Diodes in feedback loop vs. diodes to ground with Inverting gain stage

[drolo]

Hi,

This is my first ever topic, although i have been following this forum for some years now. (BTW thanks you all for the precious knowledge shared here, some of you guys are earning enough good Karma to have an incredible time in your future lifes ;-)) I have never NOT found an answer to my questions before but this time it seems that i can't find what i am looking for.

I have just understood the difference between a non-inverting clipping stage (TS) and an inverting one (Blues Breaker). I have been breadboarding a Blues Breaker for a project and was playing around with different clipping combinations.

I seemed to find very little difference between the diodes in the feedback loop and diodes to ground.

Now as i understood, the inverting gain stage has the diodes acting like they are clipping to ground. Thjis made me wonder what the benefit would be to have diodes to ground, like the Guv'nor or even the KOT where you can switch between the 2 clipping modes? Are there maybe differences i did not grasp?

Thanks in advance

[Bill Mountain]

Hi,

This is my first ever topic, although i have been following this forum for some years now. (BTW thanks you all for the precious knowledge shared here, some of you guys are earning enough good Karma to have an incredible time in your future lifes ;-)) I have never NOT found an answer to my questions before but this time it seems that i can't find what i am looking for.

I have just undertood the difference between a non-inverting clipping stage (TS) and an inverting one (Blues Breaker). I have been breadboarding a Blues Breaker for a project and was playing around with different clipping combinations.

I seemed to find very little difference between the diodes in the feedback loop and diodes to ground.

Now as i understood, the inverting gain stage has the diodes acting like they are clipping to ground. Thjis made me wonder what the benefit would be to have diodes to ground, like the Guv'nor or even the KOT where you can switch between the 2 clipping modes? Are there maybe differences i did not grasp?

Thanks in advance

I don't know any hardcore theory but I do know that the neg. input acts as a vitual ground so you shouldn't hear any differences.  If you want to make something switchable you could try switching diode types or cap sizes.

Good luck!

[Kesh]

You can build an inverting or non-inverting stage with diode clipping through feedback. The tube screamer is an example of the latter. And you can have inverting or non inverting stages driving clipping to ground.

But I don't fully get how it actually works either.

In particular one thing I don't understand about diodes are figuring out when they're clipping and when they're clamping. Lack of caps seems to be the theory behind clipping, but most of the effects pedal circuits have caps in the mix. so ought to clamp? Particularly with asymmetrical clipping, which in some circuits is probably not asymmetrical at all, the diodes just re-bias then clip evenly around the waveforms middle, as it were.

And sorry if I'm hijacking with more questions than answers.

[Mark Hammer]

First things ifrst.

How, conceptualy do op-amps work?

The op-amp's natural inclination is to go pedal to the metal.  The feedback from the output puts the brakes on that.  The amount of effective feedback you provide from the output back to the input will determine how much gain is essentially removed, and from what parts of the signal.

The feedback path and the "ground leg" on a non-inverting op-amp function like a voltage-divider with the "wiper" being the "-" input/pin; which is why gain depends on both of them.  If I have a straight wire path from the output back to the inverting input, and nothing between that inverting input and ground, I essentially have the equivalent of a pot with zero ohms on one side of the wiper, and infinite resistance on the other side.  I other words, I am applying ALL possible negative feedback from the output.  The result of that is unity gain, or no amplification.

If I was to provide a 100k feedback resistor, and a 100k resistance to ground, I am essentially presenting a 200k pot, with the wiper (the inverting input) halfway, to that feedback signal.  That is, I'm letting some of it through, and bleeding off a little.  If I make the feedback resistor larger in value, OR the ground leg smaller, it's like moving the wiper of our assumed pot closer to ground, thus reducing how much negative feedback is being applied.  Remember, the op-amp wants to go full-tilt unless directed otherwise by the negative feedback.  So, less feedback at the inverting pin = less "brakes" = more gain.

If I add a cap in parallel with the feedback resistor, I am providing a path for the negative feedback that is generally high resistance, but as the frequency goes up higher, it becomes a progressively lower resistance path for that feedback.  And of course, if the feedback path becomes lower resistance than the ground leg, then you are conserving more feedback, which means you are stepping onm the brakes harder...for higher-frequency stuff.  The content below that frequency range, however, is governed by the relationship between the resistors.  This is why a cap in the feedback loop reduces treble; less gain is applied to the treble than other parts of the signal because more negative feedback is conserved.

So far, so good.

What happens when there are diodes in the feedback loop?  The diodes also act as a kind of selective path, as a cap does, but instead of being lower resistance for higher frequencies, they become lower resistance for higher voltages.  As the signal approaches the forward voltage of the diode, it begins to conduct.  And, according to the principle, creates/conserves more negative feedback from the output, and more feedback = less gain.  Keep in mind this is happening on a signal-swing by signal-swing basis, which is how it results in "clipping".

