Tone circuit placement

[Hotcreme]

Hey friends,

I've been Googlin' and a-forum-searchin' for info on where within a circuit (before/after buffer, boost, output trim) and how the impedance seen by the tone circuit affects the values and relationships of resistors and capacitors. 

At this point I've read enough to know how the shelving filters work, but often the diagrams are taken out of circuit, or placed so as to make distinguishing where the rest of the pedal circuit ends and where the tone circuit begins, or the parts are given predetermined values without the theory behind the relationship between components (so it's difficult to modify to your desire).

Besides AMZ's fabulous Big Muff & Presence tone circuit, is there a crash course you can recommend for organizing the sequential signal path and theory of different "stages" in a circuit? I just want to know how it needs (or best) fits together. Still learning this weird shit.

Thanks for your help, loving this forum.

[Hotcreme]

extra context: specifically where one would insert an active bandpass filter and the electrical values it requires to "see" to operate properly.

e.g. second-order active high pass filter (opamp unity gain buffer) tone circuit feeding or following a booster/buffer (for clean boosting to unity gain)? Or incorporating the filter within the booster/buffer itself? Just want to build a unity gain high pass filter but not experienced enough just yet.

xoxo

[GibsonGM]

Have you checked out "duncan't tone stack calculator"?  It is free, and allows you to model what will happen with varying degrees of source impedance, and whatnot...

http://www.duncanamps.com/tsc/


A lot of what you're "worrying about" is settled by experimenting and seeing what the various combinations of things will do.   Say, a booster/O.D. that you want to add a stack to....what do you WANT from it?

We know any circuit needs an input stage, to amplify our signal above the noise floor...cool.  We do that, by whatever means.  Now - do we want to get some distortion?  OK - add another stage, or some other thing that creates non-linear output (diodes...etc).     

Maybe we want to put our tone control after this stage...we could put in a buffer and do so.  Or, try it without the buffer!  The results will be a little different, of course...you might lose highs with no buffer....BUT - let's circle back again for a sec...>>>

When you added the 2nd stage to your input stage, you had a chance to modify some things.   Just like an input cap can trim bass for you, you can use coupling networks between stages to modify your tone!   You can use fixed or variable controls when coupling, just like tone controls!    Cut bass, boost mids, mid cut...etc.     

When you generate distortion, you also generate harmonics of the original signal...some of which can be irritating, lol.   So you may cut them off just after this stage.   OR, you might wish to preferentially boost treble in a stage, so you cut bass just prior to hitting it hard!   Maybe you have some impedance mismatch at that tone stack, and it sucks some of the highs out...but you may WANT that loading, if you've generated harsh overtones!

I'm sure you can see how most of this falls into individual preference, and also how there is no 'right answer', altho there ARE a few technical concepts in play.   Experience dictates where and when you will cut, boost, or do nothing....and trial & error, looking at other designs...   

Short answer:  Generally, any audio circuit would like to be fed with a matching impedance.  Output impedance of driving stage should be near the input impedance to a filter or what have you.  Buffers are great at making up for the almost-inevitable mis-match, ha ha.   Sometimes loading a signal down is not bad, though!  So, you have to use some theory mixed with common sense/preference...

welcome to the forum, HC  Awesome question.

**Edit** to hit your ?  more specifically...an opamp-based active filter will likely be designed to have such a high input impedance so as to make most anything feeding it "transparent"...I mean, if you have an input Z of 1Meg, it really won't make much difference what's going on leading up to it...AUDIBLY, it will - depending on the above stuff - different results from different placement, much based on preference...certainly you need an input stage before the filter, in most (but not all) cases.

Look on the net for a stand-alone HPF booster - 2nd order is 'a lot'...most boosters can easily be tailored for MORE HPF action (they tend to be 1st order, anyway...).  As you say, it's incorporated into most boosters via input cap and input Z...

[garcho]

Generally, any audio circuit would like to be fed with a matching impedance.  Output impedance of driving stage should be near the input impedance to a filter or what have you.

that's not true, in fact it's the opposite. generally, load should have high impedance, source should have low impedance. the matching impedance days came from power lines and matching source to load for power transfer, not audio signal voltage, yes?

Buffers are great at making up for the almost-inevitable mis-match

isn't that because of their high impedance and low output impedance?

maybe i've misunderstood you

[GibsonGM]

I know it's quite conventional and 'the way it's done', to desire a low output impedance into a high input impedance...just, the way I've always understood it is that the "best" situation is an equal impedance - a match.  That's not 'the opposite' of high>low...that would be low feeding high, which is generally a good recipe for crap sound, LOL.    Unless of course you want to load down a stage to remove high content, which IS done at times.   

