Zener voltage divider to turn on an NPN transistor.

[Yamin]

Hi guys,
I was wondering if a voltage divider with a Zener be setup like shown below, to turn on a NPN transistor when the VCC goes above Zener breakdown voltage.

Practically I know it works, but I don't see the same setup around the web or books. Is there are reason for it? Could it be that the base current is not limited like this?

I see couple of schematic with the resistor and Zener places reversed (resistor on top and Zener on the bottom) but those circuit operate different from what I want. If the resistor is connected to the VCC and even though the Zener is not conducting the LED would be turned on.

Also I am noticing that a lot of Zener circuits has got the series resistor connected to its cathode and the anode to ground. Why can't it be the other way round (I know why it can't be done on shunt regulator)

Thanks for the help.
P.S I know that what I'm trying to do can be done easily with a pnp transistor, but the question is more to with learning about designs.

[TimFox]

setup drawing is missing

A problem with that circuit is for high Vcc, you have a Zener in series with the BE junction of the transistor, and nothing limits the current through that path except the (low) resistances of the two diodes.

[Zero999]

Move R1 to the other side of D1.

[fourfathom]

Move R1 to the other side of D1.

That limits the current, but the soft Zener knee, or just the leakage, will turn on the NPN slowly and not in a well controlled manner.
It's better to leave R1 where it is, and add another resistor between R1 and the base of the NPN.

[mawyatt]

Another version we've used with or without the LED is to use a lower voltage zener in the emitter of the NPN, then another resistor R2 to supply or collector if no LED. Overall this "looks" like a larger valued effective zener voltage:

Vzeffective ~ (Vz + Vbe)(1+R2/R1)

One can control the overall Effective Zener TC by using a Zener Diode voltage (usually somewhere between 6.2~6.4V) with TC ~2mv/C to compensate the Vbe ~ -2mv/C. Since the resistors are a ratio their TC cancels to a 1st order.

Anyway, we've used this configuration many times for various reasons.

Best

[Benta]

What TimFox said:
You need at least a base resistor to limit Zener/BE current. But your turn-on curve will be quite soft.

[fourfathom]

What TimFox said:
You need at least a base resistor to limit Zener/BE current. But your turn-on curve will be quite soft.

True, good point.  But of course this only matters if it does.  Sometimes it doesn't.  Certainly it's less soft than a simple resistive divider (which can often be good enough).  And the temperature coefficient of the base-emitter junction may also be an issue.
So what are the requirements of the application?  These are all factors that the OP should consider when working on a design.  Since they said: "but the question is more to with learning about designs", all this discussion will be useful.

[Benta]

So what are the requirements of the application?

No one knows.
Personally, I'd just use a TL431 (with appropriate input voltage divider) for this.
But let's see.

[MK14]

Possible simpler circuit, using the LED as a voltage reference (zener).
R1 and R2 to set the voltage level, of the emitter follower circuit, minimum voltage approximately the LED voltage (usually colour dependent) plus the Vbe junction voltage (around a diode drop of perhaps 650 millivolts).  With R3 setting the LED brightness (current) level.

[magic]

This is what it is about. TL431 was already suggested and even 1.25V variants thereof, to deal with requirements that the OP didn't bother (then and now again) to disclose immediately.
 

https://www.eevblog.com/forum/projects/transistor-self-current-limiting-why/msg5659527

[magic]

Move R1 to the other side of D1.

That limits the current, but the soft Zener knee, or just the leakage, will turn on the NPN slowly and not in a well controlled manner.

You can have both. R1 (as shown) sets zener current, R2 (in series with D1) adds an arbitrary fraction of Q1 Vbe to the reference voltage. You can tweak R2 to get an oddball voltage not available in discrete zeners or to cancel D1 tempco (which exceeds +2mV/°C in high voltage zeners).

Analog uses this trick to laser trim thermal drift in AD587 (note: resistor numbers are reversed there).

[David Hess]

Practically I know it works, but I don't see the same setup around the web or books. Is there are reason for it? Could it be that the base current is not limited like this?

It is hidden inside every thyristor crowbar circuit.

The base current needs to be limited by a resistance in series with either the base or zener diode.

[fourfathom]

to deal with requirements that the OP didn't bother (then and now again) to disclose immediately.
 

I don't think we have any reason to beat up the OP for not disclosing specific requirements.  They did tell us their goal was "learning about designs", so there probably isn't an actual requirement.  In that case, I think that exploring the range of [simple / sloppy / complicated / precise / stable / discrete / integrated] designs is worthwhile.

