Noisy problem

[nichapat]

Hello guys , I have build Tonemender preamp and Tube simulate follow these schematics. But there is too noisy.Even through  I have try run pream and simulation separately,there is noisy. But when connect  preamp to  simulator it seem double noisy.
How to improve?

[techno-rachit]

Hi, ensure that the power supply you are using is filtered well, one reason for noise may be the power supply... Wat power supply are you using?

[phatt]

Hi nichapat,
Oh dear you poor soul,, Ouch yes that is a potential noise maker.

Hi Z circuits with high gain really need extra isolation or share the gain between the two stages.

This might help you to get your head around the noise problem.
http://www.ssguitar.com/index.php?topic=2013.0
first Schematic on that page is my Tone circuit which will be much quieter than this one.

Even then,, the tube sound emulator might also be way hot and that can lead to even more problems.
Phil.

[J M Fahey]

Besides power supply hum, you probably have shielding and grounding problems.
Please post a couple pictures of what you built, how you connect your guitar to it and where you finally send that signal to.
I think you just have the boards "on the table", with no shielding/screening at all.
In that way, they are open to all outside interference and noise.
Good luck.
PS: start thinking about making a simple metallic chassis (an L-bent piece of aluminum) or a metal box or at least, a plastic box covered on its inside with glued aluminum foil (kitchen/cooking type).

[Kaz Kylheku]

This is regarding the Tonemender.

When we examine the schematic closely, it is a silly circuit in a number of ways.

The overall topology is that the signal goes through an op-amp buffer (an amplifier with unity gain), then through some passive tone controls, and through another op-amp amplifier.

This topology means that the tone control is not suitable for instrument level signals, but only for line level signals. (Just like, for instance, an equalizer component from your home stereo: plug your axe into that, and you will get noise, by design.)

Surely enough, the buffer creates a high input impedance, isolating the input from the tone controls, ensuring that the tone controls won't load down your guitar pickup. That's very good. But because the buffer has unity gain, it means that if the input is some weak signal like 200 mV, the signal going through the tone control is at the same level. This means that your tone control has to be completely shielded, just like the conductor in your coax cable and your guitar's cavities.

What you want to do is reverse these two elements. The first thing the device should do is boost the signal through an amplifying op-amp to get it from instrument level to line level as early in the chain as possible. This is where you can have your variable gain also. The buffer can be placed after the tone control, to isolate the tone control's impedance from the next device in the chain.

Secondly, a number of things stand out. Firstly, most of the pots used are 500K. Also, on the gain control, there is a suspicious 47 pF capacitor. Basically, this device looks like it was designed by a guitar player. (``500K knobs are nice and bright in my guitar, so they must be great everywhere.  A tone bleed on my strat's volume knob ensures bright highs when I turn down the volume, so I can just transplant into this op-amp circuit and it will do exactly the same thing!'')

I wouldn't touch the pot values in the tone control because they will change the sound. But the 47 pF cap in the gain circuit will cause some loss of high frequencies when you crank up the gain! The capacitive reactance of 47 pF with respect to a 10 Khz signal is about 339K. That is significant in regard to the 500K impedance of the pot. The cap provides somewhat of an easier path for high frequencies around the pot, thereby increasing the amount of negative feedback for those frequencies, thereby quieting them down. Anyway, you could use a 10K pot here (with a compensating for the 4.7K resistor going to ground).

Bigger resistances invite noise because less current flows through them. One source of noise are resistors themselves: they create what is called "thermal noise". It comes from the random movement of electrons in the material itself, and when it is amplified, it sounds like hiss.  The way you fight thermal noise is to have a decent signal current flowing, so that the thermal noise is small in comparison to the current.  Now in an op-amp feedback circuit, you could achieve the same amount of gain using a 1K resistor over a 2K resistor, or using a 100K resistor over a 200K resistor, or  using a 1M resistor over a 2M resistor. It's the voltage divider ratio that matters. However, the absolute resistance also matters: more resistance means lower quality feedback. The op-amp's input does not have infinite impedance; some current flows through the feedback into it.  Bottom line: smaller pot.