But, my background, how I learned electronics, is by studying RADIO stuff.  The old ARRL books were around for years, well before any sort of 'how to' for our stuff, so I turned to them.  This is before the internet 

To be honest, I can't tell ya why 'equal' would be better than 'high to low', other than maybe having to do with some sort of fidelity concern? More Hi-Fi stuff that doesn't relate to our hobby, which is really quite low-fi?   Someone more technically knowledgeable (PRR? R.G.?) could explain this better.  You could easily be on to something, Garcho, where optimum power transfer is achieved this way, and it may NOT relate to audio.

Yup, for everyday work, I do automatically set up situations where you are going low out to high in, using buffers and so on, you are totally correct that is why they are used, and I'm in no way trying to suggest otherwise - altho a buffer will simply allow the next stage to 'take what it wants' within its ability to supply current...the buffer matches the impedance desired by the next stage...if the output of stage1 did that, wouldn't the buffer be a moot piece of circuitry? (If the feeding stage output Z = next stage input Z)   It is easier to use a buffer than create this situation tho, LOL.     

Anyway, you certainly CAN feed tone stacks etc. with 'less than ideal' (somewhat high) output impedances...there are many 12AX7 gain stages, typically higher in output Z than desirable, that feed tone stacks just fine.  That's all I was pointing out.   Valve Wizard (merlinb) talked at some length about just this in his book, and gave some circuit examples optimized for this purpose.   If you know what is happening, you can correct for it.

[anotherjim]

My impression from all the circuits I've seen is a "rule of thumb" is in use - a ratio of 1:10. So if source is 10k, then a tone stacks R's will be sized to give loads about 100k with caps chosen to suit those magnitudes, then the following stage with 1M input. Definitely matching for most voltage transfer.

Like Gibson I remember teaching that called for impedance matching for maximum power transfer. It depends on the subject bias I suppose. Mine was Industrial Electronics. Equally, I don't remember using decibels for amplitude was ever taught, it was always for power at 10log a/b, never 20log. So I was totally confused when audio engineers blithely say -6dB for 1/2 instead of -3dB!

[slashandburn]

Equally, I don't remember using decibels for amplitude was ever taught, it was always for power at 10log a/b, never 20log. So I was totally confused when audio engineers blithely say -6dB for 1/2 instead of -3dB!

Ah, wonderful. Now I'm confused. That all sounds familiar. All I can remember is something about it still being the same, there is no double-standard etc etc.

Edit: I think I encountered that going the other direction. I thought my college lecturers were just making things up sometimes. "Don't worry about that, 6, 3, same thing, its all good"

[GibsonGM]

Don't forget your "line level" and all that, ha ha! 

Nice rule of thumb, Jim...the 10x source output Z...sounds great, helpful.  I could stand to remember that myself      

[ashcat_lt]

All we actually care about is voltage.  Source and Load impedance make a Voltage Divider.  If they match, you get half the voltage at the receiving end (half "wasted"), 6db down, uncool.  10:1 makes it less than 1db, which is considered acceptable.

In a passive high-pass filter, the load is parallel to the R that helps set the cutoff frequency.  If they match, effective R is half, cutoff frequency is doubled.  10:1 makes the load contribution within resistor tolerance, so it can be ignored.  There's also the broadband divider of Source over R parallel to Load.  We'd like Source to be 1:10 to R, and R 1:10 to Load.

In a passive low-pass, the source is in series with the R that helps set the cutoff.  If they match, effective R is double, cutoff is halved.  1:10 again makes it negligible.  There's also the Source + R over Load divider...

[PRR]

> always understood it is that the "best" situation is an equal impedance - a match.

That has fooled many smart people.

"Match" puts as much waste power in the source as in the load.

"Match" is sometimes used to tame wild impedances in long cables. In audio, that would be miles long.

The special problem with "match" in user-twiddled "tone" controls is that the tone control impedance varies all over the place.

Recall that all caps are "open" at zero freq and "short" at infinite freq. (Coils, though rare in audio, go the other way.) Clearly a tone control with just R and C *MUST* have a declining impedance as frequency rises. If the tone-control is going to do much, efficiently, it "has" to have significant change of impedance across the audio band.

Start Duncan TSC, switch to "James". That control will be 1.11Meg below 100Hz, 110K midband, 90K at the top of the band. Over 10:1 change of impedance. Where do we "match"?

