Noob question: Diodes in parallel, in the same direction?

[Nick C.]

Does using more than one diode in each direction for clippers have any effect on the turn on voltage?

My three guesses:

1. My hope would be that it would lower Vf. Mmmm squishy.
2. Maybe if different diode types were used it would change the harmonics produced as they would clip at different levels.
3. No effect, the 1st to clip wins.

If the answer is #3, how about having a silicon clipping stage to ground, followed by an isolating cap, followed by a Ge clipping stage. To get progressive clipping.

Any thoughts on this?

I'll have to do some testing.

[R.G.]

3 and then 2, but not 1.

The first to *start conducting* does everything until its voltage rises to the start-conducting point of the second and succeeding ones.

Diodes do not suddenly start conducting at one voltage. They have an exponential relationship between current and voltage. The current is quite near zero, but steadily increasing until the place where the exponential hits 1, then it turns for the sky, but never abruptly. In effect, there is a rounded "knee" on the conduction curve where over a tenth of a volt or so, it suddenly starts conducting.

The first one does all the conducting until its voltage rises to where the second one joins in. The two in parallel have the same start voltage as the earliest conducting one, and clip more sharply than either alone because the additional current in the conduction area clips more sharply than either diode by itself.

[Nick C.]

Looks like we might get a more complex clipping model. If pairs of identical diodes are used it looks like it will be a harder clip at the same voltage, if I understand correctly. More sonic possibilities.

Surely someone must have tried this in the past. Is it used in designs or does it make any audible contribution?

I will do some tests asap.

[Caferacernoc]

"If the answer is #3, how about having a silicon clipping stage to ground, followed by an isolating cap, followed by a Ge clipping stage. To get progressive clipping."

I think that would work with an isolating resistor also. Easy to check, put a switch on the second set of diodes and listen for a difference.

[R.G.]

In most cases, the germanium would prevent the voltage from ever getting high enough to make the si's conduct. It takes a brutal amount of current to make the voltage across a Ge diode that started conducting at 0.2V get up to 0.5V to kick the silicon on.

Still, go do the testing. There could be a diamond under that rock.

[alanlan]

You'd probably get more mileage out of experimenting with differing combinations of resistors, cap or even inductors in series with a diode and then putting the lot in parallel with another different combination in the same (diode) direction.  And then perhaps something slightly different again in the other direction.  The sky's the limit when it comes to this sort of experimentation.

It has to be said that you can gain a lot of insight into this kind of thing by using PSpice or similar as well as playing around with the real thing.

Good Luck.

[Nick C.]

Last night I plugged some diodes into the Electra distortion that I've been playing with.

1 ge vs 1 ge > fuzzy, low output
1 1n4004 vs 1 1n4004 > louder, harder edged
1 ge + 1 1n4004 in parallel vs 1 ge + 1 1n4004 in parallel > output same as ge's but with a bit more crunch, which I would think supports R.G. response.

So I would think that if one wants the clipping and compression of a Ge with a little more edge that running a Ge+Si in parallel is a valid option.

The other day I had been messing around with LTSpice, thanks to Gaussmarkov's great tutorials, and modeled the Electra. The stock library doesn't have Ge diodes but I ran some tests of parallel 1n4148s.

With 1 in each direction the output was 500mv.
With 2 in each direction the output was 450mv.
With 4 in each direction the output was 420mv.

Small drops for what it's worth.

The LTSpice is a lot of fun, but what does it really sound like? A bit analytical for a noob? I'm a mechanical engineer, but brand new to electronics and FX.

I did try some Spice models with caps which gave some pretty wild wave forms. I'll have to plug in some resistors too.

[R.G.]

This is where the small differences get you.

It is excruciatingly difficult to evaluate "a bit more crunch". To really see the difference between 1-Ge and 1Ge+1Si, you need to put the the Ge pair output on an oscilloscope and switch the Si pair in and out to see the differences.

