unity gain lowpass

[snk]

Hello,
I have a circuit where i would like to add a lowpass at the end, but without any volume drop.
It seems that what i am after is a "voltage follower" design, but i am unsure about the technical terms.
From what i have found, the "easiest and best" would be to use an opamp setup as unity gain, am i right ?

The helpful resources I have found so far ar these ones :
http://sound.whsites.net/dwopa2.htm#s7

https://electricdruid.net/designing-the-hard-bargain-distortion-pedal/
(obviously, here, this is a tilt eq, not a lowpass)
the opamp buffer from here (obviously here, this is a buffer, not a LPF) : http://www.muzique.com/lab/buffers.htm


I tried to convert the schematics into a layout, and modify it according to my needs, but while i have a little experience building circuits from layouts, i'm a newbie when it comes to design stuff, even the simplest, so i might have made obvious mistakes 



My question would be the following (keeping in mind that this little circuit should be part of a bigger design, so i have no input/output capacitors and all) :
1- First, am I going into the right direction, or is it wrong from the start ?
2- Then, is the wiring correct ? Am I missing anything ?


3- As i am trying to design a stereo effect, am I right assuming that i can use one single TL072 for both channels of the circuit ?
4- Is the last layout set up in such a way that i would get a unity gain lowpass filter (i set up the cap and pot values so i would get a 3.2Khz cutoff point) ?

Thank you in advance !

[Ben N]

Depending on how exacting your "no volume drop" requirement is, there may be passive solution. Do a search here and on Jack Orman's "AMZ" site for Mark Hammer's SWTC ("Stupidly Wonderful Tone Control") which is designed to maintain even volume across its range. (Which is not to say that you can't also tack on a follower afterward, but that is a solution to a different problem.) I imagine Mark will pop up here before long.
BTW, in the tilt control you linked to, the opamp is acting as a virtual-earth mixer, if I am not mistaken, which would not be appropriate for your purpose. If you want a voltage follower, you should be using an opamp in non-inverting configuration (feeding signal into the "+" input, with a loop from output to the "-" input), like the buffer schematic you posted, although a simple device like a BJT or FET would also work fine. Check the buffers article at AMZ.

[snk]

Hello, Ben.
Thank you very much for your reply.

Depending on how exacting your "no volume drop" requirement is, there may be passive solution. Do a search here and on Jack Orman's "AMZ" site for Mark Hammer's SWTC ("Stupidly Wonderful Tone Control") which is designed to maintain even volume across its range.

I always thought that the SWTC had a noticeable volume loss (around -20dB), due to its passive design. It may have been wrong, which is a good news
As my circuit is a light overdrive aimed at boosting a bit a rather weak signal, i wouldn't want a volume drop coming from the final stage of my circuit (the smooth LPF).
The boost provided by the circuit is quite low (around +3dB, maybe +6dB). The harmonics make the sound fuller and richer, but the volume increase is not huge (and I like it like that).

BTW, in the tilt control you linked to, the opamp is acting as a virtual-earth mixer, if I am not mistaken, which would not be appropriate for your purpose.

Ah, sorry, i might not have been clear with the ressources i provided : I provided them to see where i took some informations from, but i do not plan to follow them "step by step" : I was looking for a way to understand how to wire an opamp in a non-inverting configuration (i build several circuits from veroboard layouts, but i have very little experience in designing my own, and i am very much a newbie when it comes to understanding schematics).

If you want a voltage follower, you should be using an opamp in non-inverting configuration (feeding signal into the "+" input, with a loop from output to the "-" input), like the buffer schematic you posted,

Ok. Is it what i actually did in my layout (that was my initial intention) ?

although a simple device like a BJT or FET would also work fine.

Ok, it's good to know. The initial circuit is a simple one, derived from the Electra distorsion, so maybe that a full transistor design would make sense ?
How would i wire it ? [edit : I'll check the buffers article at AMZ, as suggested

[Ben N]

I have never built it into anything, but my understanding is not that the SWTC doesn't introduce any loss (any passive circuit necessarily includes signal loss) but that the signal loss is even across the range of the control, so that one doesn't perceive any change in volume depending on the setting of the tone control. If you are concerned about overall loss due to a tone control, the simplest solution is to add a make-up gain stage, rather than a unity-gain buffer, after the tone control, which you may take directly from any standard Big Muff or LPB-1 schematic. (If you want to use an op-amp, you can, by adding resistance to the feedback loop of your non-inverting opamp and a resister & cap from the non-inverting input to ground. Google how to calculate values.)

[GGBB]

You have 3 ways to approach this - reduce treble, boost non treble, or a bit of both. You'll have to experiment to figure out what works best for you. You'll also have to figure out what qualifies as treble and non treble.

