it as one of the first things in the path.

Your basic linear audio tools are-

* loss
* gain
* high cut
* low cut

_IF_ you can keep the signal away from overload or hiss (non-linearity), you can put these in any order or combination which is convenient.

However all passive sources work-out to be not a huge bunch above Universal Hiss. And in trying to bring them to useful levels, we are also going toward distortion. So while many arrangements are possible, in practice a few will stand-out as "better" in some way or other.

[upspoon12]

This is all making more and more sense. i think what it boils down to is just getting myself into predicaments where i learn first hand when something needs to be moved to acheive a certain goal. Thanks for the insight guys i think i've just soaked it all in like a sponge and its put a few things into perspective for me. Just need to keep going and learn first hand the little nuances of "this should go here but in this instance its better if its here" etc.

That link is awesome!  A Lot of good knowledge and something that you can reference the information to and see it broken down. I wish there was more of that.
between that link and some electronic engineering stuff i printed and the rc cutoff and impedence matching i've got my work cut out for me this weekend. haha.

Thanks again guys. This learning experience is turning out to be a pleasant one with your assistance!

[GibsonGM]

Awesome!  Do you have any simulation software?  I use LT Spice - it's free, powerful and available if you search for it.

Then you could set up, for ex., a "filter"  (tone control....) and see that math in action as you alter component values.  I find it invaluable for seeing if an idea is even going to work, ha ha...

[upspoon12]

Not yet! I've heard quite a bit about spice actually and was considering going to download it. I will have to do this when i get home from work. I suppose i could even set it up at work. I can see a simulation program definitely being invaluable. I could also see that being an awesome learning tool as well. I have breadboards and prototyping boards and stuff at home and wanted to start trying some stuff but with no clue where to begin to get something doing, i was hesitant. At least with simulation software you don't need to worry about blowing yourself up when you FIRST try something out., aha

[teemuk]

You are more than halfway there because you understand the fundamentals. After that it's just looking at how the signal gets processed stage by stage.

If it's signal processing the effects of each little subcircuit basically cascade sequentially. What usually is important is the order of that sequence because changing the order can have drastic effects to  overall outcome. For example, when it comes to circuits that distort the signal there are usually quite different effects on applying frequency response -shaping signal processing before or after the distortion. Similarly there are huge effects on whether you introduce gain before the distorting circuit block or after it. And similarly there are huge effects in what order you apply gain stages or attenuating stages because effects of each cascade and get multiplied.

At some points you may need gain, at some other points you may need attenuation, sometimes at full effective bandwidth, sometimes in more "frequency selective" form. But even then it's just either amplifying, attenuating or filtering. So, you care about the order of such tasks because they have a specific function for overall circuit performance.

Other subcircuits usually do tasks like provide external inputs such as reference voltages, power supply voltages, or perhaps inputs to drive elements like variable gain controls, etc. These may have a similar sequential circuit path. Take for example rectified and regulated power supply: You can't really build that in a random order: There's little point to filter or regulate before the AC is rectified into DC, then there's little point to regulate until that "pulsating" DC is rectified to solid DC level. At that point you can finally regulate.

The circuitry in signal path usually follows similar logical sequence; It must perform this and that before there's sense to perform something else. Like that interactivity of distortion and frequency response, or interactivity of clipping distortion and gain/attenuation, or interactivity of gain vs. noise levels, or interactivity of frequency response vs. circuit stability or interference. Everything always tends to have a purpose.

Electronics follow strict laws of electricity and what's nice about that is that everything is both calculatable and most importantly perfectly logical. Everything is in perfectly logical order and performing perfectly logical tasks concerning the overall circuit function.

Input filters, output filters and such... well.. they are just generic terms that imply where such filters are located. Both in practice are merely filters. Kinda like saying one stage is an input stage while another is an output stage. Tone control, volume control, gain control...? Not really rocket science: Does the control affect the frequency response; if yes, it's a tone control. Does it affect signal magnitude or perhaps gain ratio of a gain stage, then it's a volume or gain control. Sometimes the line gets blurred and the control actually performs both tasks. It's really just semantics. If you understand what the "circuit block" does then you don't really care what the proper name for it is, and if you know what it does, then you probably have a good clue what the name could be.

[Thecomedian]

> look at it like math.

It is all calculable.

But in many practical audio systems, the tractable linear equations blow-up into non-linear equations.

Say you want a gain variable 0.1 to 10. While it is possible to design an amp to do this, in practice it is often simpler to apply a fixed active gain of 10 and a passive variable divider (pot) 1:1 to 100:1.

Which comes first? The gain or the loss? In the ivory tower, they both come out the same.

But in some Real World.....

If we apply a 1V signal to a gain of 10 we get 10V out. But if the amp only goes to 9V, the signal must get distorted (non-linear), and that is usually bad. So when we might have a strong signal, we usually put the pot first and then the gain.