Further reading: http://en.wikipedia.org/wiki/Resistor#Electrical_and_thermal_noise

The coupling capacitors have wacky values. The 22nF input cap is very low. The purpose of the big resistors around it is to mitigate the loss of bass frequency response that this causes, but it won't work very well. The 10M resistor is in parallel with the output impedance of the previous device (e.g. guitar pickup and volume knob). Your 500K guitar volume knob provides an easier path for bass than the 22 nF cap, and the 10 M resistor might as well not even be there. You should remove these silly resistors and put in a 10 uF capacitor.

The 470 nF coupling capacitor at the output is equally silly, as is the 470 nF capacitor at the bottom of the feedback voltage divider. These both suck bass response. Make them 10 uF.

The 100K output resistor is pointless.  It basically means that your device's output faces an impedance that is no higher than 100K, which is a waste if the next device in the chain has a nice and high impedance in the megohm range. Furthermore, your op-amp's output impedance should already be pretty low, so the 100K won't do anything to lower it. You want our output impedance to be as low as possible (ideally zero), and the input impedance to be as high as possible (ideally infinite), for the best possible voltage bridging. Any silly resistors going to the ground at the input or output are just sucking away tone. Coupling caps should always be at least 10 microfarads. You want to block DC, and it's okay to block some very low frequency that nobody can hear, like 3 Hz. But 20Hz and up should nicely pass through.

Lastly, I did a google for this Tonemender and found its home page. Whatever you do, do not build it using the plain perfboard method that is illustrated. This is completely silly. Nobody in their right mind uses plain perfboard, because there exist PC perfboards which not only have copper pads, but also traces connecting those pads in a kind of "breadboard-compatible" layout. You can lay out your components on a breadboard and test the circuit, and then transfer the same layout to this type of board, where you neatly solder everything to the board. The existing traces minimize the jumpers that you have to use, and when you do need a jumper, it is nicely installed through the board like a resistor.

On the other hand, I can't believe someone produced a PC board layout for this device, without correcting any of its obvious flaws. :loco

[teemuk]

I have recently noticed that people have this trend of using more or less useless buffer elements in their circuits just for the sake of using a "buffer".

First it's wortwhile to realize the primary function of buffer, it is a high input impedance stage with low output impedance and decent current driving capacity to moderately low impedances. If we think about any basic OpAmp stage it has those same characteristics whether its configured to provide unity gain or higher gain....

Therefore, in circuits sensitive to noise, such as those working at low signal levels and especially those that work on low signal levels and introduce plenty of attenuation, it's wortwhile to boost signal level before attenuation happens.

Thus, the stupid buffered input stage of eg. Tonemender could be configured for some voltage gain, which then would compensate the losses of the passive tone stack following that stage. Additionally, with this mod the gain recovery stage could work with much lesser gain.

Also, since these are solid-state circuits and well capable to operating at moderately low impedances there is simply no reason to use values derived from tube circuits mandatorily working at high impedances. For example, all resistive elements in the tonestack can be well scaled down by a factor of ten (and the capacitors scaled appropriately to match). Lower resistance means lower noise introduced by the component.

Voila, noise level and overall performance significantly improved but without changing the loading effects or buffering capacity of the input stage at all.

Yep, pretty much what Kaz said already.

[phatt]

Hi ,,I Hope we have not scared off *nichapat* with over complexity.

If I may diverge slightly,

Well at least 3 good minds working on this one,, so if I may ask you to disect my *PhAbbTone Circuit* as well since it's also an oddball and may look to some as rather pointless and just another Amateur attempt. :lmao:

If it helps *nichapat*,,,Amateur is what I am anyway  
so I'm willing for my attempts to be dissected by better minds.

As a LIVE GIG player and having built quite a few units that reap very positive results I'm happy but if there are any possible ways to improve the circuit I'm always interested in how the experts would approach designing such a circuit.

The Basic concept was; Just pull the Hi-Z passive tone trick then bring it back up to the same level. 1/1 gain.
(These Hi Z tone circuits do something that low Z tone circuits just don't seem to be able to produce)

Negatives,  I already know;
I'm very aware the S/N penalty for no active front end but that can lead to distortion very easy,, and some pedals/preamps pump out way too hot a signal for a 9 volt circuit. So active front end seemed to have a negative effect.

Be brutal if you feel the urge as I'm just not the sensitive kind.