Okay, let's move to the other stuff.  ALL op-amps have limits to how much gain they can apply at max.  This is expressed in many ways but one of them is "voltage-swing".  Looking at a datsheet for a JRC4558 right now, I see that if it is powered by +/-15V, it ts output can only swing up +/-12V in most instances; i.e., the output cannot quite reach the difference between the power rails.  Now, I mentioned this in another thread the other day, but will mention it again here.  Using a 9v supply (effectively +/-4.5v, it is likely that the signal cannot "swing" to moe than +/-3.5v.  Given that a guitar input signal is likely to be in the range of 60-100mv a lot of the time, you can't boost it by very much before it hits the limits of what the chip can do cleanly. Plus/minus 100mv only fits into +/-3.5V thirty-five times.  More gain than that applied to a signal like that and the chip itself clips.

But wait a sec!  If we have diodes in the feedback path, won't they start conducting well before that maximum voltage-swing point is reached?  You learn quickly, grasshopper.  Yes, if a diode-pair starts conducting around 500mv, they will introduce clipping, and a "ceiling" on the signal at less than the amplitude which might produce clipping from the chip itself.

Maybe you can see where this is going.  If the clipping diodes are on the output of the chip and going to ground, then the op-amp can be pushed to a gain which reaches the maximum voltage-swing of that chip first, and is THEN clipped by the diodes.  Double clipping, or rather clipping of a signal that has already been clipped in another manner.  The TS architecture only clips the once, where the Rat/Dist+/DOD250 architecture effectively runs up against headroom limits twice.

One of several reasons why the latter form tends to sound more "ragged".  There, you're smarter now.

[amptramp]

There are some simple distinctions to be aware of here.

The Tube Screamer is a non-inverting stage where clipping is in the feedback.  Once the clipping starts, the gain for the incoming signal is reduced to unity but some of the incoming signal is always riding over the clipped or unclipped signal.  This tends to limit the harshness compared to a circuit topology with diodes to ground at the output of an amplifier like the DOD250 where the clipping stops all signal from getting through.

With an inverting stage, there is no way to bypass the feedback - if the feedback resistance goes to zero, nothing of the input can get through.  If the diodes are to ground, the amplifier can operate normally but the diodes will drag the output down to their breakover voltage at the current they are carrying.  But since the diodes are outside of the feedback loop in this case, the current through the diodes can go up to the current limit of the op amp, permitting more diode current and using the diode V-I characteristics to colour the sound.  With a feedback clipper, the feedback current is the same as the input current which is normally very low and sets the diode response to the low-current region of its V-I curve.

Hard clipping tends to give a muddy sound.  Imagine a sine wave with a higher harmonic riding on top of it.  When it goes through the clipper, the waveform is held to an upper and lower limt.  Any excursion beyond those limits is held to a fixed value.  The harmonics that were riding on the signal are eliminated entirely for the duration of the clipping.  You have some series resistance in the clipping diodes and this may permit some gain in the clipping region, but this will depend on temperature, type of diode etc.  You may approximate the TS gain - or not.

[drolo]

Thanks for all the answers

I still have a bit of trouble understanding how the diodes in the Inverting gain stage feedback act as diodes to ground.

I don't know any hardcore theory but I do know that the neg. input acts as a vitual ground so you shouldn't hear any differences.

Does this mean the inverting input has the same potential as Vref? and the same as the non-inverting input?

With an inverting stage, there is no way to bypass the feedback - if the feedback resistance goes to zero, nothing of the input can get through.  If the diodes are to ground, the amplifier can operate normally but the diodes will drag the output down to their breakover voltage at the current they are carrying.  But since the diodes are outside of the feedback loop in this case, the current through the diodes can go up to the current limit of the op amp, permitting more diode current and using the diode V-I characteristics to colour the sound.  With a feedback clipper, the feedback current is the same as the input current which is normally very low and sets the diode response to the low-current region of its V-I curve.

that is quite interresting. Does that mean that, like with a non-inverting op-amp the type of diodes in the feedback of the inverting amp does not really alter the sound, only their forward voltage? But they do when used as diodes to ground?

[ashcat_lt]

Now wait a minute!  In the bluesbreaker the diodes don't connect directly to the inverting input, but rather to a resistor which then hits the input. 

The gain of an inverting stage is equal to the feedback resistor divided by the input resistor.  Think of the diodes as a switch.  When they are open (because the output level is below their forward voltage) the feedback resistor is the one large one, giving comparatively large gain.  When the diodes conduct, you now have this much smaller resistor parallel to that large resistor.  Gain is reduced pretty drastically.  It will tend to act like tubescreamer clipping.

[Mark Hammer]

When diodes are in series with a resistor in the feedback loop, they are often described as providing "soft clipping" or "soft limiting".  I can't speak to where the boundary is between hard and soft, but it isn't quite as hard a clip.  If you look at a variety of flangers, like the BF-2 and the A/DA, you'll see such a "soft limiting" use of diodes in the stage where the feedback signal comes back to.  The intent is ti prevent the combination of the feedback signal and the input signal from being too hot for the BBD, whilst still permitting lots of feedback/regeneration.  The series resistor allows for it to not be quite so harsh-sounding.

[drolo]

Thanks all,

I guess my thinking "diodes in feedback loop of inverting opamp = diodes to ground" was a bit of a shortcut... i understand the difference better now.