OTOH, if the signal is very weak, potting it down makes it so weak it may get lost in universal hiss.

Also, as Mike is saying: Say you have 0.1V of guitar and 1V of the local radio station at 1MHz. You can't hear 1MHz, so what is the problem? Problem is that 1V of AM radio into "any" diode will "detect" (non-linear) the audio. Amplifiers are full of hidden diodes. Now you have news or sports-talk or oldies along with your guitar. If you just want to stop the 1MHz from getting to your main amplifier, the high-cut could go "anywhere". But since you want to cut the 1MHz *before* any diode in your box, you generally put it as one of the first things in the path.

Your basic linear audio tools are-

* loss
* gain
* high cut
* low cut

_IF_ you can keep the signal away from overload or hiss (non-linearity), you can put these in any order or combination which is convenient.

However all passive sources work-out to be not a huge bunch above Universal Hiss. And in trying to bring them to useful levels, we are also going toward distortion. So while many arrangements are possible, in practice a few will stand-out as "better" in some way or other.

I actually have a question regarding circuit noise. I have a book "Amature tests and measurements" by Lous Dezettel, 1969 print, page 103 states that input impedance with regard to an antenna and a receiver can be seen as signal generator and load respectively, and that when impedance is equal, maximum power transfer occurs with the least amount of circuit noise developed.

Does this follow as a general rule for all circuits, in which it may actually not be useful or desirable to buffer everything? If an impedance "mismatch" causes excess circuit noise, then are buffers a source of excess hiss in an end product?

[merlinb]

when impedance is equal, maximum power transfer occurs with the least amount of circuit noise developed.

Does this follow as a general rule for all circuits,

No, because we need voltage transfer, not power transfer. Low-noise design in radio circuits follows slightly different rules from audio circuits.  For minimum noise in an audio circuit, you always minimise series (i.e. source) resistance, and maximise shunt (i.e. load) resistance. The ultimate expression of this is a straight bit of wire! Unfortunately, bits of wire don't have gain.

[GibsonGM]

^^  Right. Which is why, rather than have the "make impedances match" thing going, we tend to look at our stuff from the perspective of "make your output impedance low, and input impedance high".    A little different than the HAM radio thing, which I too started out with.   Most things we use take little to drive (high input Z), but in the wrong circumstances (low input Z) will really degrade our signals if we don't pay attention to this 'rule'....AC transfer is like a different set of 'rules' on TOP of the other 'rules'....sort of. 

We're far less concerned with power output than HAMs; that is until you start designing output stages for amps

[Thecomedian]

Gotcha. Thanks for the clarification. 

[PRR]

> Does this follow as a general rule for all circuits

What Merlin and Mike said.

Moreover: when we do match for low hiss we must MATCH, not just pad. Throwing a 50 ohm resistor across a hi-Z antenna input will degrade signal and thus S/N.

The concept was fuzzy in tube radios. The grid input is hi-Z, but the tuned circuit losses would be significant, and "matching" to these might be wise.

The distinction became clearer in common-base BJT radio inputs. The emitter input might really be near 50 Ohms, and -all- power dissipated in that resistance not only modulated the device, it then added to the output power. Then we'd work to match to the transistor and not-match to the tuned-circuit losses.

(Nevertheless, you can often beat this with a common-emitter connection.)

Audio is different in part because efficient impedance transformation is very problematic. Audio transformers suck.

On paper you get the best S/N from a transformer+tube preamp if you can match the mike impedance to the >100Meg of the grid.

However over the whole audio range the tube is more like 50pFd so falling from 100Meg at 30Hz to 200K at 15KHz. Even if you had an ideal transformer, where would you match?

But real transformers have parasitic inductance (and capacitance). The order of magnitude is such that you really can't get effective impedance much higher than 100K at 5KHz or 15K beyond 20KHz. Once the stray inductance and the stray capacitances come together, you get a peak and then a 12dB/Oct roll-off. So from 150 Ohm mikes you rarely take more than 1:7 or 1:10 ratio for studio, 1:14 for PA. (There are exceptions; their cost/rarity proves the rule.)

Every device has an hiss-optimum noise impedance. With BJTs this varies widely with current. Taking pretty good current in input devices, the OSI may be 2K. Coming from 200 Ohm mikes, we'd use a 1:3 transformer. The actual input impedance may be 40K (4K to the mike), a hi-Z loading.

Guitar+cord impedance is all over the place. Experience shows that 34K in series and 500K in shunt does "little" harm. Going further may improve S/N, but we are well into Diminishing Returns. And many-K in series is handy in many ways (radio reduction, accident protection). The shunt may be limited by your pick of device, or by resistors in your drawers (hence 1Meg everywhere; that used to be the highest super-available value). However I've never heard that Ampegs were quieter than Fenders (Ampeg often used ~~3Meg). And 99.9% of guitars have a Volume pot under 1Meg, so little point in going to extreme.