LED to bias JFETs??? (R.G.)

[Plectrum]

Came across this article by R.G. describing an adaptable circuit...

http://www.geofex.com/Article_Folders/FXbus/fxbmod01.htm

It mentions biasing a JFET with a LED... um, how does that work???

[aron]

In the schematic section, check out the Shaka Braddah.

ttp://www.diystompboxes.com/pedals/schematics.html
http://www.diystompboxes.com/pedals/sbnotes.html

[amz-fx]

-Jack

[mac]

I built a 3 stage jfet distortion circuit biasing the first two BF245A with 1N4148s from source to gnd and near 10k from drain to vcc.
The last BF245A have a OA91 Ge diode at the source and near 3.9k at the drain. Using two Ge diodes drain resistor is near 10k.
All drains are about 4.5v. The 2.2uf polyester bypass caps adds a little more gain and highs.
If someone wants to take a look at the schem send me a msg.

mac

[R.G.]

Depletion mode JFETs can be biased by **anything** that jacks their source a few volts above their base. That can be:
(1) a resistor, in the classic "self bias" circuit
(2) any semiconductor junction with the right voltage: stack of one or more diodes, an LED, a low voltage zener, etc.
(3) an active or passive voltage souce like
- a single transistor hooked up as a Vbe multiplier; with this one you can literally dial in the bias
- a resistor votlage divider
- an opamp output set to the right voltage

All you are doing is raising the source higher than the gate, which is held at ground by the gate resistor.

[brett]

Hi.
Anyone interested in using LEDS for biasing might want to check out John Hollis' designs.  The Ring Frobnicator is just one elegant design.
cheers

[DDD]

Hi men,
As per the above schematic: is the 2,2 uF cap shunting LED to be necessary? LEDs have very low dynamic resistance, which is many times less than the impedance of the cap. Or maybe there are any sensitive subtleties?

[davebungo]

Hi men,
As per the above schematic: is the 2,2 uF cap shunting LED to be necessary? LEDs have very low dynamic resistance, which is many times less than the impedance of the cap. Or maybe there are any sensitive subtleties?

http://www.oksolar.com/led/led_color_chart.htm
seems to show between 6 and 20 ohms, so yes it does make one wonder.  A 2.2uF cap is 180 ohms at 400Hz, so the LED will dominate things.

then again...I think there may be a snag or two.  Depending on the actual device, setting a VGS of -1.8V (typical for a LED), may result in a very small ID.  With such a small current, you may be right on the cusp of the LED forward characteristic where the gradient of the V/I curve may be greater but then again, I haven't tested this.  Something to try out.

[amz-fx]

Or maybe there are any sensitive subtleties?

There are...

It is a 22uf cap, btw.

-Jack

[Plectrum]

Many thanks for the info guys...


Grant

[davebungo]

Or maybe there are any sensitive subtleties?

There are...

It is a 22uf cap, btw.

-Jack

Oops failed to read the circuit diagram before I jumped in but on the other hand, DDD does have a point which is worth exploring a little, so care to expand on what these subtleties are?  My previous point on bias may be one but I'm not sure if there is something else I'm missing.

[R.G.]

...but on the other hand, DDD does have a point which is worth exploring a little, so care to expand on what these subtleties are?

I'm not sure exactly what DDD was thinking of, but the dynamic impedance of the LED is certainly one of them.

If you look at the V-I curves of any component, you can calculate an incremental resistance.

A real resistor follows the rule that the current it lets through is just the applied voltage divided by the resistance - or as Ohm said, I =V/R. For a resistor, that's true everywhere.

For, let's say, a diode, the current it allows through increases by an exponential law: something like I = ke^nV, where k and n are some constants reflecting the exact diode. The term "e to the nV power" is what makes the current rise very slowly at first, then take a rounded turn for the stars, eventually rising at a rate limited only by the resistance of the lead wires.

For a silicon diode, the k and n terms are such that the current rises only infinitesimally until about 0.5V, then over the next 0.2V, it goes to only a few ohms. But this change is rounded, not abrupt. If you take tiny, tiny voltage differences, say 0.6000V to 0.6001V, then you can calculate an effective resistance between those two points of the difference in the voltages and the difference in the currents. To signals smaller than a mllivolt, the diode looks just like a resistor of that value.  This is the "dynamic resistance" of the diode.

There are no instantaneous changes between resistance, except perhaps for a mechanical switch. Every two-terminal, variable conduction device shows the dynamic resistance phenomena to small enough signals. This is, in fact, how many RF signals are modulated - tiny signals across diodes which have their DC conditions changed to change the dynamic resistance.

