diodes with little to no current dependent forward voltage

[iainpunk]

so im looking for diodes that have a stable forward voltage. i found that most diode's forward voltage is somewhat depending on the current going through the diode, this changing of the forward voltage means that at different currents, the diode i use has different forward voltages.

at 650uA the 1n4148 has a forward voltage of 617mV, and at 65uA, the forward voltage drops to 466mV
i was wondering if there are types of diodes that don't exhibit such vast differences in Uf at these ranges of current.

cheers

[ElectricDruid]

Isn't this sort of regulation job what zener diodes are for? What are you trying to do, and why are you trying to do it with diodes that don't want to do what you're apparently trying to make them do?!? 

Every other sort of diode is going to vary a bit with current, as far as I know. To be honest, a drop of about 35% in Vf for a 10:1 change in current doesn't seem that bad to me!

[antonis]

Just feed any diode you like with a constant current..

[PRR]

The serious way is an amplifier to run the reference (diode, battery cell, thermocouple...) at dead-constant current, and buffer that to the thing which needs a regulated voltage.

> diodes that don't exhibit such vast differences in Uf

200mV per decade is typical; by the way electrons vibrate. Your "vast" 151mV seems a little low (agrressive doping can do that to some degree).

[iainpunk]

the diodes are clipping, and the current of the stage in dependent on a kind of filter circuit. i don't want to loose more volume than necessary if i dull down the sound a little. at this point, im losing about 3 to 4 dB when turning down the treble from sharp to dull. without the clipping diodes contributing to the volume level, i'd be losing about half a dB of signal with this same amount of turning down the filter.

am i just worried over a nothing burger? is 4dB of loss enough to be worried about from a circuit design standpoint? the volume loss is quite severe in person.

cheers

[PRR]

Separate the filter from the clipper.

Sometimes a brilliant mind sees a way to do both, no compromise. Meanwhile, break up the issues.

[ElectricDruid]

+1 what PRR said.

-4dB loss is definitely not a lot. Add some make-up gain later on. People lose *way* more than that putting some famous passive tonestack in their pedal.

Clipping diodes reduce the signal anyway, just due to the clipping - no current-related effect necessary. For that, you can use LEDs or series combinations of diodes to raise the clipping threshold.

[Fancy Lime]

Essentially, you are looking for diodes with a sharp knee and low series resistance. 1N4148 are not any if those things. Body diodes of BS170 or 2N7000 have a significantly sharper knee and slightly less series resistance. That's what makes them "sound like a tube amp" when used as clippers   Similar but cheaper are the forward diodes of Zeners. Not sure if Zener voltage makes much of a difference, I would assume not. If you aren't married to the clipping threshold of around 0.6 V, then LEDs are even more sharp when you account for the higher threshold. IR LEDs may be the hardest clippers of all simple diodes to ground. That being said, all of these are slight to medium improvements for your issue, nothing drastic. If any of them is going to cut it is something that you'd have to try.

Is redesigning to use opamp clipping at the rails an option? That can be pretty dang hard.

Cheers,
Andy

[iainpunk]

the clipping is just to limit the output level of a slewed square wave. i thought i had found an elegant circuit to use, it limits the output of the integrator (which slews the edges of the squarewaves from an 555 timer.). using the rails to clip and then rely on a separate level reduction after the opamp to bring it down from 9v pk-pk to something like 1.3v pk-pk.

i might even get creative with diodes, add switchable crossover distortion to add some grind to the sound if the slew is low enough.

this is what i have right now:


im going to experiment with some clipping options, like i used to do a lot when i started... at some point i stopped caring and just put 4148's in everything. maybe a bit of mojo is what this square wave monster needs.

cheers

[Rob Strand]

A transistor with b connected to c is about a sharp a "knee" you can have (from a pn junction), provided the current isn't so high that the internal resistance start coming into play:

The exponential is part of the physics of diodes,
https://www.elprocus.com/diode-approximation-types-and-diode-models/

Change in VD from operating at I1 to I2 is d_VD = VD2 - VD1 = n * (k T / q) * ln(I2/I1);  kT/q ~ 26mV
Transistors have n=1 whereas common diodes are around 1.5 to 2.0, so less change.

For a sharper curve you need to use a zener >= 5.6V which follows breakdown not the exponential.

Another trick to give the impression of a sharper characteristic is to put a diode in series with a voltage source.
While the d_Vd is the same for a given change in current in relative terms the charge looks sharper on a say a 5V scale than the diode on a 0.6V scale,
https://interviewmania.com/discussion/97076-analog-circuits-analog-circuits-miscellaneous

You also have precision rectifiers which have no voltage drop and (in theory) a very sharp knee.  If you clamp against a known voltage you can set the clip thesholds:
https://btbm.ch/precision-positive-negative-clipper-using-any-op-amp/

[antonis]

this is what i have right now:

Just after a rough estimation, you should need a feedback resistor (across 100nF cap and diode pair) to set the gain together with 555 OUT series resistance to set the non-limiting gain or the op-amp will be forced to its open-loop gain..
(of course, it will almost immediately clip but I shouldn't rely upon it..)

P.S.
IMHO, the below configuration might be more suitable to your needs..



At least, it exhibits smooth transitions from the linear to the limiting behavior..

[amptramp]

In this application note, you have a precision diode and a precision clamp on page 17:

http://natasha.eng.usf.edu/gilbert/courses/instrumentsystems/roadmaps/0_1page_opamps/13_0b_op_amp_config%20notes.pdf

This is expanded to precision fullwave rectifiers.  You don't need D2 or C1 if you are not using the LM101/LM201/LM301.  I have used a pair of precision clamps in a fuzz design of my own and you don't have to worry about diode voltage - the op amp provides enough voltage to make the diode drop within a few millivolts of zero during conduction.

[mac]

The exponential is part of the physics of diodes,

+1*exp(inf) 

https://en.wikipedia.org/wiki/Partition_function_(statistical_mechanics)

mac