# Kas,, A Q if I may.
I understand the logic of 10uf Larger coupling caps but that is not always going to be what is needed. fine if you want hifi bandwidth but most guitar circuits have substantial rolloff below the 100Hz region.
In my world close attention to limiting bandwidth reaps great rewards,, though success does depend where and how it's implemented.

Interesting to note that some of the most revered Valve Amps have almost DC bandwidth at the first triode but by the time the signal comes out the power stages can be down 60dB @ 10 Hz.   A lot of SS guitar Amps tend to be the reverse in that the front end has a fair cut while the power stage is often flat response, even when implementing the defined impedance mentioned before a SS power stage does not (to my knowledge) pull extreme roll offs as those found in some Valve power stages. 

anyway have fun with it.
PhAbbtone Circuit below.
Phil.

[J M Fahey]

Let's see it step by step

This topology means that the tone control is not suitable for instrument level signals, but only for line level signals ........ Surely enough, the buffer creates a high input impedance, isolating the input from the tone controls, ensuring that the tone controls won't load down your guitar pickup. That's very good.

Agree.So it's not "useless".

But because the buffer has unity gain, it means that if the input is some weak signal like 200 mV, the signal going through the tone control is at the same level. This means that your tone control has to be completely shielded, just like the conductor in your coax cable and your guitar's cavities.
What you want to do is reverse these two elements. The first thing the device should do is boost the signal through an amplifying op-amp to get it from instrument level to line level as early in the chain as possible. This is where you can have your variable gain also. The buffer can be placed after the tone control, to isolate the tone control's impedance from the next device in the chain.

Agree, with a slight tweak.
I would not place the *full* gain stage ahead, because of the very certain probability of distortion, but would either put a trimmer there , adjustable up to 10X , maybe some simple Led peak indicator (best but complex) or simply set it for a 4 or 5X fixed gain, which is reasonable for that mythical "average guitar".

Secondly, a number of things stand out. Firstly, most of the pots used are 500K. Also, on the gain control, there is a suspicious 47 pF capacitor. Basically, this device looks like it was designed by a guitar player. (``500K knobs are nice and bright in my guitar, so they must be great everywhere.  A tone bleed on my strat's volume knob ensures bright highs when I turn down the volume, so I can just transplant into this op-amp circuit and it will do exactly the same thing!'')

Well, here we start to disagree.
I have no quibble with 500K pots, regularly use 1M ones with no problems at all.
The 47pF capacitor is not "suspicious" but good engineering practice, lowering gain outside the guitar range and preventing instabilities.
It's not a "tone bleed" but exactly the opposite.
Designed by a guitar player? Quite common in this business.
I agree that this tone network impedances can scaled 10X, keeping the same sound, but I *guess* that those typical "guitar" pot values were chosen precisely because of that, they are more easily available, even at Music shops, instead of Electronics ones.
In fact that allows to easily upgrade a conventional guitar with this circuit, keeping most of the original pots, specially considering that Audio taper ones are getting scarcer , these ones have the correct bushing and axis length, etc.; not to forget the main attraction: they are already there.
They are even available with push-pull switches, very convenient.
So I don't think the designer used them "just because he knows nothing else"

But the 47 pF cap in the gain circuit will cause some loss of high frequencies when you crank up the gain! The capacitive reactance of 47 pF with respect to a 10 Khz signal is about 339K. That is significant in regard to the 500K impedance of the pot. The cap provides somewhat of an easier path for high frequencies around the pot, thereby increasing the amount of negative feedback for those frequencies, thereby quieting them down.

That's exactly why it's there for.

Anyway, you could use a 10K pot here (with a compensating for the 4.7K resistor going to ground).

Said compensation would be a 100 ohm resistor, which at unity gain would be loading the Op Amp output.
Said Op Amp drives happily down to a 2K load; 100 ohms would be close to a short for it.
The 4k7 resistor is comfortably driven and a good choice.

Bigger resistances invite noise because less current flows through them. One source of noise are resistors themselves: they create what is called "thermal noise". It comes from the random movement of electrons in the material itself, and when it is amplified, it sounds like hiss.  The way you fight thermal noise is to have a decent signal current flowing, so that the thermal noise is small in comparison to the current.  Now in an op-amp feedback circuit, you could achieve the same amount of gain using a 1K resistor over a 2K resistor, or using a 100K resistor over a 200K resistor, or  using a 1M resistor over a 2M resistor. It's the voltage divider ratio that matters. However, the absolute resistance also matters: more resistance means lower quality feedback. The op-amp's input does not have infinite impedance; some current flows through the feedback into it.  Bottom line: smaller pot.

