# Why can't a P-N junction diode be used as a voltage regulator or reference in forward bias?

Usually, a Zener or avalanche breakdown region for a suitably doped P-N junction is used for voltage regulation as a substantial variation in the current (due to minority carriers) requires a negligibly slight variation in reverse bias voltage.

From the I-V characteristics of a P-N junction, it appears that when the forward bias voltage is much greater than the knee voltage, a significant variation in forward current also corresponds to a very slight variation in the forward voltage. Even if this knee voltage is very small compared to the reverse breakdown voltage, for small voltage ranges, a P-N junction in forward biasing should work as a voltage regulator.

Can a P-N junction diode be used as a voltage regulator in forward biasing for smaller voltage ranges?

• when the forward bias voltage is much greater than the knee voltage - Normally we'd like to avoid wasting large amounts of current (thus power in a dropper resistor or current source) just to get a lowish-stability voltage reference. A diode alone wouldn't be a regulator either, but could be a reference. (I suggested a tag/title edit). Jul 12, 2022 at 1:10
• This is actually a pretty common method to provide a simple bias to bipolar transistors, see the schematic of a Class-AB output stage for example. Using multiple diodes in series at the base of a transistor is also common. Jul 12, 2022 at 5:24
• @PeterCordes There might be another factor for not considering the P-N junction in forward bias as a suitable voltage regulator or reference. That is, in the forward bias, the change in in the forward current is dependent on the available charge carriers. No new carrier is getting created as in Zener breakdown or avalanche breakdown. Am I correct? Jul 12, 2022 at 16:57

The H11L1 chip uses four diodes in series as reference for its very simple voltage regulator:

It does not even try to supply a constant current, so its characteristics vary depending on supply voltage and temperature:

(source)

If you need a voltage that actually can be called "constant", you need a better reference.

It is quite often used to bias a current source.

On the left, the LED (a PN junction) creates a 1.8V reference for the transistor. It is used as a shunt regulator, just like a Zener diode, but with lower voltage and lower noise.

In theory, the temperature coefficient of the LED Vf somewhat compensates for the tempco of the transistor Vbe. In practice, not so much.

On the right, Q3 Vbe is the voltage reference, and also sets the tempco.

Both are cheap ways to make a current source. Not really accurate or super stable, but if you need a current source to bias something, it's good enough.

It's pretty common to use one Vbe as shunt regulator. The temperature dependence is quite high. Over -20 to +85°C you'd get 0.2V drift, on a 0.6V reference, that's a lot. But if you actually want to use this property to create a temperature-dependent voltage, for example to compensate for the temperature-dependent Vbe of other transistors, then it becomes very useful.

It can be used for the very reasons you mentioned. It is often used inside ICs to generate voltage drops that track that of transistors, for example, in current mirrors (usually realized as diode-connected transistors:

simulate this circuit – Schematic created using CircuitLab

## Why is it then not used much for discrete designs?

• For decrete designs, both Zeners and resistors are easy to come by.
• Just like Zeners, the voltage is not that stable, but varies with temperature and current. It is better than a fixed resistor divider, but not great
• While for Zeners their variation happens on top of a nominal voltage drop of a few volts, the nominal drop of the PN diode is pretty low, which makes the variation especially detrimental.

So in essence, its simplicity doesn't outweigh its bad performance. If you can deal with very bad regulation, a resistor divider is even simpler. If you need better performance, than Zeners or even a TL431 are not really that much more expensive.

If you look at the graph you posted in the same aspect ratio for both quadrants, you will see why the forward-biased diode is a mediocre voltage reference.

On the other hand, I remember at least 2 use cases (pointless in the modern state of art, but pretty much reasonable 2-3 decades ago):

1. As a temperature compensation of a Zener diode. Your Zener has positive temperature coefficient ~2mv/degree C ? Just add a forward-biased diode. The compensation is not perfect, some second-order temperature dependency remains, but the result can be much better than with the Zener alone.

2. A Red LED (the older ones that had ~1.5V forward voltage) as a voltage limiter for a NiCd cell charger. When connected in parallel to the cell, it serves as an overcharge protection and as a "ready" indicator at once. Of course, the charging current is to be limited to whatever the LED can (e.g. 10-20mA).

And an anecdotal use from personal experience: I had to hack an 1.5V stable-ish power source for a microphone from the only available 9V battery. The LED from the amplifier power switch together with its current limiting resistor did the trick pretty well.

• When you say "you will see", I think you're pointing out the (lack of) sharpness of the curve at the knee? Or the slope after the knee? The question did say when the forward bias voltage is much greater than the knee voltage, but the problem with that is wasting huge amounts of power in a dropper resistor or current source. Jul 12, 2022 at 1:16

It can be used as long as constant current is applied, so the voltage is predictable.

As of why it cannot be used, because the voltage drop changes as the current and temperature changes, but typical current changes produce larger voltage difference than typical temperature changes and temperature changes have effects on zener diode as well.