Perceived volume is both subjective and conditional. If your ear is focussed on mids or lows of a full range signal, you may not perceive treble cut as volume loss. The same treble cut on a guitar signal played at the upper end of the fretboard on the high b and e strings might come across as significant volume loss. Or if the full range signal is in a mix, the treble cut well stand out more since the lower ranges tend to be less obvious because of the band.

So it all depends on what you want. How are you judging overall volume? Only you know that.

A very high treble cut can probably be done passively without any perceived volume loss - that might be all you need.

[snk]

Hi,

You'll also have to figure out what qualifies as treble and non treble.

The effect will be used with a synth, put between the synth and a mixer.
What I want to do is just be able to cut gently the amplified noise and upper harmonics, so a 1st order lowpass set around 2.5-3.2KhZ should be fine. It allows most of the frequencies to pass through while reducing the residual noise a bit (at least, the way i need it to be).
So, in this specific case, i konw quite well what qualifies as treble and not treble 
As you wrote, my ears will be "focussed on [...] a full range signal", and the gentle high roll-off should not be "perceived as volume loss" (and from my tests, it works fine).

How are you judging overall volume? Only you know that.

In order to have the effect i am building (a light overdrive to increase a rather weak signal, and drive the tone a bit) not defeated in its purpose, i need the processed signal to be slightly over unity gain (+3dB boost is ok, +6dB would be icing on the cake). As the overdrive i have tested allows / is set up for a boost around +3dB, if the tonestack reduce the signal by at least the same amount, it would be a fail (I hope that makes sense ?) 

I have made a working (mono, so far) build of the orverdrive, and i like the core tone.
I have also tried to add a basic, passive lowpass, and either it was reducing the volume way too much, or it was interacting with the overdrive too much and affecting the distorsion 
It's been a couple months since i made the tests, and i'm slowly finishing other projects i had on my bench since much too long, but now i would like to learn a bit the various ways to fix that little issue.

[anotherjim]

I think that if you're sure the single pole RC will do the right filtering then you are on the right track. If it needs buffering to avoid loading effects, then yes, a unity voltage follower (buffer) is the right answer.

If it was a guitar amp/pedal input it was feeding with 1M ohm impedance, a  passive filter based on a 10k resistor won't feel any noticeable volume drop in the passband. The magic for this to work is when the following load impedance is about x10 the value of the filter series resistance (plus the preceding source impedance which should be negligible but is safe count as high as 100R usually). For a mixer input that could be 10kohm input, you would need around a 1k filter resistor and a correspondingly larger capacitor. However, that gives you something that, although completely passive, will still lose level if you ever have to drive anything with less than 10k impedance such as some microphone preamps.

I think a decent passive solution is R=1.2k and C=47nF giving 2.8kHz. You have an amplified instrument so a little loss in the passband really is negligible in practice and you can probably fit the filter in a jack plug.

[GGBB]

a 1st order lowpass set around 2.5-3.2KhZ should be fine ...  (and from my tests, it works fine).
So, in this specific case, i konw quite well what qualifies as treble and not treble 

I'm not surprised that you know, but wasn't sure why you seemed to be keeping it secret. Good to see the full picture now.

A unity gain opamp circuit is a buffer. It won't prevent volume loss, but it will prevent the filter from being affected by whatever is on the other side of the buffer and vice versa. If your filter drops overall volume, you will need some make-up gain to correct for it.

If the adjustment of your filter varies the volume loss, your makeup gain needs to also be variable. having a single knob do both is at least complicated if not very difficult to get exactly right. (And beyond my capabilities.)

Your last circuit is (on paper) ~3.2k only when fully clockwise - it is a variable frequency filter from very very high down to ~3.2k. Maybe that's right for you, maybe not - either way it gives you no room to drop to ~2.5k or lower (on paper anyway).

And it probably drops overall volume variably as the treble cut is increased. To avoid that, tie the cap only to the wiper, and tie the opamp input to the unused other end of the pot. This puts a constant resistance against most of the signal, and a variable resistance against the treble.

These are perhaps the simplest variable low pass filters of all, and with them we usually limit the "very very high" frequency to something by adding a fixed resistance before the pot, thereby giving you a fixed range. For example ~2-22k: 4.7k resistor, 50k pot, 1.5n cap.



You could also build a "fixed" frequency adjustable low pass shelf filter - that might work better for you. That looks like guitar tone controls: fixed resistance then cap and variable resistance (pot) to ground. You can add a pot to the fixed resistance part to make that a variable frequency adjustable shelf filter.



