Diode Clipping...trying to realllly understand this

[ryangobie]

ok, so i read through about 20 search pages and found some stuff and read over GM Arts page and i feel like i know a little more, but just a few questions...

GM Arts page discusses the difference between soft clipping and hard clipping and it makes sense germanium would clip harder as it lets through less voltage...but does the configuration of the clipping circuit have any factor on this...for example, will diodes shunt to ground clip harder than clipping diodes in a negative feedback circuit when using the same componants...

it seems in a lot of schematics, the negative feedback configuration is used primarily with op amps but the schematic i saw for the EH big muff used them with a transistor...is the later a more unorthodox method? will it sound different?

and just finally, how exactly does the signal flow through these circuits...2 diodes, each for one half of the waveform right? but when the diodes are shunt to ground, wouldn't the clipped waveform be shunt to ground? i know that's not what happens...curiousity got the best of me...thanks guys/gals

[R.G.]

You are wise beyond your years.

[Mark Hammer]

The "hardness" of clipping is easily misunderstood (and/or perhaps difficult to convey), as is evident from your posting.

Think of a diode as being like a bathtub, with the water running and the plug open.  If the amount of water coming in is fast and furious enough, excess spills out over the top.  If the bathtub has very high walls, the drain is wide and unplugged, and the water isn't coming in too fast, then everything that comes out of the faucet also goes down the drain, with none of it spilling onto the floor.  Of course if you lower the sides of the bathtub, then there is a strong likelihood that the amount of water inflow will be too much for the drain system, and lots will flow over the sides and be lost.

The diodes to ground set a kinf of maximum limit on signal.  As long as the incoming signal is well below the "sides of the tub", there will be no signal loss and everything coming in will also leave via the proper exit.  Should there be any momentary "excess" signal, though, it will overflow the diodes and be leaked to ground.  What you lose with diodes are those peaks that exceed what can be safely passed at that instant.  This is why a diode pair  will appear to provide a constant level despite how hard you slam against the strings.

Different types of diodes are analogous to bathtub walls of different heights.  Think of LEDs as being like those old luxurious "clawfoot" tubs with very high walls that you can sink right into.  Silicon diodes are like standard apartment bathtubs with modestly high walls.  Germanium diodes are like those little raised sides you find on university dormitory showers.

Each of these diode types results in "spillage" of signal being easier or harder to achieve.  The easier it is to produce such spillage, the more consistently distorted the signal will be and the easier it will be to achieve spillage.  In this sense, with a signal of same amplitude, Ge diodes distort "easier" than Si or LED.  Of course the samew way you can always turn the water pressure up or down to create or avoid water spillage, you can always increase or decrease signal amplitude and gain in the pedal itself to bring the "water level" dangerously up to the side of the tub, no matter what the tub type.

What often gets described as "hard" or "soft" clipping is really something different: the speed with which the diode commences the "spill" as the signal level approaches the threshold, whatever that threshold happens to be.

Part of what results in so much confusion about diodes is all the mistaken beliefs stemming mixing up HOW EASILY it clips with what it does WHEN it clips.  The hardness or softness of clipping is about what happens when it clips.

[brett]

Hi.  A while back I calculated a measure of "softness" of clipping.  This was the rate of increase in conduction (conduction=inverse of resistance) per unit of signal applied.

From memory, the results for different diodes were;
Ge = about 4 (mA/V/V ??)
Si = 2.5 (1N400X=2.4, 1N914/4148=2.6)
LEDs = 1.4

Anyway, if I don't have the exact figures right, I'm sure they are ok in a relative sense.

These results basically come about because the *shape* of the voltage vs conduction curve is very similar for different diodes, but the the fixed amount of voltage that is unclipped varies (Ge=0.3V, Si=0.7V,LEDs=2.2V).  Therefore, the amount of clipping per amount of signal ("softness") is directly related to the inverse of the diode's threshhold voltage.  

To look at it another way, for a given amount of signal clipping, a circuit with Ge diodes have only a small amount of unclipped signal (0.3V), while LEDs with the *same* amount of clipping will have 2.2V of unclipped signal as part of it.  These clipping threshholds remain the same for each type of diode, no matter what the level of signal is.  So there's a constant difference between the two types of diodes when they're used in the same type of circuit.  Maybe I've made this sound complicated, but it's not really  :wink:

have fun with your diodes and things.
PS One of my favourite things is using 2 diodes in one direction and 1 in the other, so that the clipping threshhold is doubled for one half of the waveform and the output is asymetrical (3 x 1N4148s are good in a wide variety of circuits).

[ryangobie]

great analogy mark...thanks a bunch...i can really visualize a diode as a spill hole in a sink or something like that...when you refer to speed are you talking about how fast it lumps it off...when the instant a transient hits it get cut off vs maybe a more gradual cut off (not instantaneous, some of it getting through...dynamics?)

brett...i understand that different diodes are going to let more or less through...hmm...when a signal is fed through each of those diodes, it's going to have some part clipped assuming it goes beyond the diodes threshold...for instance, a 3v signal is only going to have a small portion of the wave clipped going through an LED whereas that same signal is going through a germanium is going to be close to a square wave...so hard and soft in these terms is how much gets clipped...a lot being hard, only a little being soft? but anything could be made to clip hard or soft assuming it sees the proper voltage?

one more thing...if the diodes are shunting the "excess" to ground, in the negative feedback circuit, is it the signal after clipping or the clipped portion going back to the input? am i thinking about this totally wrong?