Agree in theory, but in practice, the network impedance seen by the Op Amp inverting input is the parallel of the pot resistance and 4K7; worst case=4k7.
The input impedance in these Fet input Op Amps is 1 Million Megohms (check the datasheet).
Even a bipolar input Op Amp such as an RC4558 would have no problem with this gain network.

Further reading: http://en.wikipedia.org/wiki/Resistor#Electrical_and_thermal_noise

Interesting, thanks for the link.

The coupling capacitors have wacky values. The 22nF input cap is very low. The purpose of the big resistors around it is to mitigate the loss of bass frequency response that this causes, but it won't work very well.

Who says so? As shown the low frequency cutoff is 8Hz, still dangerously low.
In guitar amps it's normal to cut below 60Hz, to avoid farting at high volume.

The 10M resistor is in parallel with the output impedance of the previous device (e.g. guitar pickup and volume knob). Your 500K guitar volume knob provides an easier path for bass than the 22 nF cap, and the 10 M resistor might as well not even be there.

Indeed, *if* said volume potis always there, as in an internal preamp.
If it's in a pedal, connected through an input jack, the 10M resistor to ground is an anti-pop measure, ensuring a DC path to earth for the .022 capacitor. I guess they tried to cover both construction possibilities.

You should remove these silly resistors and put in a 10 uF capacitor.

10uF? Why? I almost spat my coffee through my nose.
This is not an eartquake detector.
Why do you want this to be flat down to 0.016 Hz?
As an undesirable side effect, this preamp will turn on slowe than a tube preamp heating its filaments, it takes a *long* time to charge those 10uF through a 1M resistor.
And if you plug your guitar there (supposing that, as you suggested, the builder removed the 10M resistor), you will get the pop/thump of the century.

The 470 nF coupling capacitor at the output is equally silly

I wouldn't call it "silly", it's quite adequate considering it is expected to drive a guitar amp input or a regular pedal.
Anyway, here I will agree with you and also suggest a 10uF cap, not because of the next stage impedance but because it will lower the output impedance and help lower hum if you use a loooong cable after this.
For the same reasons I would add a series 100 ohm resistor, to isolate output wire capacitance (if too long) from the Op Amp output, for stability.

, as is the 470 nF capacitor at the bottom of the feedback voltage divider. These both suck bass response. Make them 10 uF.

As is, the LF cutoff is 64Hz, standard good practice in Guitar amplification.
No use for anything lower.

The 100K output resistor is pointless.  It basically means that your device's output faces an impedance that is no higher than 100K, which is a waste if the next device in the chain has a nice and high impedance in the megohm range. Furthermore, your op-amp's output impedance should already be pretty low, so the 100K won't do anything to lower it. You want our output impedance to be as low as possible (ideally zero), and the input impedance to be as high as possible (ideally infinite), for the best possible voltage bridging. Any silly resistors going to the ground at the input or output are just sucking away tone.

No, that 100K resistor is there to provide a DC path for the output capacitor, for anti-plop reasons.
It's *not* a load resistor.
It does not affect output impedance.

Coupling caps should always be at least 10 microfarads.

Why that fixation on 10uF?

But 20Hz and up should nicely pass through.

Not in the Guitar world (not even in the Bass world either)

Whatever you do, do not build it using the plain perfboard method that is illustrated. This is completely silly. Nobody in their right mind uses plain perfboard, because there exist PC perfboards which not only have copper pads, but also traces connecting those pads in a kind of  breadboard-compatible" layout. You can lay out your components on a breadboard and test the circuit, and then transfer the same layout to this type of board, where you neatly solder everything to the board. The existing traces minimize the jumpers that you have to use, and when you do need a jumper, it is nicely installed through the board like a resistor.