(Lots of ideas here.)

The individual pieces and the math are straightforward - figuring out which pieces to use and making them work in your circuit the way you want is the tricky part.

Use buffers before and after to eliminate interaction. Figure out what kind of filter and what frequency (range) works best for you (doesn't alter overall volume?). Then figure out makeup gain and add it to one of the buffers (usually the first one).

Experiment.

[snk]

Thank you very much, both of you : there are plenty informations, and this is exactly what i need (even if it will take some time for me to understand, put them into perspective, experiment  ) !
I will think about it (pros and cons of each solutions, "easyness" to do and/or adapt, understanding of the consequences and the way to implement it, etc). That's really food for thought !

I hadn't realized that a unity gain opamp circuit is a buffer. Now that you told it, it might seem obvious, but this is something which never occured to me, and it's good to know.

Some early comments (but i might take a few days to give it a deeper thinking) :

If the adjustment of your filter varies the volume loss, your makeup gain needs to also be variable. having a single knob do both is at least complicated if not very difficult to get exactly right. (And beyond my capabilities.)

I don't think i will go that route  I think easier ways will be fine, ah ah !

Your last circuit is (on paper) ~3.2k only when fully clockwise - it is a variable frequency filter from very very high down to ~3.2k. Maybe that's right for you, maybe not - either way it gives you no room to drop to ~2.5k or lower (on paper anyway).

Yes, but i think it should be good. From my tests with various parametric eqs, it seems that cutting gently (6dB/oct slope) around 2.5-3.5khZ is fine to get rid of the background hiss without changing the tone too much.
However, i might experiment further with resistors and caps values if i am not satisfied.

And it probably drops overall volume variably as the treble cut is increased. To avoid that, tie the cap only to the wiper, and tie the opamp input to the unused other end of the pot. This puts a constant resistance against most of the signal, and a variable resistance against the treble.

Perfect ! Thank you for the tip !

The shelf filter idea is worth investigating. As you wrote : "Lots of ideas here", I will keep it and think about it.

I think that if you're sure the single pole RC will do the right filtering then you are on the right track.

Thank you, Jim 

I think a decent passive solution is R=1.2k and C=47nF giving 2.8kHz. You have an amplified instrument so a little loss in the passband really is negligible in practice

I will consider this, too (changing the R & C values to suit better the input impedance).


Also, one thing which i had considered is integrating a little Fet preamp (like the Escobedo Duende Jfet) before the tonestack, with a trimmer (so that i can set it to compensate for the volume loss). The idea would be to have it compensate for the volume loss (due to the passive design of the filter), and act as a buffer after the overdrive section (so the filter will not interact with the distorsion).
It is a low parts count design, easy to build, and it would not take too much space on the veroboard.
Do you think it would work (and sound good) ?


Also, was I right assuming that i can use one single TL072 to process both channels of my circuit (and is the wiring on my layout correct) ?

[ElectricDruid]

Yes, you can use the two op-amps in the TL072 to process stereo, and yes, this looks fine to me:

I'd echo what GGBB said though - that filter spends a lot of the knob travel up in the ultrasonic range doing nothing useful, and then only comes down as far as 3.4KHz at maximum. I'd bump the caps up to 2n2, which puts the bit of the range you're interested in more in the middle of the pot.
Personally, I'd also put a 1K resistor or similar in series with each pot because the "minimum" value of a pot is not well-defined, and I like to know what I'm going to get. That gives a range 1.2KHz to 74KHz for the filter.
Finally, it might make sense to use a log pot for this job - or at least, bear it in mind and maybe try it if you don't like the "feel" of the linear control.

[snk]

Thank you, Electrid Druid (and thank you also for your website !)

I'm still figuring out all that has been said above 
Some early replies :

- What i like with the opamp design is that one single chip allows for a stereo use, so it seems to be a very simple solution. So, I'm glad that it can be a worthwhile solution.

- In the meantime, i had thought about using instead a small Jfet booster, and i made a simple layout gathering a lowpass filter (with values suggested above) with the Duende Jfet booster. It is a very simple design, so i think i could try it to test if i like it and if it works the way intended. Is there any pro or con between opamp vs jfet booster ?

- Do I need to put an opamp (or a buffer) before AND after the tonestack ? From what i understand, it seems that it is the way to go if i want to avoid interactions between the overdrive section and what i will be plugging the effect into, but it also seems a bit "overkill" (two buffers for a single tonestack), isn't it ?
If not, I understand that it would be better to put the buffer before the tonestack (= after the orverdrive), right ?