The LED bias thing is no exception. An LED in forward conduction has a forward resistance that varies from quite high when it's non-conducting to quite low when it's conducting. So the point is accurate - if there's not enough current flowing through the LED, it will not hold a fixed voltage across it, and so the dynamic resistance will change, changing the effective gain of the JFET. If the point is to get the LED to stabilize the JFET gain, an LED is not a great way to do it unless it has enough current flowing in it to keep the dynamic resistance low enough.

A cap may or may not help, depending on how low the JFET dynamic resistance is. You're talking about needing a cap that's lower impedance than a dynamic impedance of a couple of hundred ohms. So the cap would possibly need to be on the order of 10 ohms at the low end of your guitar range if you're trying not to use this as a tone control.

The cap value then needs to be aroudn 10 ohms at 82Hz, or C=1/2*pi*82*10, or 194uF to get there. A 22uF is only adequate for flat response at around 820Hz. Anything less than about 220uF will rolll off bass in that region. That's one subtlety.

The LED current affects the dynamic resistance and the LED current is changed by the signal through the device. This leads to a distortion effect that's distinctly one-directional or assymetrical compared to the signal. If the signal levels are just right, you get asymmetrical distortion.

If the signals are tiny, you can take advantage of the fact that the current level of the LED doesn't change much with increased current, so it does not change the JFET bias point much. So you can put in an external current source at the top of the LED and change the LED's dynamic resistance with the external current, while not changing the LED voltage much, nor the JFET operation. This modulates the gain of the JFET, and could result in either an external gain change or a tremolo setup if you were careful about how you did it.

Finally, the JFET is quite dependent on the LED (or other device in that positions) having the right voltage to make the JFET bias properly. Ideally, the LED voltage would put the JFET somewhere in the middle of its conduction region.That means that the LED voltage needs to be on the order of half the Vgsoff for the JFET. JFETS come in all kinds of Vgsoff values, from as little as 0.1V to about 20V. Once again, the JFET variance is what has kept many, many useful circuits from being made with JFETs. You have to select JFETs to work with whatever you stick in there.

There is a trick to getting wider variation in dynamic resistance that involves stacking many diode junctions to get a wider range of dynamic resistances and an easier time controlling them for bigger signals. If, instead of an LED, you use three or four silicon diodes, the final voltage is about the same but you have a much wider, rounder "knee" that will modulate bigger signals more easily than the single-junction performance of an LED. That might help make a better amplitude modulator.

There are probably others. These are the ones that occur to me off the top of my head and aren't really subtleties if you're familiar with JFETs, diodes, and dynamic resistances.

[mac]

Correct me if I am wrong.
the diode riases the source over the gate setting the bias, and being practically a short it is like a bypass cap. The diode have some small resistance I guess, not an expert in electronics, so a big electro cap can be added to get extra gain and possible a better freq response.

BTW, I tested different diodes with different jfets and sometimes the led voltage of about 1.8v is to high. For example, a BF245A
with Vgsoff <2V does not work with leds, but a silicon is fine.

mac

[DDD]

Jack,
I'm sorry for my mistake. The cap is actually 22uF :-(
What about other (possible) subtleties:
- we know that "good" distortion box distorts different frequencies in the different manner. For me, I think that lower frequency should be more distorted than mids and highs. Moreover, power chords (two lower strings playing together) require some rate of intermodulation distortion.  So, additional non-linearity in the low frequency zone may be just what we need.
Maybe, the 22uF cap do the work.
On the other hand, there is an opinion on the Russian DIY guitar electronics sites that  caps shunting source resistor or diode usually plays a bad role for the sound. What for me, I always use these caps to get more gain from FET stages and never mentioned the negative cosequences.

[R.G.]

the diode riases the source over the gate setting the bias,

Yep, 's'what I said, see my first note.

and being practically a short it is like a bypass cap. The diode have some small resistance I guess, not an expert in electronics, so a big electro cap can be added to get extra gain and possible a better freq response.

My second note was a detailed discusion of that.

BTW, I tested different diodes with different jfets and sometimes the led voltage of about 1.8v is to high. For example, a BF245A
with Vgsoff <2V does not work with leds, but a silicon is fine.

As I said in my second note,

Finally, the JFET is quite dependent on the LED (or other device in that positions) having the right voltage to make the JFET bias properly. Ideally, the LED voltage would put the JFET somewhere in the middle of its conduction region.That means that the LED voltage needs to be on the order of half the Vgsoff for the JFET. JFETS come in all kinds of Vgsoff values, from as little as 0.1V to about 20V. Once again, the JFET variance is what has kept many, many useful circuits from being made with JFETs. You have to select JFETs to work with whatever you stick in there.

[Plexi]

...and when I was thinking that re-biasing a J201 in a Echoplex preamp was all to discover about Jfets...

[mac]

...and when I was thinking that re-biasing a J201 in a Echoplex preamp was all to discover about Jfets...

... I discovered I can use Fets in the power stage of a Deacy Amp 

mac