# zener trouble

Discussion in 'Electronic Basics' started by Erik Durling, Oct 28, 2004.

1. ### Erik DurlingGuest

We're a couple of students doing an electronics project. We are to
construct a battery indicator circuit. We had a finished circuit on
paper and we had simulated it with Multisim 7. But when we tested the
circuit in real life, it didn't work at all like we had planned (or
according to simulation). The circuit was based upon the idea of
zenerdiodes having a constant voltage drop, and not conducting if the
circuit isn't able to supply the needed voltage across it. But the
zener-diodes doesn't stop conducting at all, and the voltage drop is
far from constant.

This is the main circuit: http://ersker.com/circuit.gif

The zener diodes used in real life are: BZX55 and not BZC55, but that
ought not change anything (right?).

We did a DC Transfer Analysis in the simulator and got three nice
curves showing the voltage at the three collectors (compared to
ground) as the voltage from the source goes from 0-4.5 volts, and
everything seems right there: http://ersker.com/dctrans.gif

So, the problem we're having with our real circuit is that the LEDs
never turn off (the transistors are always saturated). The zener
diodes doesn't stop conducting when the voltage across them goes under
their specified value.

What's our mistake/misconception?

Do the zeners only operate correctly at a certain current?

2. ### Roger JohanssonGuest

Zeners conduct at all voltages, but the current is very small when the
voltage is under the zener voltage.
(It is actually an exponential relationship between voltage and current)

With your circuit the hFE of the transistors amplify the base current
very much, so the little current through the zeners is enough to turn on
the LEDs.
You should design a better circuit.
The one you have is based on faulty assumptions.

(The reason why it worked in multisim is probably that is uses simplified
simulation which does not work in real life when you create a design
based on using components in a non-regular way.)

Zeners are made to provide a fairly constant voltage, not current.
Do not use zeners to control current.

3. ### John LarkinGuest

Low voltage zeners, below maybe 5 volts, are terrible. Their
conduction knee is sloppy as hell, which your simulator probably
doesn't model accurately. And your open bases are *very* sensitive to
small leakages; open bases are considered bad form by most designers.

Add b-e resistors to divert some of the leakage current, and maybe go
to bandgap pseudo-zeners (LM4040 types) which behave much better.

But I'm impressed that you're actually verifying your simulations by
experiment; some people just quit when the sim works.

John

4. ### John PopelishGuest

How low does the battery voltage have to go before the LEDs turn
essentially off? I am suspicious that you may have the zeners
backward.

That said, such low voltage zeners are not really switches that
suddenly conduct when the voltage across them rises above the zener
knee voltage, but are exponential devices, much like forward biased
junctions. In fact, you may get a better knee out of a green LED,
forward biased. To improve the switching effect, you could load
(parallel) the base emitter junction with some resistance to divert
zener current while keeping the base ot emitter voltage below about .5
volts up to the supply voltage that you want the LED to light.

But using a comparator (comparing your divider voltage to a fixed
reference instead of a transistor sensing zener current) would give a
lot cleaner threshold.

5. ### John LarkinGuest

This brings up some interesting issues:

Why would anyone simulate a circuit this simple?

If something this simple doesn't simulate correctly, what hope is
there for a serious circuit?

Did your instructor deliberately give you a circuit that would
simulate correctly but not work in real life? If you're very lucky,
yes.

Why not look at a zener data sheet?

When you built it and it didn't work, why did you resort to a
newsgroup for help? Why not measure and analyze the voltages and
currents and figure out what's actually happening? The methodology
here is: fiddle with the simulation until it works; then fiddle with
the circuit until it works, or ask for help; no thinking required.

John

6. ### Fred BartoliGuest

This is the better case.
Some other people run into prod when the sim works.

Anyone having some nice story about direct run from sim to prod ?

On the contrary, I once did the expertise of a GTO firing hybrid that had
probably been designed with the pencil/paper method and not enough tested.
Few of them had the interesting feature of being permanently on at elevated
temperature.
As those were for railways traction, I've been nice fireworks stories.
2 hours simulation work sorted the pb out. Then 2 more weeks to convince the
customers staff.

7. ### petrus bitbyterGuest

You made some wrong assumptions about the zeners. They indeed need some
current for the function you want them to fulfil but there is more. One
thing you can do to get a better insight is making a real life I-V plot of
(one of) the zeners you used. Pay special attention to the part below zener
voltage. You will measure microamps but that's still current. Multiplied by
beta (or Hfe) will give enough current through the LED to make it light.
Keep in mind that although a general purpose LED likes a 20mA for optimal
funtioning most of them start to light already at 1mA.