Figure out what the tonestack designer had in mind. Baxandall wrote-up his thoughts and you can find them online. James' work is harder to find and less clear, but a little poking at it shows the source and load impedances need to be significantly less/more than the network resistors. The -20dB midband loss is set by the 100K:10K ratio. If the source is not 1/10th of 110K, error will be over 1dB. James (using TSC default values) wants <10K source and >10Meg load. (Still some residual error due to 10:1 not being 1/10 but 1/11.)

"Match" is a useful tool when you can NOT get gain. Passive telephones. "Match" became obsolete when the cost of a tube came down to less than a day's wages.

[PRR]

> totally confused when audio engineers blithely say -6dB for 1/2 instead of -3dB!
> something about it still being the same, there is no double-standard etc etc.

Audio guys find a volt-meter more convenient than a power meter.

"audio engineers blithely say -6dB for 1/2 VOLTAGE (not Power).

This tends to work when impedances are well controlled OR when impedance is unimportant (non-Match systems).

You don't have to break the connection to measure Voltage.

Because audio systems are generally NOT "matched", we don't really know or care what Power is in there.

Radio engineers can't blithely assume impedances (SWR and all that). They can not easily afford extra Power Gain to run non-Match interfaces (much). They break the connection and measure the actual Power in there.

Audio engineers use "-3dB" as a benchmark for "significant fall-off". Whether they know this is "half power" or not may be moot. -3dB comes right out of reactance theory (X=R).

[jatalahd]

In amplifiers one typically has a gain stage before tone control stage. Often some gain is lost in tone control stage, so a makeup gain stage is put after the tone control (or just directly the power amplifier stage).

When thinking (audio)circuits as a collection of blocks with different functions, then it makes more sense to have a general rule of "a block should have high input impedance and low output impedance". If you design a transistor amplifier stage as a pre-amp block, the output impedance will indirectly affect the gain of the amplifier stage.  Any circuit block connected after the pre-amp block will have its input impedance connected parallel to the output impedance of the pre-amp. Parallel connection of low-Z and high-Z equals very close to low-Z, so the gain of the pre-amp stays as designed even when additional blocks are connected after it.

A good point from ashcat_lt was to note that "Source and Load impedance make a Voltage Divider". With low-Z-out and high-Z-in the signal attenuation is very small.

Then a bit off topic: PRR mentioned about the standing wave ratio (SWR). I understand that in audio circuits the actual standing wave phenomena does not occur due to long wavelengths, but how about the general theory of signal reflection in the boundary of two different impedances. Just wondering does this occur at all in audio circuits, and if it does, can it be seen in measurements somehow(?) Not trying to steal the topic of the original question...

[GibsonGM]

Jarmo, you mean buffer before the tone control?  Cathode follower? 

[amptramp]

Since this is about tone circuit placement, one topic that should come up is linear and non-linear circuits.  An amplifier (ideally) is linear.  A tone control is linear.  What do we mean by linear?

1. If you have an output of X from a device corresponding to input A and you multiply the input A by B, the output is also multiplied  to become B*X.
2. If you have an output X from a device corresponding to input A and you have an output Y from a device corresponding to input B, the response to an input A+B will be X+Y.

Linear circuits do not introduce new harmonics.  Non-linear circuits do.  An example of a non-linear circuit would be a fuzz/distortion whether implemented as a clipper or a mis-biased amplifier like a Fuzz Face.  In these cases, you get harmonics of the original signal.  With some frequency doublers, you can get rid of the fundamental entirely.  There are octave down circuits that produce sub-harmonics of the input frequency.

It doesn't matter if you put a tone control ahead or after a linear stage (other than for noise and level considerations).  Because of the new frequencies generated by a non-linear stage, putting a tone control before or after makes a big difference.  Many fuzz/distortion pedals use a tone control after the stage to reduce the added high frequencies.  Since these frequencies do not exist at the input, it is more effective on the output.  Octave down circuits are also affected by the added frequencies in the bass direction.

This may not be related to the question you were asking, but it should be kept in mind.

[PRR]

 > standing wave ratio (SWR). I understand that in audio circuits the actual standing wave phenomena does not occur due to long wavelengths, but how about the general theory of signal reflection in the boundary of two different impedances. Just wondering does this occur at all in audio circuits, and if it does, can it be seen in measurements somehow(?)

It does happen.

The "echo" happens MUCH too quick to hear. (For "reasonable" line length.)

One Foot (303mm) is about One Nano-Second. Different in free space or in a wire, but that was a guide back in 1960 (when it "proved" that computers could only go so fast because they were many feet across) and still true-enough now.