Agree, but please consider the average non-equipped musician-turned-builder,
I *hate* those unclad perfboards, but I am used to designing and making my own PCBs, silkscreen both them and my front panels, etc.
For a one-timer builder, perfboard still is an option.
After all, even the simplest method, drawing straight on copper clad board with a Sharpie, then etching, then drilling, then scrubbing and preferrably flux-covering can take the better part of an afternoon, with certain possibilities of failure the first times, while with perfboard this can be built in 2 hours, tops.

On the other hand, I can't believe someone produced a PC board layout for this device, without correcting any of its obvious flaws.

*Is there* a suggested PCB for this?
I'll check it later.

Dear nichapat: don't worry about this discussion among us, we love it, please show us what you did and we probably can suggest something useful for you.
That's what this Forum is about.
*Maybe* both projects together on full blast are too much , but at least each of them in its own should work properly.
Good luck.

[nichapat]

Thank you so much for all of your advice. This is what I call the power of network. We are friends.
This is the original unit with noisy problem . Sound like "SSSS....." .
I have reversed the circuit as your comment. Sound is tighter and less noise.  . But it is not enough for me so I have to do something further.

[J M Fahey]

Hey Nichapat, it is *very* well built !!!  :tu:
Maybe even overbuilt !! 
Please take another somewhat larger and better illuminated picture of everything you have in that chassis and explain what you have there.
What I don't understand is why you have so many heatsinks there.
What are you using as a power amp and speakers?
I had thought you had some unclad perfboards on your dinner table 
Now, where's that emoticon where I hit my own head with a brick?   

[Kaz Kylheku]

Let's see it step by step
The 47pF capacitor is not "suspicious" but good engineering practice, lowering gain outside the guitar range and preventing instabilities.

I thought so too, but it's value is too aggressive, relative to the resistance in the circuit, to be doing (just) that. To provide more stabilizing feedback at high frequencies, you can use quite a bit smaller cap value so that this feature does its job outside of the audio range.

Since the circuit already has a treble control, why allow the gain to mess with treble?  If you want to control the upper highs with this box, just add another pot and call it "presence", letting gain just be gain.

You're right about the input coupling cap. It may be the case that the 0.022 is sufficient due to the high input impedance there. The 470 nF going out, though, turns into a bass sucker if it faces a poor impedance in the next device. Suppose the next device has only, say, an 8K impedance. Then 470 nF coupling means 3dB loss at 42 Hz.  Maybe okay for guitar, but not bass.

If the guitar amp already cuts bass, that's fine, but the boxes in front of it don't also have to do that. If you want to re-purpose the boxes for something else, you will be glad they were more generally designed.

People design these kind of boxes with regard to their one single deployment scenario. "It works exactly how I like in my chain in between these two other units, for my instrument, and I will have you know that even the presence-cutting gain is a deliberate feature". 

[J M Fahey]

Yes, even the best designs often just don't work well, if at all, when paired to other circuits/instruments/speakers, etc. which, by themselves, work like a charm.
Oh well, the software guys, specially in the Windows world, face much harder incompatibilities than us. :tu:

[Kaz Kylheku]

Thank you so much for all of your advice. This is what I call the power of network. We are friends.
This is the original unit with noisy problem . Sound like "SSSS....." .
I have reversed the circuit as your comment. Sound is tighter and less noise.  . But it is not enough for me so I have to do something further.

Great to hear that there is an improvement.

Still not quiet enough, though? Could it be that the op-amp chip you selected for this project is noisy?

Here is an idea: replace your op-amp chip with an IC socket, so you can "plug and play" different chips that are pin-compatible. You can do an instant A-B listening comparison of different ones.

[joecool85]

Oh well, the software guys, specially in the Windows world, face much harder incompatibilities than us. :tu:

Boy howdy!  I've not done a lot of software programming but I do Web Development and websites need to be cross compatible with many browsers and operating systems which can be a giant pain in the butt!

[nichapat]

Please take another somewhat larger and better illuminated picture of everything you have in that chassis and explain what you have there.
What I don't understand is why you have so many heatsinks there.
What are you using as a power amp and speakers?

There are pre-amp , tube simulation , SMPS and class D amp. But I think classD amp does not work well for this application. I have  to go back to AB amp.
And for speaker it is normal PA type,15" .

I have just reduce the gain of first OP-amp (Reversed circuit) . And leave the last OP-amp of tube simulation as buffer. So it seem OK for me now.