[Yamin]

Thanks everyone for the help lot of information to digest. Like I said before this post is mostly about trying to figure out design in general.
It started out with having to design an over voltage circuit and while researching I have come across a lot of variation of the circuit which consists of zener and pnp transistor (schematic attached). I have also attached a schematic of the type of circuit I mentioned in my first post which has got a pnp in too (its missing a current limiting resistor at the base too its needed right?)

Then I thought to myself why aren't there any circuits with npn transistors - I have breadboarded the circuit I attached in my first post with an additional resistor for the base and it does seem to work as well as the pnp variant of the circuit. Is there any other reason for not connecting the Zener's cathode directly to VCC?

Forgive me in advance for the noob question I am about to ask now -
 

A problem with that circuit is for high Vcc, you have a Zener in series with the BE junction of the transistor, and nothing limits the current through that path except the (low) resistances of the two diodes.

   
By the low resistance of two diode you mean the zener and BE junction?

Practically I know it works, but I don't see the same setup around the web or books. Is there are reason for it? Could it be that the base current is not limited like this?

It is hidden inside every thyristor crowbar circuit.

The base current needs to be limited by a resistance in series with either the base or zener diode.

Exactly, in the crowbar circuits the Zener is on the top right? I have seen crowbar circuits using SCR with that Zener/resistor setup. Shouldn't the gate current be limited too?

Move R1 to the other side of D1.

That limits the current, but the soft Zener knee, or just the leakage, will turn on the NPN slowly and not in a well controlled manner.
It's better to leave R1 where it is, and add another resistor between R1 and the base of the NPN.

So in this case the led should slowly fade in correct?

This is what it is about. TL431 was already suggested and even 1.25V variants thereof, to deal with requirements that the OP didn't bother (then and now again) to disclose immediately.
 

https://www.eevblog.com/forum/projects/transistor-self-current-limiting-why/msg5659527

Hehehe thanks magic for your suggestion of the TL431, this started off with some college assignments when we were doing transistor module. The whole idea was to learn about different applications, and hence why the questions about transistors and related components connecting to it. Like on my other post as well I was trying to understand what parameter of transistor caused it to behave how it did.

Again thanks everyone for your help, very happy with all the information.

[TimFox]

Yes, I was referring to the series connection of the conducting Zener and the forward-biased BE junction.
As you increase the voltage across the series connection, the current increases quickly.
(The resistances are non-linear, but is not high for normal operating currents of the two devices.)
In some circuits (not your original one), there is substantial resistance in series with the Vcc feed to the whole circuit that limits the current.
With a simple series resistor added there, you have a shunt regulator circuit where the bulk of the current flows through the NPN transistor.

[Yamin]

Yes, I was referring to the series connection of the conducting Zener and the forward-biased BE junction.
As you increase the voltage across the series connection, the current increases quickly.
(The resistances are non-linear, but is not high for normal operating currents of the two devices.)
In some circuits (not your original one), there is substantial resistance in series with the Vcc feed to the whole circuit that limits the current.
With a simple series resistor added there, you have a shunt regulator circuit where the bulk of the current flows through the NPN transistor.

Thanks Tim, incoming noob question again-
Could you please explain a little bit more the parts you said
"(The resistances are non-linear, but is not high for normal operating currents of the two devices.)"
and
"In some circuits (not your original one), there is substantial resistance in series with the Vcc feed to the whole circuit that limits the current.
With a simple series resistor added there, you have a shunt regulator circuit where the bulk of the current flows through the NPN transistor."
By "series resistor added there" do you mean the one below the zener in my circuit. So a base resistor added should do the trick no?

Also how would I size the base resistor? The transistor is 2N3904 and the Zener is 5.1V zener diode.

Thanks again

[TimFox]

Yes, I was referring to the series connection of the conducting Zener and the forward-biased BE junction.
As you increase the voltage across the series connection, the current increases quickly.
(The resistances are non-linear, but is not high for normal operating currents of the two devices.)
In some circuits (not your original one), there is substantial resistance in series with the Vcc feed to the whole circuit that limits the current.
With a simple series resistor added there, you have a shunt regulator circuit where the bulk of the current flows through the NPN transistor.

Thanks Tim, incoming noob question again-
Could you please explain a little bit more the parts you said
"(The resistances are non-linear, but is not high for normal operating currents of the two devices.)"
and
"In some circuits (not your original one), there is substantial resistance in series with the Vcc feed to the whole circuit that limits the current.
With a simple series resistor added there, you have a shunt regulator circuit where the bulk of the current flows through the NPN transistor."
By "series resistor added there" do you mean the one below the zener in my circuit. So a base resistor added should do the trick no?