- I will take into consideration what GGBB said about adding resistors to define better the musical range of the pot (I need a little time to think about it).
What the 1K resistor does (i mean : does it limit the overall range, or does it limit the upper or lower frequency range ? How would i do to have the filter work in the 2K-20K range ? Would i need another resistor value, or another resistor somewhere ?)


This is the little layout i did. It is using a Jfet, so i should be able to see if i can trim the boost the way i need to compensate for the filter volume loss, and if the filter works fine with these values. I have added a diode and an electrolytic cap, which wouldn't be required if i integrate this design into my circuit, but as it, it should work as an "autonomous" design for testing purposes...

[ElectricDruid]

- Do I need to put an opamp (or a buffer) before AND after the tonestack ? From what i understand, it seems that it is the way to go if i want to avoid interactions between the overdrive section and what i will be plugging the effect into, but it also seems a bit "overkill" (two buffers for a single tonestack), isn't it ?
If not, I understand that it would be better to put the buffer before the tonestack (= after the overdrive), right ?

It depends what you're trying to do. If the filter/tone control is going to be a stand-alone effect, a buffer on the front would be a good idea because you never know what is going to be plugged into it. On the other hand, if you're adding it to an existing overdrive stage, you know *exactly* what it's going to be connected to, and you can make sure there's no need for the buffer. But you still don't know what's going to be connected to the *output*, so the buffer *after* the filter is a good idea.

- I will take into consideration what GGBB said about adding resistors to define better the musical range of the pot (I need a little time to think about it).
What the 1K resistor does (i mean : does it limit the overall range, or does it limit the upper or lower frequency range ? How would i do to have the filter work in the 2K-20K range ? Would i need another resistor value, or another resistor somewhere ?)

The extra 1K limits the upper frequency, since even with the pot at minimum the value cannot go below that value.
If you wanted to limit the upper frequency to 20KHz with a 2n2 cap, you'd need a 3K6 resistor:

https://electricdruid.net/rc-filter-calc/?f=20KHz&r=&r_series=3&r_errors=1&c=2n2&c_series=1&c_error=10

That'd give you 50+3.6K = 53.6K for the maximum value, which gives around 1.4KHz (pretty close to 2K):

https://electricdruid.net/rc-filter-calc/?f=&r=53.6k&r_series=3&r_errors=1&c=2n2&c_series=1&c_error=10

[snk]

It depends what you're trying to do. If the filter/tone control is going to be a stand-alone effect, a buffer on the front would be a good idea because you never know what is going to be plugged into it. On the other hand, if you're adding it to an existing overdrive stage, you know *exactly* what it's going to be connected to, and you can make sure there's no need for the buffer. But you still don't know what's going to be connected to the *output*, so the buffer *after* the filter is a good idea.

Ok.
So it should go after the filter. But when i tried, a couple months ago, i remember that tweaking the filter affected the distorsion, and this is why i was suggested to put a buffer in between...
Is there any easy way to avoid any interaction between both parts of the circuit, or is the only safe way is to have two buffers ?
(the core part of the circuit is based on the EQD Speaker Cranker -which is based on the Electra Distorsion-, with some tweaks, and in stereo) ?

The extra 1K limits the upper frequency, since even with the pot at minimum the value cannot go below that value.
If you wanted to limit the upper frequency to 20KHz with a 2n2 cap, you'd need a 3K6 resistor:

Your RC filter calculator is excellent !
Thank you also for the technical insight, i learn a lot.

[ElectricDruid]

Ok.
So it should go after the filter. But when i tried, a couple months ago, i remember that tweaking the filter affected the distorsion, and this is why i was suggested to put a buffer in between...
Is there any easy way to avoid any interaction between both parts of the circuit, or is the only safe way is to have two buffers ?
(the core part of the circuit is based on the EQD Speaker Cranker -which is based on the Electra Distorsion-, with some tweaks, and in stereo) ?

Ok, it's possible that the filter will interact with a simple gain stage like the Electra/speaker cranker. The output has diodes to ground for the clipping, and anything that loads those much will affect the tone. That's ok if you're feeding a high impedance, but the tone control filter isn't a high impedance. In fact, with it dialled right up to maximum, we'd have something that looks a lot like a 3K6 resistor to ground (for AC signals, the capacitor acts like a wire). So you might need a buffer before and after, or you might be able to tweak the gain stage a little once the filter is attached to compensate for the filter loading.

Simple circuits are great, but there's also a reason people build more complicated things - a lot of the simple ones aren't that "well behaved" and interact with other stuff around them. If your gain stage had been a tubescreamer-based thing, you'd probably have got away with this, since the op-amp's output would easily drive the tone filter, but a single transistor with clipping diodes hanging off it is a different story.