For a circuit like this I would use one zener or bandgap as a reference
voltage. A comparator (an LM339 contains four of them) and a voltage
divider. The reference voltage can be used for all three comparators. The
voltage divider will set the voltage at which the comparator will switch.

Below I give an idea to make something like that using discretes. Beware
it's just an idea.

+-------+--------+
| | |
.-. .-. .-.
| | | | | |
| | | | | |
'-' '-' '-'
| | |
| | |
| | V LED
| | -
| | |
| | |
| | |/
+-------|------| Si
| | |>
| | |
| +---->|--+
| | Si |
.-. | .-.
| | - | |
| | \ Z | |
'-' ^ '-'
| | |
+-------+--------+
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

petrus bitbyter

John

9. ### meGuest

But I'm impressed that you're actually verifying your simulations by
Don't be. If they didn't have to build it to demonstrate it in the lab
they would have called it good after the simulation.

10. ### John LarkinGuest

I resently had a tour of the EE department at Cornell University. I
figure that computer screens outnumber oscilloscope screens by roughly
15:1.

John

11. ### Erik DurlingGuest

I think I explained the purpose of the zener improperly. We have it
there to increase the potential needed to saturate the transistor so
it needs a smaller change in the source to turn off the LED.
But the "leakage" was still the big problem eitherway.

---

Would it work better by connecting ordinary diodes in series?

What's meant by "open bases"?

Thanks! That made a huge difference.

---
The fresh-voltage is about 4.5, then one LED turn off every 0.5 volt.
Though we have potentiometers connected between the two resistors in
the picture (and the values differ a little), but I left that out on
purpose, those values aren't that important for the "problem".

Isn't that the point with zeners (that they're backwards)?

That was what we planned to use at first, but they themselves needed a
higher voltage than the battery was able to provide (less than 3 volts
at minimum).

---

Cause we wanted to know what components we needed before we spent our
money.
yes.

Actually, our instructor doesn't know shit about electronics (he
thinks he does though).

We trusted in the simulator (won't do that again =) )
newsgroup for help?

12. ### Robert MonsenGuest

Well. Here is a funny thing. I just simulated a circuit with a variable
voltage source across a 1k resistor, connected to a 3.3V zener, and then
to ground. The voltage source goes from 0 to 10V in 10 seconds. Oddly,
the junction between the resistor and zener starts out at 2.3V, and goes
up slowly from there. Thus, it appears that spice believes that the
zener diode is a voltage source This happens for both circuitmaker and
LTSpice. There is actually current flowing backwards through the
resistor into the voltage source until the dc source gets to equal the
'voltage source' of the zener. I'd say that the spice implementation of
zener diodes of low value is seriously flawed when used with a voltage
lower than the actual zener voltage minimum.

Here is the spice netlist:
* node0--[V1=PWL]--node3--[R1=1k]--node4--[XD1=1N5226B]--node0--GND
*
*Spice netlist for Circuit: C:\CM60S\Circuits\UNTITLED.CKT
V1 3 0 DC 0 PWL( 0 0 10 10)
XD1 0 4 X1N5226B
R1 4 3 1k
..SAVE V(3) V(4) @v1[p] v1#branch @r1[p] @r1
*BKGND=RGB 0 0 0
*BINARY RAW FILE

* Selected Circuit Analyses :
..TRAN 20m 10 0 20m

* Models/Subcircuits Used:

*1N5226B 3.3V 500mW Si pkgIODE0.4 A,K
..SUBCKT X1N5226B 1 2
D1 1 2 DF
DZ 3 1 DR
VZ 2 3 2.536
..MODEL DF D (IS=2.51N RS=84M N=1.7 CJO=182P VJ=0.75 M=0.33 TT=50.1N)
..MODEL DR D (IS=5.59M RS=8.4 N=15)
..ENDS X1N5226B
..END

--
Regards,
Robert Monsen

"Your Highness, I have no need of this hypothesis."
- Pierre Laplace (1749-1827), to Napoleon,
on why his works on celestial mechanics make no mention of God.