Also how would I size the base resistor? The transistor is 2N3904 and the Zener is 5.1V zener diode.

Thanks again

1.  The two diodes will show increasing voltage across each as the current increases:  that is a non-linear resistance.
Zeners are specified for their incremental resistance at a given current level, for design purposes.
Normal diodes are governed by the Schockley diode equation, found in textbooks, for the relationship between voltage and current.
For your circuit, the problem is that if you put, say 10 V, into the circuit from a source without current limit, the current through the combination of Zener and BE junction will be huge and probably overheat the devices.

2.  Sometimes, one needs a simple circuit that acts like a Zener diode with a higher current capability than the Zener itself.  Your circuit will function in that way, giving a voltage across it as the sum of the Zener and BE voltages (at a relatively low current), while the equivalent Zener current is the collector current of the transistor which is much higher than the base current.
Just like with a regular Zener, you must externally limit the current, usually with a series resistor from the voltage source, or the Zener will conduct too heavily.

3.  Personally, I don't like the base-resistor option, so I won't comment.

[fourfathom]

3.  Personally, I don't like the base-resistor option, so I won't comment.

Tim, which base resistor is this?  In the original circuit I claim the base resistor or other base-current limiting modification is necessary, and the base resistor is probably the best simple option.

In some of the "amplified Zener" circuits, a base resistor is not useful.

[mawyatt]

For those that like to play around with circuits and such, here's a LTspice simulation of the circuit mentioned in #4 above.

This is a ~21V Low TC Equivalent Zener that follows:

Vzeffective ~ (Vz + Vbe)(1+R2/R1), where Vz is a 6.2V Zener, Vbe is from 2N3904, R1 is 10K and R2 is 20K.

Tweaking the Effective Zener Current to ~6.87mA with Rload finds a relatively low TC as shown with Vzeffective ~ 21.07V.

Anyway, hope some find this useful.

Best

[Zero999]

You'll never get a sharp snap-on action with a single transistor. It'll also flicker lots around the threshold.

Adding another transistor will help, by adding some hysteresis, i.e. it will turn on at a higher voltage, than it takes to turn off.

When the voltage is too low for D1 to conduct, Q1 will be off. The voltage on its collector will be near the supply voltage, minus the tiny voltage drop across D2 and R1, which will turn on Q2. When the voltage is high enough for D1 to conduct, some of the base current for Q1 will be diverted to 0V via Q2 and R3, making it a little more difficult to turn on. When the voltage is high enough for the current flowing through R3 to develop a voltage across which exceeds the turn-on threshold for Q1 (about 0.6V) Q1 will turn on. When Q1 is on, its collector voltage will be near-zero and all of the base current for Q2 will be diverted to 0V, causing Q2 to turn off. With Q2 off, Q1 will receive even more base drive current, since Q2 is no longer diverting some of it to 0V, via R3. This means the voltage will have to drop much lower, in order for Q1 to turn off.

Here's a simulation. The red plot is the power supply voltage (left-hand scale in Volts) and the green is the current trough the LED (right-hand scale). The supply voltage is increased from 10.5V to 12V, then decreased to 10.5V. Note the points when the LED turns on and off are different, depending on whether the LED was already on or off.

[TimFox]

3.  Personally, I don't like the base-resistor option, so I won't comment.

Tim, which base resistor is this?  In the original circuit I claim the base resistor or other base-current limiting modification is necessary, and the base resistor is probably the best simple option.

In some of the "amplified Zener" circuits, a base resistor is not useful.

Yes, adding a resistor in series with the base will limit the current, but I don't like that circuit since it "softens" the turn-on.
That's why I did not comment about the size of a resistor in series with the BE.

[David Hess]

Practically I know it works, but I don't see the same setup around the web or books. Is there are reason for it? Could it be that the base current is not limited like this?

It is hidden inside every thyristor crowbar circuit.

The base current needs to be limited by a resistance in series with either the base or zener diode.

Exactly, in the crowbar circuits the Zener is on the top right? I have seen crowbar circuits using SCR with that Zener/resistor setup. Shouldn't the gate current be limited too?

The crowbar circuit examples I was thinking of include a low value of series resistance to limit the gate current.

[David Hess]

Yes, adding a resistor in series with the base will limit the current, but I don't like that circuit since it "softens" the turn-on.
That's why I did not comment about the size of a resistor in series with the BE.

The value of the resistor can be pretty small, and the change in base current is multiplied by the current gain.

An alternative is to use a differential pair for the switching instead of a single transistor.

[youngda9]

If you want a sharp corner and precise threshold, use a comparator.  You can get a tiny one with a built in reference even for this application.  Check out the LT6703-2 as an example.