[petemoore]

great analogy mark...thanks a bunch...i can really visualize a diode as a spill hole in a sink or something like that...when you refer to speed are you talking about how fast it lumps it off...when the instant a transient hits it get cut off vs maybe a more gradual cut off (not instantaneous, some of it getting through...dynamics?)
 >>>Yupp Mark types real good don't he lol

brett...i understand that different diodes are going to let more or less through...hmm...when a signal is fed through each of those diodes, it's going to have some part clipped assuming it goes beyond the diodes threshold...for instance, a 3v signal is only going to have a small portion of the wave clipped going through an LED whereas that same signal is going through a germanium is going to be close to a square wave...so hard and soft in these terms is how much gets clipped...a lot being hard, only a little being soft? but anything could be made to clip hard or soft assuming it sees the proper voltage?
 >>>Sounds like you pretty much got it...

one more thing...if the diodes are shunting the "excess" to ground, in the negative feedback circuit, is it the signal after clipping or the clipped portion going back to the input? am i thinking about this totally wrong?

Basically ground stays at ground or 0V. when the signal gets a voltage difference greater than the diodes clipping threshold, the diode starts clipping, that energy is shunted to ground, which stays at 0V.
 There are lots of ways to influence clipping
 Change diode type/threshold
 Change circuit: Shunt to ground or feedback loop diodes
 Use the gain knob or increase input voltage [say add a booster/turn the guitar up]
 Check out AMZ's Lab Notebook...Adding caps or resistances over, or adjacent to clipping diodes can adjust the amount of clipping or what frequencies are clipped.

[brett]

Hi.
I knew I made it confusing... :oops:

for instance, a 3v signal is only going to have a small portion of the wave clipped going through an LED whereas that same signal is going through a germanium is going to be close to a square wave...so hard and soft in these terms is how much gets clipped...a lot being hard, only a little being soft?

Actually, when you convert them back to the same size of signal (e.g. 0.1V), the LED clipping looks really square because before being scaled back to a nominal size, it had really tall "sides" and a top that was no rounder than a Ge diode.  From this you can probably see that it's the height of the unclipped "sides" of the waveform that will detmine the overall squareness (all other things being the same!).  So LEDs with "tall" clipping at 2.2 to 3V make for "square" (more like rectangular, actually) and "hard" clipped waveforms, while the short little ones from Ge diodes are "rounded" and "soft" clipped, because they don't get scaled back down as much.

This seems much clearer now... Apologies if I added any confusion.

[ryangobie]

thanks again guys...anyone else want to chime in?

[Outlaws]

thanks again guys...anyone else want to chime in?

LOL  Yes.


So if you have a back to back diode setup, then you can just double up the each direction to get double the clearance before clipping?

Would you suggest leaving the silicon with a normal back to back setup and then using 4 GE diodes if you were swapping them out?

[Mark Hammer]

Well, here's the thing.  The threshold at which clipping begins can be altered by change of diode type and/or by diode number/quantity.  The individual voltage drops are summed.

So, if you have a silicon diode with a threshold/drop pf 543mv, and a germanium with a 228mv drop/threshold, in series they would start to clip the relevant half-cycle at around 771mv.  The particulars of what happens when such hybrids are used and examined on a scope is outside the perimeter of my experience.

Because change of diode type also implies change of threshold, one of the perennial problems that needs addressing is what to do about level changes.  For instance, I stuck a toggle and two sets of diodes on both my Superfuzz and Distortion+, to select between Si, Ge and no diodes.  IN each instance, we are talking about comparing a simple 1+1 back to back pair of one type against the other.  Switching to Si from Ge provides more crunch than buzz because the threshold is raised, but raising the threshold/ceiling also means an overall increase in volume level, which is not necessarily a problem but can be awkward for gigging purposes.

There are a few ways around this problem.  One is to simply use combinations of diodes to achieve the same general threshold.  So, for instance, a well-selected 2+2 pair of Ge diodes CAN come close to the same threshold as a 1+1 pair of Si (measure the parts in your bins and select). This is not necessarily the prefered strategy because often what people want a diode change for is not the quality of clipping when it occurs, but the *likelihood* of clipping.  Adding a bunch of Ge diodes in series will decrease the likelihood of achieving the hard fuzz that the person may be seeking.  

In which case, maybe the smart thing to do is to use a gain recovery circuit, post-diodes, and incorporate into the switching some means of holding output level constant when diodes are switched.  So, for instance, say I had a DOD 250 or Dist+ and wanted to be able to flick a switch and get extra buzz with germanium, OR nice crunch with silicon, without having to futz around with volume adjustments OR sacrifice the extra drive afforded by a pedal with more output.  Simply tack a a non-inverting gain stage after the diodes, and using a DPDT toggle or slide switch, you could arrange for one set of contacts to lift the ground connection of the one set of diodes or the other, at the same time as the OTHER set of contacts changed the value of the feedback resistance for the op-amp to get *more* post-clip gain for the Ge diodes, and less for the Si diodes.  You may need to use a trimpot to tweak to perfection, but in principle it ought to be able to change the amount of "hair" on the dog, without changing the "size" of the dog.