13. ### John FieldsGuest

---
I've read some of the replies to your original post, above, but your
last post seems to show that you're still pretty much in the dark, so
here goes:

If you hook up this very simple circuit:

+VIN>---+
|
|K
[Z4.3V]
|
+---->VOUT>---+
| |
[R1] [VOLTMETER]
| |
GND>----+-------------+

You'll notice that if you start with +VIN at 0V and then make it more
and more positive, the voltage across R1 will start to increase long
before +VIN gets to 4.3V. The reason for that is because the Zener
isn't a perfect switch and it will start allowing current to flow
through itself before its Zener voltage is reached. Zeners are
designed to be shunt regulators, and if you look at a data sheet for a
Zener you'll find that the Zener voltage is only guaranteed to be
within a certain range of voltages if the current through the Zener is
the "test current", usually 20mA for 1/2 watt diodes with a Zener
voltage of 12V or less.

Now, if we take a look at your circuit:

VIN>-----+-------------+
| |
| [LED]
[1K] |
| [150R]
| |
| C
+--[<ZENER]--B Q1
| E
[10K] |
| |
GND>-----+-------------+

we can see that as VIN starts to go more and more positive, more and
more current will start to flow through the base-to-emitter junction
of the transistor, just because the Zener isn't a switch and will
start to conduct well below its Zener voltage. If you have a
transistor with a reasonably high beta (100 to 300) then the current
which is flowing through the Zener (and also through the b-e
junction)will cause a collector current to flow which will be 100 to
300 times higher than that. So, even if you're way below the Zener's
knee and you have, say, 10µA of reverse current flowing, the collector
current will be somewhere between 1mA and 3mA, which will be enough to
light the LED through that 150 ohm resistor.

If what you're trying to do is build something with LEDs which light
sequentially as the supply voltage increases, then you need to use
three comparators, a voltage reference, three LEDs and a handful of
resistors.

Want a schematic?

14. ### John PopelishGuest

These work down to 2.7 volts:
http://cache.national.com/ds/LM/LMV331.pdf

15. ### Erik DurlingGuest

We understand this would be easier to build with comparators and
voltage reference components, but we really want to build it using
only discreete components.

16. ### John LarkinGuest

A base-emitter resistor will help a lot.

John

17. ### john jardineGuest

Something like this has quite a clean switch-on action when the battery
drops below 3.0V.
Tr1 compares the fixed 2.5V supply with a divided down version of the
battery voltage and starts turning on when its base voltage goes below 1.9V.
The "stable 2.5V" can come from a zener or a reference.
(the 1k and 10k at Tr2 are there just to kill any leakage currents)

Battery Volts
IN
| Stable 2.5V
| ,---------,
.-. | |
| | | .-.
5k6| | | | |
'-' | | |47
| |< Tr1 '-'
+---------| BC556 |
| |\ |
.-. | V Led
| | .-. -
10k| | | | |
'-' | | |
| '-'1k |/ Tr2
| +-------| BC546
| .-. |>
| | | |
| | | |
| '-'10k |
| | |
=== === ===
GND GND GND

created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

18. ### Bill BowdenGuest

Maybe you can just put the LEDs in the emitter side of the
transistor, so they light at some base voltage set by a
couple resistors? No zener needed since the LEDs light
at around 2 volts. Something like this:

+ Battery +
| |
| \
| / R3
\ \
R1 / |
\ C
| |/
+-------B| NPN
| |\
\ E
R2 / |
\ LED
| |
GND GND

-Bill

19. ### Robert MonsenGuest

As I've pointed out elsewhere, Zener models appear to suck.

Here is a simple circuit that will do what you need:

Vin
|
|
o------.
| |
| |
e |
.---b [1k]
| c |
| | |
| | |
{10k] [1k] |
| | e
| o-----b
| | c
zener | |
| [10k] LED
| | |
'----o------'
GND

When Vin is above the zener voltage + a bit, the left transistor will
conduct. If its conducting, then the base of the right transistor will
be higher than the emitter + 0.7, that transistor will be off, turning
the LED off.

When the voltage drops below the zener voltage, the left PNP transitor
will turn off, causing the right PNP transistor to turn 'sharply' on. As
the voltage continues to decay, the LED will remain on, but will get dimmer.

--
Regards,
Robert Monsen

"Your Highness, I have no need of this hypothesis."
- Pierre Laplace (1749-1827), to Napoleon,
on why his works on celestial mechanics make no mention of God.

20. ### CBarn24050Guest

Subject: Re: zener trouble
Hi, if you look at the datasheet you will see why it doesn't work. Why you
would even consider simulating such a circiut is beyond me.