It's possible that there could be an effect of a silicon in parallel with a Ge, but the conduction voltages are so far apart, I'd expect no change at all. I could be wrong, of course. I guess I can go out to the garage and do the experiment.

The slight drops in LT spice for paralleled silicons are about right acccording to my internal "simulator". In addition, in a simulator, all devices are mathematically identical. In the real world, even diodes on the same wafer vary a little bit. So getting three 1N4148s that are perfection identical would be hard to do. The individual diodes themselves could vary by 20-30 millivolts

[Nick C.]

I did the swapping pretty quickly with alligator clips and I could hear a small difference. A preconceived notion?

I wouldn't claim to have golden ears, but my hearing always tests well despite some hyperacusis and tinnitus.

[R.G.]

Hard to say. Could be. Could be expectation.

[Caferacernoc]

So if you had si diodes to ground, followed by say a 10k coupling resistor, then ge diodes to ground, would the signal not get clipped by the si diodes and then the remaining signal get clipped again by the ge diodes?

[R.G.]

It would indeed. It would be clipped at the same place where it would have been clipped if you left the silicons off entirely, if you think about it for a moment.

[brett]

Hi
there have been some circuit ideas posted over the years with mutiple diode +resistor systems.
e.g. 1N4148, 1N4148+2k, 1N4148+4.7k, 1N4148+10k (NB I'm not suggesting these are good values for your setup.)

The idea has usually been to "soften" clipping by allowing the clipping voltage to rise through several stages, especially in diode-to-ground systems. 
cheers

[Paul Perry (Frostwave)]

Brett's right, if you have enough diodes and resistors, you can simulate ANY kind of curve (that is what they used to do in the old analog computer days ).
And while people are looking for odd diodes to try - feel free to use either of the junctions in a transistor! If you exceed the reverse breakdown voltage, it will be even more 'interesting'

[Caferacernoc]

It would indeed. It would be clipped at the same place where it would have been clipped if you left the silicons off entirely, if you think about it for a moment.

Yep. You are right.

[Caferacernoc]

"Brett's right, if you have enough diodes and resistors, you can simulate ANY kind of curve (that is what they used to do in the old analog computer days )."

Well, golly, than why do we fool around with those darn tubes, jfets, and germanium things?  ;-)

[R.G.]

Because you can simulate any curve you want, so long as you can us a signal amplified up big enough to make the actual diode drops be insignificant.

That allows you maximum flexibility in curve shaping. But it means you need a big power supply voltage, and usually a bipolar one, although this can be worked around.

If it's a one-transistor circuit with a 9V battery and ten other parts given that the transistor is germanium, or a AC wall powered supply at +/-15V, four opamps, twenty diodes and a batch of adjustments, which one gets built and sold to musicians to use at gigs if they sound the same?

[Nick C.]

Last night I tried a softening resistor in series with the diodes to ground, on both sides. I was a little surprised at first by the jump in volume, but it makes sense that there is a voltage drop across the resistor and the diode sees a lower voltage and will clip later. So, by adding a series resistor we can raise the effective forward voltage of any diode. If this is true, we can change the Vf of a Ge to that of an LED, but with a different knee curve. Interesting. Would the curve be the same just at a higher voltage?

[R.G.]

I posted the idea of a softening resistor for diode clipping about seven years ago, including a variable clipping circuit for the MXR D+.

Jack Orman amplified endlessly on this with his warp controls.

The diode Vf does not change - only that a voltage equal to the diode current times the resistor is added to it. This does sound much softer. The bigger the voltage and resistance, the less the diode matters.

[johngreene]

I posted the idea of a softening resistor for diode clipping about seven years ago, including a variable clipping circuit for the MXR D+.

I suggested the progressive clipping/shaping about 8 years ago.
http://archive.ampage.org/articles/1/fxsb/042458/Re_Seemingly_good_idea_but_anudder_idea.html

--john