A 1,000 foot (305 meter) wire has a "length" of 1,000 nS or 1 micro Second.

Frequency of 1 uS is 1MHz. We don't hear sounds that short. The ear has nearly no response to sounds of 50uS.

So you can see it (on a 'scope), but you won't hear it. The echo of each audio transient is shorter than the transient.

(And guys who think 1MHz is "low" have to work different.)

It *can* matter on LONG lines. The early telephone line across the US had echo. Not only was it over 3,000 miles, the cable used to get tolerable loss had extra low wave-speed. Some thumb-counting says over 15mS. Of course the one-way echo could not be detected (compared to what?). But the "matching" at the far end (and joint-points between) can never be perfect so the speech from NYC would "bounce" at SF and come back to NYC. The talker hears a 30mS echo. That's far enough into the Haas area (give or take the actual line speed and length) to be upsetting. The whole long-line telephony technology gets very complicated.

That's *after* you optimize the whole Telegrapher's Equation. R L C and R limit line speed different ways. The first trans-Atlantic telegraphs could hardly pass 100 words an HOUR. Press the key in NYC, the voltage in Ireland rose VERY slowly. If sent too fast, the far end just averaged out to steady zero. We still face this problem (simplified) when we run a passive guitar over a too-long cord.

While matched 600:600 lines "work", for some decades the convention on long broadcast lines (before everything went internet) has been (was) 60 ohm source and 100K load. Cable (not open-wire) pairs really run 90-130 ohm ultimate impedance. 60 Ohms will knock-down much of the echo on the first bounce, to about nothing after a few bounces, long before the echo tail is long enough to be an audible "echo". The other advantage is that you can inter-patch true-600 with 60/100K connections with 1dB error, which is not-bad un-trimmed (less than the uncertain line-loss) and well within the studio's gain-trim range.

[jatalahd]

Thank you Paul, you know everything

Jarmo, you mean buffer before the tone control?  Cathode follower? 

Basically yes, any kind of active follower can help to prevent the preceding circuit from interacting with the following stage.

If you simulate a stand-alone tone control circuit with SPICE, you will most likely feed it with an ideal voltage source (zero internal impedance) and terminate it with infinite impedance (nothing connected after the tone control). To make the simulated tone control behave the same way in any circuit, you need to create similar environment within the circuit. If placing the tone control stage between two op-amp voltage buffers (unity gain followers, for example), you'll have a signal source with small internal impedance and the termination is utilizing the large input impedance of the op-amp. When looking at the small-signal model of a BJT, FET or a tube, the controlled source often acts as the "isolator" component, in op-amps the isolation becomes from the large input impedance.

Trying an example:
You have analyzed the output signal from the guitar, measuring it by connecting the guitar cable to a spectrum analyzer. You see that strongest harmonics are accumulating in the low end, but you would like to get more prominent higher harmonics. You plan to design a high-pass filter pedal with simple 1st order RC filter (input directly in series to C, output across R to ground). Simulate it as a textbook example, notice that cut-off is at 1/(2*pi*R*C), slope 6 dB/octave. Results look good, easy to build, ready to rock.

Pluck the guitar in to the high-pass filter pedal and measure the spectrum again from the pedal output, expect more highs, get disappointed. Why? The guitar pickup is a signal source where the internal impedance is significantly high and is already a 2nd order filter on its own. The guitar tone control and the cable each add one more order. The high-pass filter pedal is then directly connected to this to make a 5th order filter, all components interacting with each other. It will change the harmonic content of course, but most likely not the way as the simulations were showing.

When adding a simple op-amp voltage buffer in the pedal before the simple RC filter, the guitar signal remains the same at the input of the pedal due to the isolation and the pedal works more like simulated. However, when connecting from the pedal to the amplifier, the RC filter R is parallel with the input-Z of the amplifier, so this still alters the behaviour of the high-pass filter...

Anyway, I guess my point was that interactions between circuit stages can be reduced by using active components (as buffers or to add gain) in between. Furthermore if a (passive, pot-controlled) tone control stage would be the first (or last) stage of a circuit, tuning it from the pot would change the input (or output) impedance of the whole circuit and the connection interface to other devices would not be well-defined.

[GibsonGM]

Good example Jarmo!

I understand this, I was thinking more about the idea of a gain stage prior to a passive tone network (tone stack)...many tube gain stages have pretty high output impedances, 40K and the like.   So I wanted to get the buffer idea out there for the OP.

Altho a buffer IS technically a gain stage, offering less than unity voltage gain but POWER gain!